==================================================Ascend ============================= test session starts ============================== platform linux -- Python 3.9.19, pytest-6.2.5, py-1.11.0, pluggy-1.5.0 rootdir: /home/jenkins/mindspore/testcases/testcases/tests/st/mint, configfile: ../../../../../../sault/virtual_test/virtualenv_008/sault/config/pytest.ini plugins: mock-3.14.0, hydra-core-1.3.2, forked-1.6.0, anyio-4.9.0, xdist-1.32.0 collected 2 items test_functional_mul.py TotalTime = 0.452126, [21] [bootstrap]: 0.00075472 [type_inference]: 0.269538 [event_method]: 4.033e-05 [auto_monad]: 0.00011661 [graph_reusing]: 7.33e-06 [inline]: 1.96998e-06 [add_attr]: 0.00821227, [1] [add_attr_with_inline]: 0.00819771, [1] [Cycle 1]: 0.00011404, [2] [tag_attr]: 3.567e-05 [meta_addattr_fg_expand]: 7.38e-06 [parallel-infer-symbol]: 3.86001e-06 [pre_auto_parallel]: 6.861e-05 [insert-virtual-dataset]: 2.58998e-06 [parallel-infer-symbol-second]: 8.30012e-07 [dataset_repeat_opt]: 2.19999e-06 [pipeline_split]: 1.71998e-06 [optimize]: 0.172069, [53] [py_interpret_to_execute]: 3.51e-05 [rewriter_before_opt_a]: 0.00012542 [opt_a]: 0.00320395, [2] [Cycle 1]: 0.00253631, [45] [expand_dump_flag]: 3.94002e-06 [switch_simplify]: 5.434e-05 [loop_unroll]: 3.833e-05 [a_1]: 0.00089072 [with_stream_mark]: 1.84e-05 [recompute_prepare]: 8.33001e-06 [updatestate_depend_eliminate]: 4.33999e-06 [updatestate_assign_eliminate]: 3.62002e-06 [updatestate_loads_eliminate]: 2.76e-06 [parameter_eliminate]: 2.31e-06 [a_2]: 8.212e-05 [accelerated_algorithm]: 6.82002e-06 [shard]: 2.12999e-06 [meta_shard_fg_expand]: 2.15002e-06 [shard_inline]: 7.23999e-06 [merge_send_recv]: 9.44e-06 [auto_parallel]: 7.51999e-06 [parallel]: 4.371e-05 [flash_sp]: 1.909e-05 [merge_comm]: 4.05e-06 [allreduce_fusion]: 3.55e-06 [matmul_add_comm_reduction]: 1.081e-05 [allreduce_slice_to_reducescatter]: 1.12e-06 [virtual_shard_identity]: 8.17998e-06 [virtual_dataset]: 6.73998e-06 [get_grad_eliminate_]: 5.79e-06 [virtual_output]: 5.91e-06 [merge_forward]: 4.27998e-06 [cell_reuse_recompute_pass]: 1.44e-06 [offload_activation]: 1.079e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.283e-05 [merge_recompute_call_nodes]: 2.04e-06 [before_grad]: 1.013e-05 [set_forward_comm_id_for_comm_node_pass]: 3.55998e-06 [meta_fg_expand]: 3.58999e-06 [flash_sp_send_recv_attached]: 2.63e-06 [receive_attached]: 1.136e-05 [after_resolve]: 1.145e-05 [a_after_grad]: 9.71e-06 [renormalize]: 0.00082954 [add_forward_monad_depend]: 6.59001e-06 [auto_monad_grad]: 2.74001e-06 [auto_monad_eliminator]: 1.573e-05 [cse]: 4.071e-05 [a_3]: 4.486e-05 [Cycle 2]: 0.00065651, [45] [expand_dump_flag]: 1.97999e-06 [switch_simplify]: 7.98001e-06 [loop_unroll]: 5.89999e-06 [a_1]: 0.00013027 [with_stream_mark]: 1.137e-05 [recompute_prepare]: 6.04001e-06 [updatestate_depend_eliminate]: 3.45003e-06 [updatestate_assign_eliminate]: 2.46e-06 [updatestate_loads_eliminate]: 2.89999e-06 [parameter_eliminate]: 1.02998e-06 [a_2]: 8.301e-05 [accelerated_algorithm]: 6.29001e-06 [shard]: 1.20999e-06 [meta_shard_fg_expand]: 1.49e-06 [shard_inline]: 5.96e-06 [merge_send_recv]: 4.67e-06 [auto_parallel]: 5.85002e-06 [parallel]: 4.87e-06 [flash_sp]: 3.90998e-06 [merge_comm]: 3.28e-06 [allreduce_fusion]: 2.79001e-06 [matmul_add_comm_reduction]: 5.65001e-06 [allreduce_slice_to_reducescatter]: 3.89991e-07 [virtual_shard_identity]: 6.72002e-06 [virtual_dataset]: 5.55001e-06 [get_grad_eliminate_]: 5.82001e-06 [virtual_output]: 5.19998e-06 [merge_forward]: 2.73e-06 [cell_reuse_recompute_pass]: 1.39e-06 [offload_activation]: 7.23e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.066e-05 [merge_recompute_call_nodes]: 7.09988e-07 [before_grad]: 8.37998e-06 [set_forward_comm_id_for_comm_node_pass]: 3.88999e-06 [meta_fg_expand]: 2.09e-06 [flash_sp_send_recv_attached]: 9.79984e-07 [receive_attached]: 1.15999e-06 [after_resolve]: 1.058e-05 [a_after_grad]: 9.25001e-06 [renormalize]: 8.00064e-08 [add_forward_monad_depend]: 1.69e-06 [auto_monad_grad]: 1.05999e-06 [auto_monad_eliminator]: 8.39998e-06 [cse]: 1.521e-05 [a_3]: 3.575e-05 [py_interpret_to_execute_after_opt_a]: 9.47001e-06 [slice_cell_reuse_recomputed_activation]: 2.91e-06 [rewriter_after_opt_a]: 0.00027843 [convert_after_rewriter]: 1.063e-05 [order_py_execute_after_rewriter]: 6.33002e-06 [mutable_eliminate]: 0.00056804 [opt_b]: 0.00020118, [1] [Cycle 1]: 0.00019414, [7] [b_1]: 0.00012041 [b_2]: 8e-06 [updatestate_depend_eliminate]: 6.44001e-06 [updatestate_assign_eliminate]: 2.42001e-06 [updatestate_loads_eliminate]: 2.32001e-06 [renormalize]: 4.69998e-07 [cse]: 1.892e-05 [optimize_parallel_all_gather_comm]: 1.731e-05 [overlap_param_gather]: 4.77e-06 [cconv]: 2.675e-05 [loop_unroll]: 0.00043001 [opt_after_cconv]: 0.00010243, [1] [Cycle 1]: 9.659e-05, [7] [c_1]: 2.943e-05 [parameter_eliminate]: 3.21999e-06 [updatestate_depend_eliminate]: 5.49998e-06 [updatestate_assign_eliminate]: 2.56e-06 [updatestate_loads_eliminate]: 2.21e-06 [cse]: 1.887e-05 [renormalize]: 3.09985e-07 [remove_dup_value]: 1.546e-05 [tuple_transform]: 7.577e-05, [1] [Cycle 1]: 7.101e-05, [4] [d_1]: 4.328e-05 [none_parameter_eliminate]: 1.49e-06 [renormalize]: 2.40019e-07 [switch_simplify]: 7.12002e-06 [partial_unused_args_eliminate]: 1.96e-06 [add_recomputation]: 0.0001213 [cse_after_recomputation]: 7.151e-05, [1] [Cycle 1]: 5.867e-05, [1] [cse]: 4.151e-05 [environ_conv]: 2.332e-05 [swap_dp_allreduce_reducescatter]: 6.37001e-06 [bias_add_comm_swap]: 3.21001e-06 [label_micro_interleaved_index]: 8.75001e-06 [label_fine_grained_interleaved_index]: 2.87002e-06 [merge_cast_opt]: 1.37999e-06 [slice_recompute_activation]: 2.61e-06 [micro_interleaved_order_control]: 2.51e-06 [assign_add_opt]: 2.24001e-06 [ForceFp32Comm]: 1.17e-06 [remove_cast_before_assign_add]: 1.52999e-06 [full_micro_interleaved_order_control]: 2.20002e-06 [reorder_send_recv_between_fp_bp]: 3.26001e-06 [comm_op_add_attrs]: 1.25999e-06 [add_comm_op_reuse_tag]: 1.05999e-06 [interleave_split_concat_branches]: 1.60001e-06 [interleave_parallel_branches]: 1.22e-06 [overlap_opt_shard_in_pipeline]: 1.283e-05 [overlap_opt_shard_grad_in_pipeline]: 1.72001e-06 [control_data_broadcast_order]: 1.979e-05 [grouped_pairwise_exchange_alltoall]: 1.74e-06 [offloading_packed_experts]: 4.67e-06 [overlap_recompute_and_grad_model_parallel]: 5.32999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.35001e-06 [overlap_recompute_allgather_and_fa_grad]: 1.57001e-06 [overlap_recompute_comm]: 2.91e-06 [overlap_grad_ring_attention]: 4.27e-06 [overlap_grad_flash_sp]: 2.465e-05 [begin_end_overlap_inline]: 8.99978e-07 [split_matmul_comm_elemetwise]: 2.46e-06 [split_layernorm_comm]: 1.81998e-06 [handle_group_info]: 1.07e-06 [symbol_engine_optimizer]: 0.00011489, [1] [Cycle 1]: 0.00010928, [6] [build]: 4.42e-06 [elim_shapecalc]: 2.756e-05 [elim_not_effective]: 2.398e-05 [opt_reshape]: 7.7e-06 [fold_const_symbol]: 1.063e-05 [renormalize]: 3.89991e-07 [detach_backward]: 2.66e-06 [pipeline_parallel_scheduler]: 1.88002e-06 [auto_monad_reorder]: 2.978e-05 [get_jit_bprop_graph]: 1.96998e-06 [rewriter_after_jit_bprop_graph]: 0.00021753 [opt_after_jit_grad]: 0.00081353 [validate]: 6.344e-05 Sums bootstrap : 0.000755s : 0.27% type_inference : 0.269538s : 97.43% event_method : 0.000040s : 0.01% auto_monad : 0.000117s : 0.04% graph_reusing : 0.000007s : 0.00% inline : 0.000002s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000036s : 0.01% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.00% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000069s : 0.02% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000035s : 0.01% optimize.rewriter_before_opt_a : 0.000125s : 0.05% optimize.opt_a.expand_dump_flag : 0.000006s : 0.00% optimize.opt_a.switch_simplify : 0.000062s : 0.02% optimize.opt_a.loop_unroll : 0.000044s : 0.02% optimize.opt_a.a_1 : 0.001021s : 0.37% optimize.opt_a.with_stream_mark : 0.000030s : 0.01% optimize.opt_a.recompute_prepare : 0.000014s : 0.01% optimize.opt_a.updatestate_depend_eliminate : 0.000008s : 0.00% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.00% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.00% optimize.opt_a.parameter_eliminate : 0.000003s : 0.00% optimize.opt_a.a_2 : 0.000165s : 0.06% optimize.opt_a.accelerated_algorithm : 0.000013s : 0.00% optimize.opt_a.shard : 0.000003s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000004s : 0.00% optimize.opt_a.shard_inline : 0.000013s : 0.00% optimize.opt_a.merge_send_recv : 0.000014s : 0.01% optimize.opt_a.auto_parallel : 0.000013s : 0.00% optimize.opt_a.parallel : 0.000049s : 0.02% optimize.opt_a.flash_sp : 0.000023s : 0.01% optimize.opt_a.merge_comm : 0.000007s : 0.00% optimize.opt_a.allreduce_fusion : 0.000006s : 0.00% optimize.opt_a.matmul_add_comm_reduction : 0.000016s : 0.01% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000015s : 0.01% optimize.opt_a.virtual_dataset : 0.000012s : 0.00% optimize.opt_a.get_grad_eliminate_ : 0.000012s : 0.00% optimize.opt_a.virtual_output : 0.000011s : 0.00% optimize.opt_a.merge_forward : 0.000007s : 0.00% optimize.opt_a.cell_reuse_recompute_pass : 0.000003s : 0.00% optimize.opt_a.offload_activation : 0.000018s : 0.01% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000023s : 0.01% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.00% optimize.opt_a.before_grad : 0.000019s : 0.01% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000007s : 0.00% optimize.opt_a.meta_fg_expand : 0.000006s : 0.00% optimize.opt_a.flash_sp_send_recv_attached : 0.000004s : 0.00% optimize.opt_a.receive_attached : 0.000013s : 0.00% optimize.opt_a.after_resolve : 0.000022s : 0.01% optimize.opt_a.a_after_grad : 0.000019s : 0.01% optimize.opt_a.renormalize : 0.000830s : 0.30% optimize.opt_a.add_forward_monad_depend : 0.000008s : 0.00% optimize.opt_a.auto_monad_grad : 0.000004s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000024s : 0.01% optimize.opt_a.cse : 0.000056s : 0.02% optimize.opt_a.a_3 : 0.000081s : 0.03% optimize.py_interpret_to_execute_after_opt_a : 0.000009s : 0.00% optimize.slice_cell_reuse_recomputed_activation : 0.000003s : 0.00% optimize.rewriter_after_opt_a : 0.000278s : 0.10% optimize.convert_after_rewriter : 0.000011s : 0.00% optimize.order_py_execute_after_rewriter : 0.000006s : 0.00% optimize.mutable_eliminate : 0.000568s : 0.21% optimize.opt_b.b_1 : 0.000120s : 0.04% optimize.opt_b.b_2 : 0.000008s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000006s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000019s : 0.01% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.01% optimize.overlap_param_gather : 0.000005s : 0.00% optimize.cconv : 0.000027s : 0.01% optimize.loop_unroll : 0.000430s : 0.16% optimize.opt_after_cconv.c_1 : 0.000029s : 0.01% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000005s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000019s : 0.01% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000015s : 0.01% optimize.tuple_transform.d_1 : 0.000043s : 0.02% optimize.tuple_transform.none_parameter_eliminate : 0.000001s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000121s : 0.04% optimize.cse_after_recomputation.cse : 0.000042s : 0.02% optimize.environ_conv : 0.000023s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.00% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000009s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000003s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.00% optimize.assign_add_opt : 0.000002s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000002s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000002s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000013s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000020s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000005s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000002s : 0.00% optimize.overlap_recompute_comm : 0.000003s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000025s : 0.01% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000004s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000028s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000024s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000008s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000011s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000030s : 0.01% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000218s : 0.08% opt_after_jit_grad : 0.000814s : 0.29% validate : 0.000063s : 0.02% Time group info: ------[substitution.] 0.000294 40 11.94% : 0.000035s : 8: substitution.arithmetic_simplify 0.96% : 0.000003s : 2: substitution.elim_not_effective 0.59% : 0.000002s : 2: substitution.fold_const_symbol 2.24% : 0.000007s : 4: substitution.graph_param_transform 73.35% : 0.000216s : 6: substitution.inline 1.01% : 0.000003s : 4: substitution.j_node_and_user_rematch 1.68% : 0.000005s : 4: substitution.remove_not_recompute_node 1.40% : 0.000004s : 4: substitution.replace_old_param 6.84% : 0.000020s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.269374 2 99.58% : 0.268244s : 1: type_inference.infer 0.42% : 0.001131s : 1: type_inference.specialize ------[replace.] 0.000093 12 60.75% : 0.000056s : 6: replace.inline 39.25% : 0.000036s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000230 12 92.45% : 0.000213s : 6: match.inline 7.55% : 0.000017s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000238 1534 0.96% : 0.000002s : 17: predicate.accumulaten_eliminater 0.81% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.39% : 0.000001s : 8: predicate.addn_check_dump 0.93% : 0.000002s : 17: predicate.addn_zero_filter 0.88% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.21% : 0.000005s : 25: predicate.arithmetic_simplify 1.00% : 0.000002s : 17: predicate.cast_eliminate 0.50% : 0.000001s : 8: predicate.check_bprop_eliminate 0.39% : 0.000001s : 8: predicate.compare_switch_simplify 0.16% : 0.000000s : 4: predicate.const_output_eliminate 0.49% : 0.000001s : 8: predicate.depend_value_elim 1.07% : 0.000003s : 17: predicate.dict_get_item_const_eliminator 1.19% : 0.000003s : 17: predicate.dict_get_item_eliminator 0.98% : 0.000002s : 17: predicate.dict_set_item_eliminator 1.00% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.53% : 0.000001s : 4: predicate.elim_not_effective 0.78% : 0.000002s : 4: predicate.elim_shapecalc_of_broadcastargs 1.30% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.14% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.12% : 0.000003s : 21: predicate.environ_get_depend_swap 1.63% : 0.000004s : 29: predicate.environ_get_eliminate 1.13% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.66% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.55% : 0.000006s : 29: predicate.float_depend_g_call 0.41% : 0.000001s : 8: predicate.float_environ_get_switch 0.61% : 0.000001s : 12: predicate.float_tuple_getitem_switch 0.14% : 0.000000s : 4: predicate.fold_const_symbol 0.53% : 0.000001s : 8: predicate.get_grad_eliminate 0.27% : 0.000001s : 4: predicate.graph_param_transform 0.48% : 0.000001s : 8: predicate.incorporate_call 0.39% : 0.000001s : 8: predicate.incorporate_call_switch 6.12% : 0.000015s : 70: predicate.inline 0.67% : 0.000002s : 8: predicate.inline_without_move 0.29% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.76% : 0.000002s : 8: predicate.less_batch_normalization 1.98% : 0.000005s : 31: predicate.list_to_tuple_eliminator_ 2.68% : 0.000006s : 48: predicate.load_eliminater 0.72% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.63% : 0.000006s : 47: predicate.loop_unroll_before_grad 1.60% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.44% : 0.000001s : 8: predicate.merge_addn 0.47% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.47% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.86% : 0.000002s : 17: predicate.minmaximum_grad 1.03% : 0.000002s : 4: predicate.mutable_eliminate 0.35% : 0.000001s : 4: predicate.opt_reshape 0.39% : 0.000001s : 4: predicate.parallel_virtual_node 2.04% : 0.000005s : 29: predicate.partial_defer_inline 1.66% : 0.000004s : 27: predicate.partial_eliminate 0.93% : 0.000002s : 17: predicate.print_const_string_wrapper 0.49% : 0.000001s : 8: predicate.reduce_all_const_elim 1.23% : 0.000003s : 17: predicate.reduce_eliminate 2.69% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.52% : 0.000001s : 8: predicate.remove_not_recompute_node 1.41% : 0.000003s : 31: predicate.replace_applicator 0.38% : 0.000001s : 8: predicate.replace_old_param 0.23% : 0.000001s : 4: predicate.reset_defer_inline 1.05% : 0.000003s : 17: predicate.reshape_eliminate 0.52% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.38% : 0.000001s : 4: predicate.row_tensor_eliminate 0.66% : 0.000002s : 8: predicate.same_eliminate 0.39% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.67% : 0.000002s : 8: predicate.shard_identity_eliminate 0.64% : 0.000002s : 8: predicate.special_op_eliminate 0.59% : 0.000001s : 8: predicate.specialize_transform 0.79% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.63% : 0.000001s : 8: predicate.stack_unstack_eliminate 0.29% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.75% : 0.000004s : 29: predicate.switch_defer_inline 2.21% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.46% : 0.000013s : 88: predicate.switch_simplify 0.98% : 0.000002s : 17: predicate.tile_eliminate 1.00% : 0.000002s : 17: predicate.transpose_eliminate 1.70% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.54% : 0.000004s : 25: predicate.tuple_list_get_item_const_eliminator 1.63% : 0.000004s : 25: predicate.tuple_list_get_item_depend_reorder 3.17% : 0.000008s : 39: predicate.tuple_list_get_item_eliminator 1.38% : 0.000003s : 25: predicate.tuple_list_get_set_item_eliminator 2.12% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 1.84% : 0.000004s : 31: predicate.tuple_to_list_eliminator_ 2.54% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.20% : 0.000008s : 56: predicate.updatestate_useless_node_eliminater 0.27% : 0.000001s : 4: predicate.value_based_eliminate 0.59% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.59% : 0.000001s : 8: predicate.virtual_output_eliminate 0.23% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.50% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000849 15 53.46% : 0.000454s : 7: func_graph_cloner_run.FuncGraphClonerGraph 46.54% : 0.000395s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.634793 192 0.00% : 0.000004s : 1: ForceFp32Comm 1.29% : 0.008219s : 1: add_attr 1.29% : 0.008202s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.02% : 0.000139s : 1: add_recomputation 0.00% : 0.000005s : 1: assign_add_opt 0.02% : 0.000123s : 1: auto_monad 0.01% : 0.000034s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000007s : 1: bias_add_comm_swap 0.12% : 0.000785s : 1: bootstrap 0.00% : 0.000030s : 1: cconv 0.00% : 0.000006s : 1: comm_op_add_attrs 0.00% : 0.000024s : 1: control_data_broadcast_order 0.00% : 0.000015s : 1: convert_after_rewriter 0.01% : 0.000075s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000007s : 1: detach_backward 0.00% : 0.000027s : 1: environ_conv 0.01% : 0.000048s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000005s : 1: get_jit_bprop_graph 0.00% : 0.000011s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000005s : 1: inline 0.00% : 0.000008s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000005s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000012s : 1: label_micro_interleaved_index 0.07% : 0.000438s : 1: loop_unroll 0.00% : 0.000006s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.09% : 0.000577s : 1: mutable_eliminate 0.00% : 0.000009s : 1: offloading_packed_experts 0.00% : 0.000013s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000015s : 1: opt.transform.mutable_eliminate 0.23% : 0.001461s : 78: opt.transform.opt_a 0.00% : 0.000028s : 1: opt.transform.opt_after_cconv 0.00% : 0.000025s : 1: opt.transform.opt_after_jit_grad 0.02% : 0.000100s : 28: opt.transform.opt_b 0.01% : 0.000048s : 2: opt.transform.opt_trans_graph 0.01% : 0.000062s : 4: opt.transform.symbol_engine_opt 0.51% : 0.003207s : 1: opt_a 0.02% : 0.000106s : 1: opt_after_cconv 0.13% : 0.000823s : 1: opt_after_jit_grad 0.03% : 0.000205s : 1: opt_b 27.11% : 0.172076s : 1: optimize 0.00% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000010s : 1: order_py_execute_after_rewriter 0.00% : 0.000028s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000016s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000008s : 1: overlap_param_gather 0.00% : 0.000005s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000006s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 26.18% : 0.166207s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.01% : 0.000074s : 1: pre_auto_parallel 0.01% : 0.000040s : 1: py_interpret_to_execute 0.00% : 0.000013s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000005s : 1: remove_cast_before_assign_add 0.00% : 0.000019s : 1: remove_dup_value 0.07% : 0.000438s : 1: renormalize.infer 0.06% : 0.000382s : 1: renormalize.specialize 0.00% : 0.000007s : 1: reorder_send_recv_between_fp_bp 0.04% : 0.000225s : 1: rewriter_after_jit_bprop_graph 0.05% : 0.000288s : 1: rewriter_after_opt_a 0.02% : 0.000130s : 1: rewriter_before_opt_a 0.00% : 0.000006s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000004s : 1: split_layernorm_comm 0.00% : 0.000007s : 1: split_matmul_comm_elemetwise 0.00% : 0.000010s : 1: swap_dp_allreduce_reducescatter 0.02% : 0.000118s : 1: symbol_engine_optimizer 0.01% : 0.000079s : 1: tuple_transform 42.46% : 0.269556s : 1: type_inference ************************************************************************************************** *************************************TEST_OP_GENERALIZE Start************************************* ************************************************************************************************** *****************Start Generalize test PYNATIVE_MODE DiscontiguousInput case***************** [INFO]: TEST_OP_GENERALIZE generate discontiguous input with shape (10, 10, 8), is_contiguous False, origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE generate discontiguous input with shape (10, 10, 8), is_contiguous False, origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE Start to test discontiguous input [INFO]: TEST_OP_GENERALIZE Start to test contiguous input [INFO]: TEST_OP_GENERALIZE Start to compare discontiguous out with contiguous out Compare ['Success'] for 0'th output. *****************End Generalize test PYNATIVE_MODE DiscontiguousInput case***************** *****************Start Generalize test PYNATIVE_MODE ViewTensor case***************** [INFO]: TEST_OP_GENERALIZE generate view input with shape (10, 10, 8), origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE generate view input with shape (10, 10, 8), origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE Start to test no view input [INFO]: TEST_OP_GENERALIZE Start to test view input [INFO]: TEST_OP_GENERALIZE Start to compare view out with no view out Compare ['Success'] for 0'th output. *****************End Generalize test PYNATIVE_MODE ViewTensor case***************** *****************Start Generalize test PYNATIVE_MODE EmptyTensor case***************** [WARNING]: TEST_OP_GENERALIZE PYNATIVE_MODE 'EmptyTensor' in 'disable_case', EmptyTensor case is skipped. *****************End Generalize test PYNATIVE_MODE EmptyTensor case***************** *****************Start Generalize test PYNATIVE_MODE ScalarTensor case***************** [INFO]: TEST_OP_GENERALIZE Start to test scalar tensor [INFO]: TEST_OP_GENERALIZE Start to test 1d tensor [INFO]: TEST_OP_GENERALIZE Start to compare scalar tensor out with 1d tensor out Compare ['Success'] for 0'th output. *****************End Generalize test PYNATIVE_MODE ScalarTensor case***************** *****************Start Generalize test PYNATIVE_MODE GradByRequirement case***************** [INFO]: TEST_OP_GENERALIZE Start to test all grads [INFO]: TEST_OP_GENERALIZE Start to test grad by requirement on position 0 [INFO]: TEST_OP_GENERALIZE Start to test grad by requirement on position 1 [INFO]: TEST_OP_GENERALIZE Start to compare grad by requirement with all grads Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. *****************End Generalize test PYNATIVE_MODE GradByRequirement case***************** *****************Start Generalize test PYNATIVE_MODE Deterministic case***************** [INFO]: TEST_OP_GENERALIZE test deterministic case loop 0 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 1 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 2 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 3 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 4 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. *****************End Generalize test PYNATIVE_MODE Deterministic case***************** *****************Start Generalize test GRAPH_MODE_O0 DiscontiguousInput case***************** [WARNING]: TEST_OP_GENERALIZE GRAPH_MODE_O0 'DiscontiguousInput' is skipped temporarily with 'GRAPH_MODE_O0' and 'GRAPH_MODE_GE', it's will cause accuracy issue after 'Tensor storage refactor'. *****************End Generalize test GRAPH_MODE_O0 DiscontiguousInput case***************** *****************Start Generalize test GRAPH_MODE_O0 ViewTensor case***************** [INFO]: TEST_OP_GENERALIZE generate view input with shape (10, 10, 8), origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE generate view input with shape (10, 10, 8), origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE Start to test no view input [INFO]: TEST_OP_GENERALIZE Start to test view input [INFO]: TEST_OP_GENERALIZE TotalTime = 1.0516, [21] [bootstrap]: 0.00054382 [type_inference]: 0.0103673 [event_method]: 6.529e-05 [auto_monad]: 8.413e-05 [graph_reusing]: 7.06999e-06 [inline]: 2.89001e-06 [add_attr]: 1.03419, [1] [add_attr_with_inline]: 1.03418, [1] [Cycle 1]: 8.338e-05, [2] [tag_attr]: 3.182e-05 [meta_addattr_fg_expand]: 7.31001e-06 [parallel-infer-symbol]: 4.12003e-06 [pre_auto_parallel]: 5.292e-05 [insert-virtual-dataset]: 3.14001e-06 [parallel-infer-symbol-second]: 9.39996e-07 [dataset_repeat_opt]: 2.25002e-06 [pipeline_split]: 1.92999e-06 [optimize]: 0.00552913, [53] [py_interpret_to_execute]: 3.033e-05 [rewriter_before_opt_a]: 0.00011469 [opt_a]: 0.0033021, [2] [Cycle 1]: 0.00260136, [45] [expand_dump_flag]: 3.65e-06 [switch_simplify]: 5.504e-05 [loop_unroll]: 3.913e-05 [a_1]: 0.00092314 [with_stream_mark]: 1.774e-05 [recompute_prepare]: 1.023e-05 [updatestate_depend_eliminate]: 4.32e-06 [updatestate_assign_eliminate]: 3.63e-06 [updatestate_loads_eliminate]: 3.41999e-06 [parameter_eliminate]: 2.03002e-06 [a_2]: 8.463e-05 [accelerated_algorithm]: 7.39002e-06 [shard]: 1.92999e-06 [meta_shard_fg_expand]: 1.71998e-06 [shard_inline]: 6.42001e-06 [merge_send_recv]: 8.82999e-06 [auto_parallel]: 7.31001e-06 [parallel]: 3.539e-05 [flash_sp]: 9.32999e-06 [merge_comm]: 4.23001e-06 [allreduce_fusion]: 3.63999e-06 [matmul_add_comm_reduction]: 9.76e-06 [allreduce_slice_to_reducescatter]: 6.10016e-07 [virtual_shard_identity]: 8.62998e-06 [virtual_dataset]: 7.02002e-06 [get_grad_eliminate_]: 6.01998e-06 [virtual_output]: 6.43e-06 [merge_forward]: 4.55999e-06 [cell_reuse_recompute_pass]: 1.33002e-06 [offload_activation]: 1.166e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.275e-05 [merge_recompute_call_nodes]: 1.63002e-06 [before_grad]: 1.029e-05 [set_forward_comm_id_for_comm_node_pass]: 3.66999e-06 [meta_fg_expand]: 2.77002e-06 [flash_sp_send_recv_attached]: 2.58003e-06 [receive_attached]: 2.37001e-06 [after_resolve]: 1.133e-05 [a_after_grad]: 1.045e-05 [renormalize]: 0.00089969 [add_forward_monad_depend]: 6.01e-06 [auto_monad_grad]: 2.88e-06 [auto_monad_eliminator]: 1.665e-05 [cse]: 3.04e-05 [a_3]: 4.761e-05 [Cycle 2]: 0.00069037, [45] [expand_dump_flag]: 1.22e-06 [switch_simplify]: 8.21002e-06 [loop_unroll]: 6.53e-06 [a_1]: 0.00014048 [with_stream_mark]: 1.264e-05 [recompute_prepare]: 6.89999e-06 [updatestate_depend_eliminate]: 1.132e-05 [updatestate_assign_eliminate]: 2.40002e-06 [updatestate_loads_eliminate]: 2.49001e-06 [parameter_eliminate]: 1.30999e-06 [a_2]: 7.86e-05 [accelerated_algorithm]: 6.01e-06 [shard]: 1.45999e-06 [meta_shard_fg_expand]: 1.49e-06 [shard_inline]: 6.33002e-06 [merge_send_recv]: 4.84998e-06 [auto_parallel]: 6.68e-06 [parallel]: 4.57e-06 [flash_sp]: 4.13001e-06 [merge_comm]: 3.82002e-06 [allreduce_fusion]: 3.11999e-06 [matmul_add_comm_reduction]: 5.82999e-06 [allreduce_slice_to_reducescatter]: 5.3001e-07 [virtual_shard_identity]: 7.28e-06 [virtual_dataset]: 6.48998e-06 [get_grad_eliminate_]: 5.85002e-06 [virtual_output]: 5.71e-06 [merge_forward]: 3.39001e-06 [cell_reuse_recompute_pass]: 1.94e-06 [offload_activation]: 7.23e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.163e-05 [merge_recompute_call_nodes]: 7.99977e-07 [before_grad]: 9.61e-06 [set_forward_comm_id_for_comm_node_pass]: 3.72998e-06 [meta_fg_expand]: 1.76e-06 [flash_sp_send_recv_attached]: 9.09989e-07 [receive_attached]: 9.5999e-07 [after_resolve]: 1.081e-05 [a_after_grad]: 9.39e-06 [renormalize]: 1.00001e-07 [add_forward_monad_depend]: 1.55001e-06 [auto_monad_grad]: 1.31002e-06 [auto_monad_eliminator]: 7.03e-06 [cse]: 1.476e-05 [a_3]: 3.469e-05 [py_interpret_to_execute_after_opt_a]: 9.64e-06 [slice_cell_reuse_recomputed_activation]: 2.39001e-06 [rewriter_after_opt_a]: 4.183e-05 [convert_after_rewriter]: 7.45e-06 [order_py_execute_after_rewriter]: 5.44998e-06 [mutable_eliminate]: 0.00061753 [opt_b]: 0.00020181, [1] [Cycle 1]: 0.0001948, [7] [b_1]: 0.00011924 [b_2]: 8.07998e-06 [updatestate_depend_eliminate]: 5.59998e-06 [updatestate_assign_eliminate]: 2.46e-06 [updatestate_loads_eliminate]: 2.54001e-06 [renormalize]: 6.60017e-07 [cse]: 1.904e-05 [optimize_parallel_all_gather_comm]: 1.745e-05 [overlap_param_gather]: 2.07001e-06 [cconv]: 2.766e-05 [loop_unroll]: 0.00047079 [opt_after_cconv]: 0.00010388, [1] [Cycle 1]: 9.757e-05, [7] [c_1]: 3.037e-05 [parameter_eliminate]: 2.39001e-06 [updatestate_depend_eliminate]: 5.15999e-06 [updatestate_assign_eliminate]: 2.64999e-06 [updatestate_loads_eliminate]: 2.32001e-06 [cse]: 1.882e-05 [renormalize]: 3.50003e-07 [remove_dup_value]: 1.586e-05 [tuple_transform]: 7.248e-05, [1] [Cycle 1]: 6.79e-05, [4] [d_1]: 4.11e-05 [none_parameter_eliminate]: 1.83002e-06 [renormalize]: 1.40019e-07 [switch_simplify]: 6.73998e-06 [partial_unused_args_eliminate]: 2.14e-06 [add_recomputation]: 5.139e-05 [cse_after_recomputation]: 2.172e-05, [1] [Cycle 1]: 1.711e-05, [1] [cse]: 1.131e-05 [environ_conv]: 6.60997e-06 [swap_dp_allreduce_reducescatter]: 5.46e-06 [bias_add_comm_swap]: 2.64999e-06 [label_micro_interleaved_index]: 4e-06 [label_fine_grained_interleaved_index]: 3.21001e-06 [merge_cast_opt]: 1.54998e-06 [slice_recompute_activation]: 2.34001e-06 [micro_interleaved_order_control]: 2.43e-06 [assign_add_opt]: 1.24e-06 [ForceFp32Comm]: 1.26997e-06 [remove_cast_before_assign_add]: 1.05999e-06 [full_micro_interleaved_order_control]: 2.39999e-06 [reorder_send_recv_between_fp_bp]: 2.88e-06 [comm_op_add_attrs]: 1.36002e-06 [add_comm_op_reuse_tag]: 1.01997e-06 [interleave_split_concat_branches]: 1.52001e-06 [interleave_parallel_branches]: 1.09e-06 [overlap_opt_shard_in_pipeline]: 1.19e-06 [overlap_opt_shard_grad_in_pipeline]: 1.76e-06 [control_data_broadcast_order]: 1.252e-05 [grouped_pairwise_exchange_alltoall]: 1.69e-06 [offloading_packed_experts]: 3.78001e-06 [overlap_recompute_and_grad_model_parallel]: 4.74e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.35999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.39e-06 [overlap_recompute_comm]: 2.59999e-06 [overlap_grad_ring_attention]: 4.53999e-06 [overlap_grad_flash_sp]: 2.127e-05 [begin_end_overlap_inline]: 5.89993e-07 [split_matmul_comm_elemetwise]: 2.44001e-06 [split_layernorm_comm]: 1.81e-06 [handle_group_info]: 1.29e-06 [symbol_engine_optimizer]: 7.852e-05, [1] [Cycle 1]: 7.416e-05, [6] [build]: 3.01001e-06 [elim_shapecalc]: 1.016e-05 [elim_not_effective]: 1.355e-05 [opt_reshape]: 6.74999e-06 [fold_const_symbol]: 1.016e-05 [renormalize]: 2.3999e-07 [detach_backward]: 1.94e-06 [pipeline_parallel_scheduler]: 1.86e-06 [auto_monad_reorder]: 1.743e-05 [get_jit_bprop_graph]: 1.45999e-06 [rewriter_after_jit_bprop_graph]: 4.2e-06 [opt_after_jit_grad]: 0.0004986 [validate]: 4.483e-05 Sums bootstrap : 0.000544s : 3.31% type_inference : 0.010367s : 63.18% event_method : 0.000065s : 0.40% auto_monad : 0.000084s : 0.51% graph_reusing : 0.000007s : 0.04% inline : 0.000003s : 0.02% add_attr.add_attr_with_inline.tag_attr : 0.000032s : 0.19% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.04% parallel-infer-symbol : 0.000004s : 0.03% pre_auto_parallel : 0.000053s : 0.32% insert-virtual-dataset : 0.000003s : 0.02% parallel-infer-symbol-second : 0.000001s : 0.01% dataset_repeat_opt : 0.000002s : 0.01% pipeline_split : 0.000002s : 0.01% optimize.py_interpret_to_execute : 0.000030s : 0.18% optimize.rewriter_before_opt_a : 0.000115s : 0.70% optimize.opt_a.expand_dump_flag : 0.000005s : 0.03% optimize.opt_a.switch_simplify : 0.000063s : 0.39% optimize.opt_a.loop_unroll : 0.000046s : 0.28% optimize.opt_a.a_1 : 0.001064s : 6.48% optimize.opt_a.with_stream_mark : 0.000030s : 0.19% optimize.opt_a.recompute_prepare : 0.000017s : 0.10% optimize.opt_a.updatestate_depend_eliminate : 0.000016s : 0.10% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.04% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.04% optimize.opt_a.parameter_eliminate : 0.000003s : 0.02% optimize.opt_a.a_2 : 0.000163s : 0.99% optimize.opt_a.accelerated_algorithm : 0.000013s : 0.08% optimize.opt_a.shard : 0.000003s : 0.02% optimize.opt_a.meta_shard_fg_expand : 0.000003s : 0.02% optimize.opt_a.shard_inline : 0.000013s : 0.08% optimize.opt_a.merge_send_recv : 0.000014s : 0.08% optimize.opt_a.auto_parallel : 0.000014s : 0.09% optimize.opt_a.parallel : 0.000040s : 0.24% optimize.opt_a.flash_sp : 0.000013s : 0.08% optimize.opt_a.merge_comm : 0.000008s : 0.05% optimize.opt_a.allreduce_fusion : 0.000007s : 0.04% optimize.opt_a.matmul_add_comm_reduction : 0.000016s : 0.10% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.01% optimize.opt_a.virtual_shard_identity : 0.000016s : 0.10% optimize.opt_a.virtual_dataset : 0.000014s : 0.08% optimize.opt_a.get_grad_eliminate_ : 0.000012s : 0.07% optimize.opt_a.virtual_output : 0.000012s : 0.07% optimize.opt_a.merge_forward : 0.000008s : 0.05% optimize.opt_a.cell_reuse_recompute_pass : 0.000003s : 0.02% optimize.opt_a.offload_activation : 0.000019s : 0.12% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000024s : 0.15% optimize.opt_a.merge_recompute_call_nodes : 0.000002s : 0.01% optimize.opt_a.before_grad : 0.000020s : 0.12% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000007s : 0.05% optimize.opt_a.meta_fg_expand : 0.000005s : 0.03% optimize.opt_a.flash_sp_send_recv_attached : 0.000003s : 0.02% optimize.opt_a.receive_attached : 0.000003s : 0.02% optimize.opt_a.after_resolve : 0.000022s : 0.13% optimize.opt_a.a_after_grad : 0.000020s : 0.12% optimize.opt_a.renormalize : 0.000900s : 5.48% optimize.opt_a.add_forward_monad_depend : 0.000008s : 0.05% optimize.opt_a.auto_monad_grad : 0.000004s : 0.03% optimize.opt_a.auto_monad_eliminator : 0.000024s : 0.14% optimize.opt_a.cse : 0.000045s : 0.28% optimize.opt_a.a_3 : 0.000082s : 0.50% optimize.py_interpret_to_execute_after_opt_a : 0.000010s : 0.06% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.01% optimize.rewriter_after_opt_a : 0.000042s : 0.25% optimize.convert_after_rewriter : 0.000007s : 0.05% optimize.order_py_execute_after_rewriter : 0.000005s : 0.03% optimize.mutable_eliminate : 0.000618s : 3.76% optimize.opt_b.b_1 : 0.000119s : 0.73% optimize.opt_b.b_2 : 0.000008s : 0.05% optimize.opt_b.updatestate_depend_eliminate : 0.000006s : 0.03% optimize.opt_b.updatestate_assign_eliminate : 0.000002s : 0.01% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.02% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000019s : 0.12% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.11% optimize.overlap_param_gather : 0.000002s : 0.01% optimize.cconv : 0.000028s : 0.17% optimize.loop_unroll : 0.000471s : 2.87% optimize.opt_after_cconv.c_1 : 0.000030s : 0.19% optimize.opt_after_cconv.parameter_eliminate : 0.000002s : 0.01% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000005s : 0.03% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.02% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.01% optimize.opt_after_cconv.cse : 0.000019s : 0.11% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000016s : 0.10% optimize.tuple_transform.d_1 : 0.000041s : 0.25% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.01% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.04% optimize.partial_unused_args_eliminate : 0.000002s : 0.01% optimize.add_recomputation : 0.000051s : 0.31% optimize.cse_after_recomputation.cse : 0.000011s : 0.07% optimize.environ_conv : 0.000007s : 0.04% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.03% optimize.bias_add_comm_swap : 0.000003s : 0.02% optimize.label_micro_interleaved_index : 0.000004s : 0.02% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.02% optimize.merge_cast_opt : 0.000002s : 0.01% optimize.slice_recompute_activation : 0.000002s : 0.01% optimize.micro_interleaved_order_control : 0.000002s : 0.01% optimize.assign_add_opt : 0.000001s : 0.01% optimize.ForceFp32Comm : 0.000001s : 0.01% optimize.remove_cast_before_assign_add : 0.000001s : 0.01% optimize.full_micro_interleaved_order_control : 0.000002s : 0.01% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.02% optimize.comm_op_add_attrs : 0.000001s : 0.01% optimize.add_comm_op_reuse_tag : 0.000001s : 0.01% optimize.interleave_split_concat_branches : 0.000002s : 0.01% optimize.interleave_parallel_branches : 0.000001s : 0.01% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.01% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.01% optimize.control_data_broadcast_order : 0.000013s : 0.08% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.01% optimize.offloading_packed_experts : 0.000004s : 0.02% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.03% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.01% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.01% optimize.overlap_recompute_comm : 0.000003s : 0.02% optimize.overlap_grad_ring_attention : 0.000005s : 0.03% optimize.overlap_grad_flash_sp : 0.000021s : 0.13% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.01% optimize.split_layernorm_comm : 0.000002s : 0.01% optimize.handle_group_info : 0.000001s : 0.01% optimize.symbol_engine_optimizer.build : 0.000003s : 0.02% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000010s : 0.06% optimize.symbol_engine_optimizer.elim_not_effective : 0.000014s : 0.08% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.04% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.06% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.01% pipeline_parallel_scheduler : 0.000002s : 0.01% auto_monad_reorder : 0.000017s : 0.11% get_jit_bprop_graph : 0.000001s : 0.01% rewriter_after_jit_bprop_graph : 0.000004s : 0.03% opt_after_jit_grad : 0.000499s : 3.04% validate : 0.000045s : 0.27% Time group info: ------[substitution.] 0.000293 40 12.06% : 0.000035s : 8: substitution.arithmetic_simplify 0.77% : 0.000002s : 2: substitution.elim_not_effective 0.48% : 0.000001s : 2: substitution.fold_const_symbol 2.15% : 0.000006s : 4: substitution.graph_param_transform 73.02% : 0.000214s : 6: substitution.inline 1.11% : 0.000003s : 4: substitution.j_node_and_user_rematch 1.86% : 0.000005s : 4: substitution.remove_not_recompute_node 1.56% : 0.000005s : 4: substitution.replace_old_param 6.98% : 0.000020s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.010292 2 90.10% : 0.009273s : 1: type_inference.infer 9.90% : 0.001019s : 1: type_inference.specialize ------[replace.] 0.000091 12 60.63% : 0.000055s : 6: replace.inline 39.37% : 0.000036s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000227 12 92.37% : 0.000210s : 6: match.inline 7.63% : 0.000017s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000250 1534 1.02% : 0.000003s : 17: predicate.accumulaten_eliminater 0.62% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.44% : 0.000001s : 8: predicate.addn_check_dump 1.24% : 0.000003s : 17: predicate.addn_zero_filter 0.90% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.24% : 0.000006s : 25: predicate.arithmetic_simplify 1.02% : 0.000003s : 17: predicate.cast_eliminate 0.56% : 0.000001s : 8: predicate.check_bprop_eliminate 0.41% : 0.000001s : 8: predicate.compare_switch_simplify 0.16% : 0.000000s : 4: predicate.const_output_eliminate 0.52% : 0.000001s : 8: predicate.depend_value_elim 1.00% : 0.000003s : 17: predicate.dict_get_item_const_eliminator 1.12% : 0.000003s : 17: predicate.dict_get_item_eliminator 1.10% : 0.000003s : 17: predicate.dict_set_item_eliminator 0.81% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.19% : 0.000000s : 4: predicate.elim_not_effective 0.41% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.26% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.18% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.15% : 0.000003s : 21: predicate.environ_get_depend_swap 1.75% : 0.000004s : 29: predicate.environ_get_eliminate 1.20% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.75% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.58% : 0.000006s : 29: predicate.float_depend_g_call 0.48% : 0.000001s : 8: predicate.float_environ_get_switch 0.68% : 0.000002s : 12: predicate.float_tuple_getitem_switch 0.13% : 0.000000s : 4: predicate.fold_const_symbol 0.52% : 0.000001s : 8: predicate.get_grad_eliminate 0.37% : 0.000001s : 4: predicate.graph_param_transform 0.48% : 0.000001s : 8: predicate.incorporate_call 0.39% : 0.000001s : 8: predicate.incorporate_call_switch 5.96% : 0.000015s : 70: predicate.inline 0.65% : 0.000002s : 8: predicate.inline_without_move 0.27% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.87% : 0.000002s : 8: predicate.less_batch_normalization 1.79% : 0.000004s : 31: predicate.list_to_tuple_eliminator_ 2.71% : 0.000007s : 48: predicate.load_eliminater 0.87% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.68% : 0.000007s : 47: predicate.loop_unroll_before_grad 1.61% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.44% : 0.000001s : 8: predicate.merge_addn 0.48% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.52% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.92% : 0.000002s : 17: predicate.minmaximum_grad 0.88% : 0.000002s : 4: predicate.mutable_eliminate 0.32% : 0.000001s : 4: predicate.opt_reshape 0.54% : 0.000001s : 4: predicate.parallel_virtual_node 2.34% : 0.000006s : 29: predicate.partial_defer_inline 1.64% : 0.000004s : 27: predicate.partial_eliminate 0.98% : 0.000002s : 17: predicate.print_const_string_wrapper 0.56% : 0.000001s : 8: predicate.reduce_all_const_elim 1.25% : 0.000003s : 17: predicate.reduce_eliminate 2.72% : 0.000007s : 48: predicate.redundant_stop_gradient_eliminater 0.29% : 0.000001s : 8: predicate.remove_not_recompute_node 1.44% : 0.000004s : 31: predicate.replace_applicator 0.34% : 0.000001s : 8: predicate.replace_old_param 0.22% : 0.000001s : 4: predicate.reset_defer_inline 1.21% : 0.000003s : 17: predicate.reshape_eliminate 0.54% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.29% : 0.000001s : 4: predicate.row_tensor_eliminate 0.58% : 0.000001s : 8: predicate.same_eliminate 0.40% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.66% : 0.000002s : 8: predicate.shard_identity_eliminate 0.72% : 0.000002s : 8: predicate.special_op_eliminate 0.61% : 0.000002s : 8: predicate.specialize_transform 0.80% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.60% : 0.000001s : 8: predicate.stack_unstack_eliminate 0.26% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.74% : 0.000004s : 29: predicate.switch_defer_inline 2.26% : 0.000006s : 37: predicate.switch_layer_defer_inline 5.23% : 0.000013s : 88: predicate.switch_simplify 1.01% : 0.000003s : 17: predicate.tile_eliminate 1.10% : 0.000003s : 17: predicate.transpose_eliminate 1.55% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.49% : 0.000004s : 25: predicate.tuple_list_get_item_const_eliminator 1.44% : 0.000004s : 25: predicate.tuple_list_get_item_depend_reorder 3.12% : 0.000008s : 39: predicate.tuple_list_get_item_eliminator 1.44% : 0.000004s : 25: predicate.tuple_list_get_set_item_eliminator 2.03% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 1.81% : 0.000005s : 31: predicate.tuple_to_list_eliminator_ 2.59% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.24% : 0.000008s : 56: predicate.updatestate_useless_node_eliminater 0.31% : 0.000001s : 4: predicate.value_based_eliminate 0.62% : 0.000002s : 8: predicate.virtual_dataset_eliminate 0.59% : 0.000001s : 8: predicate.virtual_output_eliminate 0.26% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.50% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000812 15 54.85% : 0.000446s : 7: func_graph_cloner_run.FuncGraphClonerGraph 45.15% : 0.000367s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 2.093854 192 0.00% : 0.000004s : 1: ForceFp32Comm 49.39% : 1.034198s : 1: add_attr 49.39% : 1.034182s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.00% : 0.000056s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.00% : 0.000090s : 1: auto_monad 0.00% : 0.000021s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 0.03% : 0.000576s : 1: bootstrap 0.00% : 0.000031s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000016s : 1: control_data_broadcast_order 0.00% : 0.000011s : 1: convert_after_rewriter 0.00% : 0.000025s : 1: cse_after_recomputation 0.00% : 0.000006s : 1: dataset_repeat_opt 0.00% : 0.000005s : 1: detach_backward 0.00% : 0.000010s : 1: environ_conv 0.00% : 0.000075s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000005s : 1: get_jit_bprop_graph 0.00% : 0.000011s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000009s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000005s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000007s : 1: label_micro_interleaved_index 0.02% : 0.000479s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.03% : 0.000627s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.00% : 0.000015s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000015s : 1: opt.transform.mutable_eliminate 0.07% : 0.001516s : 78: opt.transform.opt_a 0.00% : 0.000029s : 1: opt.transform.opt_after_cconv 0.00% : 0.000024s : 1: opt.transform.opt_after_jit_grad 0.00% : 0.000100s : 28: opt.transform.opt_b 0.00% : 0.000046s : 2: opt.transform.opt_trans_graph 0.00% : 0.000037s : 4: opt.transform.symbol_engine_opt 0.16% : 0.003305s : 1: opt_a 0.01% : 0.000108s : 1: opt_after_cconv 0.02% : 0.000507s : 1: opt_after_jit_grad 0.01% : 0.000205s : 1: opt_b 0.26% : 0.005534s : 1: optimize 0.00% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000009s : 1: order_py_execute_after_rewriter 0.00% : 0.000025s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000009s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.00% : 0.000058s : 1: pre_auto_parallel 0.00% : 0.000034s : 1: py_interpret_to_execute 0.00% : 0.000014s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000019s : 1: remove_dup_value 0.02% : 0.000437s : 1: renormalize.infer 0.02% : 0.000453s : 1: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000007s : 1: rewriter_after_jit_bprop_graph 0.00% : 0.000046s : 1: rewriter_after_opt_a 0.01% : 0.000119s : 1: rewriter_before_opt_a 0.00% : 0.000006s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000006s : 1: slice_recompute_activation 0.00% : 0.000006s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.00% : 0.000081s : 1: symbol_engine_optimizer 0.00% : 0.000075s : 1: tuple_transform 0.50% : 0.010387s : 1: type_inference TotalTime = 0.790399, [21] [bootstrap]: 0.00041599 [type_inference]: 0.77906 [event_method]: 4.188e-05 [auto_monad]: 7.48e-05 [graph_reusing]: 7.26999e-06 [inline]: 2.69999e-06 [add_attr]: 0.00461492, [1] [add_attr_with_inline]: 0.00460333, [1] [Cycle 1]: 7.604e-05, [2] [tag_attr]: 2.98e-05 [meta_addattr_fg_expand]: 7.16999e-06 [parallel-infer-symbol]: 3.80998e-06 [pre_auto_parallel]: 4.681e-05 [insert-virtual-dataset]: 2.71e-06 [parallel-infer-symbol-second]: 7.80012e-07 [dataset_repeat_opt]: 2.07001e-06 [pipeline_split]: 1.84e-06 [optimize]: 0.00539737, [53] [py_interpret_to_execute]: 3.332e-05 [rewriter_before_opt_a]: 0.00010828 [opt_a]: 0.00318549, [2] [Cycle 1]: 0.00253107, [45] [expand_dump_flag]: 3.16999e-06 [switch_simplify]: 5.153e-05 [loop_unroll]: 3.74e-05 [a_1]: 0.00088005 [with_stream_mark]: 3.434e-05 [recompute_prepare]: 1.022e-05 [updatestate_depend_eliminate]: 4.95001e-06 [updatestate_assign_eliminate]: 3.88999e-06 [updatestate_loads_eliminate]: 3.23998e-06 [parameter_eliminate]: 2.44999e-06 [a_2]: 8.013e-05 [accelerated_algorithm]: 7.11001e-06 [shard]: 2.45002e-06 [meta_shard_fg_expand]: 2.22999e-06 [shard_inline]: 5.97999e-06 [merge_send_recv]: 9.10999e-06 [auto_parallel]: 8.33001e-06 [parallel]: 2.013e-05 [flash_sp]: 8.38001e-06 [merge_comm]: 4.28999e-06 [allreduce_fusion]: 3.38999e-06 [matmul_add_comm_reduction]: 9.64e-06 [allreduce_slice_to_reducescatter]: 8.60018e-07 [virtual_shard_identity]: 7.46999e-06 [virtual_dataset]: 6.34001e-06 [get_grad_eliminate_]: 5.86e-06 [virtual_output]: 5.85002e-06 [merge_forward]: 3.96001e-06 [cell_reuse_recompute_pass]: 1.42e-06 [offload_activation]: 1.19e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.197e-05 [merge_recompute_call_nodes]: 1.54e-06 [before_grad]: 9.42001e-06 [set_forward_comm_id_for_comm_node_pass]: 3.53e-06 [meta_fg_expand]: 2.64001e-06 [flash_sp_send_recv_attached]: 2.76e-06 [receive_attached]: 2.10002e-06 [after_resolve]: 1.065e-05 [a_after_grad]: 9.52999e-06 [renormalize]: 0.0008883 [add_forward_monad_depend]: 5.71998e-06 [auto_monad_grad]: 2.79999e-06 [auto_monad_eliminator]: 1.75e-05 [cse]: 3.1e-05 [a_3]: 4.721e-05 [Cycle 2]: 0.0006432, [45] [expand_dump_flag]: 1.53002e-06 [switch_simplify]: 7.99002e-06 [loop_unroll]: 5.77001e-06 [a_1]: 0.00013094 [with_stream_mark]: 1.24e-05 [recompute_prepare]: 6.22001e-06 [updatestate_depend_eliminate]: 3.11001e-06 [updatestate_assign_eliminate]: 2.43e-06 [updatestate_loads_eliminate]: 2.81e-06 [parameter_eliminate]: 1.07e-06 [a_2]: 6.984e-05 [accelerated_algorithm]: 5.77001e-06 [shard]: 1.27e-06 [meta_shard_fg_expand]: 1.50001e-06 [shard_inline]: 6.21e-06 [merge_send_recv]: 5.02e-06 [auto_parallel]: 6.83998e-06 [parallel]: 4.77e-06 [flash_sp]: 3.58e-06 [merge_comm]: 3.16999e-06 [allreduce_fusion]: 3.02002e-06 [matmul_add_comm_reduction]: 5.80002e-06 [allreduce_slice_to_reducescatter]: 4.10015e-07 [virtual_shard_identity]: 9.96e-06 [virtual_dataset]: 5.43002e-06 [get_grad_eliminate_]: 5.12e-06 [virtual_output]: 5.20999e-06 [merge_forward]: 3.68e-06 [cell_reuse_recompute_pass]: 1.77999e-06 [offload_activation]: 7.46001e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.12e-05 [merge_recompute_call_nodes]: 1.21002e-06 [before_grad]: 8.85999e-06 [set_forward_comm_id_for_comm_node_pass]: 3.41001e-06 [meta_fg_expand]: 1.84e-06 [flash_sp_send_recv_attached]: 9.20001e-07 [receive_attached]: 1.69e-06 [after_resolve]: 1.021e-05 [a_after_grad]: 8.43999e-06 [renormalize]: 8.00064e-08 [add_forward_monad_depend]: 1.23002e-06 [auto_monad_grad]: 1.35001e-06 [auto_monad_eliminator]: 7.54002e-06 [cse]: 1.38e-05 [a_3]: 3.209e-05 [py_interpret_to_execute_after_opt_a]: 1.161e-05 [slice_cell_reuse_recomputed_activation]: 2.21e-06 [rewriter_after_opt_a]: 3.622e-05 [convert_after_rewriter]: 7.22002e-06 [order_py_execute_after_rewriter]: 5.17e-06 [mutable_eliminate]: 0.00064191 [opt_b]: 0.00019214, [1] [Cycle 1]: 0.00018519, [7] [b_1]: 0.00011043 [b_2]: 8.17e-06 [updatestate_depend_eliminate]: 7.09001e-06 [updatestate_assign_eliminate]: 2.64001e-06 [updatestate_loads_eliminate]: 2.79001e-06 [renormalize]: 8.2e-07 [cse]: 1.932e-05 [optimize_parallel_all_gather_comm]: 1.705e-05 [overlap_param_gather]: 1.81e-06 [cconv]: 2.759e-05 [loop_unroll]: 0.00044129 [opt_after_cconv]: 9.939e-05, [1] [Cycle 1]: 9.348e-05, [7] [c_1]: 2.889e-05 [parameter_eliminate]: 3.04999e-06 [updatestate_depend_eliminate]: 6.06e-06 [updatestate_assign_eliminate]: 2.55002e-06 [updatestate_loads_eliminate]: 2.14999e-06 [cse]: 1.738e-05 [renormalize]: 2.50002e-07 [remove_dup_value]: 1.448e-05 [tuple_transform]: 7.288e-05, [1] [Cycle 1]: 6.839e-05, [4] [d_1]: 4.208e-05 [none_parameter_eliminate]: 1.58002e-06 [renormalize]: 2.60014e-07 [switch_simplify]: 6.69001e-06 [partial_unused_args_eliminate]: 2.17001e-06 [add_recomputation]: 4.871e-05 [cse_after_recomputation]: 2.224e-05, [1] [Cycle 1]: 1.725e-05, [1] [cse]: 1.209e-05 [environ_conv]: 5.39e-06 [swap_dp_allreduce_reducescatter]: 5.39e-06 [bias_add_comm_swap]: 3.08e-06 [label_micro_interleaved_index]: 4.72998e-06 [label_fine_grained_interleaved_index]: 2.81e-06 [merge_cast_opt]: 1.50999e-06 [slice_recompute_activation]: 2.17001e-06 [micro_interleaved_order_control]: 2.53e-06 [assign_add_opt]: 1.65001e-06 [ForceFp32Comm]: 1.04e-06 [remove_cast_before_assign_add]: 1.05999e-06 [full_micro_interleaved_order_control]: 2.23998e-06 [reorder_send_recv_between_fp_bp]: 2.81e-06 [comm_op_add_attrs]: 1.25999e-06 [add_comm_op_reuse_tag]: 1.04e-06 [interleave_split_concat_branches]: 1.52999e-06 [interleave_parallel_branches]: 1.05001e-06 [overlap_opt_shard_in_pipeline]: 1.19e-06 [overlap_opt_shard_grad_in_pipeline]: 1.86e-06 [control_data_broadcast_order]: 1.217e-05 [grouped_pairwise_exchange_alltoall]: 2.03002e-06 [offloading_packed_experts]: 3.98999e-06 [overlap_recompute_and_grad_model_parallel]: 4.85001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.41002e-06 [overlap_recompute_allgather_and_fa_grad]: 1.48002e-06 [overlap_recompute_comm]: 2.71e-06 [overlap_grad_ring_attention]: 4.42e-06 [overlap_grad_flash_sp]: 2.051e-05 [begin_end_overlap_inline]: 5.39992e-07 [split_matmul_comm_elemetwise]: 2.46e-06 [split_layernorm_comm]: 1.91998e-06 [handle_group_info]: 1.72999e-06 [symbol_engine_optimizer]: 7.617e-05, [1] [Cycle 1]: 7.175e-05, [6] [build]: 3.16001e-06 [elim_shapecalc]: 9.81e-06 [elim_not_effective]: 1.322e-05 [opt_reshape]: 6.87002e-06 [fold_const_symbol]: 9.57001e-06 [renormalize]: 2.80008e-07 [detach_backward]: 2.06998e-06 [pipeline_parallel_scheduler]: 1.64e-06 [auto_monad_reorder]: 1.758e-05 [get_jit_bprop_graph]: 1.71002e-06 [rewriter_after_jit_bprop_graph]: 4.12e-06 [opt_after_jit_grad]: 0.00049165 [validate]: 4.126e-05 Sums bootstrap : 0.000416s : 0.05% type_inference : 0.779060s : 99.27% event_method : 0.000042s : 0.01% auto_monad : 0.000075s : 0.01% graph_reusing : 0.000007s : 0.00% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000030s : 0.00% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.00% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000047s : 0.01% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000033s : 0.00% optimize.rewriter_before_opt_a : 0.000108s : 0.01% optimize.opt_a.expand_dump_flag : 0.000005s : 0.00% optimize.opt_a.switch_simplify : 0.000060s : 0.01% optimize.opt_a.loop_unroll : 0.000043s : 0.01% optimize.opt_a.a_1 : 0.001011s : 0.13% optimize.opt_a.with_stream_mark : 0.000047s : 0.01% optimize.opt_a.recompute_prepare : 0.000016s : 0.00% optimize.opt_a.updatestate_depend_eliminate : 0.000008s : 0.00% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.00% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.00% optimize.opt_a.parameter_eliminate : 0.000004s : 0.00% optimize.opt_a.a_2 : 0.000150s : 0.02% optimize.opt_a.accelerated_algorithm : 0.000013s : 0.00% optimize.opt_a.shard : 0.000004s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000004s : 0.00% optimize.opt_a.shard_inline : 0.000012s : 0.00% optimize.opt_a.merge_send_recv : 0.000014s : 0.00% optimize.opt_a.auto_parallel : 0.000015s : 0.00% optimize.opt_a.parallel : 0.000025s : 0.00% optimize.opt_a.flash_sp : 0.000012s : 0.00% optimize.opt_a.merge_comm : 0.000007s : 0.00% optimize.opt_a.allreduce_fusion : 0.000006s : 0.00% optimize.opt_a.matmul_add_comm_reduction : 0.000015s : 0.00% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000017s : 0.00% optimize.opt_a.virtual_dataset : 0.000012s : 0.00% optimize.opt_a.get_grad_eliminate_ : 0.000011s : 0.00% optimize.opt_a.virtual_output : 0.000011s : 0.00% optimize.opt_a.merge_forward : 0.000008s : 0.00% optimize.opt_a.cell_reuse_recompute_pass : 0.000003s : 0.00% optimize.opt_a.offload_activation : 0.000019s : 0.00% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000023s : 0.00% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.00% optimize.opt_a.before_grad : 0.000018s : 0.00% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000007s : 0.00% optimize.opt_a.meta_fg_expand : 0.000004s : 0.00% optimize.opt_a.flash_sp_send_recv_attached : 0.000004s : 0.00% optimize.opt_a.receive_attached : 0.000004s : 0.00% optimize.opt_a.after_resolve : 0.000021s : 0.00% optimize.opt_a.a_after_grad : 0.000018s : 0.00% optimize.opt_a.renormalize : 0.000888s : 0.11% optimize.opt_a.add_forward_monad_depend : 0.000007s : 0.00% optimize.opt_a.auto_monad_grad : 0.000004s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000025s : 0.00% optimize.opt_a.cse : 0.000045s : 0.01% optimize.opt_a.a_3 : 0.000079s : 0.01% optimize.py_interpret_to_execute_after_opt_a : 0.000012s : 0.00% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000036s : 0.00% optimize.convert_after_rewriter : 0.000007s : 0.00% optimize.order_py_execute_after_rewriter : 0.000005s : 0.00% optimize.mutable_eliminate : 0.000642s : 0.08% optimize.opt_b.b_1 : 0.000110s : 0.01% optimize.opt_b.b_2 : 0.000008s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000007s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000019s : 0.00% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.00% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000028s : 0.00% optimize.loop_unroll : 0.000441s : 0.06% optimize.opt_after_cconv.c_1 : 0.000029s : 0.00% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000017s : 0.00% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000014s : 0.00% optimize.tuple_transform.d_1 : 0.000042s : 0.01% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000049s : 0.01% optimize.cse_after_recomputation.cse : 0.000012s : 0.00% optimize.environ_conv : 0.000005s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.00% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.00% optimize.assign_add_opt : 0.000002s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000002s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000012s : 0.00% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000003s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000021s : 0.00% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000002s : 0.00% optimize.symbol_engine_optimizer.build : 0.000003s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000010s : 0.00% optimize.symbol_engine_optimizer.elim_not_effective : 0.000013s : 0.00% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000018s : 0.00% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.00% opt_after_jit_grad : 0.000492s : 0.06% validate : 0.000041s : 0.01% Time group info: ------[substitution.] 0.000290 40 12.28% : 0.000036s : 8: substitution.arithmetic_simplify 0.71% : 0.000002s : 2: substitution.elim_not_effective 0.62% : 0.000002s : 2: substitution.fold_const_symbol 2.27% : 0.000007s : 4: substitution.graph_param_transform 73.00% : 0.000212s : 6: substitution.inline 1.25% : 0.000004s : 4: substitution.j_node_and_user_rematch 1.72% : 0.000005s : 4: substitution.remove_not_recompute_node 1.69% : 0.000005s : 4: substitution.replace_old_param 6.46% : 0.000019s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.778992 2 99.87% : 0.777997s : 1: type_inference.infer 0.13% : 0.000995s : 1: type_inference.specialize ------[replace.] 0.000089 12 60.27% : 0.000054s : 6: replace.inline 39.73% : 0.000035s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000225 12 92.78% : 0.000209s : 6: match.inline 7.22% : 0.000016s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000229 1534 1.01% : 0.000002s : 17: predicate.accumulaten_eliminater 0.60% : 0.000001s : 4: predicate.ad_related_special_op_eliminate 0.38% : 0.000001s : 8: predicate.addn_check_dump 1.02% : 0.000002s : 17: predicate.addn_zero_filter 0.98% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.11% : 0.000005s : 25: predicate.arithmetic_simplify 1.08% : 0.000002s : 17: predicate.cast_eliminate 0.52% : 0.000001s : 8: predicate.check_bprop_eliminate 0.40% : 0.000001s : 8: predicate.compare_switch_simplify 0.15% : 0.000000s : 4: predicate.const_output_eliminate 0.58% : 0.000001s : 8: predicate.depend_value_elim 1.08% : 0.000002s : 17: predicate.dict_get_item_const_eliminator 1.10% : 0.000003s : 17: predicate.dict_get_item_eliminator 1.13% : 0.000003s : 17: predicate.dict_set_item_eliminator 0.81% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.28% : 0.000001s : 4: predicate.elim_not_effective 0.41% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.27% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.14% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.16% : 0.000003s : 21: predicate.environ_get_depend_swap 1.64% : 0.000004s : 29: predicate.environ_get_eliminate 1.13% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.77% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.48% : 0.000006s : 29: predicate.float_depend_g_call 0.46% : 0.000001s : 8: predicate.float_environ_get_switch 0.70% : 0.000002s : 12: predicate.float_tuple_getitem_switch 0.15% : 0.000000s : 4: predicate.fold_const_symbol 0.60% : 0.000001s : 8: predicate.get_grad_eliminate 0.35% : 0.000001s : 4: predicate.graph_param_transform 0.54% : 0.000001s : 8: predicate.incorporate_call 0.37% : 0.000001s : 8: predicate.incorporate_call_switch 6.08% : 0.000014s : 70: predicate.inline 0.63% : 0.000001s : 8: predicate.inline_without_move 0.26% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.77% : 0.000002s : 8: predicate.less_batch_normalization 1.96% : 0.000004s : 31: predicate.list_to_tuple_eliminator_ 2.59% : 0.000006s : 48: predicate.load_eliminater 0.79% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.70% : 0.000006s : 47: predicate.loop_unroll_before_grad 1.77% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.43% : 0.000001s : 8: predicate.merge_addn 0.47% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.54% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.92% : 0.000002s : 17: predicate.minmaximum_grad 0.90% : 0.000002s : 4: predicate.mutable_eliminate 0.33% : 0.000001s : 4: predicate.opt_reshape 0.30% : 0.000001s : 4: predicate.parallel_virtual_node 2.28% : 0.000005s : 29: predicate.partial_defer_inline 1.68% : 0.000004s : 27: predicate.partial_eliminate 1.02% : 0.000002s : 17: predicate.print_const_string_wrapper 0.49% : 0.000001s : 8: predicate.reduce_all_const_elim 1.38% : 0.000003s : 17: predicate.reduce_eliminate 2.65% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.42% : 0.000001s : 8: predicate.remove_not_recompute_node 1.47% : 0.000003s : 31: predicate.replace_applicator 0.38% : 0.000001s : 8: predicate.replace_old_param 0.22% : 0.000001s : 4: predicate.reset_defer_inline 1.21% : 0.000003s : 17: predicate.reshape_eliminate 0.58% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.36% : 0.000001s : 4: predicate.row_tensor_eliminate 0.59% : 0.000001s : 8: predicate.same_eliminate 0.37% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.83% : 0.000002s : 8: predicate.shard_identity_eliminate 0.62% : 0.000001s : 8: predicate.special_op_eliminate 0.54% : 0.000001s : 8: predicate.specialize_transform 0.97% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.75% : 0.000002s : 8: predicate.stack_unstack_eliminate 0.31% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.81% : 0.000004s : 29: predicate.switch_defer_inline 2.21% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.54% : 0.000013s : 88: predicate.switch_simplify 0.98% : 0.000002s : 17: predicate.tile_eliminate 1.11% : 0.000003s : 17: predicate.transpose_eliminate 1.64% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.36% : 0.000003s : 25: predicate.tuple_list_get_item_const_eliminator 1.39% : 0.000003s : 25: predicate.tuple_list_get_item_depend_reorder 3.09% : 0.000007s : 39: predicate.tuple_list_get_item_eliminator 1.40% : 0.000003s : 25: predicate.tuple_list_get_set_item_eliminator 1.85% : 0.000004s : 33: predicate.tuple_list_set_item_eliminator 2.01% : 0.000005s : 31: predicate.tuple_to_list_eliminator_ 2.59% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.18% : 0.000007s : 56: predicate.updatestate_useless_node_eliminater 0.30% : 0.000001s : 4: predicate.value_based_eliminate 0.55% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.52% : 0.000001s : 8: predicate.virtual_output_eliminate 0.21% : 0.000000s : 4: predicate.virtual_view_grad_eliminate 0.32% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.770490 15 99.96% : 0.770147s : 7: func_graph_cloner_run.FuncGraphClonerGraph 0.04% : 0.000343s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.802848 192 0.00% : 0.000004s : 1: ForceFp32Comm 0.58% : 0.004621s : 1: add_attr 0.57% : 0.004608s : 1: add_attr_with_inline 0.00% : 0.000005s : 1: add_comm_op_reuse_tag 0.01% : 0.000053s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.01% : 0.000080s : 1: auto_monad 0.00% : 0.000021s : 1: auto_monad_reorder 0.00% : 0.000003s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 0.06% : 0.000445s : 1: bootstrap 0.00% : 0.000031s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000015s : 1: control_data_broadcast_order 0.00% : 0.000010s : 1: convert_after_rewriter 0.00% : 0.000025s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000005s : 1: detach_backward 0.00% : 0.000009s : 1: environ_conv 0.01% : 0.000049s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000005s : 1: get_jit_bprop_graph 0.00% : 0.000012s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000006s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000020s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.06% : 0.000449s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.08% : 0.000652s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.00% : 0.000014s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000015s : 1: opt.transform.mutable_eliminate 0.18% : 0.001434s : 78: opt.transform.opt_a 0.00% : 0.000028s : 1: opt.transform.opt_after_cconv 0.00% : 0.000023s : 1: opt.transform.opt_after_jit_grad 0.01% : 0.000092s : 28: opt.transform.opt_b 0.01% : 0.000047s : 2: opt.transform.opt_trans_graph 0.00% : 0.000036s : 4: opt.transform.symbol_engine_opt 0.40% : 0.003189s : 1: opt_a 0.01% : 0.000103s : 1: opt_after_cconv 0.06% : 0.000501s : 1: opt_after_jit_grad 0.02% : 0.000196s : 1: opt_b 0.67% : 0.005402s : 1: optimize 0.00% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000008s : 1: order_py_execute_after_rewriter 0.00% : 0.000024s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000005s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000009s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000006s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.01% : 0.000052s : 1: pre_auto_parallel 0.00% : 0.000037s : 1: py_interpret_to_execute 0.00% : 0.000015s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000018s : 1: remove_dup_value 0.06% : 0.000458s : 1: renormalize.infer 0.05% : 0.000421s : 1: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000007s : 1: rewriter_after_jit_bprop_graph 0.00% : 0.000040s : 1: rewriter_after_opt_a 0.01% : 0.000113s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000006s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.01% : 0.000079s : 1: symbol_engine_optimizer 0.01% : 0.000076s : 1: tuple_transform 97.04% : 0.779083s : 1: type_inference TotalTime = 2.78441, [21] [bootstrap]: 0.00045167 [type_inference]: 0.812383 [event_method]: 5.754e-05 [auto_monad]: 0.00012489 [graph_reusing]: 8.07e-06 [inline]: 3.01001e-06 [add_attr]: 1.09179, [1] [add_attr_with_inline]: 1.09178, [1] [Cycle 1]: 0.0001006, [2] [tag_attr]: 4.985e-05 [meta_addattr_fg_expand]: 1.024e-05 [parallel-infer-symbol]: 4.16001e-06 [pre_auto_parallel]: 6.792e-05 [insert-virtual-dataset]: 2.64001e-06 [parallel-infer-symbol-second]: 8.80013e-07 [dataset_repeat_opt]: 2.15002e-06 [pipeline_split]: 1.76003e-06 [optimize]: 0.0157221, [53] [py_interpret_to_execute]: 4.699e-05 [rewriter_before_opt_a]: 0.0001911 [opt_a]: 0.0132575, [3] [Cycle 1]: 0.00854209, [45] [expand_dump_flag]: 3.54002e-06 [switch_simplify]: 8.614e-05 [loop_unroll]: 7.056e-05 [a_1]: 0.0016256 [with_stream_mark]: 2.972e-05 [recompute_prepare]: 2.39e-05 [updatestate_depend_eliminate]: 9.62001e-06 [updatestate_assign_eliminate]: 7.76001e-06 [updatestate_loads_eliminate]: 7.6e-06 [parameter_eliminate]: 2.66999e-06 [a_2]: 0.00027187 [accelerated_algorithm]: 3.293e-05 [shard]: 1.89e-06 [meta_shard_fg_expand]: 3.68999e-06 [shard_inline]: 1.669e-05 [merge_send_recv]: 1.872e-05 [auto_parallel]: 1.255e-05 [parallel]: 2.073e-05 [flash_sp]: 1.062e-05 [merge_comm]: 1.023e-05 [allreduce_fusion]: 9.08002e-06 [matmul_add_comm_reduction]: 3.244e-05 [allreduce_slice_to_reducescatter]: 6.69999e-07 [virtual_shard_identity]: 1.893e-05 [virtual_dataset]: 1.677e-05 [get_grad_eliminate_]: 1.611e-05 [virtual_output]: 1.652e-05 [merge_forward]: 9.76e-06 [cell_reuse_recompute_pass]: 1.48002e-06 [offload_activation]: 2.116e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.226e-05 [merge_recompute_call_nodes]: 1.99999e-06 [before_grad]: 3.144e-05 [set_forward_comm_id_for_comm_node_pass]: 1.113e-05 [meta_fg_expand]: 0.00174609 [flash_sp_send_recv_attached]: 7.07002e-06 [receive_attached]: 2.44001e-06 [after_resolve]: 6.985e-05 [a_after_grad]: 9.323e-05 [renormalize]: 0.00317706 [add_forward_monad_depend]: 1.104e-05 [auto_monad_grad]: 6.51999e-06 [auto_monad_eliminator]: 5.815e-05 [cse]: 0.00024582 [a_3]: 0.00033421 [Cycle 2]: 0.00383266, [45] [expand_dump_flag]: 2.57001e-06 [switch_simplify]: 4.836e-05 [loop_unroll]: 4.274e-05 [a_1]: 0.00157116 [with_stream_mark]: 1.641e-05 [recompute_prepare]: 1.085e-05 [updatestate_depend_eliminate]: 4.13999e-06 [updatestate_assign_eliminate]: 2.95002e-06 [updatestate_loads_eliminate]: 2.41e-06 [parameter_eliminate]: 2.09e-06 [a_2]: 6.772e-05 [accelerated_algorithm]: 1.143e-05 [shard]: 1.79e-06 [meta_shard_fg_expand]: 2.02999e-06 [shard_inline]: 5.89999e-06 [merge_send_recv]: 6.78e-06 [auto_parallel]: 8.45999e-06 [parallel]: 6.67002e-06 [flash_sp]: 4.08999e-06 [merge_comm]: 3.31001e-06 [allreduce_fusion]: 3.93999e-06 [matmul_add_comm_reduction]: 7.03e-06 [allreduce_slice_to_reducescatter]: 3.89991e-07 [virtual_shard_identity]: 7.36999e-06 [virtual_dataset]: 6.66e-06 [get_grad_eliminate_]: 6.51e-06 [virtual_output]: 5.66998e-06 [merge_forward]: 3.58999e-06 [cell_reuse_recompute_pass]: 1.10001e-06 [offload_activation]: 8.84e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.085e-05 [merge_recompute_call_nodes]: 1.11002e-06 [before_grad]: 8.72e-06 [set_forward_comm_id_for_comm_node_pass]: 3.87998e-06 [meta_fg_expand]: 0.00011682 [flash_sp_send_recv_attached]: 1.37e-06 [receive_attached]: 1.87999e-06 [after_resolve]: 1.47e-05 [a_after_grad]: 9.17999e-06 [renormalize]: 0.00145482 [add_forward_monad_depend]: 5.42999e-06 [auto_monad_grad]: 2.43e-06 [auto_monad_eliminator]: 1.486e-05 [cse]: 2.018e-05 [a_3]: 4.32e-05 [Cycle 3]: 0.00086646, [45] [expand_dump_flag]: 1.87999e-06 [switch_simplify]: 7.33e-06 [loop_unroll]: 5.79999e-06 [a_1]: 0.00011965 [with_stream_mark]: 1.212e-05 [recompute_prepare]: 5.74999e-06 [updatestate_depend_eliminate]: 3.56001e-06 [updatestate_assign_eliminate]: 2.38002e-06 [updatestate_loads_eliminate]: 2.44999e-06 [parameter_eliminate]: 1.10001e-06 [a_2]: 7.344e-05 [accelerated_algorithm]: 1.543e-05 [shard]: 1.14e-06 [meta_shard_fg_expand]: 1.50999e-06 [shard_inline]: 6.14001e-06 [merge_send_recv]: 8.08001e-06 [auto_parallel]: 7.73001e-06 [parallel]: 6.73e-06 [flash_sp]: 1.15999e-06 [merge_comm]: 3.21999e-06 [allreduce_fusion]: 3.05002e-06 [matmul_add_comm_reduction]: 6.18998e-06 [allreduce_slice_to_reducescatter]: 3.30008e-07 [virtual_shard_identity]: 6.60002e-06 [virtual_dataset]: 5.52999e-06 [get_grad_eliminate_]: 5.83002e-06 [virtual_output]: 5.20001e-06 [merge_forward]: 3.43999e-06 [cell_reuse_recompute_pass]: 1.57001e-06 [offload_activation]: 8.1e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.12e-05 [merge_recompute_call_nodes]: 9.09989e-07 [before_grad]: 8.79998e-06 [set_forward_comm_id_for_comm_node_pass]: 3.2e-06 [meta_fg_expand]: 2.81e-06 [flash_sp_send_recv_attached]: 1.09998e-06 [receive_attached]: 1.20999e-06 [after_resolve]: 1.057e-05 [a_after_grad]: 8.1e-06 [renormalize]: 7.99773e-08 [add_forward_monad_depend]: 2.68998e-06 [auto_monad_grad]: 1.59e-06 [auto_monad_eliminator]: 9.77999e-06 [cse]: 1.743e-05 [a_3]: 3.321e-05 [py_interpret_to_execute_after_opt_a]: 1.329e-05 [slice_cell_reuse_recomputed_activation]: 3.01001e-06 [rewriter_after_opt_a]: 4.166e-05 [convert_after_rewriter]: 7.26001e-06 [order_py_execute_after_rewriter]: 5.66e-06 [mutable_eliminate]: 0.0006775 [opt_b]: 0.00020925, [1] [Cycle 1]: 0.00020046, [7] [b_1]: 0.00012175 [b_2]: 7.68001e-06 [updatestate_depend_eliminate]: 7.13e-06 [updatestate_assign_eliminate]: 2.69999e-06 [updatestate_loads_eliminate]: 2.48e-06 [renormalize]: 5.60016e-07 [cse]: 2.063e-05 [optimize_parallel_all_gather_comm]: 1.815e-05 [overlap_param_gather]: 1.99e-06 [cconv]: 2.845e-05 [loop_unroll]: 0.00048476 [opt_after_cconv]: 0.00010716, [1] [Cycle 1]: 0.0001006, [7] [c_1]: 2.989e-05 [parameter_eliminate]: 3.36001e-06 [updatestate_depend_eliminate]: 6.81999e-06 [updatestate_assign_eliminate]: 2.59999e-06 [updatestate_loads_eliminate]: 2.53e-06 [cse]: 2.009e-05 [renormalize]: 3.4002e-07 [remove_dup_value]: 1.391e-05 [tuple_transform]: 7.597e-05, [1] [Cycle 1]: 7.121e-05, [4] [d_1]: 4.402e-05 [none_parameter_eliminate]: 1.80001e-06 [renormalize]: 2.00002e-07 [switch_simplify]: 6.58998e-06 [partial_unused_args_eliminate]: 1.97999e-06 [add_recomputation]: 5.093e-05 [cse_after_recomputation]: 2.261e-05, [1] [Cycle 1]: 1.76e-05, [1] [cse]: 1.206e-05 [environ_conv]: 1.016e-05 [swap_dp_allreduce_reducescatter]: 5.57001e-06 [bias_add_comm_swap]: 2.51e-06 [label_micro_interleaved_index]: 5.00999e-06 [label_fine_grained_interleaved_index]: 3.18e-06 [merge_cast_opt]: 1.54e-06 [slice_recompute_activation]: 2.17999e-06 [micro_interleaved_order_control]: 2.40002e-06 [assign_add_opt]: 1.40001e-06 [ForceFp32Comm]: 1.25001e-06 [remove_cast_before_assign_add]: 1.07998e-06 [full_micro_interleaved_order_control]: 2.22999e-06 [reorder_send_recv_between_fp_bp]: 2.94001e-06 [comm_op_add_attrs]: 1.18001e-06 [add_comm_op_reuse_tag]: 1.04e-06 [interleave_split_concat_branches]: 1.19e-06 [interleave_parallel_branches]: 1.06002e-06 [overlap_opt_shard_in_pipeline]: 1.50001e-06 [overlap_opt_shard_grad_in_pipeline]: 1.96e-06 [control_data_broadcast_order]: 1.3e-05 [grouped_pairwise_exchange_alltoall]: 1.98002e-06 [offloading_packed_experts]: 3.68999e-06 [overlap_recompute_and_grad_model_parallel]: 4.90999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.31002e-06 [overlap_recompute_allgather_and_fa_grad]: 1.49e-06 [overlap_recompute_comm]: 2.65002e-06 [overlap_grad_ring_attention]: 4.54002e-06 [overlap_grad_flash_sp]: 2.203e-05 [begin_end_overlap_inline]: 5.8001e-07 [split_matmul_comm_elemetwise]: 2.43e-06 [split_layernorm_comm]: 1.94e-06 [handle_group_info]: 1.24e-06 [symbol_engine_optimizer]: 0.00010219, [1] [Cycle 1]: 9.751e-05, [6] [build]: 2.763e-05 [elim_shapecalc]: 9.99001e-06 [elim_not_effective]: 1.292e-05 [opt_reshape]: 6.44001e-06 [fold_const_symbol]: 9.42001e-06 [renormalize]: 2.10013e-07 [detach_backward]: 2.51e-06 [pipeline_parallel_scheduler]: 1.64e-06 [auto_monad_reorder]: 1.957e-05 [get_jit_bprop_graph]: 1.84998e-06 [rewriter_after_jit_bprop_graph]: 4.45e-06 [opt_after_jit_grad]: 0.00050533 [validate]: 0.863042 Sums bootstrap : 0.000452s : 0.03% type_inference : 0.812383s : 48.04% event_method : 0.000058s : 0.00% auto_monad : 0.000125s : 0.01% graph_reusing : 0.000008s : 0.00% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000050s : 0.00% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000010s : 0.00% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000068s : 0.00% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000047s : 0.00% optimize.rewriter_before_opt_a : 0.000191s : 0.01% optimize.opt_a.expand_dump_flag : 0.000008s : 0.00% optimize.opt_a.switch_simplify : 0.000142s : 0.01% optimize.opt_a.loop_unroll : 0.000119s : 0.01% optimize.opt_a.a_1 : 0.003316s : 0.20% optimize.opt_a.with_stream_mark : 0.000058s : 0.00% optimize.opt_a.recompute_prepare : 0.000040s : 0.00% optimize.opt_a.updatestate_depend_eliminate : 0.000017s : 0.00% optimize.opt_a.updatestate_assign_eliminate : 0.000013s : 0.00% optimize.opt_a.updatestate_loads_eliminate : 0.000012s : 0.00% optimize.opt_a.parameter_eliminate : 0.000006s : 0.00% optimize.opt_a.a_2 : 0.000413s : 0.02% optimize.opt_a.accelerated_algorithm : 0.000060s : 0.00% optimize.opt_a.shard : 0.000005s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000007s : 0.00% optimize.opt_a.shard_inline : 0.000029s : 0.00% optimize.opt_a.merge_send_recv : 0.000034s : 0.00% optimize.opt_a.auto_parallel : 0.000029s : 0.00% optimize.opt_a.parallel : 0.000034s : 0.00% optimize.opt_a.flash_sp : 0.000016s : 0.00% optimize.opt_a.merge_comm : 0.000017s : 0.00% optimize.opt_a.allreduce_fusion : 0.000016s : 0.00% optimize.opt_a.matmul_add_comm_reduction : 0.000046s : 0.00% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000033s : 0.00% optimize.opt_a.virtual_dataset : 0.000029s : 0.00% optimize.opt_a.get_grad_eliminate_ : 0.000028s : 0.00% optimize.opt_a.virtual_output : 0.000027s : 0.00% optimize.opt_a.merge_forward : 0.000017s : 0.00% optimize.opt_a.cell_reuse_recompute_pass : 0.000004s : 0.00% optimize.opt_a.offload_activation : 0.000038s : 0.00% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000054s : 0.00% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.00% optimize.opt_a.before_grad : 0.000049s : 0.00% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000018s : 0.00% optimize.opt_a.meta_fg_expand : 0.001866s : 0.11% optimize.opt_a.flash_sp_send_recv_attached : 0.000010s : 0.00% optimize.opt_a.receive_attached : 0.000006s : 0.00% optimize.opt_a.after_resolve : 0.000095s : 0.01% optimize.opt_a.a_after_grad : 0.000111s : 0.01% optimize.opt_a.renormalize : 0.004632s : 0.27% optimize.opt_a.add_forward_monad_depend : 0.000019s : 0.00% optimize.opt_a.auto_monad_grad : 0.000011s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000083s : 0.00% optimize.opt_a.cse : 0.000283s : 0.02% optimize.opt_a.a_3 : 0.000411s : 0.02% optimize.py_interpret_to_execute_after_opt_a : 0.000013s : 0.00% optimize.slice_cell_reuse_recomputed_activation : 0.000003s : 0.00% optimize.rewriter_after_opt_a : 0.000042s : 0.00% optimize.convert_after_rewriter : 0.000007s : 0.00% optimize.order_py_execute_after_rewriter : 0.000006s : 0.00% optimize.mutable_eliminate : 0.000678s : 0.04% optimize.opt_b.b_1 : 0.000122s : 0.01% optimize.opt_b.b_2 : 0.000008s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000007s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000021s : 0.00% optimize.optimize_parallel_all_gather_comm : 0.000018s : 0.00% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000028s : 0.00% optimize.loop_unroll : 0.000485s : 0.03% optimize.opt_after_cconv.c_1 : 0.000030s : 0.00% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000007s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.cse : 0.000020s : 0.00% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000014s : 0.00% optimize.tuple_transform.d_1 : 0.000044s : 0.00% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000051s : 0.00% optimize.cse_after_recomputation.cse : 0.000012s : 0.00% optimize.environ_conv : 0.000010s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.00% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000002s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000013s : 0.00% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000003s : 0.00% optimize.overlap_grad_ring_attention : 0.000005s : 0.00% optimize.overlap_grad_flash_sp : 0.000022s : 0.00% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000028s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000010s : 0.00% optimize.symbol_engine_optimizer.elim_not_effective : 0.000013s : 0.00% optimize.symbol_engine_optimizer.opt_reshape : 0.000006s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000009s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000020s : 0.00% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.00% opt_after_jit_grad : 0.000505s : 0.03% validate : 0.863042s : 51.04% Time group info: ------[substitution.] 0.000907 190 9.71% : 0.000088s : 6: substitution.arithmetic_simplify 0.22% : 0.000002s : 2: substitution.elim_not_effective 0.62% : 0.000006s : 6: substitution.float_depend_g_call 0.53% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.18% : 0.000002s : 2: substitution.fold_const_symbol 0.66% : 0.000006s : 4: substitution.graph_param_transform 0.32% : 0.000003s : 2: substitution.incorporate_call 0.27% : 0.000002s : 2: substitution.incorporate_call_switch 56.25% : 0.000510s : 16: substitution.inline 2.26% : 0.000021s : 2: substitution.inline_without_move 1.09% : 0.000010s : 14: substitution.j_node_and_user_rematch 2.05% : 0.000019s : 3: substitution.less_batch_normalization 1.58% : 0.000014s : 11: substitution.minmaximum_grad 0.71% : 0.000006s : 6: substitution.partial_eliminate 1.34% : 0.000012s : 14: substitution.remove_not_recompute_node 3.12% : 0.000028s : 10: substitution.replace_applicator 1.27% : 0.000012s : 12: substitution.replace_old_param 0.28% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.33% : 0.000030s : 11: substitution.tuple_list_convert_item_index_to_positive 1.62% : 0.000015s : 11: substitution.tuple_list_get_item_const_eliminator 2.18% : 0.000020s : 11: substitution.tuple_list_get_item_depend_reorder 8.28% : 0.000075s : 30: substitution.tuple_list_get_item_eliminator 2.13% : 0.000019s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.812281 2 99.76% : 0.810360s : 1: type_inference.infer 0.24% : 0.001921s : 1: type_inference.specialize ------[replace.] 0.000257 34 4.35% : 0.000011s : 2: replace.arithmetic_simplify 55.70% : 0.000143s : 16: replace.inline 39.94% : 0.000103s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.000596 34 9.49% : 0.000057s : 2: match.arithmetic_simplify 84.03% : 0.000501s : 16: match.inline 6.48% : 0.000039s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.000699 4766 1.10% : 0.000008s : 58: predicate.accumulaten_eliminater 0.21% : 0.000001s : 4: predicate.ad_related_special_op_eliminate 0.42% : 0.000003s : 23: predicate.addn_check_dump 1.15% : 0.000008s : 58: predicate.addn_zero_filter 1.07% : 0.000007s : 58: predicate.adjust_all_reduce_mul_add 1.97% : 0.000014s : 83: predicate.arithmetic_simplify 1.19% : 0.000008s : 60: predicate.cast_eliminate 1.02% : 0.000007s : 55: predicate.check_bprop_eliminate 0.44% : 0.000003s : 23: predicate.compare_switch_simplify 0.05% : 0.000000s : 4: predicate.const_output_eliminate 0.44% : 0.000003s : 23: predicate.depend_value_elim 1.23% : 0.000009s : 60: predicate.dict_get_item_const_eliminator 1.38% : 0.000010s : 60: predicate.dict_get_item_eliminator 1.23% : 0.000009s : 60: predicate.dict_set_item_eliminator 0.26% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.06% : 0.000000s : 4: predicate.elim_not_effective 0.12% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.19% : 0.000008s : 64: predicate.environ_add_const_eliminate 1.15% : 0.000008s : 64: predicate.environ_get_add_eliminate 1.20% : 0.000008s : 64: predicate.environ_get_depend_swap 1.65% : 0.000012s : 87: predicate.environ_get_eliminate 1.18% : 0.000008s : 64: predicate.environ_get_set_eliminate 1.83% : 0.000013s : 92: predicate.exchange_switch_depend_value 2.45% : 0.000017s : 92: predicate.float_depend_g_call 0.44% : 0.000003s : 23: predicate.float_environ_get_switch 0.57% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.05% : 0.000000s : 4: predicate.fold_const_symbol 0.49% : 0.000003s : 23: predicate.get_grad_eliminate 0.09% : 0.000001s : 4: predicate.graph_param_transform 0.43% : 0.000003s : 23: predicate.incorporate_call 0.38% : 0.000003s : 23: predicate.incorporate_call_switch 5.34% : 0.000037s : 204: predicate.inline 1.21% : 0.000008s : 46: predicate.inline_without_move 0.26% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.60% : 0.000004s : 23: predicate.less_batch_normalization 1.79% : 0.000013s : 84: predicate.list_to_tuple_eliminator_ 2.64% : 0.000018s : 142: predicate.load_eliminater 0.36% : 0.000003s : 4: predicate.loop_unroll_after_grad 2.54% : 0.000018s : 128: predicate.loop_unroll_before_grad 1.36% : 0.000010s : 68: predicate.make_slice_get_slice_eliminator 0.44% : 0.000003s : 23: predicate.merge_addn 1.07% : 0.000008s : 55: predicate.micro_step_allgather_replace 1.10% : 0.000008s : 55: predicate.mini_step_allgather_replace 1.21% : 0.000008s : 60: predicate.minmaximum_grad 0.33% : 0.000002s : 4: predicate.mutable_eliminate 0.09% : 0.000001s : 4: predicate.opt_reshape 0.16% : 0.000001s : 4: predicate.parallel_virtual_node 2.38% : 0.000017s : 92: predicate.partial_defer_inline 1.70% : 0.000012s : 80: predicate.partial_eliminate 1.15% : 0.000008s : 58: predicate.print_const_string_wrapper 0.41% : 0.000003s : 23: predicate.reduce_all_const_elim 1.36% : 0.000010s : 60: predicate.reduce_eliminate 2.67% : 0.000019s : 142: predicate.redundant_stop_gradient_eliminater 0.36% : 0.000002s : 23: predicate.remove_not_recompute_node 1.85% : 0.000013s : 131: predicate.replace_applicator 0.60% : 0.000004s : 46: predicate.replace_old_param 0.08% : 0.000001s : 4: predicate.reset_defer_inline 1.24% : 0.000009s : 60: predicate.reshape_eliminate 1.04% : 0.000007s : 55: predicate.row_tensor_add_zeros_like 0.08% : 0.000001s : 4: predicate.row_tensor_eliminate 1.29% : 0.000009s : 55: predicate.same_eliminate 0.31% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.54% : 0.000004s : 23: predicate.shard_identity_eliminate 0.21% : 0.000001s : 8: predicate.special_op_eliminate 0.58% : 0.000004s : 23: predicate.specialize_transform 1.21% : 0.000008s : 55: predicate.split_environ_get_set_with_tuple_value 1.23% : 0.000009s : 46: predicate.stack_unstack_eliminate 0.12% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.99% : 0.000014s : 92: predicate.switch_defer_inline 2.96% : 0.000021s : 147: predicate.switch_layer_defer_inline 5.38% : 0.000038s : 247: predicate.switch_simplify 1.18% : 0.000008s : 60: predicate.tile_eliminate 1.14% : 0.000008s : 60: predicate.transpose_eliminate 1.43% : 0.000010s : 68: predicate.tuple_list_convert_item_index_to_positive 1.55% : 0.000011s : 68: predicate.tuple_list_get_item_const_eliminator 1.30% : 0.000009s : 68: predicate.tuple_list_get_item_depend_reorder 3.52% : 0.000025s : 107: predicate.tuple_list_get_item_eliminator 1.42% : 0.000010s : 68: predicate.tuple_list_get_set_item_eliminator 1.88% : 0.000013s : 91: predicate.tuple_list_set_item_eliminator 1.66% : 0.000012s : 84: predicate.tuple_to_list_eliminator_ 2.59% : 0.000018s : 142: predicate.updatestate_pure_node_eliminater 3.11% : 0.000022s : 165: predicate.updatestate_useless_node_eliminater 0.15% : 0.000001s : 4: predicate.value_based_eliminate 0.50% : 0.000003s : 23: predicate.virtual_dataset_eliminate 0.56% : 0.000004s : 23: predicate.virtual_output_eliminate 0.08% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.27% : 0.000002s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002171 37 60.22% : 0.001307s : 17: func_graph_cloner_run.FuncGraphClonerGraph 39.78% : 0.000864s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 3.038460 233 0.00% : 0.000004s : 1: ForceFp32Comm 35.93% : 1.091798s : 1: add_attr 35.93% : 1.091782s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.00% : 0.000055s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.00% : 0.000133s : 1: auto_monad 0.00% : 0.000023s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 0.02% : 0.000485s : 1: bootstrap 0.00% : 0.000032s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000017s : 1: control_data_broadcast_order 0.00% : 0.000010s : 1: convert_after_rewriter 0.00% : 0.000026s : 1: cse_after_recomputation 0.00% : 0.000006s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000013s : 1: environ_conv 0.00% : 0.000066s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000006s : 1: get_jit_bprop_graph 0.00% : 0.000012s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.02% : 0.000493s : 1: loop_unroll 0.00% : 0.000005s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.02% : 0.000707s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.00% : 0.000016s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000017s : 1: opt.transform.mutable_eliminate 0.16% : 0.004817s : 117: opt.transform.opt_a 0.00% : 0.000029s : 1: opt.transform.opt_after_cconv 0.00% : 0.000026s : 1: opt.transform.opt_after_jit_grad 0.00% : 0.000103s : 28: opt.transform.opt_b 0.00% : 0.000049s : 2: opt.transform.opt_trans_graph 0.00% : 0.000035s : 4: opt.transform.symbol_engine_opt 0.44% : 0.013261s : 1: opt_a 0.00% : 0.000111s : 1: opt_after_cconv 0.02% : 0.000515s : 1: opt_after_jit_grad 0.01% : 0.000213s : 1: opt_b 0.52% : 0.015729s : 1: optimize 0.00% : 0.000022s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000009s : 1: order_py_execute_after_rewriter 0.00% : 0.000026s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000006s : 1: overlap_recompute_comm 0.00% : 0.000009s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000006s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.00% : 0.000073s : 1: pre_auto_parallel 0.00% : 0.000051s : 1: py_interpret_to_execute 0.00% : 0.000017s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000017s : 1: remove_dup_value 0.07% : 0.002049s : 2: renormalize.infer 0.08% : 0.002564s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000008s : 1: rewriter_after_jit_bprop_graph 0.00% : 0.000045s : 1: rewriter_after_opt_a 0.01% : 0.000196s : 1: rewriter_before_opt_a 0.00% : 0.000006s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.00% : 0.000105s : 1: symbol_engine_optimizer 0.00% : 0.000079s : 1: tuple_transform 26.74% : 0.812409s : 1: type_inference TotalTime = 0.841662, [21] [bootstrap]: 0.00048432 [type_inference]: 0.626733 [event_method]: 7.295e-05 [auto_monad]: 0.00012864 [graph_reusing]: 9.22001e-06 [inline]: 3.03e-06 [add_attr]: 0.0039936, [1] [add_attr_with_inline]: 0.00398237, [1] [Cycle 1]: 9.693e-05, [2] [tag_attr]: 4.856e-05 [meta_addattr_fg_expand]: 9.24e-06 [parallel-infer-symbol]: 3.89002e-06 [pre_auto_parallel]: 6.351e-05 [insert-virtual-dataset]: 2.69999e-06 [parallel-infer-symbol-second]: 7.60017e-07 [dataset_repeat_opt]: 2.02999e-06 [pipeline_split]: 1.66002e-06 [optimize]: 0.209435, [53] [py_interpret_to_execute]: 4.555e-05 [rewriter_before_opt_a]: 0.00017282 [opt_a]: 0.207021, [3] [Cycle 1]: 0.202982, [45] [expand_dump_flag]: 3.98001e-06 [switch_simplify]: 7.875e-05 [loop_unroll]: 6.293e-05 [a_1]: 0.00148003 [with_stream_mark]: 2.604e-05 [recompute_prepare]: 2.145e-05 [updatestate_depend_eliminate]: 8.47998e-06 [updatestate_assign_eliminate]: 6.76e-06 [updatestate_loads_eliminate]: 6.36e-06 [parameter_eliminate]: 3.04001e-06 [a_2]: 0.00021347 [accelerated_algorithm]: 1.438e-05 [shard]: 2.46e-06 [meta_shard_fg_expand]: 3.51001e-06 [shard_inline]: 1.356e-05 [merge_send_recv]: 1.695e-05 [auto_parallel]: 1.251e-05 [parallel]: 1.994e-05 [flash_sp]: 1.07e-05 [merge_comm]: 8.60001e-06 [allreduce_fusion]: 7.54002e-06 [matmul_add_comm_reduction]: 2.787e-05 [allreduce_slice_to_reducescatter]: 9.00007e-07 [virtual_shard_identity]: 1.567e-05 [virtual_dataset]: 1.363e-05 [get_grad_eliminate_]: 1.394e-05 [virtual_output]: 1.375e-05 [merge_forward]: 8.52998e-06 [cell_reuse_recompute_pass]: 1.25999e-06 [offload_activation]: 1.708e-05 [cell_reuse_handle_not_recompute_node_pass]: 2.534e-05 [merge_recompute_call_nodes]: 1.55999e-06 [before_grad]: 2.457e-05 [set_forward_comm_id_for_comm_node_pass]: 8.28999e-06 [meta_fg_expand]: 0.00158644 [flash_sp_send_recv_attached]: 4.02e-06 [receive_attached]: 2.71e-06 [after_resolve]: 5.914e-05 [a_after_grad]: 7.981e-05 [renormalize]: 0.19804 [add_forward_monad_depend]: 1.37e-05 [auto_monad_grad]: 8.95001e-06 [auto_monad_eliminator]: 6.47e-05 [cse]: 0.00024863 [a_3]: 0.00034134 [Cycle 2]: 0.00354717, [45] [expand_dump_flag]: 2.86e-06 [switch_simplify]: 4.473e-05 [loop_unroll]: 4.065e-05 [a_1]: 0.00200998 [with_stream_mark]: 2.29e-05 [recompute_prepare]: 9.66e-06 [updatestate_depend_eliminate]: 4.05e-06 [updatestate_assign_eliminate]: 2.56e-06 [updatestate_loads_eliminate]: 2.29001e-06 [parameter_eliminate]: 2.30002e-06 [a_2]: 4.887e-05 [accelerated_algorithm]: 5.05001e-06 [shard]: 2.50002e-06 [meta_shard_fg_expand]: 3.53999e-06 [shard_inline]: 4.23001e-06 [merge_send_recv]: 8.27e-06 [auto_parallel]: 1.045e-05 [parallel]: 9.10999e-06 [flash_sp]: 4.79e-06 [merge_comm]: 2.85002e-06 [allreduce_fusion]: 2.54001e-06 [matmul_add_comm_reduction]: 8.47e-06 [allreduce_slice_to_reducescatter]: 7.7e-07 [virtual_shard_identity]: 5.79999e-06 [virtual_dataset]: 5.96e-06 [get_grad_eliminate_]: 4.18001e-06 [virtual_output]: 4.74002e-06 [merge_forward]: 3.5e-06 [cell_reuse_recompute_pass]: 1.67001e-06 [offload_activation]: 9.72001e-06 [cell_reuse_handle_not_recompute_node_pass]: 8.90999e-06 [merge_recompute_call_nodes]: 1.81998e-06 [before_grad]: 8.32e-06 [set_forward_comm_id_for_comm_node_pass]: 3.05998e-06 [meta_fg_expand]: 0.00015742 [flash_sp_send_recv_attached]: 1.92999e-06 [receive_attached]: 2.96999e-06 [after_resolve]: 1.141e-05 [a_after_grad]: 6.74999e-06 [renormalize]: 0.00071474 [add_forward_monad_depend]: 5.42001e-06 [auto_monad_grad]: 2.52001e-06 [auto_monad_eliminator]: 1.109e-05 [cse]: 2.035e-05 [a_3]: 3.448e-05 [Cycle 3]: 0.00047169, [45] [expand_dump_flag]: 1.74e-06 [switch_simplify]: 5.54998e-06 [loop_unroll]: 4.48999e-06 [a_1]: 6.495e-05 [with_stream_mark]: 8.80999e-06 [recompute_prepare]: 4.28001e-06 [updatestate_depend_eliminate]: 2.20002e-06 [updatestate_assign_eliminate]: 1.62999e-06 [updatestate_loads_eliminate]: 1.42999e-06 [parameter_eliminate]: 8.79983e-07 [a_2]: 4.225e-05 [accelerated_algorithm]: 4.08999e-06 [shard]: 1.30001e-06 [meta_shard_fg_expand]: 1.28002e-06 [shard_inline]: 3.97e-06 [merge_send_recv]: 4.50001e-06 [auto_parallel]: 6.09999e-06 [parallel]: 5.30999e-06 [flash_sp]: 1.20999e-06 [merge_comm]: 2.22001e-06 [allreduce_fusion]: 2.04e-06 [matmul_add_comm_reduction]: 4.22e-06 [allreduce_slice_to_reducescatter]: 4.59986e-07 [virtual_shard_identity]: 5.09e-06 [virtual_dataset]: 4.18001e-06 [get_grad_eliminate_]: 4.11001e-06 [virtual_output]: 3.86999e-06 [merge_forward]: 2.31998e-06 [cell_reuse_recompute_pass]: 1.57001e-06 [offload_activation]: 5.64998e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.065e-05 [merge_recompute_call_nodes]: 8.70001e-07 [before_grad]: 6.13002e-06 [set_forward_comm_id_for_comm_node_pass]: 2.51e-06 [meta_fg_expand]: 1.74e-06 [flash_sp_send_recv_attached]: 9.39996e-07 [receive_attached]: 1.05999e-06 [after_resolve]: 8.55999e-06 [a_after_grad]: 5.52999e-06 [renormalize]: 9.00181e-08 [add_forward_monad_depend]: 1.22999e-06 [auto_monad_grad]: 1.24e-06 [auto_monad_eliminator]: 3.91999e-06 [cse]: 9.89001e-06 [a_3]: 2.259e-05 [py_interpret_to_execute_after_opt_a]: 1.108e-05 [slice_cell_reuse_recomputed_activation]: 2.17001e-06 [rewriter_after_opt_a]: 3.152e-05 [convert_after_rewriter]: 6.54999e-06 [order_py_execute_after_rewriter]: 4.80999e-06 [mutable_eliminate]: 0.00076778 [opt_b]: 0.00016943, [1] [Cycle 1]: 0.00016136, [7] [b_1]: 9.154e-05 [b_2]: 6.39001e-06 [updatestate_depend_eliminate]: 4.79e-06 [updatestate_assign_eliminate]: 2.02999e-06 [updatestate_loads_eliminate]: 2.21e-06 [renormalize]: 4.60015e-07 [cse]: 1.661e-05 [optimize_parallel_all_gather_comm]: 1.665e-05 [overlap_param_gather]: 2.09999e-06 [cconv]: 2.789e-05 [loop_unroll]: 0.00052949 [opt_after_cconv]: 8.919e-05, [1] [Cycle 1]: 8.238e-05, [7] [c_1]: 2.02e-05 [parameter_eliminate]: 2.58e-06 [updatestate_depend_eliminate]: 4.25e-06 [updatestate_assign_eliminate]: 1.85001e-06 [updatestate_loads_eliminate]: 1.63002e-06 [cse]: 1.67e-05 [renormalize]: 6.60017e-07 [remove_dup_value]: 1.285e-05 [tuple_transform]: 6.035e-05, [1] [Cycle 1]: 5.562e-05, [4] [d_1]: 3.019e-05 [none_parameter_eliminate]: 1.81003e-06 [renormalize]: 3.19997e-07 [switch_simplify]: 4.80001e-06 [partial_unused_args_eliminate]: 1.82001e-06 [add_recomputation]: 3.849e-05 [cse_after_recomputation]: 1.925e-05, [1] [Cycle 1]: 1.434e-05, [1] [cse]: 8.12e-06 [environ_conv]: 4.86002e-06 [swap_dp_allreduce_reducescatter]: 4.39002e-06 [bias_add_comm_swap]: 3.98999e-06 [label_micro_interleaved_index]: 4.70001e-06 [label_fine_grained_interleaved_index]: 2.91999e-06 [merge_cast_opt]: 1.57001e-06 [slice_recompute_activation]: 2.43e-06 [micro_interleaved_order_control]: 2.58e-06 [assign_add_opt]: 1.34e-06 [ForceFp32Comm]: 1.10001e-06 [remove_cast_before_assign_add]: 1.09003e-06 [full_micro_interleaved_order_control]: 2.09e-06 [reorder_send_recv_between_fp_bp]: 2.92002e-06 [comm_op_add_attrs]: 1.42e-06 [add_comm_op_reuse_tag]: 1.13001e-06 [interleave_split_concat_branches]: 1.15001e-06 [interleave_parallel_branches]: 1.09e-06 [overlap_opt_shard_in_pipeline]: 1.24e-06 [overlap_opt_shard_grad_in_pipeline]: 1.84998e-06 [control_data_broadcast_order]: 1.064e-05 [grouped_pairwise_exchange_alltoall]: 1.66e-06 [offloading_packed_experts]: 3.21999e-06 [overlap_recompute_and_grad_model_parallel]: 3.73999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.50999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.34e-06 [overlap_recompute_comm]: 2.37999e-06 [overlap_grad_ring_attention]: 3.13e-06 [overlap_grad_flash_sp]: 1.839e-05 [begin_end_overlap_inline]: 5.89993e-07 [split_matmul_comm_elemetwise]: 2.51e-06 [split_layernorm_comm]: 1.76e-06 [handle_group_info]: 1.19998e-06 [symbol_engine_optimizer]: 6.762e-05, [1] [Cycle 1]: 6.285e-05, [6] [build]: 2.98e-06 [elim_shapecalc]: 8.64e-06 [elim_not_effective]: 9.91e-06 [opt_reshape]: 5.24998e-06 [fold_const_symbol]: 6.96001e-06 [renormalize]: 2.10013e-07 [detach_backward]: 2.31998e-06 [pipeline_parallel_scheduler]: 1.82001e-06 [auto_monad_reorder]: 1.524e-05 [get_jit_bprop_graph]: 2.21e-06 [rewriter_after_jit_bprop_graph]: 3.76999e-06 [opt_after_jit_grad]: 0.00047751 [validate]: 3.63e-05 Sums bootstrap : 0.000484s : 0.06% type_inference : 0.626733s : 74.94% event_method : 0.000073s : 0.01% auto_monad : 0.000129s : 0.02% graph_reusing : 0.000009s : 0.00% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000049s : 0.01% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000009s : 0.00% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000064s : 0.01% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000046s : 0.01% optimize.rewriter_before_opt_a : 0.000173s : 0.02% optimize.opt_a.expand_dump_flag : 0.000009s : 0.00% optimize.opt_a.switch_simplify : 0.000129s : 0.02% optimize.opt_a.loop_unroll : 0.000108s : 0.01% optimize.opt_a.a_1 : 0.003555s : 0.43% optimize.opt_a.with_stream_mark : 0.000058s : 0.01% optimize.opt_a.recompute_prepare : 0.000035s : 0.00% optimize.opt_a.updatestate_depend_eliminate : 0.000015s : 0.00% optimize.opt_a.updatestate_assign_eliminate : 0.000011s : 0.00% optimize.opt_a.updatestate_loads_eliminate : 0.000010s : 0.00% optimize.opt_a.parameter_eliminate : 0.000006s : 0.00% optimize.opt_a.a_2 : 0.000305s : 0.04% optimize.opt_a.accelerated_algorithm : 0.000024s : 0.00% optimize.opt_a.shard : 0.000006s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000008s : 0.00% optimize.opt_a.shard_inline : 0.000022s : 0.00% optimize.opt_a.merge_send_recv : 0.000030s : 0.00% optimize.opt_a.auto_parallel : 0.000029s : 0.00% optimize.opt_a.parallel : 0.000034s : 0.00% optimize.opt_a.flash_sp : 0.000017s : 0.00% optimize.opt_a.merge_comm : 0.000014s : 0.00% optimize.opt_a.allreduce_fusion : 0.000012s : 0.00% optimize.opt_a.matmul_add_comm_reduction : 0.000041s : 0.00% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000027s : 0.00% optimize.opt_a.virtual_dataset : 0.000024s : 0.00% optimize.opt_a.get_grad_eliminate_ : 0.000022s : 0.00% optimize.opt_a.virtual_output : 0.000022s : 0.00% optimize.opt_a.merge_forward : 0.000014s : 0.00% optimize.opt_a.cell_reuse_recompute_pass : 0.000005s : 0.00% optimize.opt_a.offload_activation : 0.000032s : 0.00% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000045s : 0.01% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.00% optimize.opt_a.before_grad : 0.000039s : 0.00% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000014s : 0.00% optimize.opt_a.meta_fg_expand : 0.001746s : 0.21% optimize.opt_a.flash_sp_send_recv_attached : 0.000007s : 0.00% optimize.opt_a.receive_attached : 0.000007s : 0.00% optimize.opt_a.after_resolve : 0.000079s : 0.01% optimize.opt_a.a_after_grad : 0.000092s : 0.01% optimize.opt_a.renormalize : 0.198755s : 23.77% optimize.opt_a.add_forward_monad_depend : 0.000020s : 0.00% optimize.opt_a.auto_monad_grad : 0.000013s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000080s : 0.01% optimize.opt_a.cse : 0.000279s : 0.03% optimize.opt_a.a_3 : 0.000398s : 0.05% optimize.py_interpret_to_execute_after_opt_a : 0.000011s : 0.00% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000032s : 0.00% optimize.convert_after_rewriter : 0.000007s : 0.00% optimize.order_py_execute_after_rewriter : 0.000005s : 0.00% optimize.mutable_eliminate : 0.000768s : 0.09% optimize.opt_b.b_1 : 0.000092s : 0.01% optimize.opt_b.b_2 : 0.000006s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000005s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000017s : 0.00% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.00% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000028s : 0.00% optimize.loop_unroll : 0.000529s : 0.06% optimize.opt_after_cconv.c_1 : 0.000020s : 0.00% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000004s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000017s : 0.00% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000013s : 0.00% optimize.tuple_transform.d_1 : 0.000030s : 0.00% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000005s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000038s : 0.00% optimize.cse_after_recomputation.cse : 0.000008s : 0.00% optimize.environ_conv : 0.000005s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000004s : 0.00% optimize.bias_add_comm_swap : 0.000004s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000011s : 0.00% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000003s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000004s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000002s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000003s : 0.00% optimize.overlap_grad_flash_sp : 0.000018s : 0.00% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000003s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000003s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000009s : 0.00% optimize.symbol_engine_optimizer.elim_not_effective : 0.000010s : 0.00% optimize.symbol_engine_optimizer.opt_reshape : 0.000005s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000007s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000015s : 0.00% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.00% opt_after_jit_grad : 0.000478s : 0.06% validate : 0.000036s : 0.00% Time group info: ------[substitution.] 0.000956 143 7.82% : 0.000075s : 5: substitution.arithmetic_simplify 0.16% : 0.000002s : 1: substitution.elim_not_effective 0.60% : 0.000006s : 6: substitution.float_depend_g_call 0.47% : 0.000004s : 2: substitution.float_tuple_getitem_switch 0.14% : 0.000001s : 1: substitution.fold_const_symbol 0.49% : 0.000005s : 2: substitution.graph_param_transform 0.31% : 0.000003s : 2: substitution.incorporate_call 0.27% : 0.000003s : 2: substitution.incorporate_call_switch 64.07% : 0.000612s : 16: substitution.inline 1.70% : 0.000016s : 2: substitution.inline_without_move 0.85% : 0.000008s : 10: substitution.j_node_and_user_rematch 1.41% : 0.000014s : 7: substitution.minmaximum_grad 0.95% : 0.000009s : 6: substitution.partial_eliminate 1.04% : 0.000010s : 10: substitution.remove_not_recompute_node 3.84% : 0.000037s : 10: substitution.replace_applicator 1.14% : 0.000011s : 10: substitution.replace_old_param 0.28% : 0.000003s : 1: substitution.set_cell_output_no_recompute 2.71% : 0.000026s : 7: substitution.tuple_list_convert_item_index_to_positive 1.15% : 0.000011s : 7: substitution.tuple_list_get_item_const_eliminator 1.73% : 0.000016s : 7: substitution.tuple_list_get_item_depend_reorder 7.23% : 0.000069s : 22: substitution.tuple_list_get_item_eliminator 1.62% : 0.000016s : 7: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.626612 2 2.24% : 0.014059s : 1: type_inference.infer 97.76% : 0.612553s : 1: type_inference.specialize ------[replace.] 0.000268 30 2.23% : 0.000006s : 1: replace.arithmetic_simplify 57.57% : 0.000154s : 16: replace.inline 40.20% : 0.000108s : 13: replace.tuple_list_get_item_eliminator ------[match.] 0.000679 30 5.61% : 0.000038s : 1: match.arithmetic_simplify 88.80% : 0.000603s : 16: match.inline 5.59% : 0.000038s : 13: match.tuple_list_get_item_eliminator ------[predicate.] 0.000572 3865 1.28% : 0.000007s : 48: predicate.accumulaten_eliminater 0.23% : 0.000001s : 2: predicate.ad_related_special_op_eliminate 0.39% : 0.000002s : 17: predicate.addn_check_dump 1.19% : 0.000007s : 48: predicate.addn_zero_filter 1.04% : 0.000006s : 48: predicate.adjust_all_reduce_mul_add 2.09% : 0.000012s : 66: predicate.arithmetic_simplify 1.22% : 0.000007s : 49: predicate.cast_eliminate 1.23% : 0.000007s : 48: predicate.check_bprop_eliminate 0.40% : 0.000002s : 17: predicate.compare_switch_simplify 0.04% : 0.000000s : 2: predicate.const_output_eliminate 0.41% : 0.000002s : 17: predicate.depend_value_elim 1.26% : 0.000007s : 49: predicate.dict_get_item_const_eliminator 1.25% : 0.000007s : 49: predicate.dict_get_item_eliminator 1.16% : 0.000007s : 49: predicate.dict_set_item_eliminator 0.27% : 0.000002s : 4: predicate.dumpgradient_eliminate 0.05% : 0.000000s : 2: predicate.elim_not_effective 0.16% : 0.000001s : 2: predicate.elim_shapecalc_of_broadcastargs 1.24% : 0.000007s : 51: predicate.environ_add_const_eliminate 1.11% : 0.000006s : 51: predicate.environ_get_add_eliminate 1.09% : 0.000006s : 51: predicate.environ_get_depend_swap 1.68% : 0.000010s : 68: predicate.environ_get_eliminate 1.07% : 0.000006s : 51: predicate.environ_get_set_eliminate 1.85% : 0.000011s : 78: predicate.exchange_switch_depend_value 2.78% : 0.000016s : 78: predicate.float_depend_g_call 0.44% : 0.000003s : 17: predicate.float_environ_get_switch 0.45% : 0.000003s : 19: predicate.float_tuple_getitem_switch 0.04% : 0.000000s : 2: predicate.fold_const_symbol 0.56% : 0.000003s : 17: predicate.get_grad_eliminate 0.07% : 0.000000s : 2: predicate.graph_param_transform 0.39% : 0.000002s : 17: predicate.incorporate_call 0.35% : 0.000002s : 17: predicate.incorporate_call_switch 5.34% : 0.000031s : 164: predicate.inline 1.32% : 0.000008s : 40: predicate.inline_without_move 0.23% : 0.000001s : 17: predicate.j_node_and_user_rematch 0.55% : 0.000003s : 17: predicate.less_batch_normalization 1.65% : 0.000009s : 66: predicate.list_to_tuple_eliminator_ 2.50% : 0.000014s : 114: predicate.load_eliminater 0.26% : 0.000001s : 2: predicate.loop_unroll_after_grad 2.83% : 0.000016s : 121: predicate.loop_unroll_before_grad 1.36% : 0.000008s : 53: predicate.make_slice_get_slice_eliminator 0.50% : 0.000003s : 17: predicate.merge_addn 1.17% : 0.000007s : 48: predicate.micro_step_allgather_replace 1.24% : 0.000007s : 48: predicate.mini_step_allgather_replace 1.13% : 0.000006s : 49: predicate.minmaximum_grad 0.31% : 0.000002s : 2: predicate.mutable_eliminate 0.07% : 0.000000s : 2: predicate.opt_reshape 0.16% : 0.000001s : 2: predicate.parallel_virtual_node 2.67% : 0.000015s : 78: predicate.partial_defer_inline 1.64% : 0.000009s : 64: predicate.partial_eliminate 1.07% : 0.000006s : 48: predicate.print_const_string_wrapper 0.43% : 0.000002s : 17: predicate.reduce_all_const_elim 1.42% : 0.000008s : 49: predicate.reduce_eliminate 2.56% : 0.000015s : 114: predicate.redundant_stop_gradient_eliminater 0.31% : 0.000002s : 17: predicate.remove_not_recompute_node 2.07% : 0.000012s : 110: predicate.replace_applicator 0.58% : 0.000003s : 40: predicate.replace_old_param 0.07% : 0.000000s : 2: predicate.reset_defer_inline 1.27% : 0.000007s : 49: predicate.reshape_eliminate 1.21% : 0.000007s : 48: predicate.row_tensor_add_zeros_like 0.13% : 0.000001s : 2: predicate.row_tensor_eliminate 1.38% : 0.000008s : 48: predicate.same_eliminate 0.31% : 0.000002s : 17: predicate.set_cell_output_no_recompute 0.57% : 0.000003s : 17: predicate.shard_identity_eliminate 0.28% : 0.000002s : 4: predicate.special_op_eliminate 0.49% : 0.000003s : 17: predicate.specialize_transform 1.49% : 0.000009s : 48: predicate.split_environ_get_set_with_tuple_value 1.24% : 0.000007s : 40: predicate.stack_unstack_eliminate 0.06% : 0.000000s : 2: predicate.switch_call_monad_eliminater 2.04% : 0.000012s : 78: predicate.switch_defer_inline 3.09% : 0.000018s : 126: predicate.switch_layer_defer_inline 5.34% : 0.000031s : 218: predicate.switch_simplify 1.19% : 0.000007s : 49: predicate.tile_eliminate 1.16% : 0.000007s : 49: predicate.transpose_eliminate 1.45% : 0.000008s : 53: predicate.tuple_list_convert_item_index_to_positive 1.38% : 0.000008s : 53: predicate.tuple_list_get_item_const_eliminator 1.12% : 0.000006s : 53: predicate.tuple_list_get_item_depend_reorder 2.90% : 0.000017s : 83: predicate.tuple_list_get_item_eliminator 1.35% : 0.000008s : 53: predicate.tuple_list_get_set_item_eliminator 1.78% : 0.000010s : 70: predicate.tuple_list_set_item_eliminator 1.72% : 0.000010s : 66: predicate.tuple_to_list_eliminator_ 2.54% : 0.000015s : 114: predicate.updatestate_pure_node_eliminater 2.90% : 0.000017s : 131: predicate.updatestate_useless_node_eliminater 0.08% : 0.000000s : 2: predicate.value_based_eliminate 0.60% : 0.000003s : 17: predicate.virtual_dataset_eliminate 0.52% : 0.000003s : 17: predicate.virtual_output_eliminate 0.05% : 0.000000s : 2: predicate.virtual_view_grad_eliminate 0.13% : 0.000001s : 2: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002198 37 55.91% : 0.001229s : 17: func_graph_cloner_run.FuncGraphClonerGraph 44.09% : 0.000969s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 1.258668 233 0.00% : 0.000004s : 1: ForceFp32Comm 0.32% : 0.003999s : 1: add_attr 0.32% : 0.003987s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.00% : 0.000043s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.01% : 0.000136s : 1: auto_monad 0.00% : 0.000019s : 1: auto_monad_reorder 0.00% : 0.000005s : 1: begin_end_overlap_inline 0.00% : 0.000007s : 1: bias_add_comm_swap 0.04% : 0.000516s : 1: bootstrap 0.00% : 0.000032s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000014s : 1: control_data_broadcast_order 0.00% : 0.000010s : 1: convert_after_rewriter 0.00% : 0.000022s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000008s : 1: environ_conv 0.01% : 0.000083s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000005s : 1: get_jit_bprop_graph 0.00% : 0.000014s : 1: graph_reusing 0.00% : 0.000006s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000007s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.04% : 0.000538s : 1: loop_unroll 0.00% : 0.000005s : 1: merge_cast_opt 0.00% : 0.000006s : 1: micro_interleaved_order_control 0.06% : 0.000778s : 1: mutable_eliminate 0.00% : 0.000006s : 1: offloading_packed_experts 0.00% : 0.000012s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000013s : 1: opt.transform.mutable_eliminate 0.38% : 0.004783s : 117: opt.transform.opt_a 0.00% : 0.000019s : 1: opt.transform.opt_after_cconv 0.00% : 0.000018s : 1: opt.transform.opt_after_jit_grad 0.01% : 0.000071s : 28: opt.transform.opt_b 0.00% : 0.000033s : 2: opt.transform.opt_trans_graph 0.00% : 0.000027s : 4: opt.transform.symbol_engine_opt 16.45% : 0.207025s : 1: opt_a 0.01% : 0.000093s : 1: opt_after_cconv 0.04% : 0.000487s : 1: opt_after_jit_grad 0.01% : 0.000173s : 1: opt_b 16.64% : 0.209442s : 1: optimize 0.00% : 0.000020s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000009s : 1: order_py_execute_after_rewriter 0.00% : 0.000022s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000006s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000007s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.01% : 0.000068s : 1: pre_auto_parallel 0.00% : 0.000050s : 1: py_interpret_to_execute 0.00% : 0.000014s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000017s : 1: remove_dup_value 15.63% : 0.196772s : 2: renormalize.infer 0.16% : 0.001961s : 2: renormalize.specialize 0.00% : 0.000007s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000008s : 1: rewriter_after_jit_bprop_graph 0.00% : 0.000035s : 1: rewriter_after_opt_a 0.01% : 0.000177s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000007s : 1: swap_dp_allreduce_reducescatter 0.01% : 0.000070s : 1: symbol_engine_optimizer 0.01% : 0.000063s : 1: tuple_transform 49.80% : 0.626766s : 1: type_inference TotalTime = 0.613496, [21] [bootstrap]: 0.00045601 [type_inference]: 0.582187 [event_method]: 6.605e-05 [auto_monad]: 0.00011995 [graph_reusing]: 8.76002e-06 [inline]: 3.01001e-06 [add_attr]: 0.00353451, [1] [add_attr_with_inline]: 0.00352558, [1] [Cycle 1]: 8.716e-05, [2] [tag_attr]: 4.127e-05 [meta_addattr_fg_expand]: 9.88998e-06 [parallel-infer-symbol]: 3.81999e-06 [pre_auto_parallel]: 5.737e-05 [insert-virtual-dataset]: 2.93e-06 [parallel-infer-symbol-second]: 7.39994e-07 [dataset_repeat_opt]: 2.04e-06 [pipeline_split]: 1.64998e-06 [optimize]: 0.0263265, [53] [py_interpret_to_execute]: 4.121e-05 [rewriter_before_opt_a]: 0.00016217 [opt_a]: 0.0240394, [3] [Cycle 1]: 0.00793414, [45] [expand_dump_flag]: 4.12e-06 [switch_simplify]: 7.617e-05 [loop_unroll]: 6.116e-05 [a_1]: 0.00139155 [with_stream_mark]: 2.591e-05 [recompute_prepare]: 2.113e-05 [updatestate_depend_eliminate]: 8.80999e-06 [updatestate_assign_eliminate]: 6.93998e-06 [updatestate_loads_eliminate]: 6.39999e-06 [parameter_eliminate]: 3.20998e-06 [a_2]: 0.00021045 [accelerated_algorithm]: 1.465e-05 [shard]: 1.96e-06 [meta_shard_fg_expand]: 3.32002e-06 [shard_inline]: 1.351e-05 [merge_send_recv]: 1.622e-05 [auto_parallel]: 1.173e-05 [parallel]: 2.547e-05 [flash_sp]: 9.56998e-06 [merge_comm]: 8.67998e-06 [allreduce_fusion]: 8.04997e-06 [matmul_add_comm_reduction]: 2.876e-05 [allreduce_slice_to_reducescatter]: 5.89993e-07 [virtual_shard_identity]: 1.583e-05 [virtual_dataset]: 1.363e-05 [get_grad_eliminate_]: 1.298e-05 [virtual_output]: 1.284e-05 [merge_forward]: 9.26002e-06 [cell_reuse_recompute_pass]: 1.03001e-06 [offload_activation]: 1.695e-05 [cell_reuse_handle_not_recompute_node_pass]: 2.545e-05 [merge_recompute_call_nodes]: 1.64998e-06 [before_grad]: 3.727e-05 [set_forward_comm_id_for_comm_node_pass]: 9.17001e-06 [meta_fg_expand]: 0.00190804 [flash_sp_send_recv_attached]: 3.91999e-06 [receive_attached]: 2.88e-06 [after_resolve]: 6.288e-05 [a_after_grad]: 8.079e-05 [renormalize]: 0.00289065 [add_forward_monad_depend]: 1.13e-05 [auto_monad_grad]: 6.21998e-06 [auto_monad_eliminator]: 5.56e-05 [cse]: 0.00019566 [a_3]: 0.00031158 [Cycle 2]: 0.015498, [45] [expand_dump_flag]: 2.31e-06 [switch_simplify]: 4.178e-05 [loop_unroll]: 3.908e-05 [a_1]: 0.00125268 [with_stream_mark]: 1.864e-05 [recompute_prepare]: 7.74002e-06 [updatestate_depend_eliminate]: 3.97998e-06 [updatestate_assign_eliminate]: 2.06998e-06 [updatestate_loads_eliminate]: 2.06e-06 [parameter_eliminate]: 2.36e-06 [a_2]: 4.539e-05 [accelerated_algorithm]: 4.53001e-06 [shard]: 1.75001e-06 [meta_shard_fg_expand]: 1.39998e-06 [shard_inline]: 4.37e-06 [merge_send_recv]: 5.57999e-06 [auto_parallel]: 7.21999e-06 [parallel]: 7.55e-06 [flash_sp]: 3.91999e-06 [merge_comm]: 2.96999e-06 [allreduce_fusion]: 2.36e-06 [matmul_add_comm_reduction]: 6.58e-06 [allreduce_slice_to_reducescatter]: 4.69998e-07 [virtual_shard_identity]: 5.42999e-06 [virtual_dataset]: 4.2e-06 [get_grad_eliminate_]: 4.25999e-06 [virtual_output]: 3.8e-06 [merge_forward]: 3.01999e-06 [cell_reuse_recompute_pass]: 9.89996e-07 [offload_activation]: 7.88999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.009e-05 [merge_recompute_call_nodes]: 1.05001e-06 [before_grad]: 6.94001e-06 [set_forward_comm_id_for_comm_node_pass]: 3.19001e-06 [meta_fg_expand]: 0.00012055 [flash_sp_send_recv_attached]: 1.30999e-06 [receive_attached]: 1.89e-06 [after_resolve]: 1.117e-05 [a_after_grad]: 5.84e-06 [renormalize]: 0.0134173 [add_forward_monad_depend]: 1.368e-05 [auto_monad_grad]: 3.71999e-06 [auto_monad_eliminator]: 2.521e-05 [cse]: 3.199e-05 [a_3]: 5.647e-05 [Cycle 3]: 0.00058824, [45] [expand_dump_flag]: 2.83e-06 [switch_simplify]: 6.78e-06 [loop_unroll]: 5.10999e-06 [a_1]: 7.825e-05 [with_stream_mark]: 1.685e-05 [recompute_prepare]: 4.57998e-06 [updatestate_depend_eliminate]: 3.45e-06 [updatestate_assign_eliminate]: 2.46e-06 [updatestate_loads_eliminate]: 2.21e-06 [parameter_eliminate]: 2.50002e-06 [a_2]: 4.518e-05 [accelerated_algorithm]: 4.48001e-06 [shard]: 3.31001e-06 [meta_shard_fg_expand]: 2.64999e-06 [shard_inline]: 4.13001e-06 [merge_send_recv]: 8.18999e-06 [auto_parallel]: 9.26002e-06 [parallel]: 9.89001e-06 [flash_sp]: 1.822e-05 [merge_comm]: 3.58999e-06 [allreduce_fusion]: 2.58998e-06 [matmul_add_comm_reduction]: 9.30001e-06 [allreduce_slice_to_reducescatter]: 8.70001e-07 [virtual_shard_identity]: 7.25e-06 [virtual_dataset]: 5.09e-06 [get_grad_eliminate_]: 4.23001e-06 [virtual_output]: 4.58001e-06 [merge_forward]: 4.22998e-06 [cell_reuse_recompute_pass]: 4.03999e-06 [offload_activation]: 9.04e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.758e-05 [merge_recompute_call_nodes]: 1.42999e-06 [before_grad]: 8.00999e-06 [set_forward_comm_id_for_comm_node_pass]: 3.00002e-06 [meta_fg_expand]: 1.89e-06 [flash_sp_send_recv_attached]: 2.01e-06 [receive_attached]: 2.37999e-06 [after_resolve]: 1.123e-05 [a_after_grad]: 5.82999e-06 [renormalize]: 9.00181e-08 [add_forward_monad_depend]: 1.72001e-06 [auto_monad_grad]: 1.16002e-06 [auto_monad_eliminator]: 5.30999e-06 [cse]: 1.203e-05 [a_3]: 2.088e-05 [py_interpret_to_execute_after_opt_a]: 1.835e-05 [slice_cell_reuse_recomputed_activation]: 1.99e-06 [rewriter_after_opt_a]: 3.739e-05 [convert_after_rewriter]: 6.23002e-06 [order_py_execute_after_rewriter]: 5.72999e-06 [mutable_eliminate]: 0.00075442 [opt_b]: 0.00016258, [1] [Cycle 1]: 0.00014128, [7] [b_1]: 7.971e-05 [b_2]: 6.01998e-06 [updatestate_depend_eliminate]: 4.47e-06 [updatestate_assign_eliminate]: 1.79e-06 [updatestate_loads_eliminate]: 1.50999e-06 [renormalize]: 6.69999e-07 [cse]: 1.231e-05 [optimize_parallel_all_gather_comm]: 1.5e-05 [overlap_param_gather]: 1.111e-05 [cconv]: 3.163e-05 [loop_unroll]: 0.00042652 [opt_after_cconv]: 8.576e-05, [1] [Cycle 1]: 8.012e-05, [7] [c_1]: 2.003e-05 [parameter_eliminate]: 2.51e-06 [updatestate_depend_eliminate]: 4.25e-06 [updatestate_assign_eliminate]: 1.87999e-06 [updatestate_loads_eliminate]: 1.51002e-06 [cse]: 1.492e-05 [renormalize]: 2.80008e-07 [remove_dup_value]: 1.467e-05 [tuple_transform]: 5.78e-05, [1] [Cycle 1]: 5.338e-05, [4] [d_1]: 2.84e-05 [none_parameter_eliminate]: 1.57001e-06 [renormalize]: 1.40019e-07 [switch_simplify]: 4.70999e-06 [partial_unused_args_eliminate]: 2.12001e-06 [add_recomputation]: 3.732e-05 [cse_after_recomputation]: 1.738e-05, [1] [Cycle 1]: 1.271e-05, [1] [cse]: 7.31001e-06 [environ_conv]: 4.74002e-06 [swap_dp_allreduce_reducescatter]: 4.43999e-06 [bias_add_comm_swap]: 2.96001e-06 [label_micro_interleaved_index]: 4.33001e-06 [label_fine_grained_interleaved_index]: 2.93003e-06 [merge_cast_opt]: 1.42999e-06 [slice_recompute_activation]: 2.14e-06 [micro_interleaved_order_control]: 2.64999e-06 [assign_add_opt]: 1.29e-06 [ForceFp32Comm]: 9.20001e-07 [remove_cast_before_assign_add]: 1.15001e-06 [full_micro_interleaved_order_control]: 3.06001e-06 [reorder_send_recv_between_fp_bp]: 3.08e-06 [comm_op_add_attrs]: 1.20999e-06 [add_comm_op_reuse_tag]: 1.07e-06 [interleave_split_concat_branches]: 1.44003e-06 [interleave_parallel_branches]: 1.05001e-06 [overlap_opt_shard_in_pipeline]: 4.60001e-06 [overlap_opt_shard_grad_in_pipeline]: 1.83997e-06 [control_data_broadcast_order]: 9.96998e-06 [grouped_pairwise_exchange_alltoall]: 1.68002e-06 [offloading_packed_experts]: 2.94999e-06 [overlap_recompute_and_grad_model_parallel]: 3.98999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.51002e-06 [overlap_recompute_allgather_and_fa_grad]: 1.37e-06 [overlap_recompute_comm]: 2.06998e-06 [overlap_grad_ring_attention]: 3.3e-06 [overlap_grad_flash_sp]: 1.906e-05 [begin_end_overlap_inline]: 5.69999e-07 [split_matmul_comm_elemetwise]: 2.19001e-06 [split_layernorm_comm]: 1.84e-06 [handle_group_info]: 1.22e-06 [symbol_engine_optimizer]: 6.685e-05, [1] [Cycle 1]: 6.231e-05, [6] [build]: 3.33998e-06 [elim_shapecalc]: 7.68001e-06 [elim_not_effective]: 1.031e-05 [opt_reshape]: 4.80999e-06 [fold_const_symbol]: 6.68998e-06 [renormalize]: 1.69995e-07 [detach_backward]: 2.84999e-06 [pipeline_parallel_scheduler]: 1.60999e-06 [auto_monad_reorder]: 1.458e-05 [get_jit_bprop_graph]: 2.05002e-06 [rewriter_after_jit_bprop_graph]: 4.05e-06 [opt_after_jit_grad]: 0.00048509 [validate]: 3.992e-05 Sums bootstrap : 0.000456s : 0.07% type_inference : 0.582187s : 95.66% event_method : 0.000066s : 0.01% auto_monad : 0.000120s : 0.02% graph_reusing : 0.000009s : 0.00% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000041s : 0.01% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000010s : 0.00% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000057s : 0.01% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000041s : 0.01% optimize.rewriter_before_opt_a : 0.000162s : 0.03% optimize.opt_a.expand_dump_flag : 0.000009s : 0.00% optimize.opt_a.switch_simplify : 0.000125s : 0.02% optimize.opt_a.loop_unroll : 0.000105s : 0.02% optimize.opt_a.a_1 : 0.002722s : 0.45% optimize.opt_a.with_stream_mark : 0.000061s : 0.01% optimize.opt_a.recompute_prepare : 0.000033s : 0.01% optimize.opt_a.updatestate_depend_eliminate : 0.000016s : 0.00% optimize.opt_a.updatestate_assign_eliminate : 0.000011s : 0.00% optimize.opt_a.updatestate_loads_eliminate : 0.000011s : 0.00% optimize.opt_a.parameter_eliminate : 0.000008s : 0.00% optimize.opt_a.a_2 : 0.000301s : 0.05% optimize.opt_a.accelerated_algorithm : 0.000024s : 0.00% optimize.opt_a.shard : 0.000007s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000007s : 0.00% optimize.opt_a.shard_inline : 0.000022s : 0.00% optimize.opt_a.merge_send_recv : 0.000030s : 0.00% optimize.opt_a.auto_parallel : 0.000028s : 0.00% optimize.opt_a.parallel : 0.000043s : 0.01% optimize.opt_a.flash_sp : 0.000032s : 0.01% optimize.opt_a.merge_comm : 0.000015s : 0.00% optimize.opt_a.allreduce_fusion : 0.000013s : 0.00% optimize.opt_a.matmul_add_comm_reduction : 0.000045s : 0.01% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000029s : 0.00% optimize.opt_a.virtual_dataset : 0.000023s : 0.00% optimize.opt_a.get_grad_eliminate_ : 0.000021s : 0.00% optimize.opt_a.virtual_output : 0.000021s : 0.00% optimize.opt_a.merge_forward : 0.000017s : 0.00% optimize.opt_a.cell_reuse_recompute_pass : 0.000006s : 0.00% optimize.opt_a.offload_activation : 0.000034s : 0.01% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000053s : 0.01% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.00% optimize.opt_a.before_grad : 0.000052s : 0.01% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000015s : 0.00% optimize.opt_a.meta_fg_expand : 0.002030s : 0.33% optimize.opt_a.flash_sp_send_recv_attached : 0.000007s : 0.00% optimize.opt_a.receive_attached : 0.000007s : 0.00% optimize.opt_a.after_resolve : 0.000085s : 0.01% optimize.opt_a.a_after_grad : 0.000092s : 0.02% optimize.opt_a.renormalize : 0.016308s : 2.68% optimize.opt_a.add_forward_monad_depend : 0.000027s : 0.00% optimize.opt_a.auto_monad_grad : 0.000011s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000086s : 0.01% optimize.opt_a.cse : 0.000240s : 0.04% optimize.opt_a.a_3 : 0.000389s : 0.06% optimize.py_interpret_to_execute_after_opt_a : 0.000018s : 0.00% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000037s : 0.01% optimize.convert_after_rewriter : 0.000006s : 0.00% optimize.order_py_execute_after_rewriter : 0.000006s : 0.00% optimize.mutable_eliminate : 0.000754s : 0.12% optimize.opt_b.b_1 : 0.000080s : 0.01% optimize.opt_b.b_2 : 0.000006s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000004s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000012s : 0.00% optimize.optimize_parallel_all_gather_comm : 0.000015s : 0.00% optimize.overlap_param_gather : 0.000011s : 0.00% optimize.cconv : 0.000032s : 0.01% optimize.loop_unroll : 0.000427s : 0.07% optimize.opt_after_cconv.c_1 : 0.000020s : 0.00% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000004s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000015s : 0.00% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000015s : 0.00% optimize.tuple_transform.d_1 : 0.000028s : 0.00% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000005s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000037s : 0.01% optimize.cse_after_recomputation.cse : 0.000007s : 0.00% optimize.environ_conv : 0.000005s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000004s : 0.00% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000004s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000003s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000005s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000010s : 0.00% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000003s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000004s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000002s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000003s : 0.00% optimize.overlap_grad_flash_sp : 0.000019s : 0.00% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000003s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000008s : 0.00% optimize.symbol_engine_optimizer.elim_not_effective : 0.000010s : 0.00% optimize.symbol_engine_optimizer.opt_reshape : 0.000005s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000007s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000015s : 0.00% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.00% opt_after_jit_grad : 0.000485s : 0.08% validate : 0.000040s : 0.01% Time group info: ------[substitution.] 0.000773 143 7.76% : 0.000060s : 5: substitution.arithmetic_simplify 0.24% : 0.000002s : 1: substitution.elim_not_effective 0.65% : 0.000005s : 6: substitution.float_depend_g_call 0.48% : 0.000004s : 2: substitution.float_tuple_getitem_switch 0.14% : 0.000001s : 1: substitution.fold_const_symbol 0.69% : 0.000005s : 2: substitution.graph_param_transform 0.37% : 0.000003s : 2: substitution.incorporate_call 0.25% : 0.000002s : 2: substitution.incorporate_call_switch 60.99% : 0.000471s : 16: substitution.inline 2.30% : 0.000018s : 2: substitution.inline_without_move 1.06% : 0.000008s : 10: substitution.j_node_and_user_rematch 1.41% : 0.000011s : 7: substitution.minmaximum_grad 0.81% : 0.000006s : 6: substitution.partial_eliminate 1.39% : 0.000011s : 10: substitution.remove_not_recompute_node 3.72% : 0.000029s : 10: substitution.replace_applicator 1.75% : 0.000013s : 10: substitution.replace_old_param 0.35% : 0.000003s : 1: substitution.set_cell_output_no_recompute 2.89% : 0.000022s : 7: substitution.tuple_list_convert_item_index_to_positive 1.34% : 0.000010s : 7: substitution.tuple_list_get_item_const_eliminator 1.90% : 0.000015s : 7: substitution.tuple_list_get_item_depend_reorder 7.69% : 0.000059s : 22: substitution.tuple_list_get_item_eliminator 1.82% : 0.000014s : 7: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.582078 2 2.23% : 0.012972s : 1: type_inference.infer 97.77% : 0.569106s : 1: type_inference.specialize ------[replace.] 0.000231 30 2.72% : 0.000006s : 1: replace.arithmetic_simplify 57.11% : 0.000132s : 16: replace.inline 40.18% : 0.000093s : 13: replace.tuple_list_get_item_eliminator ------[match.] 0.000526 30 5.85% : 0.000031s : 1: match.arithmetic_simplify 87.79% : 0.000462s : 16: match.inline 6.36% : 0.000033s : 13: match.tuple_list_get_item_eliminator ------[predicate.] 0.000550 3865 1.12% : 0.000006s : 48: predicate.accumulaten_eliminater 0.18% : 0.000001s : 2: predicate.ad_related_special_op_eliminate 0.38% : 0.000002s : 17: predicate.addn_check_dump 1.16% : 0.000006s : 48: predicate.addn_zero_filter 1.06% : 0.000006s : 48: predicate.adjust_all_reduce_mul_add 2.01% : 0.000011s : 66: predicate.arithmetic_simplify 1.27% : 0.000007s : 49: predicate.cast_eliminate 1.19% : 0.000007s : 48: predicate.check_bprop_eliminate 0.38% : 0.000002s : 17: predicate.compare_switch_simplify 0.03% : 0.000000s : 2: predicate.const_output_eliminate 0.42% : 0.000002s : 17: predicate.depend_value_elim 1.19% : 0.000007s : 49: predicate.dict_get_item_const_eliminator 1.28% : 0.000007s : 49: predicate.dict_get_item_eliminator 1.20% : 0.000007s : 49: predicate.dict_set_item_eliminator 0.25% : 0.000001s : 4: predicate.dumpgradient_eliminate 0.08% : 0.000000s : 2: predicate.elim_not_effective 0.09% : 0.000000s : 2: predicate.elim_shapecalc_of_broadcastargs 1.16% : 0.000006s : 51: predicate.environ_add_const_eliminate 1.11% : 0.000006s : 51: predicate.environ_get_add_eliminate 1.21% : 0.000007s : 51: predicate.environ_get_depend_swap 1.71% : 0.000009s : 68: predicate.environ_get_eliminate 1.12% : 0.000006s : 51: predicate.environ_get_set_eliminate 1.89% : 0.000010s : 78: predicate.exchange_switch_depend_value 2.58% : 0.000014s : 78: predicate.float_depend_g_call 0.38% : 0.000002s : 17: predicate.float_environ_get_switch 0.48% : 0.000003s : 19: predicate.float_tuple_getitem_switch 0.03% : 0.000000s : 2: predicate.fold_const_symbol 0.46% : 0.000003s : 17: predicate.get_grad_eliminate 0.08% : 0.000000s : 2: predicate.graph_param_transform 0.40% : 0.000002s : 17: predicate.incorporate_call 0.36% : 0.000002s : 17: predicate.incorporate_call_switch 5.60% : 0.000031s : 164: predicate.inline 1.30% : 0.000007s : 40: predicate.inline_without_move 0.23% : 0.000001s : 17: predicate.j_node_and_user_rematch 0.55% : 0.000003s : 17: predicate.less_batch_normalization 1.59% : 0.000009s : 66: predicate.list_to_tuple_eliminator_ 2.51% : 0.000014s : 114: predicate.load_eliminater 0.29% : 0.000002s : 2: predicate.loop_unroll_after_grad 2.82% : 0.000016s : 121: predicate.loop_unroll_before_grad 1.33% : 0.000007s : 53: predicate.make_slice_get_slice_eliminator 0.40% : 0.000002s : 17: predicate.merge_addn 1.17% : 0.000006s : 48: predicate.micro_step_allgather_replace 1.28% : 0.000007s : 48: predicate.mini_step_allgather_replace 1.08% : 0.000006s : 49: predicate.minmaximum_grad 0.29% : 0.000002s : 2: predicate.mutable_eliminate 0.10% : 0.000001s : 2: predicate.opt_reshape 0.16% : 0.000001s : 2: predicate.parallel_virtual_node 2.33% : 0.000013s : 78: predicate.partial_defer_inline 1.69% : 0.000009s : 64: predicate.partial_eliminate 1.12% : 0.000006s : 48: predicate.print_const_string_wrapper 0.43% : 0.000002s : 17: predicate.reduce_all_const_elim 1.58% : 0.000009s : 49: predicate.reduce_eliminate 2.64% : 0.000015s : 114: predicate.redundant_stop_gradient_eliminater 0.28% : 0.000002s : 17: predicate.remove_not_recompute_node 2.17% : 0.000012s : 110: predicate.replace_applicator 0.67% : 0.000004s : 40: predicate.replace_old_param 0.07% : 0.000000s : 2: predicate.reset_defer_inline 1.23% : 0.000007s : 49: predicate.reshape_eliminate 1.23% : 0.000007s : 48: predicate.row_tensor_add_zeros_like 0.11% : 0.000001s : 2: predicate.row_tensor_eliminate 1.42% : 0.000008s : 48: predicate.same_eliminate 0.33% : 0.000002s : 17: predicate.set_cell_output_no_recompute 0.57% : 0.000003s : 17: predicate.shard_identity_eliminate 0.19% : 0.000001s : 4: predicate.special_op_eliminate 0.48% : 0.000003s : 17: predicate.specialize_transform 1.49% : 0.000008s : 48: predicate.split_environ_get_set_with_tuple_value 1.21% : 0.000007s : 40: predicate.stack_unstack_eliminate 0.06% : 0.000000s : 2: predicate.switch_call_monad_eliminater 2.08% : 0.000011s : 78: predicate.switch_defer_inline 3.19% : 0.000018s : 126: predicate.switch_layer_defer_inline 5.50% : 0.000030s : 218: predicate.switch_simplify 1.15% : 0.000006s : 49: predicate.tile_eliminate 1.14% : 0.000006s : 49: predicate.transpose_eliminate 1.41% : 0.000008s : 53: predicate.tuple_list_convert_item_index_to_positive 1.48% : 0.000008s : 53: predicate.tuple_list_get_item_const_eliminator 1.25% : 0.000007s : 53: predicate.tuple_list_get_item_depend_reorder 2.96% : 0.000016s : 83: predicate.tuple_list_get_item_eliminator 1.33% : 0.000007s : 53: predicate.tuple_list_get_set_item_eliminator 1.97% : 0.000011s : 70: predicate.tuple_list_set_item_eliminator 1.60% : 0.000009s : 66: predicate.tuple_to_list_eliminator_ 2.53% : 0.000014s : 114: predicate.updatestate_pure_node_eliminater 2.94% : 0.000016s : 131: predicate.updatestate_useless_node_eliminater 0.07% : 0.000000s : 2: predicate.value_based_eliminate 0.52% : 0.000003s : 17: predicate.virtual_dataset_eliminate 0.49% : 0.000003s : 17: predicate.virtual_output_eliminate 0.05% : 0.000000s : 2: predicate.virtual_view_grad_eliminate 0.10% : 0.000001s : 2: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002031 37 57.43% : 0.001166s : 17: func_graph_cloner_run.FuncGraphClonerGraph 42.57% : 0.000864s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.663641 233 0.00% : 0.000004s : 1: ForceFp32Comm 0.53% : 0.003540s : 1: add_attr 0.53% : 0.003529s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.01% : 0.000041s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.02% : 0.000127s : 1: auto_monad 0.00% : 0.000018s : 1: auto_monad_reorder 0.00% : 0.000005s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 0.07% : 0.000490s : 1: bootstrap 0.01% : 0.000035s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000013s : 1: control_data_broadcast_order 0.00% : 0.000009s : 1: convert_after_rewriter 0.00% : 0.000020s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000008s : 1: environ_conv 0.01% : 0.000074s : 1: event_method 0.00% : 0.000006s : 1: full_micro_interleaved_order_control 0.00% : 0.000006s : 1: get_jit_bprop_graph 0.00% : 0.000013s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000007s : 1: label_fine_grained_interleaved_index 0.00% : 0.000007s : 1: label_micro_interleaved_index 0.07% : 0.000434s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000006s : 1: micro_interleaved_order_control 0.11% : 0.000762s : 1: mutable_eliminate 0.00% : 0.000006s : 1: offloading_packed_experts 0.00% : 0.000010s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000011s : 1: opt.transform.mutable_eliminate 0.60% : 0.003954s : 117: opt.transform.opt_a 0.00% : 0.000019s : 1: opt.transform.opt_after_cconv 0.00% : 0.000018s : 1: opt.transform.opt_after_jit_grad 0.01% : 0.000059s : 28: opt.transform.opt_b 0.00% : 0.000031s : 2: opt.transform.opt_trans_graph 0.00% : 0.000026s : 4: opt.transform.symbol_engine_opt 3.62% : 0.024043s : 1: opt_a 0.01% : 0.000090s : 1: opt_after_cconv 0.07% : 0.000494s : 1: opt_after_jit_grad 0.03% : 0.000166s : 1: opt_b 3.97% : 0.026332s : 1: optimize 0.00% : 0.000019s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000010s : 1: order_py_execute_after_rewriter 0.00% : 0.000022s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000006s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000008s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000015s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000007s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.01% : 0.000062s : 1: pre_auto_parallel 0.01% : 0.000045s : 1: py_interpret_to_execute 0.00% : 0.000021s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000018s : 1: remove_dup_value 0.29% : 0.001892s : 2: renormalize.infer 2.17% : 0.014394s : 2: renormalize.specialize 0.00% : 0.000007s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000008s : 1: rewriter_after_jit_bprop_graph 0.01% : 0.000041s : 1: rewriter_after_opt_a 0.03% : 0.000167s : 1: rewriter_before_opt_a 0.00% : 0.000006s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000007s : 1: swap_dp_allreduce_reducescatter 0.01% : 0.000070s : 1: symbol_engine_optimizer 0.01% : 0.000061s : 1: tuple_transform 87.73% : 0.582216s : 1: type_inference TotalTime = 0.0991993, [21] [bootstrap]: 0.00045591 [type_inference]: 0.0882606 [event_method]: 4.179e-05 [auto_monad]: 7.309e-05 [graph_reusing]: 7.03e-06 [inline]: 3.04001e-06 [add_attr]: 0.00422227, [1] [add_attr_with_inline]: 0.0042107, [1] [Cycle 1]: 7.502e-05, [2] [tag_attr]: 3.241e-05 [meta_addattr_fg_expand]: 7.08e-06 [parallel-infer-symbol]: 3.86001e-06 [pre_auto_parallel]: 4.578e-05 [insert-virtual-dataset]: 2.71e-06 [parallel-infer-symbol-second]: 7.2e-07 [dataset_repeat_opt]: 2.16998e-06 [pipeline_split]: 1.77999e-06 [optimize]: 0.00537142, [53] [py_interpret_to_execute]: 3.135e-05 [rewriter_before_opt_a]: 0.00011401 [opt_a]: 0.00321702, [2] [Cycle 1]: 0.00256655, [45] [expand_dump_flag]: 2.93e-06 [switch_simplify]: 5.136e-05 [loop_unroll]: 4.126e-05 [a_1]: 0.00093448 [with_stream_mark]: 1.506e-05 [recompute_prepare]: 8.28999e-06 [updatestate_depend_eliminate]: 3.87998e-06 [updatestate_assign_eliminate]: 3.7e-06 [updatestate_loads_eliminate]: 3.23e-06 [parameter_eliminate]: 1.91998e-06 [a_2]: 8.37e-05 [accelerated_algorithm]: 7.12002e-06 [shard]: 1.93002e-06 [meta_shard_fg_expand]: 1.81e-06 [shard_inline]: 6.72002e-06 [merge_send_recv]: 9.44e-06 [auto_parallel]: 6.91001e-06 [parallel]: 1.823e-05 [flash_sp]: 7.85e-06 [merge_comm]: 3.81999e-06 [allreduce_fusion]: 3.75998e-06 [matmul_add_comm_reduction]: 9.34e-06 [allreduce_slice_to_reducescatter]: 6.39993e-07 [virtual_shard_identity]: 7.65e-06 [virtual_dataset]: 6.22001e-06 [get_grad_eliminate_]: 5.89e-06 [virtual_output]: 5.85002e-06 [merge_forward]: 4.02002e-06 [cell_reuse_recompute_pass]: 1.55001e-06 [offload_activation]: 1.026e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.205e-05 [merge_recompute_call_nodes]: 1.90001e-06 [before_grad]: 1.214e-05 [set_forward_comm_id_for_comm_node_pass]: 3.86001e-06 [meta_fg_expand]: 2.59001e-06 [flash_sp_send_recv_attached]: 3.13e-06 [receive_attached]: 2.05002e-06 [after_resolve]: 1.153e-05 [a_after_grad]: 9.54e-06 [renormalize]: 0.00082935 [add_forward_monad_depend]: 5.72999e-06 [auto_monad_grad]: 2.57001e-06 [auto_monad_eliminator]: 1.74e-05 [cse]: 3.037e-05 [a_3]: 4.529e-05 [Cycle 2]: 0.00064024, [45] [expand_dump_flag]: 1.45999e-06 [switch_simplify]: 7.28e-06 [loop_unroll]: 5.96e-06 [a_1]: 0.00013791 [with_stream_mark]: 1.099e-05 [recompute_prepare]: 6.46999e-06 [updatestate_depend_eliminate]: 3.03998e-06 [updatestate_assign_eliminate]: 2.34999e-06 [updatestate_loads_eliminate]: 2.93998e-06 [parameter_eliminate]: 1.35999e-06 [a_2]: 7.402e-05 [accelerated_algorithm]: 6.43e-06 [shard]: 1.30001e-06 [meta_shard_fg_expand]: 1.44e-06 [shard_inline]: 6.36998e-06 [merge_send_recv]: 5.47999e-06 [auto_parallel]: 6.96999e-06 [parallel]: 5.96998e-06 [flash_sp]: 3.86999e-06 [merge_comm]: 3.03e-06 [allreduce_fusion]: 2.79001e-06 [matmul_add_comm_reduction]: 6.26e-06 [allreduce_slice_to_reducescatter]: 5.10016e-07 [virtual_shard_identity]: 6.93e-06 [virtual_dataset]: 5.27999e-06 [get_grad_eliminate_]: 6.13998e-06 [virtual_output]: 5.35001e-06 [merge_forward]: 3.33e-06 [cell_reuse_recompute_pass]: 1.84e-06 [offload_activation]: 7.70998e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.058e-05 [merge_recompute_call_nodes]: 8.30012e-07 [before_grad]: 8.67e-06 [set_forward_comm_id_for_comm_node_pass]: 3.34001e-06 [meta_fg_expand]: 2.16e-06 [flash_sp_send_recv_attached]: 9.70002e-07 [receive_attached]: 9.40025e-07 [after_resolve]: 9.52999e-06 [a_after_grad]: 8.19998e-06 [renormalize]: 8.00064e-08 [add_forward_monad_depend]: 1.10999e-06 [auto_monad_grad]: 8.00006e-07 [auto_monad_eliminator]: 6.91999e-06 [cse]: 1.332e-05 [a_3]: 3.385e-05 [py_interpret_to_execute_after_opt_a]: 9.02999e-06 [slice_cell_reuse_recomputed_activation]: 2.51e-06 [rewriter_after_opt_a]: 3.452e-05 [convert_after_rewriter]: 7.06001e-06 [order_py_execute_after_rewriter]: 6.09001e-06 [mutable_eliminate]: 0.00058741 [opt_b]: 0.00019058, [1] [Cycle 1]: 0.00018372, [7] [b_1]: 0.00011239 [b_2]: 7.15e-06 [updatestate_depend_eliminate]: 5.13002e-06 [updatestate_assign_eliminate]: 2.46e-06 [updatestate_loads_eliminate]: 2.27001e-06 [renormalize]: 3.4002e-07 [cse]: 1.702e-05 [optimize_parallel_all_gather_comm]: 1.602e-05 [overlap_param_gather]: 1.75001e-06 [cconv]: 2.54e-05 [loop_unroll]: 0.00045875 [opt_after_cconv]: 0.00011085, [1] [Cycle 1]: 9.239e-05, [7] [c_1]: 2.795e-05 [parameter_eliminate]: 3.03998e-06 [updatestate_depend_eliminate]: 6.10002e-06 [updatestate_assign_eliminate]: 2.29001e-06 [updatestate_loads_eliminate]: 2.20002e-06 [cse]: 1.785e-05 [renormalize]: 3.59985e-07 [remove_dup_value]: 1.393e-05 [tuple_transform]: 7.495e-05, [1] [Cycle 1]: 7.023e-05, [4] [d_1]: 4.317e-05 [none_parameter_eliminate]: 1.57001e-06 [renormalize]: 2.40019e-07 [switch_simplify]: 6.44001e-06 [partial_unused_args_eliminate]: 1.82001e-06 [add_recomputation]: 4.945e-05 [cse_after_recomputation]: 2.163e-05, [1] [Cycle 1]: 1.629e-05, [1] [cse]: 1.097e-05 [environ_conv]: 5.25001e-06 [swap_dp_allreduce_reducescatter]: 5.44e-06 [bias_add_comm_swap]: 2.31998e-06 [label_micro_interleaved_index]: 4.16001e-06 [label_fine_grained_interleaved_index]: 2.64001e-06 [merge_cast_opt]: 1.24e-06 [slice_recompute_activation]: 2.24001e-06 [micro_interleaved_order_control]: 2.49001e-06 [assign_add_opt]: 1.59998e-06 [ForceFp32Comm]: 8.80013e-07 [remove_cast_before_assign_add]: 1.07998e-06 [full_micro_interleaved_order_control]: 2.26e-06 [reorder_send_recv_between_fp_bp]: 3.08998e-06 [comm_op_add_attrs]: 1.13001e-06 [add_comm_op_reuse_tag]: 1.07998e-06 [interleave_split_concat_branches]: 1.38002e-06 [interleave_parallel_branches]: 1.04e-06 [overlap_opt_shard_in_pipeline]: 1.37e-06 [overlap_opt_shard_grad_in_pipeline]: 1.79e-06 [control_data_broadcast_order]: 1.311e-05 [grouped_pairwise_exchange_alltoall]: 1.93002e-06 [offloading_packed_experts]: 4.1e-06 [overlap_recompute_and_grad_model_parallel]: 4.80001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.19998e-06 [overlap_recompute_allgather_and_fa_grad]: 1.37e-06 [overlap_recompute_comm]: 2.32001e-06 [overlap_grad_ring_attention]: 4.42e-06 [overlap_grad_flash_sp]: 1.878e-05 [begin_end_overlap_inline]: 8.29983e-07 [split_matmul_comm_elemetwise]: 2.19999e-06 [split_layernorm_comm]: 1.92999e-06 [handle_group_info]: 1.66e-06 [symbol_engine_optimizer]: 7.337e-05, [1] [Cycle 1]: 6.876e-05, [6] [build]: 2.79001e-06 [elim_shapecalc]: 9.32999e-06 [elim_not_effective]: 1.201e-05 [opt_reshape]: 6.17001e-06 [fold_const_symbol]: 9.91e-06 [renormalize]: 2.00002e-07 [detach_backward]: 2.16e-06 [pipeline_parallel_scheduler]: 2.12001e-06 [auto_monad_reorder]: 1.62e-05 [get_jit_bprop_graph]: 9.90025e-07 [rewriter_after_jit_bprop_graph]: 4.07e-06 [opt_after_jit_grad]: 0.00047283 [validate]: 3.711e-05 Sums bootstrap : 0.000456s : 0.49% type_inference : 0.088261s : 93.95% event_method : 0.000042s : 0.04% auto_monad : 0.000073s : 0.08% graph_reusing : 0.000007s : 0.01% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000032s : 0.03% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.01% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000046s : 0.05% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000031s : 0.03% optimize.rewriter_before_opt_a : 0.000114s : 0.12% optimize.opt_a.expand_dump_flag : 0.000004s : 0.00% optimize.opt_a.switch_simplify : 0.000059s : 0.06% optimize.opt_a.loop_unroll : 0.000047s : 0.05% optimize.opt_a.a_1 : 0.001072s : 1.14% optimize.opt_a.with_stream_mark : 0.000026s : 0.03% optimize.opt_a.recompute_prepare : 0.000015s : 0.02% optimize.opt_a.updatestate_depend_eliminate : 0.000007s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.01% optimize.opt_a.parameter_eliminate : 0.000003s : 0.00% optimize.opt_a.a_2 : 0.000158s : 0.17% optimize.opt_a.accelerated_algorithm : 0.000014s : 0.01% optimize.opt_a.shard : 0.000003s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000003s : 0.00% optimize.opt_a.shard_inline : 0.000013s : 0.01% optimize.opt_a.merge_send_recv : 0.000015s : 0.02% optimize.opt_a.auto_parallel : 0.000014s : 0.01% optimize.opt_a.parallel : 0.000024s : 0.03% optimize.opt_a.flash_sp : 0.000012s : 0.01% optimize.opt_a.merge_comm : 0.000007s : 0.01% optimize.opt_a.allreduce_fusion : 0.000007s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000016s : 0.02% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000015s : 0.02% optimize.opt_a.virtual_dataset : 0.000012s : 0.01% optimize.opt_a.get_grad_eliminate_ : 0.000012s : 0.01% optimize.opt_a.virtual_output : 0.000011s : 0.01% optimize.opt_a.merge_forward : 0.000007s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000003s : 0.00% optimize.opt_a.offload_activation : 0.000018s : 0.02% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000023s : 0.02% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.00% optimize.opt_a.before_grad : 0.000021s : 0.02% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000007s : 0.01% optimize.opt_a.meta_fg_expand : 0.000005s : 0.01% optimize.opt_a.flash_sp_send_recv_attached : 0.000004s : 0.00% optimize.opt_a.receive_attached : 0.000003s : 0.00% optimize.opt_a.after_resolve : 0.000021s : 0.02% optimize.opt_a.a_after_grad : 0.000018s : 0.02% optimize.opt_a.renormalize : 0.000829s : 0.88% optimize.opt_a.add_forward_monad_depend : 0.000007s : 0.01% optimize.opt_a.auto_monad_grad : 0.000003s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000024s : 0.03% optimize.opt_a.cse : 0.000044s : 0.05% optimize.opt_a.a_3 : 0.000079s : 0.08% optimize.py_interpret_to_execute_after_opt_a : 0.000009s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000003s : 0.00% optimize.rewriter_after_opt_a : 0.000035s : 0.04% optimize.convert_after_rewriter : 0.000007s : 0.01% optimize.order_py_execute_after_rewriter : 0.000006s : 0.01% optimize.mutable_eliminate : 0.000587s : 0.63% optimize.opt_b.b_1 : 0.000112s : 0.12% optimize.opt_b.b_2 : 0.000007s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000005s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000017s : 0.02% optimize.optimize_parallel_all_gather_comm : 0.000016s : 0.02% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000025s : 0.03% optimize.loop_unroll : 0.000459s : 0.49% optimize.opt_after_cconv.c_1 : 0.000028s : 0.03% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000018s : 0.02% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000014s : 0.01% optimize.tuple_transform.d_1 : 0.000043s : 0.05% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000006s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000049s : 0.05% optimize.cse_after_recomputation.cse : 0.000011s : 0.01% optimize.environ_conv : 0.000005s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.01% optimize.bias_add_comm_swap : 0.000002s : 0.00% optimize.label_micro_interleaved_index : 0.000004s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000002s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000013s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000019s : 0.02% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000002s : 0.00% optimize.symbol_engine_optimizer.build : 0.000003s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000009s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000012s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000006s : 0.01% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000016s : 0.02% get_jit_bprop_graph : 0.000001s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.00% opt_after_jit_grad : 0.000473s : 0.50% validate : 0.000037s : 0.04% Time group info: ------[substitution.] 0.000289 40 12.35% : 0.000036s : 8: substitution.arithmetic_simplify 0.73% : 0.000002s : 2: substitution.elim_not_effective 0.45% : 0.000001s : 2: substitution.fold_const_symbol 2.11% : 0.000006s : 4: substitution.graph_param_transform 73.19% : 0.000212s : 6: substitution.inline 1.17% : 0.000003s : 4: substitution.j_node_and_user_rematch 1.74% : 0.000005s : 4: substitution.remove_not_recompute_node 1.56% : 0.000005s : 4: substitution.replace_old_param 6.70% : 0.000019s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.088192 2 98.92% : 0.087241s : 1: type_inference.infer 1.08% : 0.000950s : 1: type_inference.specialize ------[replace.] 0.000087 12 61.31% : 0.000053s : 6: replace.inline 38.69% : 0.000034s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000225 12 92.45% : 0.000208s : 6: match.inline 7.55% : 0.000017s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000237 1534 1.03% : 0.000002s : 17: predicate.accumulaten_eliminater 0.60% : 0.000001s : 4: predicate.ad_related_special_op_eliminate 0.48% : 0.000001s : 8: predicate.addn_check_dump 1.39% : 0.000003s : 17: predicate.addn_zero_filter 0.90% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.46% : 0.000006s : 25: predicate.arithmetic_simplify 0.99% : 0.000002s : 17: predicate.cast_eliminate 0.47% : 0.000001s : 8: predicate.check_bprop_eliminate 0.46% : 0.000001s : 8: predicate.compare_switch_simplify 0.15% : 0.000000s : 4: predicate.const_output_eliminate 0.52% : 0.000001s : 8: predicate.depend_value_elim 1.08% : 0.000003s : 17: predicate.dict_get_item_const_eliminator 1.16% : 0.000003s : 17: predicate.dict_get_item_eliminator 1.16% : 0.000003s : 17: predicate.dict_set_item_eliminator 0.81% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.16% : 0.000000s : 4: predicate.elim_not_effective 0.30% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.14% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.25% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.15% : 0.000003s : 21: predicate.environ_get_depend_swap 1.81% : 0.000004s : 29: predicate.environ_get_eliminate 1.12% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.74% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.66% : 0.000006s : 29: predicate.float_depend_g_call 0.42% : 0.000001s : 8: predicate.float_environ_get_switch 0.63% : 0.000001s : 12: predicate.float_tuple_getitem_switch 0.17% : 0.000000s : 4: predicate.fold_const_symbol 0.75% : 0.000002s : 8: predicate.get_grad_eliminate 0.16% : 0.000000s : 4: predicate.graph_param_transform 0.49% : 0.000001s : 8: predicate.incorporate_call 0.41% : 0.000001s : 8: predicate.incorporate_call_switch 6.04% : 0.000014s : 70: predicate.inline 0.68% : 0.000002s : 8: predicate.inline_without_move 0.26% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.64% : 0.000002s : 8: predicate.less_batch_normalization 1.84% : 0.000004s : 31: predicate.list_to_tuple_eliminator_ 2.60% : 0.000006s : 48: predicate.load_eliminater 0.87% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.75% : 0.000007s : 47: predicate.loop_unroll_before_grad 1.80% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.44% : 0.000001s : 8: predicate.merge_addn 0.46% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.49% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.95% : 0.000002s : 17: predicate.minmaximum_grad 0.80% : 0.000002s : 4: predicate.mutable_eliminate 0.28% : 0.000001s : 4: predicate.opt_reshape 0.43% : 0.000001s : 4: predicate.parallel_virtual_node 2.20% : 0.000005s : 29: predicate.partial_defer_inline 1.70% : 0.000004s : 27: predicate.partial_eliminate 0.97% : 0.000002s : 17: predicate.print_const_string_wrapper 0.54% : 0.000001s : 8: predicate.reduce_all_const_elim 1.42% : 0.000003s : 17: predicate.reduce_eliminate 2.70% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.42% : 0.000001s : 8: predicate.remove_not_recompute_node 1.46% : 0.000003s : 31: predicate.replace_applicator 0.37% : 0.000001s : 8: predicate.replace_old_param 0.22% : 0.000001s : 4: predicate.reset_defer_inline 0.98% : 0.000002s : 17: predicate.reshape_eliminate 0.54% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.32% : 0.000001s : 4: predicate.row_tensor_eliminate 0.76% : 0.000002s : 8: predicate.same_eliminate 0.39% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.56% : 0.000001s : 8: predicate.shard_identity_eliminate 0.50% : 0.000001s : 8: predicate.special_op_eliminate 0.64% : 0.000002s : 8: predicate.specialize_transform 0.70% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.69% : 0.000002s : 8: predicate.stack_unstack_eliminate 0.33% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.83% : 0.000004s : 29: predicate.switch_defer_inline 2.25% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.46% : 0.000013s : 88: predicate.switch_simplify 1.03% : 0.000002s : 17: predicate.tile_eliminate 1.01% : 0.000002s : 17: predicate.transpose_eliminate 1.50% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.47% : 0.000003s : 25: predicate.tuple_list_get_item_const_eliminator 1.25% : 0.000003s : 25: predicate.tuple_list_get_item_depend_reorder 3.14% : 0.000007s : 39: predicate.tuple_list_get_item_eliminator 1.43% : 0.000003s : 25: predicate.tuple_list_get_set_item_eliminator 2.13% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 2.01% : 0.000005s : 31: predicate.tuple_to_list_eliminator_ 2.60% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.23% : 0.000008s : 56: predicate.updatestate_useless_node_eliminater 0.29% : 0.000001s : 4: predicate.value_based_eliminate 0.51% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.49% : 0.000001s : 8: predicate.virtual_output_eliminate 0.23% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.43% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000752 15 60.13% : 0.000452s : 7: func_graph_cloner_run.FuncGraphClonerGraph 39.87% : 0.000300s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.111247 192 0.00% : 0.000004s : 1: ForceFp32Comm 3.80% : 0.004229s : 1: add_attr 3.79% : 0.004215s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.05% : 0.000054s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.07% : 0.000079s : 1: auto_monad 0.02% : 0.000020s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000005s : 1: bias_add_comm_swap 0.44% : 0.000492s : 1: bootstrap 0.03% : 0.000029s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.01% : 0.000016s : 1: control_data_broadcast_order 0.01% : 0.000010s : 1: convert_after_rewriter 0.02% : 0.000024s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.01% : 0.000009s : 1: environ_conv 0.04% : 0.000049s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000004s : 1: get_jit_bprop_graph 0.01% : 0.000011s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.01% : 0.000006s : 1: inline 0.01% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000005s : 1: label_fine_grained_interleaved_index 0.01% : 0.000007s : 1: label_micro_interleaved_index 0.42% : 0.000467s : 1: loop_unroll 0.00% : 0.000005s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.54% : 0.000597s : 1: mutable_eliminate 0.01% : 0.000008s : 1: offloading_packed_experts 0.01% : 0.000014s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000014s : 1: opt.transform.mutable_eliminate 1.35% : 0.001504s : 78: opt.transform.opt_a 0.02% : 0.000027s : 1: opt.transform.opt_after_cconv 0.02% : 0.000023s : 1: opt.transform.opt_after_jit_grad 0.08% : 0.000093s : 28: opt.transform.opt_b 0.04% : 0.000047s : 2: opt.transform.opt_trans_graph 0.03% : 0.000034s : 4: opt.transform.symbol_engine_opt 2.89% : 0.003220s : 1: opt_a 0.10% : 0.000114s : 1: opt_after_cconv 0.43% : 0.000483s : 1: opt_after_jit_grad 0.17% : 0.000194s : 1: opt_b 4.83% : 0.005377s : 1: optimize 0.02% : 0.000020s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000009s : 1: order_py_execute_after_rewriter 0.02% : 0.000022s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000004s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.01% : 0.000009s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.05% : 0.000051s : 1: pre_auto_parallel 0.03% : 0.000036s : 1: py_interpret_to_execute 0.01% : 0.000013s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.02% : 0.000017s : 1: remove_dup_value 0.41% : 0.000451s : 1: renormalize.infer 0.33% : 0.000369s : 1: renormalize.specialize 0.01% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000007s : 1: rewriter_after_jit_bprop_graph 0.03% : 0.000039s : 1: rewriter_after_opt_a 0.11% : 0.000118s : 1: rewriter_before_opt_a 0.01% : 0.000006s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.01% : 0.000006s : 1: split_matmul_comm_elemetwise 0.01% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.07% : 0.000076s : 1: symbol_engine_optimizer 0.07% : 0.000078s : 1: tuple_transform 79.36% : 0.088282s : 1: type_inference TotalTime = 0.529131, [21] [bootstrap]: 0.00055363 [type_inference]: 0.265489 [event_method]: 8.465e-05 [auto_monad]: 0.0001402 [graph_reusing]: 9.35001e-06 [inline]: 3.50998e-06 [add_attr]: 0.0708175, [1] [add_attr_with_inline]: 0.0707982, [1] [Cycle 1]: 0.00010987, [2] [tag_attr]: 5.007e-05 [meta_addattr_fg_expand]: 9.61998e-06 [parallel-infer-symbol]: 3.81999e-06 [pre_auto_parallel]: 6.82e-05 [insert-virtual-dataset]: 2.73e-06 [parallel-infer-symbol-second]: 8.49977e-07 [dataset_repeat_opt]: 2.14e-06 [pipeline_split]: 1.63002e-06 [optimize]: 0.183256, [53] [py_interpret_to_execute]: 5.322e-05 [rewriter_before_opt_a]: 0.00785798 [opt_a]: 0.147479, [3] [Cycle 1]: 0.0981196, [45] [expand_dump_flag]: 8.04002e-06 [switch_simplify]: 9.701e-05 [loop_unroll]: 6.484e-05 [a_1]: 0.0172561 [with_stream_mark]: 4.649e-05 [recompute_prepare]: 2.927e-05 [updatestate_depend_eliminate]: 1.033e-05 [updatestate_assign_eliminate]: 8.07e-06 [updatestate_loads_eliminate]: 7.65998e-06 [parameter_eliminate]: 3.66001e-06 [a_2]: 0.00683459 [accelerated_algorithm]: 5.551e-05 [shard]: 3.75e-06 [meta_shard_fg_expand]: 8.15e-06 [shard_inline]: 1.693e-05 [merge_send_recv]: 2.81e-05 [auto_parallel]: 2.017e-05 [parallel]: 2.293e-05 [flash_sp]: 1.406e-05 [merge_comm]: 1.018e-05 [allreduce_fusion]: 9.12999e-06 [matmul_add_comm_reduction]: 3.678e-05 [allreduce_slice_to_reducescatter]: 7.40023e-07 [virtual_shard_identity]: 1.851e-05 [virtual_dataset]: 1.576e-05 [get_grad_eliminate_]: 1.499e-05 [virtual_output]: 1.543e-05 [merge_forward]: 1.068e-05 [cell_reuse_recompute_pass]: 3.79002e-06 [offload_activation]: 1.997e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.087e-05 [merge_recompute_call_nodes]: 1.38002e-06 [before_grad]: 2.973e-05 [set_forward_comm_id_for_comm_node_pass]: 9.81e-06 [meta_fg_expand]: 0.0262225 [flash_sp_send_recv_attached]: 1.872e-05 [receive_attached]: 2.60002e-06 [after_resolve]: 0.00011925 [a_after_grad]: 0.00014126 [renormalize]: 0.0351691 [add_forward_monad_depend]: 2.737e-05 [auto_monad_grad]: 1.004e-05 [auto_monad_eliminator]: 0.0001102 [cse]: 0.00031057 [a_3]: 0.0003685 [Cycle 2]: 0.0418782, [45] [expand_dump_flag]: 2.78e-06 [switch_simplify]: 0.00877283 [loop_unroll]: 0.00010278 [a_1]: 0.0142744 [with_stream_mark]: 2.846e-05 [recompute_prepare]: 1.289e-05 [updatestate_depend_eliminate]: 4.35999e-06 [updatestate_assign_eliminate]: 3.36001e-06 [updatestate_loads_eliminate]: 2.81e-06 [parameter_eliminate]: 2.04999e-06 [a_2]: 6.978e-05 [accelerated_algorithm]: 1.283e-05 [shard]: 2.36e-06 [meta_shard_fg_expand]: 3.18998e-06 [shard_inline]: 6.93998e-06 [merge_send_recv]: 9.17999e-06 [auto_parallel]: 9.41e-06 [parallel]: 9.96e-06 [flash_sp]: 4.53001e-06 [merge_comm]: 3.36001e-06 [allreduce_fusion]: 3.12002e-06 [matmul_add_comm_reduction]: 9.84001e-06 [allreduce_slice_to_reducescatter]: 1.62001e-06 [virtual_shard_identity]: 4.438e-05 [virtual_dataset]: 8.19002e-06 [get_grad_eliminate_]: 7.30003e-06 [virtual_output]: 6.61999e-06 [merge_forward]: 1.759e-05 [cell_reuse_recompute_pass]: 3.53999e-06 [offload_activation]: 2.093e-05 [cell_reuse_handle_not_recompute_node_pass]: 2.581e-05 [merge_recompute_call_nodes]: 2.93998e-06 [before_grad]: 1.171e-05 [set_forward_comm_id_for_comm_node_pass]: 8.26002e-06 [meta_fg_expand]: 0.00016307 [flash_sp_send_recv_attached]: 2.81999e-06 [receive_attached]: 3.06001e-06 [after_resolve]: 1.833e-05 [a_after_grad]: 1.047e-05 [renormalize]: 0.0100944 [add_forward_monad_depend]: 1.191e-05 [auto_monad_grad]: 2.48e-06 [auto_monad_eliminator]: 2.43e-05 [cse]: 3.705e-05 [a_3]: 6.216e-05 [Cycle 3]: 0.0074542, [45] [expand_dump_flag]: 2.66e-06 [switch_simplify]: 9.22999e-06 [loop_unroll]: 6.56e-06 [a_1]: 0.00673484 [with_stream_mark]: 3.955e-05 [recompute_prepare]: 1.601e-05 [updatestate_depend_eliminate]: 7.10998e-06 [updatestate_assign_eliminate]: 2.97002e-06 [updatestate_loads_eliminate]: 3.13e-06 [parameter_eliminate]: 2.30002e-06 [a_2]: 8.08e-05 [accelerated_algorithm]: 2.09e-05 [shard]: 3.38999e-06 [meta_shard_fg_expand]: 2.99001e-06 [shard_inline]: 6.59001e-06 [merge_send_recv]: 1.01e-05 [auto_parallel]: 1.185e-05 [parallel]: 1.132e-05 [flash_sp]: 1.82001e-06 [merge_comm]: 3.28e-06 [allreduce_fusion]: 3.21999e-06 [matmul_add_comm_reduction]: 1.149e-05 [allreduce_slice_to_reducescatter]: 1.19e-06 [virtual_shard_identity]: 7.64002e-06 [virtual_dataset]: 6.29001e-06 [get_grad_eliminate_]: 6.00002e-06 [virtual_output]: 7.03998e-06 [merge_forward]: 3.95e-06 [cell_reuse_recompute_pass]: 3.68999e-06 [offload_activation]: 1.138e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.296e-05 [merge_recompute_call_nodes]: 1.74e-06 [before_grad]: 1.055e-05 [set_forward_comm_id_for_comm_node_pass]: 3.53e-06 [meta_fg_expand]: 2.39001e-06 [flash_sp_send_recv_attached]: 2.12999e-06 [receive_attached]: 2.96001e-06 [after_resolve]: 1.247e-05 [a_after_grad]: 9.53002e-06 [renormalize]: 8.9989e-08 [add_forward_monad_depend]: 3.22997e-06 [auto_monad_grad]: 2.69001e-06 [auto_monad_eliminator]: 1.659e-05 [cse]: 3.405e-05 [a_3]: 3.672e-05 [py_interpret_to_execute_after_opt_a]: 1.873e-05 [slice_cell_reuse_recomputed_activation]: 2.69001e-06 [rewriter_after_opt_a]: 9.565e-05 [convert_after_rewriter]: 9.41e-06 [order_py_execute_after_rewriter]: 4.91002e-06 [mutable_eliminate]: 0.00728345 [opt_b]: 0.00792056, [1] [Cycle 1]: 0.00790749, [7] [b_1]: 0.0001514 [b_2]: 0.0076154 [updatestate_depend_eliminate]: 1.507e-05 [updatestate_assign_eliminate]: 4.29002e-06 [updatestate_loads_eliminate]: 3.38999e-06 [renormalize]: 8.39995e-07 [cse]: 3.793e-05 [optimize_parallel_all_gather_comm]: 2.922e-05 [overlap_param_gather]: 1.77001e-06 [cconv]: 3.689e-05 [loop_unroll]: 0.00579812 [opt_after_cconv]: 0.0001501, [1] [Cycle 1]: 0.0001401, [7] [c_1]: 3.582e-05 [parameter_eliminate]: 6.49001e-06 [updatestate_depend_eliminate]: 1.216e-05 [updatestate_assign_eliminate]: 3.38999e-06 [updatestate_loads_eliminate]: 2.96001e-06 [cse]: 4.036e-05 [renormalize]: 9.10019e-07 [remove_dup_value]: 1.575e-05 [tuple_transform]: 0.00581438, [1] [Cycle 1]: 0.00580582, [4] [d_1]: 0.00570248 [none_parameter_eliminate]: 6.69001e-06 [renormalize]: 1.06002e-06 [switch_simplify]: 2.571e-05 [partial_unused_args_eliminate]: 3.23e-06 [add_recomputation]: 8.131e-05 [cse_after_recomputation]: 5.2e-05, [1] [Cycle 1]: 4.245e-05, [1] [cse]: 3.561e-05 [environ_conv]: 1.179e-05 [swap_dp_allreduce_reducescatter]: 5.76e-06 [bias_add_comm_swap]: 3.84002e-06 [label_micro_interleaved_index]: 8.12e-06 [label_fine_grained_interleaved_index]: 2.63998e-06 [merge_cast_opt]: 1.54e-06 [slice_recompute_activation]: 2.31e-06 [micro_interleaved_order_control]: 2.48002e-06 [assign_add_opt]: 1.59998e-06 [ForceFp32Comm]: 9.5999e-07 [remove_cast_before_assign_add]: 1.46998e-06 [full_micro_interleaved_order_control]: 2.02999e-06 [reorder_send_recv_between_fp_bp]: 2.68e-06 [comm_op_add_attrs]: 1.10999e-06 [add_comm_op_reuse_tag]: 1.45999e-06 [interleave_split_concat_branches]: 1.21997e-06 [interleave_parallel_branches]: 1.04e-06 [overlap_opt_shard_in_pipeline]: 1.16002e-06 [overlap_opt_shard_grad_in_pipeline]: 2.19001e-06 [control_data_broadcast_order]: 1.704e-05 [grouped_pairwise_exchange_alltoall]: 1.64e-06 [offloading_packed_experts]: 4.13001e-06 [overlap_recompute_and_grad_model_parallel]: 5.07999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.24e-06 [overlap_recompute_allgather_and_fa_grad]: 1.31002e-06 [overlap_recompute_comm]: 2.19001e-06 [overlap_grad_ring_attention]: 4.03001e-06 [overlap_grad_flash_sp]: 2.545e-05 [begin_end_overlap_inline]: 5.19998e-07 [split_matmul_comm_elemetwise]: 2.30002e-06 [split_layernorm_comm]: 2.04e-06 [handle_group_info]: 1.07e-06 [symbol_engine_optimizer]: 0.00010003, [1] [Cycle 1]: 9.435e-05, [6] [build]: 1.429e-05 [elim_shapecalc]: 1.257e-05 [elim_not_effective]: 1.977e-05 [opt_reshape]: 7.02002e-06 [fold_const_symbol]: 1.054e-05 [renormalize]: 1.79978e-07 [detach_backward]: 2.40002e-06 [pipeline_parallel_scheduler]: 1.54e-06 [auto_monad_reorder]: 2.016e-05 [get_jit_bprop_graph]: 2.56e-06 [rewriter_after_jit_bprop_graph]: 6.73e-06 [opt_after_jit_grad]: 0.00840467 [validate]: 6.267e-05 Sums bootstrap : 0.000554s : 0.13% type_inference : 0.265489s : 60.55% event_method : 0.000085s : 0.02% auto_monad : 0.000140s : 0.03% graph_reusing : 0.000009s : 0.00% inline : 0.000004s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000050s : 0.01% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000010s : 0.00% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000068s : 0.02% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000053s : 0.01% optimize.rewriter_before_opt_a : 0.007858s : 1.79% optimize.opt_a.expand_dump_flag : 0.000013s : 0.00% optimize.opt_a.switch_simplify : 0.008879s : 2.02% optimize.opt_a.loop_unroll : 0.000174s : 0.04% optimize.opt_a.a_1 : 0.038265s : 8.73% optimize.opt_a.with_stream_mark : 0.000114s : 0.03% optimize.opt_a.recompute_prepare : 0.000058s : 0.01% optimize.opt_a.updatestate_depend_eliminate : 0.000022s : 0.00% optimize.opt_a.updatestate_assign_eliminate : 0.000014s : 0.00% optimize.opt_a.updatestate_loads_eliminate : 0.000014s : 0.00% optimize.opt_a.parameter_eliminate : 0.000008s : 0.00% optimize.opt_a.a_2 : 0.006985s : 1.59% optimize.opt_a.accelerated_algorithm : 0.000089s : 0.02% optimize.opt_a.shard : 0.000009s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000014s : 0.00% optimize.opt_a.shard_inline : 0.000030s : 0.01% optimize.opt_a.merge_send_recv : 0.000047s : 0.01% optimize.opt_a.auto_parallel : 0.000041s : 0.01% optimize.opt_a.parallel : 0.000044s : 0.01% optimize.opt_a.flash_sp : 0.000020s : 0.00% optimize.opt_a.merge_comm : 0.000017s : 0.00% optimize.opt_a.allreduce_fusion : 0.000015s : 0.00% optimize.opt_a.matmul_add_comm_reduction : 0.000058s : 0.01% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000004s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000071s : 0.02% optimize.opt_a.virtual_dataset : 0.000030s : 0.01% optimize.opt_a.get_grad_eliminate_ : 0.000028s : 0.01% optimize.opt_a.virtual_output : 0.000029s : 0.01% optimize.opt_a.merge_forward : 0.000032s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000011s : 0.00% optimize.opt_a.offload_activation : 0.000052s : 0.01% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000070s : 0.02% optimize.opt_a.merge_recompute_call_nodes : 0.000006s : 0.00% optimize.opt_a.before_grad : 0.000052s : 0.01% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000022s : 0.00% optimize.opt_a.meta_fg_expand : 0.026388s : 6.02% optimize.opt_a.flash_sp_send_recv_attached : 0.000024s : 0.01% optimize.opt_a.receive_attached : 0.000009s : 0.00% optimize.opt_a.after_resolve : 0.000150s : 0.03% optimize.opt_a.a_after_grad : 0.000161s : 0.04% optimize.opt_a.renormalize : 0.045264s : 10.32% optimize.opt_a.add_forward_monad_depend : 0.000043s : 0.01% optimize.opt_a.auto_monad_grad : 0.000015s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000151s : 0.03% optimize.opt_a.cse : 0.000382s : 0.09% optimize.opt_a.a_3 : 0.000467s : 0.11% optimize.py_interpret_to_execute_after_opt_a : 0.000019s : 0.00% optimize.slice_cell_reuse_recomputed_activation : 0.000003s : 0.00% optimize.rewriter_after_opt_a : 0.000096s : 0.02% optimize.convert_after_rewriter : 0.000009s : 0.00% optimize.order_py_execute_after_rewriter : 0.000005s : 0.00% optimize.mutable_eliminate : 0.007283s : 1.66% optimize.opt_b.b_1 : 0.000151s : 0.03% optimize.opt_b.b_2 : 0.007615s : 1.74% optimize.opt_b.updatestate_depend_eliminate : 0.000015s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000004s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000038s : 0.01% optimize.optimize_parallel_all_gather_comm : 0.000029s : 0.01% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000037s : 0.01% optimize.loop_unroll : 0.005798s : 1.32% optimize.opt_after_cconv.c_1 : 0.000036s : 0.01% optimize.opt_after_cconv.parameter_eliminate : 0.000006s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000012s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.cse : 0.000040s : 0.01% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000016s : 0.00% optimize.tuple_transform.d_1 : 0.005702s : 1.30% optimize.tuple_transform.none_parameter_eliminate : 0.000007s : 0.00% optimize.tuple_transform.renormalize : 0.000001s : 0.00% optimize.tuple_transform.switch_simplify : 0.000026s : 0.01% optimize.partial_unused_args_eliminate : 0.000003s : 0.00% optimize.add_recomputation : 0.000081s : 0.02% optimize.cse_after_recomputation.cse : 0.000036s : 0.01% optimize.environ_conv : 0.000012s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.00% optimize.bias_add_comm_swap : 0.000004s : 0.00% optimize.label_micro_interleaved_index : 0.000008s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000002s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000017s : 0.00% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000025s : 0.01% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000014s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000013s : 0.00% optimize.symbol_engine_optimizer.elim_not_effective : 0.000020s : 0.00% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000011s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000020s : 0.00% get_jit_bprop_graph : 0.000003s : 0.00% rewriter_after_jit_bprop_graph : 0.000007s : 0.00% opt_after_jit_grad : 0.008405s : 1.92% validate : 0.000063s : 0.01% Time group info: ------[substitution.] 0.007299 190 1.54% : 0.000112s : 6: substitution.arithmetic_simplify 0.03% : 0.000002s : 2: substitution.elim_not_effective 0.10% : 0.000007s : 6: substitution.float_depend_g_call 0.07% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.02% : 0.000001s : 2: substitution.fold_const_symbol 0.09% : 0.000006s : 4: substitution.graph_param_transform 0.05% : 0.000004s : 2: substitution.incorporate_call 0.03% : 0.000002s : 2: substitution.incorporate_call_switch 15.47% : 0.001129s : 16: substitution.inline 0.59% : 0.000043s : 2: substitution.inline_without_move 0.16% : 0.000012s : 14: substitution.j_node_and_user_rematch 0.48% : 0.000035s : 3: substitution.less_batch_normalization 0.31% : 0.000023s : 11: substitution.minmaximum_grad 0.11% : 0.000008s : 6: substitution.partial_eliminate 0.18% : 0.000013s : 14: substitution.remove_not_recompute_node 0.57% : 0.000041s : 10: substitution.replace_applicator 0.26% : 0.000019s : 12: substitution.replace_old_param 0.04% : 0.000003s : 1: substitution.set_cell_output_no_recompute 0.63% : 0.000046s : 11: substitution.tuple_list_convert_item_index_to_positive 0.25% : 0.000018s : 11: substitution.tuple_list_get_item_const_eliminator 0.42% : 0.000031s : 11: substitution.tuple_list_get_item_depend_reorder 1.71% : 0.000125s : 30: substitution.tuple_list_get_item_eliminator 76.89% : 0.005612s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.265276 2 91.11% : 0.241680s : 1: type_inference.infer 8.89% : 0.023596s : 1: type_inference.specialize ------[replace.] 0.000373 34 3.04% : 0.000011s : 2: replace.arithmetic_simplify 59.58% : 0.000222s : 16: replace.inline 37.38% : 0.000139s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.001229 34 5.23% : 0.000064s : 2: match.arithmetic_simplify 90.68% : 0.001114s : 16: match.inline 4.09% : 0.000050s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.007348 4766 0.14% : 0.000010s : 58: predicate.accumulaten_eliminater 0.05% : 0.000004s : 4: predicate.ad_related_special_op_eliminate 0.05% : 0.000004s : 23: predicate.addn_check_dump 0.18% : 0.000013s : 58: predicate.addn_zero_filter 0.15% : 0.000011s : 58: predicate.adjust_all_reduce_mul_add 0.32% : 0.000024s : 83: predicate.arithmetic_simplify 0.14% : 0.000010s : 60: predicate.cast_eliminate 0.11% : 0.000008s : 55: predicate.check_bprop_eliminate 0.05% : 0.000004s : 23: predicate.compare_switch_simplify 0.01% : 0.000000s : 4: predicate.const_output_eliminate 0.05% : 0.000004s : 23: predicate.depend_value_elim 0.13% : 0.000009s : 60: predicate.dict_get_item_const_eliminator 0.14% : 0.000011s : 60: predicate.dict_get_item_eliminator 0.13% : 0.000009s : 60: predicate.dict_set_item_eliminator 0.05% : 0.000004s : 8: predicate.dumpgradient_eliminate 0.01% : 0.000001s : 4: predicate.elim_not_effective 0.01% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 0.13% : 0.000009s : 64: predicate.environ_add_const_eliminate 0.13% : 0.000010s : 64: predicate.environ_get_add_eliminate 0.12% : 0.000009s : 64: predicate.environ_get_depend_swap 0.17% : 0.000013s : 87: predicate.environ_get_eliminate 0.11% : 0.000008s : 64: predicate.environ_get_set_eliminate 0.18% : 0.000013s : 92: predicate.exchange_switch_depend_value 0.31% : 0.000023s : 92: predicate.float_depend_g_call 88.33% : 0.006491s : 23: predicate.float_environ_get_switch 0.05% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.01% : 0.000000s : 4: predicate.fold_const_symbol 0.05% : 0.000004s : 23: predicate.get_grad_eliminate 0.01% : 0.000001s : 4: predicate.graph_param_transform 0.05% : 0.000004s : 23: predicate.incorporate_call 0.04% : 0.000003s : 23: predicate.incorporate_call_switch 0.71% : 0.000052s : 204: predicate.inline 0.15% : 0.000011s : 46: predicate.inline_without_move 0.02% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.07% : 0.000005s : 23: predicate.less_batch_normalization 0.20% : 0.000015s : 84: predicate.list_to_tuple_eliminator_ 0.27% : 0.000020s : 142: predicate.load_eliminater 0.07% : 0.000005s : 4: predicate.loop_unroll_after_grad 0.45% : 0.000033s : 128: predicate.loop_unroll_before_grad 0.14% : 0.000010s : 68: predicate.make_slice_get_slice_eliminator 0.05% : 0.000004s : 23: predicate.merge_addn 0.10% : 0.000007s : 55: predicate.micro_step_allgather_replace 0.11% : 0.000008s : 55: predicate.mini_step_allgather_replace 0.11% : 0.000008s : 60: predicate.minmaximum_grad 0.06% : 0.000004s : 4: predicate.mutable_eliminate 0.01% : 0.000001s : 4: predicate.opt_reshape 0.01% : 0.000001s : 4: predicate.parallel_virtual_node 0.41% : 0.000030s : 92: predicate.partial_defer_inline 0.16% : 0.000012s : 80: predicate.partial_eliminate 0.12% : 0.000009s : 58: predicate.print_const_string_wrapper 0.05% : 0.000003s : 23: predicate.reduce_all_const_elim 0.19% : 0.000014s : 60: predicate.reduce_eliminate 0.26% : 0.000019s : 142: predicate.redundant_stop_gradient_eliminater 0.04% : 0.000003s : 23: predicate.remove_not_recompute_node 0.21% : 0.000015s : 131: predicate.replace_applicator 0.07% : 0.000005s : 46: predicate.replace_old_param 0.01% : 0.000001s : 4: predicate.reset_defer_inline 0.13% : 0.000010s : 60: predicate.reshape_eliminate 0.13% : 0.000009s : 55: predicate.row_tensor_add_zeros_like 0.05% : 0.000004s : 4: predicate.row_tensor_eliminate 0.13% : 0.000010s : 55: predicate.same_eliminate 0.03% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.14% : 0.000010s : 23: predicate.shard_identity_eliminate 0.03% : 0.000002s : 8: predicate.special_op_eliminate 0.05% : 0.000004s : 23: predicate.specialize_transform 0.13% : 0.000010s : 55: predicate.split_environ_get_set_with_tuple_value 0.16% : 0.000012s : 46: predicate.stack_unstack_eliminate 0.01% : 0.000001s : 4: predicate.switch_call_monad_eliminater 0.19% : 0.000014s : 92: predicate.switch_defer_inline 0.29% : 0.000021s : 147: predicate.switch_layer_defer_inline 0.60% : 0.000044s : 247: predicate.switch_simplify 0.13% : 0.000009s : 60: predicate.tile_eliminate 0.13% : 0.000010s : 60: predicate.transpose_eliminate 0.17% : 0.000012s : 68: predicate.tuple_list_convert_item_index_to_positive 0.16% : 0.000012s : 68: predicate.tuple_list_get_item_const_eliminator 0.18% : 0.000013s : 68: predicate.tuple_list_get_item_depend_reorder 0.44% : 0.000033s : 107: predicate.tuple_list_get_item_eliminator 0.14% : 0.000010s : 68: predicate.tuple_list_get_set_item_eliminator 0.23% : 0.000017s : 91: predicate.tuple_list_set_item_eliminator 0.18% : 0.000013s : 84: predicate.tuple_to_list_eliminator_ 0.26% : 0.000019s : 142: predicate.updatestate_pure_node_eliminater 0.32% : 0.000024s : 165: predicate.updatestate_useless_node_eliminater 0.01% : 0.000001s : 4: predicate.value_based_eliminate 0.05% : 0.000004s : 23: predicate.virtual_dataset_eliminate 0.05% : 0.000004s : 23: predicate.virtual_output_eliminate 0.01% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.01% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.056279 37 84.34% : 0.047464s : 17: func_graph_cloner_run.FuncGraphClonerGraph 15.66% : 0.008815s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.897194 233 0.00% : 0.000004s : 1: ForceFp32Comm 7.89% : 0.070825s : 1: add_attr 7.89% : 0.070803s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.01% : 0.000086s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.02% : 0.000147s : 1: auto_monad 0.00% : 0.000024s : 1: auto_monad_reorder 0.00% : 0.000003s : 1: begin_end_overlap_inline 0.00% : 0.000007s : 1: bias_add_comm_swap 0.07% : 0.000585s : 1: bootstrap 0.00% : 0.000041s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000022s : 1: control_data_broadcast_order 0.00% : 0.000013s : 1: convert_after_rewriter 0.01% : 0.000055s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000015s : 1: environ_conv 0.01% : 0.000094s : 1: event_method 0.00% : 0.000006s : 1: full_micro_interleaved_order_control 0.00% : 0.000007s : 1: get_jit_bprop_graph 0.00% : 0.000013s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000007s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000007s : 1: label_fine_grained_interleaved_index 0.00% : 0.000011s : 1: label_micro_interleaved_index 0.65% : 0.005822s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.81% : 0.007300s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.00% : 0.000035s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000029s : 1: opt.transform.mutable_eliminate 6.17% : 0.055314s : 117: opt.transform.opt_a 0.00% : 0.000034s : 1: opt.transform.opt_after_cconv 0.00% : 0.000042s : 1: opt.transform.opt_after_jit_grad 0.86% : 0.007731s : 28: opt.transform.opt_b 0.64% : 0.005717s : 2: opt.transform.opt_trans_graph 0.01% : 0.000046s : 4: opt.transform.symbol_engine_opt 16.44% : 0.147485s : 1: opt_a 0.02% : 0.000155s : 1: opt_after_cconv 0.94% : 0.008424s : 1: opt_after_jit_grad 0.88% : 0.007927s : 1: opt_b 20.43% : 0.183262s : 1: optimize 0.00% : 0.000033s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000008s : 1: order_py_execute_after_rewriter 0.00% : 0.000030s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000006s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000006s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000009s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.01% : 0.000073s : 1: pre_auto_parallel 0.01% : 0.000058s : 1: py_interpret_to_execute 0.00% : 0.000023s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000019s : 1: remove_dup_value 4.04% : 0.036218s : 2: renormalize.infer 1.00% : 0.009011s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000010s : 1: rewriter_after_jit_bprop_graph 0.01% : 0.000101s : 1: rewriter_after_opt_a 0.88% : 0.007895s : 1: rewriter_before_opt_a 0.00% : 0.000006s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.01% : 0.000103s : 1: symbol_engine_optimizer 0.65% : 0.005819s : 1: tuple_transform 29.59% : 0.265521s : 1: type_inference TotalTime = 0.158756, [21] [bootstrap]: 0.00045714 [type_inference]: 0.12333 [event_method]: 6.266e-05 [auto_monad]: 0.00013115 [graph_reusing]: 8.28001e-06 [inline]: 3.01001e-06 [add_attr]: 0.00419623, [1] [add_attr_with_inline]: 0.00418333, [1] [Cycle 1]: 0.00010021, [2] [tag_attr]: 4.686e-05 [meta_addattr_fg_expand]: 9.68002e-06 [parallel-infer-symbol]: 4.32e-06 [pre_auto_parallel]: 6.558e-05 [insert-virtual-dataset]: 2.44001e-06 [parallel-infer-symbol-second]: 8.29983e-07 [dataset_repeat_opt]: 2.01998e-06 [pipeline_split]: 1.60999e-06 [optimize]: 0.0296505, [53] [py_interpret_to_execute]: 4.718e-05 [rewriter_before_opt_a]: 0.00017366 [opt_a]: 0.0268997, [3] [Cycle 1]: 0.0086986, [45] [expand_dump_flag]: 4.23999e-06 [switch_simplify]: 7.924e-05 [loop_unroll]: 6.263e-05 [a_1]: 0.0015883 [with_stream_mark]: 3.575e-05 [recompute_prepare]: 2.863e-05 [updatestate_depend_eliminate]: 1.012e-05 [updatestate_assign_eliminate]: 8.35999e-06 [updatestate_loads_eliminate]: 7.33e-06 [parameter_eliminate]: 3.30998e-06 [a_2]: 0.00024799 [accelerated_algorithm]: 3.865e-05 [shard]: 2.14e-06 [meta_shard_fg_expand]: 4.75999e-06 [shard_inline]: 1.618e-05 [merge_send_recv]: 1.952e-05 [auto_parallel]: 1.676e-05 [parallel]: 2.041e-05 [flash_sp]: 1.271e-05 [merge_comm]: 1.047e-05 [allreduce_fusion]: 9.36002e-06 [matmul_add_comm_reduction]: 3.418e-05 [allreduce_slice_to_reducescatter]: 1.10001e-06 [virtual_shard_identity]: 2.021e-05 [virtual_dataset]: 1.553e-05 [get_grad_eliminate_]: 1.563e-05 [virtual_output]: 1.577e-05 [merge_forward]: 1.051e-05 [cell_reuse_recompute_pass]: 2.21998e-06 [offload_activation]: 2.015e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.265e-05 [merge_recompute_call_nodes]: 1.81e-06 [before_grad]: 2.998e-05 [set_forward_comm_id_for_comm_node_pass]: 1.103e-05 [meta_fg_expand]: 0.00187359 [flash_sp_send_recv_attached]: 7.77998e-06 [receive_attached]: 2.76e-06 [after_resolve]: 7.535e-05 [a_after_grad]: 9.439e-05 [renormalize]: 0.00322407 [add_forward_monad_depend]: 1.237e-05 [auto_monad_grad]: 6.53e-06 [auto_monad_eliminator]: 5.923e-05 [cse]: 0.00021277 [a_3]: 0.00034025 [Cycle 2]: 0.0174181, [45] [expand_dump_flag]: 3.5e-06 [switch_simplify]: 4.727e-05 [loop_unroll]: 6.534e-05 [a_1]: 0.00158921 [with_stream_mark]: 2.204e-05 [recompute_prepare]: 1.012e-05 [updatestate_depend_eliminate]: 4.37998e-06 [updatestate_assign_eliminate]: 3.62002e-06 [updatestate_loads_eliminate]: 2.59999e-06 [parameter_eliminate]: 1.97999e-06 [a_2]: 6.961e-05 [accelerated_algorithm]: 1.136e-05 [shard]: 2.68998e-06 [meta_shard_fg_expand]: 2.28002e-06 [shard_inline]: 6.22001e-06 [merge_send_recv]: 8.77999e-06 [auto_parallel]: 1.048e-05 [parallel]: 9.47001e-06 [flash_sp]: 3.89002e-06 [merge_comm]: 3.67002e-06 [allreduce_fusion]: 3.16999e-06 [matmul_add_comm_reduction]: 9.61e-06 [allreduce_slice_to_reducescatter]: 6.70028e-07 [virtual_shard_identity]: 7.55e-06 [virtual_dataset]: 5.51998e-06 [get_grad_eliminate_]: 5.91003e-06 [virtual_output]: 5.25999e-06 [merge_forward]: 4.17e-06 [cell_reuse_recompute_pass]: 1.54e-06 [offload_activation]: 1.027e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.12e-05 [merge_recompute_call_nodes]: 1.39e-06 [before_grad]: 9.81e-06 [set_forward_comm_id_for_comm_node_pass]: 3.68999e-06 [meta_fg_expand]: 0.00012855 [flash_sp_send_recv_attached]: 2.02001e-06 [receive_attached]: 2.51e-06 [after_resolve]: 1.486e-05 [a_after_grad]: 9.00999e-06 [renormalize]: 0.0148602 [add_forward_monad_depend]: 1.311e-05 [auto_monad_grad]: 2.88998e-06 [auto_monad_eliminator]: 2.471e-05 [cse]: 3.819e-05 [a_3]: 6.744e-05 [Cycle 3]: 0.00076298, [45] [expand_dump_flag]: 2.77002e-06 [switch_simplify]: 1.016e-05 [loop_unroll]: 7.03e-06 [a_1]: 0.00013649 [with_stream_mark]: 2.047e-05 [recompute_prepare]: 6.64001e-06 [updatestate_depend_eliminate]: 4.63999e-06 [updatestate_assign_eliminate]: 3.35e-06 [updatestate_loads_eliminate]: 2.48002e-06 [parameter_eliminate]: 1.69998e-06 [a_2]: 6.682e-05 [accelerated_algorithm]: 1.902e-05 [shard]: 2.86999e-06 [meta_shard_fg_expand]: 2.22001e-06 [shard_inline]: 5.92001e-06 [merge_send_recv]: 1.067e-05 [auto_parallel]: 1.133e-05 [parallel]: 1.129e-05 [flash_sp]: 1.86e-06 [merge_comm]: 3.55e-06 [allreduce_fusion]: 3.56999e-06 [matmul_add_comm_reduction]: 1.039e-05 [allreduce_slice_to_reducescatter]: 9.80013e-07 [virtual_shard_identity]: 9.07999e-06 [virtual_dataset]: 6.96001e-06 [get_grad_eliminate_]: 6.31e-06 [virtual_output]: 7e-06 [merge_forward]: 4.48001e-06 [cell_reuse_recompute_pass]: 5.07e-06 [offload_activation]: 1.279e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.422e-05 [merge_recompute_call_nodes]: 1.98997e-06 [before_grad]: 1.138e-05 [set_forward_comm_id_for_comm_node_pass]: 5.35001e-06 [meta_fg_expand]: 2.79001e-06 [flash_sp_send_recv_attached]: 2.48e-06 [receive_attached]: 2.31e-06 [after_resolve]: 1.395e-05 [a_after_grad]: 9.97999e-06 [renormalize]: 1.00001e-07 [add_forward_monad_depend]: 3.38999e-06 [auto_monad_grad]: 2.04e-06 [auto_monad_eliminator]: 1.299e-05 [cse]: 2.261e-05 [a_3]: 3.72e-05 [py_interpret_to_execute_after_opt_a]: 2.104e-05 [slice_cell_reuse_recomputed_activation]: 2.31e-06 [rewriter_after_opt_a]: 4.845e-05 [convert_after_rewriter]: 8.45999e-06 [order_py_execute_after_rewriter]: 6.24001e-06 [mutable_eliminate]: 0.00081071 [opt_b]: 0.0002424, [1] [Cycle 1]: 0.00023323, [7] [b_1]: 0.00013271 [b_2]: 9.36e-06 [updatestate_depend_eliminate]: 1.116e-05 [updatestate_assign_eliminate]: 3.31001e-06 [updatestate_loads_eliminate]: 3.5e-06 [renormalize]: 6.50005e-07 [cse]: 3.164e-05 [optimize_parallel_all_gather_comm]: 2.331e-05 [overlap_param_gather]: 2.04999e-06 [cconv]: 3.803e-05 [loop_unroll]: 0.00051793 [opt_after_cconv]: 0.00015881, [1] [Cycle 1]: 0.00015139, [7] [c_1]: 3.172e-05 [parameter_eliminate]: 5.30999e-06 [updatestate_depend_eliminate]: 8.70001e-06 [updatestate_assign_eliminate]: 2.90998e-06 [updatestate_loads_eliminate]: 2.32999e-06 [cse]: 6.123e-05 [renormalize]: 4.00003e-07 [remove_dup_value]: 1.672e-05 [tuple_transform]: 8.381e-05, [1] [Cycle 1]: 7.84e-05, [4] [d_1]: 5e-05 [none_parameter_eliminate]: 1.86e-06 [renormalize]: 1.8999e-07 [switch_simplify]: 7.18e-06 [partial_unused_args_eliminate]: 1.83002e-06 [add_recomputation]: 5.554e-05 [cse_after_recomputation]: 2.343e-05, [1] [Cycle 1]: 1.88e-05, [1] [cse]: 1.339e-05 [environ_conv]: 1.087e-05 [swap_dp_allreduce_reducescatter]: 5.64998e-06 [bias_add_comm_swap]: 3.53999e-06 [label_micro_interleaved_index]: 5.62999e-06 [label_fine_grained_interleaved_index]: 3.16999e-06 [merge_cast_opt]: 1.88002e-06 [slice_recompute_activation]: 2.11e-06 [micro_interleaved_order_control]: 2.26e-06 [assign_add_opt]: 1.40999e-06 [ForceFp32Comm]: 8.09989e-07 [remove_cast_before_assign_add]: 1.12e-06 [full_micro_interleaved_order_control]: 2.24001e-06 [reorder_send_recv_between_fp_bp]: 2.77002e-06 [comm_op_add_attrs]: 1.10999e-06 [add_comm_op_reuse_tag]: 9.89996e-07 [interleave_split_concat_branches]: 1.15999e-06 [interleave_parallel_branches]: 1.18001e-06 [overlap_opt_shard_in_pipeline]: 1.21002e-06 [overlap_opt_shard_grad_in_pipeline]: 1.99999e-06 [control_data_broadcast_order]: 1.453e-05 [grouped_pairwise_exchange_alltoall]: 1.57001e-06 [offloading_packed_experts]: 4.38001e-06 [overlap_recompute_and_grad_model_parallel]: 5.79999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.36998e-06 [overlap_recompute_allgather_and_fa_grad]: 1.47001e-06 [overlap_recompute_comm]: 2.45002e-06 [overlap_grad_ring_attention]: 4.58001e-06 [overlap_grad_flash_sp]: 2.32e-05 [begin_end_overlap_inline]: 5.39992e-07 [split_matmul_comm_elemetwise]: 2.79001e-06 [split_layernorm_comm]: 1.89e-06 [handle_group_info]: 1.55999e-06 [symbol_engine_optimizer]: 9.152e-05, [1] [Cycle 1]: 8.634e-05, [6] [build]: 1.257e-05 [elim_shapecalc]: 1.214e-05 [elim_not_effective]: 1.429e-05 [opt_reshape]: 6.89001e-06 [fold_const_symbol]: 9.08002e-06 [renormalize]: 1.8999e-07 [detach_backward]: 2.69999e-06 [pipeline_parallel_scheduler]: 1.45999e-06 [auto_monad_reorder]: 1.97e-05 [get_jit_bprop_graph]: 1.99e-06 [rewriter_after_jit_bprop_graph]: 6.23998e-06 [opt_after_jit_grad]: 0.00055428 [validate]: 5.944e-05 Sums bootstrap : 0.000457s : 0.30% type_inference : 0.123330s : 80.55% event_method : 0.000063s : 0.04% auto_monad : 0.000131s : 0.09% graph_reusing : 0.000008s : 0.01% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000047s : 0.03% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000010s : 0.01% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000066s : 0.04% insert-virtual-dataset : 0.000002s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000047s : 0.03% optimize.rewriter_before_opt_a : 0.000174s : 0.11% optimize.opt_a.expand_dump_flag : 0.000011s : 0.01% optimize.opt_a.switch_simplify : 0.000137s : 0.09% optimize.opt_a.loop_unroll : 0.000135s : 0.09% optimize.opt_a.a_1 : 0.003314s : 2.16% optimize.opt_a.with_stream_mark : 0.000078s : 0.05% optimize.opt_a.recompute_prepare : 0.000045s : 0.03% optimize.opt_a.updatestate_depend_eliminate : 0.000019s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000015s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000012s : 0.01% optimize.opt_a.parameter_eliminate : 0.000007s : 0.00% optimize.opt_a.a_2 : 0.000384s : 0.25% optimize.opt_a.accelerated_algorithm : 0.000069s : 0.05% optimize.opt_a.shard : 0.000008s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000009s : 0.01% optimize.opt_a.shard_inline : 0.000028s : 0.02% optimize.opt_a.merge_send_recv : 0.000039s : 0.03% optimize.opt_a.auto_parallel : 0.000039s : 0.03% optimize.opt_a.parallel : 0.000041s : 0.03% optimize.opt_a.flash_sp : 0.000018s : 0.01% optimize.opt_a.merge_comm : 0.000018s : 0.01% optimize.opt_a.allreduce_fusion : 0.000016s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000054s : 0.04% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000003s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000037s : 0.02% optimize.opt_a.virtual_dataset : 0.000028s : 0.02% optimize.opt_a.get_grad_eliminate_ : 0.000028s : 0.02% optimize.opt_a.virtual_output : 0.000028s : 0.02% optimize.opt_a.merge_forward : 0.000019s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000009s : 0.01% optimize.opt_a.offload_activation : 0.000043s : 0.03% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000058s : 0.04% optimize.opt_a.merge_recompute_call_nodes : 0.000005s : 0.00% optimize.opt_a.before_grad : 0.000051s : 0.03% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000020s : 0.01% optimize.opt_a.meta_fg_expand : 0.002005s : 1.31% optimize.opt_a.flash_sp_send_recv_attached : 0.000012s : 0.01% optimize.opt_a.receive_attached : 0.000008s : 0.00% optimize.opt_a.after_resolve : 0.000104s : 0.07% optimize.opt_a.a_after_grad : 0.000113s : 0.07% optimize.opt_a.renormalize : 0.018084s : 11.81% optimize.opt_a.add_forward_monad_depend : 0.000029s : 0.02% optimize.opt_a.auto_monad_grad : 0.000011s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000097s : 0.06% optimize.opt_a.cse : 0.000274s : 0.18% optimize.opt_a.a_3 : 0.000445s : 0.29% optimize.py_interpret_to_execute_after_opt_a : 0.000021s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000048s : 0.03% optimize.convert_after_rewriter : 0.000008s : 0.01% optimize.order_py_execute_after_rewriter : 0.000006s : 0.00% optimize.mutable_eliminate : 0.000811s : 0.53% optimize.opt_b.b_1 : 0.000133s : 0.09% optimize.opt_b.b_2 : 0.000009s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000011s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000032s : 0.02% optimize.optimize_parallel_all_gather_comm : 0.000023s : 0.02% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000038s : 0.02% optimize.loop_unroll : 0.000518s : 0.34% optimize.opt_after_cconv.c_1 : 0.000032s : 0.02% optimize.opt_after_cconv.parameter_eliminate : 0.000005s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000009s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000061s : 0.04% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000017s : 0.01% optimize.tuple_transform.d_1 : 0.000050s : 0.03% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000056s : 0.04% optimize.cse_after_recomputation.cse : 0.000013s : 0.01% optimize.environ_conv : 0.000011s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.00% optimize.bias_add_comm_swap : 0.000004s : 0.00% optimize.label_micro_interleaved_index : 0.000006s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000015s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000006s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000005s : 0.00% optimize.overlap_grad_flash_sp : 0.000023s : 0.02% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000003s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000002s : 0.00% optimize.symbol_engine_optimizer.build : 0.000013s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000012s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000014s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000009s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000001s : 0.00% auto_monad_reorder : 0.000020s : 0.01% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000006s : 0.00% opt_after_jit_grad : 0.000554s : 0.36% validate : 0.000059s : 0.04% Time group info: ------[substitution.] 0.000961 190 9.05% : 0.000087s : 6: substitution.arithmetic_simplify 0.22% : 0.000002s : 2: substitution.elim_not_effective 0.55% : 0.000005s : 6: substitution.float_depend_g_call 0.50% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.14% : 0.000001s : 2: substitution.fold_const_symbol 0.66% : 0.000006s : 4: substitution.graph_param_transform 0.32% : 0.000003s : 2: substitution.incorporate_call 0.23% : 0.000002s : 2: substitution.incorporate_call_switch 56.26% : 0.000540s : 16: substitution.inline 2.44% : 0.000023s : 2: substitution.inline_without_move 1.16% : 0.000011s : 14: substitution.j_node_and_user_rematch 2.34% : 0.000022s : 3: substitution.less_batch_normalization 1.49% : 0.000014s : 11: substitution.minmaximum_grad 0.74% : 0.000007s : 6: substitution.partial_eliminate 1.23% : 0.000012s : 14: substitution.remove_not_recompute_node 3.43% : 0.000033s : 10: substitution.replace_applicator 1.44% : 0.000014s : 12: substitution.replace_old_param 0.33% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.39% : 0.000033s : 11: substitution.tuple_list_convert_item_index_to_positive 1.55% : 0.000015s : 11: substitution.tuple_list_get_item_const_eliminator 2.14% : 0.000021s : 11: substitution.tuple_list_get_item_depend_reorder 8.37% : 0.000080s : 30: substitution.tuple_list_get_item_eliminator 2.03% : 0.000019s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.123221 2 98.41% : 0.121263s : 1: type_inference.infer 1.59% : 0.001958s : 1: type_inference.specialize ------[replace.] 0.000261 34 3.93% : 0.000010s : 2: replace.arithmetic_simplify 56.50% : 0.000148s : 16: replace.inline 39.56% : 0.000103s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.000622 34 8.16% : 0.000051s : 2: match.arithmetic_simplify 85.29% : 0.000531s : 16: match.inline 6.55% : 0.000041s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.000689 4766 1.16% : 0.000008s : 58: predicate.accumulaten_eliminater 0.28% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.51% : 0.000004s : 23: predicate.addn_check_dump 1.13% : 0.000008s : 58: predicate.addn_zero_filter 1.09% : 0.000008s : 58: predicate.adjust_all_reduce_mul_add 2.08% : 0.000014s : 83: predicate.arithmetic_simplify 1.19% : 0.000008s : 60: predicate.cast_eliminate 1.08% : 0.000007s : 55: predicate.check_bprop_eliminate 0.42% : 0.000003s : 23: predicate.compare_switch_simplify 0.06% : 0.000000s : 4: predicate.const_output_eliminate 0.46% : 0.000003s : 23: predicate.depend_value_elim 1.25% : 0.000009s : 60: predicate.dict_get_item_const_eliminator 1.22% : 0.000008s : 60: predicate.dict_get_item_eliminator 1.21% : 0.000008s : 60: predicate.dict_set_item_eliminator 0.30% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.08% : 0.000001s : 4: predicate.elim_not_effective 0.14% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.29% : 0.000009s : 64: predicate.environ_add_const_eliminate 1.16% : 0.000008s : 64: predicate.environ_get_add_eliminate 1.16% : 0.000008s : 64: predicate.environ_get_depend_swap 1.57% : 0.000011s : 87: predicate.environ_get_eliminate 1.15% : 0.000008s : 64: predicate.environ_get_set_eliminate 1.79% : 0.000012s : 92: predicate.exchange_switch_depend_value 2.47% : 0.000017s : 92: predicate.float_depend_g_call 0.41% : 0.000003s : 23: predicate.float_environ_get_switch 0.56% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.06% : 0.000000s : 4: predicate.fold_const_symbol 0.52% : 0.000004s : 23: predicate.get_grad_eliminate 0.11% : 0.000001s : 4: predicate.graph_param_transform 0.46% : 0.000003s : 23: predicate.incorporate_call 0.38% : 0.000003s : 23: predicate.incorporate_call_switch 5.58% : 0.000038s : 204: predicate.inline 1.30% : 0.000009s : 46: predicate.inline_without_move 0.24% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.80% : 0.000006s : 23: predicate.less_batch_normalization 1.72% : 0.000012s : 84: predicate.list_to_tuple_eliminator_ 2.64% : 0.000018s : 142: predicate.load_eliminater 0.43% : 0.000003s : 4: predicate.loop_unroll_after_grad 2.43% : 0.000017s : 128: predicate.loop_unroll_before_grad 1.47% : 0.000010s : 68: predicate.make_slice_get_slice_eliminator 0.45% : 0.000003s : 23: predicate.merge_addn 1.07% : 0.000007s : 55: predicate.micro_step_allgather_replace 1.04% : 0.000007s : 55: predicate.mini_step_allgather_replace 1.18% : 0.000008s : 60: predicate.minmaximum_grad 0.62% : 0.000004s : 4: predicate.mutable_eliminate 0.11% : 0.000001s : 4: predicate.opt_reshape 0.12% : 0.000001s : 4: predicate.parallel_virtual_node 2.22% : 0.000015s : 92: predicate.partial_defer_inline 1.71% : 0.000012s : 80: predicate.partial_eliminate 1.05% : 0.000007s : 58: predicate.print_const_string_wrapper 0.44% : 0.000003s : 23: predicate.reduce_all_const_elim 1.40% : 0.000010s : 60: predicate.reduce_eliminate 2.63% : 0.000018s : 142: predicate.redundant_stop_gradient_eliminater 0.29% : 0.000002s : 23: predicate.remove_not_recompute_node 1.88% : 0.000013s : 131: predicate.replace_applicator 0.64% : 0.000004s : 46: predicate.replace_old_param 0.10% : 0.000001s : 4: predicate.reset_defer_inline 1.17% : 0.000008s : 60: predicate.reshape_eliminate 1.13% : 0.000008s : 55: predicate.row_tensor_add_zeros_like 0.11% : 0.000001s : 4: predicate.row_tensor_eliminate 1.29% : 0.000009s : 55: predicate.same_eliminate 0.36% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.63% : 0.000004s : 23: predicate.shard_identity_eliminate 0.17% : 0.000001s : 8: predicate.special_op_eliminate 0.51% : 0.000003s : 23: predicate.specialize_transform 1.42% : 0.000010s : 55: predicate.split_environ_get_set_with_tuple_value 1.23% : 0.000008s : 46: predicate.stack_unstack_eliminate 0.11% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.93% : 0.000013s : 92: predicate.switch_defer_inline 2.89% : 0.000020s : 147: predicate.switch_layer_defer_inline 5.11% : 0.000035s : 247: predicate.switch_simplify 1.21% : 0.000008s : 60: predicate.tile_eliminate 1.16% : 0.000008s : 60: predicate.transpose_eliminate 1.39% : 0.000010s : 68: predicate.tuple_list_convert_item_index_to_positive 1.53% : 0.000011s : 68: predicate.tuple_list_get_item_const_eliminator 1.30% : 0.000009s : 68: predicate.tuple_list_get_item_depend_reorder 2.95% : 0.000020s : 107: predicate.tuple_list_get_item_eliminator 1.29% : 0.000009s : 68: predicate.tuple_list_get_set_item_eliminator 1.88% : 0.000013s : 91: predicate.tuple_list_set_item_eliminator 1.78% : 0.000012s : 84: predicate.tuple_to_list_eliminator_ 2.64% : 0.000018s : 142: predicate.updatestate_pure_node_eliminater 3.05% : 0.000021s : 165: predicate.updatestate_useless_node_eliminater 0.15% : 0.000001s : 4: predicate.value_based_eliminate 0.57% : 0.000004s : 23: predicate.virtual_dataset_eliminate 0.59% : 0.000004s : 23: predicate.virtual_output_eliminate 0.07% : 0.000000s : 4: predicate.virtual_view_grad_eliminate 0.14% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002438 37 59.27% : 0.001445s : 17: func_graph_cloner_run.FuncGraphClonerGraph 40.73% : 0.000993s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.215644 233 0.00% : 0.000004s : 1: ForceFp32Comm 1.95% : 0.004203s : 1: add_attr 1.94% : 0.004188s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.03% : 0.000060s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.06% : 0.000140s : 1: auto_monad 0.01% : 0.000024s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000007s : 1: bias_add_comm_swap 0.23% : 0.000490s : 1: bootstrap 0.02% : 0.000042s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.01% : 0.000019s : 1: control_data_broadcast_order 0.01% : 0.000012s : 1: convert_after_rewriter 0.01% : 0.000026s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.01% : 0.000014s : 1: environ_conv 0.03% : 0.000072s : 1: event_method 0.00% : 0.000006s : 1: full_micro_interleaved_order_control 0.00% : 0.000007s : 1: get_jit_bprop_graph 0.01% : 0.000013s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000007s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.25% : 0.000528s : 1: loop_unroll 0.00% : 0.000005s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.38% : 0.000824s : 1: mutable_eliminate 0.00% : 0.000008s : 1: offloading_packed_experts 0.01% : 0.000019s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000027s : 1: opt.transform.mutable_eliminate 2.24% : 0.004840s : 117: opt.transform.opt_a 0.01% : 0.000030s : 1: opt.transform.opt_after_cconv 0.01% : 0.000027s : 1: opt.transform.opt_after_jit_grad 0.05% : 0.000113s : 28: opt.transform.opt_b 0.03% : 0.000055s : 2: opt.transform.opt_trans_graph 0.02% : 0.000038s : 4: opt.transform.symbol_engine_opt 12.48% : 0.026904s : 1: opt_a 0.08% : 0.000163s : 1: opt_after_cconv 0.26% : 0.000567s : 1: opt_after_jit_grad 0.11% : 0.000247s : 1: opt_b 13.75% : 0.029656s : 1: optimize 0.01% : 0.000028s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000010s : 1: order_py_execute_after_rewriter 0.01% : 0.000028s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000008s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000006s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000009s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000006s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000007s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.03% : 0.000071s : 1: pre_auto_parallel 0.02% : 0.000051s : 1: py_interpret_to_execute 0.01% : 0.000025s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.01% : 0.000021s : 1: remove_dup_value 7.42% : 0.016011s : 2: renormalize.infer 0.95% : 0.002045s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000009s : 1: rewriter_after_jit_bprop_graph 0.03% : 0.000055s : 1: rewriter_after_opt_a 0.08% : 0.000179s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000006s : 1: split_matmul_comm_elemetwise 0.00% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.04% : 0.000095s : 1: symbol_engine_optimizer 0.04% : 0.000087s : 1: tuple_transform 57.21% : 0.123360s : 1: type_inference TotalTime = 0.19509, [21] [bootstrap]: 0.00043595 [type_inference]: 0.0164263 [event_method]: 5.684e-05 [auto_monad]: 0.0001247 [graph_reusing]: 8.38999e-06 [inline]: 2.33002e-06 [add_attr]: 0.159977, [1] [add_attr_with_inline]: 0.159964, [1] [Cycle 1]: 0.00011328, [2] [tag_attr]: 5.396e-05 [meta_addattr_fg_expand]: 9.96e-06 [parallel-infer-symbol]: 4.15999e-06 [pre_auto_parallel]: 6.827e-05 [insert-virtual-dataset]: 2.78e-06 [parallel-infer-symbol-second]: 8.50006e-07 [dataset_repeat_opt]: 2.30002e-06 [pipeline_split]: 1.71e-06 [optimize]: 0.0171789, [53] [py_interpret_to_execute]: 5.028e-05 [rewriter_before_opt_a]: 0.00018019 [opt_a]: 0.0146946, [3] [Cycle 1]: 0.0105901, [45] [expand_dump_flag]: 4.11001e-06 [switch_simplify]: 8.046e-05 [loop_unroll]: 6.34e-05 [a_1]: 0.00163318 [with_stream_mark]: 4.258e-05 [recompute_prepare]: 3.428e-05 [updatestate_depend_eliminate]: 1.093e-05 [updatestate_assign_eliminate]: 7.73999e-06 [updatestate_loads_eliminate]: 7.32997e-06 [parameter_eliminate]: 3.44001e-06 [a_2]: 0.00025196 [accelerated_algorithm]: 3.964e-05 [shard]: 2.19999e-06 [meta_shard_fg_expand]: 5.04e-06 [shard_inline]: 1.722e-05 [merge_send_recv]: 2.088e-05 [auto_parallel]: 1.758e-05 [parallel]: 0.00147558 [flash_sp]: 2.876e-05 [merge_comm]: 3.006e-05 [allreduce_fusion]: 1.019e-05 [matmul_add_comm_reduction]: 3.925e-05 [allreduce_slice_to_reducescatter]: 1.21997e-06 [virtual_shard_identity]: 4.163e-05 [virtual_dataset]: 1.7e-05 [get_grad_eliminate_]: 1.587e-05 [virtual_output]: 1.634e-05 [merge_forward]: 1.28e-05 [cell_reuse_recompute_pass]: 3.28e-06 [offload_activation]: 2.078e-05 [cell_reuse_handle_not_recompute_node_pass]: 4.072e-05 [merge_recompute_call_nodes]: 1.49e-06 [before_grad]: 2.943e-05 [set_forward_comm_id_for_comm_node_pass]: 1.058e-05 [meta_fg_expand]: 0.00206893 [flash_sp_send_recv_attached]: 6.43e-06 [receive_attached]: 2.80002e-06 [after_resolve]: 7.446e-05 [a_after_grad]: 9.111e-05 [renormalize]: 0.00319709 [add_forward_monad_depend]: 1.37e-05 [auto_monad_grad]: 6.79999e-06 [auto_monad_eliminator]: 6.345e-05 [cse]: 0.00022675 [a_3]: 0.0003661 [Cycle 2]: 0.00342788, [45] [expand_dump_flag]: 3.57002e-06 [switch_simplify]: 4.653e-05 [loop_unroll]: 4.179e-05 [a_1]: 0.00165615 [with_stream_mark]: 2.76e-05 [recompute_prepare]: 1.061e-05 [updatestate_depend_eliminate]: 4.89e-06 [updatestate_assign_eliminate]: 3.76001e-06 [updatestate_loads_eliminate]: 2.91e-06 [parameter_eliminate]: 2.07999e-06 [a_2]: 7.122e-05 [accelerated_algorithm]: 1.466e-05 [shard]: 2.64999e-06 [meta_shard_fg_expand]: 2.78998e-06 [shard_inline]: 6.23998e-06 [merge_send_recv]: 1.021e-05 [auto_parallel]: 1.114e-05 [parallel]: 9.89001e-06 [flash_sp]: 3.93001e-06 [merge_comm]: 3.78001e-06 [allreduce_fusion]: 3.56999e-06 [matmul_add_comm_reduction]: 9.54e-06 [allreduce_slice_to_reducescatter]: 7.00005e-07 [virtual_shard_identity]: 8.64e-06 [virtual_dataset]: 6.15002e-06 [get_grad_eliminate_]: 5.81e-06 [virtual_output]: 5.55001e-06 [merge_forward]: 4.60001e-06 [cell_reuse_recompute_pass]: 2.13998e-06 [offload_activation]: 1.049e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.303e-05 [merge_recompute_call_nodes]: 1.60999e-06 [before_grad]: 9.84999e-06 [set_forward_comm_id_for_comm_node_pass]: 4.22e-06 [meta_fg_expand]: 0.00014675 [flash_sp_send_recv_attached]: 1.92001e-06 [receive_attached]: 2.91999e-06 [after_resolve]: 1.853e-05 [a_after_grad]: 9.55001e-06 [renormalize]: 0.00084956 [add_forward_monad_depend]: 6.66e-06 [auto_monad_grad]: 2.59001e-06 [auto_monad_eliminator]: 1.616e-05 [cse]: 3.062e-05 [a_3]: 4.692e-05 [Cycle 3]: 0.0006576, [45] [expand_dump_flag]: 2.31e-06 [switch_simplify]: 7.63999e-06 [loop_unroll]: 5.91998e-06 [a_1]: 0.00012607 [with_stream_mark]: 1.291e-05 [recompute_prepare]: 5.71003e-06 [updatestate_depend_eliminate]: 3.55e-06 [updatestate_assign_eliminate]: 2.84001e-06 [updatestate_loads_eliminate]: 2.42001e-06 [parameter_eliminate]: 9.79984e-07 [a_2]: 6.387e-05 [accelerated_algorithm]: 1.321e-05 [shard]: 1.46998e-06 [meta_shard_fg_expand]: 1.51998e-06 [shard_inline]: 5.66e-06 [merge_send_recv]: 6.21998e-06 [auto_parallel]: 7.28e-06 [parallel]: 5.96998e-06 [flash_sp]: 1.87999e-06 [merge_comm]: 3.61001e-06 [allreduce_fusion]: 3.16001e-06 [matmul_add_comm_reduction]: 6.88e-06 [allreduce_slice_to_reducescatter]: 5.09986e-07 [virtual_shard_identity]: 7.41999e-06 [virtual_dataset]: 5.27001e-06 [get_grad_eliminate_]: 4.99998e-06 [virtual_output]: 5.49e-06 [merge_forward]: 3.65e-06 [cell_reuse_recompute_pass]: 2.34999e-06 [offload_activation]: 9.72999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.198e-05 [merge_recompute_call_nodes]: 1.46002e-06 [before_grad]: 9.14e-06 [set_forward_comm_id_for_comm_node_pass]: 3.5e-06 [meta_fg_expand]: 2.23998e-06 [flash_sp_send_recv_attached]: 1.78002e-06 [receive_attached]: 1.92999e-06 [after_resolve]: 1.023e-05 [a_after_grad]: 8.28999e-06 [renormalize]: 6.99947e-08 [add_forward_monad_depend]: 2.04999e-06 [auto_monad_grad]: 1.59e-06 [auto_monad_eliminator]: 8.99e-06 [cse]: 1.572e-05 [a_3]: 3.233e-05 [py_interpret_to_execute_after_opt_a]: 1.483e-05 [slice_cell_reuse_recomputed_activation]: 2.19999e-06 [rewriter_after_opt_a]: 4.313e-05 [convert_after_rewriter]: 7.58001e-06 [order_py_execute_after_rewriter]: 5.30999e-06 [mutable_eliminate]: 0.00072582 [opt_b]: 0.00021654, [1] [Cycle 1]: 0.00020851, [7] [b_1]: 0.00011259 [b_2]: 7.02002e-06 [updatestate_depend_eliminate]: 9.02999e-06 [updatestate_assign_eliminate]: 3.01001e-06 [updatestate_loads_eliminate]: 2.62001e-06 [renormalize]: 7.2e-07 [cse]: 2.634e-05 [optimize_parallel_all_gather_comm]: 2.065e-05 [overlap_param_gather]: 2.29001e-06 [cconv]: 3.385e-05 [loop_unroll]: 0.00046577 [opt_after_cconv]: 0.00011137, [1] [Cycle 1]: 0.00010397, [7] [c_1]: 3.215e-05 [parameter_eliminate]: 4.84e-06 [updatestate_depend_eliminate]: 6.73e-06 [updatestate_assign_eliminate]: 2.94001e-06 [updatestate_loads_eliminate]: 2.25002e-06 [cse]: 2.021e-05 [renormalize]: 1.59984e-07 [remove_dup_value]: 1.582e-05 [tuple_transform]: 7.581e-05, [1] [Cycle 1]: 7.102e-05, [4] [d_1]: 4.399e-05 [none_parameter_eliminate]: 1.69e-06 [renormalize]: 1.30007e-07 [switch_simplify]: 6.35002e-06 [partial_unused_args_eliminate]: 1.84e-06 [add_recomputation]: 5.061e-05 [cse_after_recomputation]: 2.237e-05, [1] [Cycle 1]: 1.715e-05, [1] [cse]: 1.122e-05 [environ_conv]: 9.51003e-06 [swap_dp_allreduce_reducescatter]: 5.87999e-06 [bias_add_comm_swap]: 3.05002e-06 [label_micro_interleaved_index]: 5.13002e-06 [label_fine_grained_interleaved_index]: 3.16999e-06 [merge_cast_opt]: 1.32999e-06 [slice_recompute_activation]: 2.04e-06 [micro_interleaved_order_control]: 2.28998e-06 [assign_add_opt]: 1.25999e-06 [ForceFp32Comm]: 9.60019e-07 [remove_cast_before_assign_add]: 1.25001e-06 [full_micro_interleaved_order_control]: 2.29001e-06 [reorder_send_recv_between_fp_bp]: 2.87002e-06 [comm_op_add_attrs]: 1.11997e-06 [add_comm_op_reuse_tag]: 9.80013e-07 [interleave_split_concat_branches]: 1.38002e-06 [interleave_parallel_branches]: 1.04e-06 [overlap_opt_shard_in_pipeline]: 1.22999e-06 [overlap_opt_shard_grad_in_pipeline]: 1.67999e-06 [control_data_broadcast_order]: 1.497e-05 [grouped_pairwise_exchange_alltoall]: 2.29001e-06 [offloading_packed_experts]: 4.06001e-06 [overlap_recompute_and_grad_model_parallel]: 4.53999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.24998e-06 [overlap_recompute_allgather_and_fa_grad]: 1.30999e-06 [overlap_recompute_comm]: 2.34999e-06 [overlap_grad_ring_attention]: 4.13001e-06 [overlap_grad_flash_sp]: 2.165e-05 [begin_end_overlap_inline]: 4.7998e-07 [split_matmul_comm_elemetwise]: 2.21e-06 [split_layernorm_comm]: 1.91e-06 [handle_group_info]: 1.21002e-06 [symbol_engine_optimizer]: 8.968e-05, [1] [Cycle 1]: 8.437e-05, [6] [build]: 1.282e-05 [elim_shapecalc]: 1.056e-05 [elim_not_effective]: 1.314e-05 [opt_reshape]: 6.76e-06 [fold_const_symbol]: 9.37999e-06 [renormalize]: 2.10013e-07 [detach_backward]: 2.61999e-06 [pipeline_parallel_scheduler]: 1.71998e-06 [auto_monad_reorder]: 1.791e-05 [get_jit_bprop_graph]: 1.92001e-06 [rewriter_after_jit_bprop_graph]: 5.89999e-06 [opt_after_jit_grad]: 0.00053125 [validate]: 5.331e-05 Sums bootstrap : 0.000436s : 1.30% type_inference : 0.016426s : 48.83% event_method : 0.000057s : 0.17% auto_monad : 0.000125s : 0.37% graph_reusing : 0.000008s : 0.02% inline : 0.000002s : 0.01% add_attr.add_attr_with_inline.tag_attr : 0.000054s : 0.16% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000010s : 0.03% parallel-infer-symbol : 0.000004s : 0.01% pre_auto_parallel : 0.000068s : 0.20% insert-virtual-dataset : 0.000003s : 0.01% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.01% pipeline_split : 0.000002s : 0.01% optimize.py_interpret_to_execute : 0.000050s : 0.15% optimize.rewriter_before_opt_a : 0.000180s : 0.54% optimize.opt_a.expand_dump_flag : 0.000010s : 0.03% optimize.opt_a.switch_simplify : 0.000135s : 0.40% optimize.opt_a.loop_unroll : 0.000111s : 0.33% optimize.opt_a.a_1 : 0.003415s : 10.15% optimize.opt_a.with_stream_mark : 0.000083s : 0.25% optimize.opt_a.recompute_prepare : 0.000051s : 0.15% optimize.opt_a.updatestate_depend_eliminate : 0.000019s : 0.06% optimize.opt_a.updatestate_assign_eliminate : 0.000014s : 0.04% optimize.opt_a.updatestate_loads_eliminate : 0.000013s : 0.04% optimize.opt_a.parameter_eliminate : 0.000006s : 0.02% optimize.opt_a.a_2 : 0.000387s : 1.15% optimize.opt_a.accelerated_algorithm : 0.000068s : 0.20% optimize.opt_a.shard : 0.000006s : 0.02% optimize.opt_a.meta_shard_fg_expand : 0.000009s : 0.03% optimize.opt_a.shard_inline : 0.000029s : 0.09% optimize.opt_a.merge_send_recv : 0.000037s : 0.11% optimize.opt_a.auto_parallel : 0.000036s : 0.11% optimize.opt_a.parallel : 0.001491s : 4.43% optimize.opt_a.flash_sp : 0.000035s : 0.10% optimize.opt_a.merge_comm : 0.000037s : 0.11% optimize.opt_a.allreduce_fusion : 0.000017s : 0.05% optimize.opt_a.matmul_add_comm_reduction : 0.000056s : 0.17% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.01% optimize.opt_a.virtual_shard_identity : 0.000058s : 0.17% optimize.opt_a.virtual_dataset : 0.000028s : 0.08% optimize.opt_a.get_grad_eliminate_ : 0.000027s : 0.08% optimize.opt_a.virtual_output : 0.000027s : 0.08% optimize.opt_a.merge_forward : 0.000021s : 0.06% optimize.opt_a.cell_reuse_recompute_pass : 0.000008s : 0.02% optimize.opt_a.offload_activation : 0.000041s : 0.12% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000066s : 0.20% optimize.opt_a.merge_recompute_call_nodes : 0.000005s : 0.01% optimize.opt_a.before_grad : 0.000048s : 0.14% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000018s : 0.05% optimize.opt_a.meta_fg_expand : 0.002218s : 6.59% optimize.opt_a.flash_sp_send_recv_attached : 0.000010s : 0.03% optimize.opt_a.receive_attached : 0.000008s : 0.02% optimize.opt_a.after_resolve : 0.000103s : 0.31% optimize.opt_a.a_after_grad : 0.000109s : 0.32% optimize.opt_a.renormalize : 0.004047s : 12.03% optimize.opt_a.add_forward_monad_depend : 0.000022s : 0.07% optimize.opt_a.auto_monad_grad : 0.000011s : 0.03% optimize.opt_a.auto_monad_eliminator : 0.000089s : 0.26% optimize.opt_a.cse : 0.000273s : 0.81% optimize.opt_a.a_3 : 0.000445s : 1.32% optimize.py_interpret_to_execute_after_opt_a : 0.000015s : 0.04% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.01% optimize.rewriter_after_opt_a : 0.000043s : 0.13% optimize.convert_after_rewriter : 0.000008s : 0.02% optimize.order_py_execute_after_rewriter : 0.000005s : 0.02% optimize.mutable_eliminate : 0.000726s : 2.16% optimize.opt_b.b_1 : 0.000113s : 0.33% optimize.opt_b.b_2 : 0.000007s : 0.02% optimize.opt_b.updatestate_depend_eliminate : 0.000009s : 0.03% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.01% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000026s : 0.08% optimize.optimize_parallel_all_gather_comm : 0.000021s : 0.06% optimize.overlap_param_gather : 0.000002s : 0.01% optimize.cconv : 0.000034s : 0.10% optimize.loop_unroll : 0.000466s : 1.38% optimize.opt_after_cconv.c_1 : 0.000032s : 0.10% optimize.opt_after_cconv.parameter_eliminate : 0.000005s : 0.01% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000007s : 0.02% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.01% optimize.opt_after_cconv.cse : 0.000020s : 0.06% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000016s : 0.05% optimize.tuple_transform.d_1 : 0.000044s : 0.13% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.01% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000006s : 0.02% optimize.partial_unused_args_eliminate : 0.000002s : 0.01% optimize.add_recomputation : 0.000051s : 0.15% optimize.cse_after_recomputation.cse : 0.000011s : 0.03% optimize.environ_conv : 0.000010s : 0.03% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.02% optimize.bias_add_comm_swap : 0.000003s : 0.01% optimize.label_micro_interleaved_index : 0.000005s : 0.02% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.01% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.01% optimize.micro_interleaved_order_control : 0.000002s : 0.01% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.01% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.01% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000015s : 0.04% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.01% optimize.offloading_packed_experts : 0.000004s : 0.01% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.01% optimize.overlap_grad_ring_attention : 0.000004s : 0.01% optimize.overlap_grad_flash_sp : 0.000022s : 0.06% optimize.begin_end_overlap_inline : 0.000000s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.01% optimize.split_layernorm_comm : 0.000002s : 0.01% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000013s : 0.04% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000011s : 0.03% optimize.symbol_engine_optimizer.elim_not_effective : 0.000013s : 0.04% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.02% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000009s : 0.03% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.01% pipeline_parallel_scheduler : 0.000002s : 0.01% auto_monad_reorder : 0.000018s : 0.05% get_jit_bprop_graph : 0.000002s : 0.01% rewriter_after_jit_bprop_graph : 0.000006s : 0.02% opt_after_jit_grad : 0.000531s : 1.58% validate : 0.000053s : 0.16% Time group info: ------[substitution.] 0.001038 190 9.13% : 0.000095s : 6: substitution.arithmetic_simplify 0.19% : 0.000002s : 2: substitution.elim_not_effective 0.59% : 0.000006s : 6: substitution.float_depend_g_call 0.52% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.18% : 0.000002s : 2: substitution.fold_const_symbol 0.60% : 0.000006s : 4: substitution.graph_param_transform 0.35% : 0.000004s : 2: substitution.incorporate_call 0.22% : 0.000002s : 2: substitution.incorporate_call_switch 57.56% : 0.000597s : 16: substitution.inline 2.35% : 0.000024s : 2: substitution.inline_without_move 1.06% : 0.000011s : 14: substitution.j_node_and_user_rematch 2.15% : 0.000022s : 3: substitution.less_batch_normalization 1.52% : 0.000016s : 11: substitution.minmaximum_grad 0.68% : 0.000007s : 6: substitution.partial_eliminate 1.17% : 0.000012s : 14: substitution.remove_not_recompute_node 3.17% : 0.000033s : 10: substitution.replace_applicator 1.37% : 0.000014s : 12: substitution.replace_old_param 0.51% : 0.000005s : 1: substitution.set_cell_output_no_recompute 3.12% : 0.000032s : 11: substitution.tuple_list_convert_item_index_to_positive 1.53% : 0.000016s : 11: substitution.tuple_list_get_item_const_eliminator 2.00% : 0.000021s : 11: substitution.tuple_list_get_item_depend_reorder 8.06% : 0.000084s : 30: substitution.tuple_list_get_item_eliminator 1.99% : 0.000021s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.016331 2 88.60% : 0.014469s : 1: type_inference.infer 11.40% : 0.001862s : 1: type_inference.specialize ------[replace.] 0.000287 34 5.02% : 0.000014s : 2: replace.arithmetic_simplify 56.60% : 0.000162s : 16: replace.inline 38.38% : 0.000110s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.000686 34 8.24% : 0.000057s : 2: match.arithmetic_simplify 85.53% : 0.000587s : 16: match.inline 6.22% : 0.000043s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.000683 4766 1.08% : 0.000007s : 58: predicate.accumulaten_eliminater 0.33% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.43% : 0.000003s : 23: predicate.addn_check_dump 1.07% : 0.000007s : 58: predicate.addn_zero_filter 1.04% : 0.000007s : 58: predicate.adjust_all_reduce_mul_add 2.16% : 0.000015s : 83: predicate.arithmetic_simplify 1.18% : 0.000008s : 60: predicate.cast_eliminate 1.08% : 0.000007s : 55: predicate.check_bprop_eliminate 0.42% : 0.000003s : 23: predicate.compare_switch_simplify 0.06% : 0.000000s : 4: predicate.const_output_eliminate 0.45% : 0.000003s : 23: predicate.depend_value_elim 1.27% : 0.000009s : 60: predicate.dict_get_item_const_eliminator 1.24% : 0.000008s : 60: predicate.dict_get_item_eliminator 1.19% : 0.000008s : 60: predicate.dict_set_item_eliminator 0.46% : 0.000003s : 8: predicate.dumpgradient_eliminate 0.09% : 0.000001s : 4: predicate.elim_not_effective 0.19% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.17% : 0.000008s : 64: predicate.environ_add_const_eliminate 1.14% : 0.000008s : 64: predicate.environ_get_add_eliminate 1.15% : 0.000008s : 64: predicate.environ_get_depend_swap 1.61% : 0.000011s : 87: predicate.environ_get_eliminate 1.21% : 0.000008s : 64: predicate.environ_get_set_eliminate 1.80% : 0.000012s : 92: predicate.exchange_switch_depend_value 2.58% : 0.000018s : 92: predicate.float_depend_g_call 0.43% : 0.000003s : 23: predicate.float_environ_get_switch 0.56% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.05% : 0.000000s : 4: predicate.fold_const_symbol 0.52% : 0.000004s : 23: predicate.get_grad_eliminate 0.08% : 0.000001s : 4: predicate.graph_param_transform 0.47% : 0.000003s : 23: predicate.incorporate_call 0.38% : 0.000003s : 23: predicate.incorporate_call_switch 5.60% : 0.000038s : 204: predicate.inline 1.36% : 0.000009s : 46: predicate.inline_without_move 0.25% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.84% : 0.000006s : 23: predicate.less_batch_normalization 1.58% : 0.000011s : 84: predicate.list_to_tuple_eliminator_ 2.65% : 0.000018s : 142: predicate.load_eliminater 0.35% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.43% : 0.000017s : 128: predicate.loop_unroll_before_grad 1.30% : 0.000009s : 68: predicate.make_slice_get_slice_eliminator 0.46% : 0.000003s : 23: predicate.merge_addn 1.02% : 0.000007s : 55: predicate.micro_step_allgather_replace 1.05% : 0.000007s : 55: predicate.mini_step_allgather_replace 1.14% : 0.000008s : 60: predicate.minmaximum_grad 0.54% : 0.000004s : 4: predicate.mutable_eliminate 0.10% : 0.000001s : 4: predicate.opt_reshape 0.14% : 0.000001s : 4: predicate.parallel_virtual_node 2.46% : 0.000017s : 92: predicate.partial_defer_inline 1.70% : 0.000012s : 80: predicate.partial_eliminate 1.15% : 0.000008s : 58: predicate.print_const_string_wrapper 0.53% : 0.000004s : 23: predicate.reduce_all_const_elim 1.38% : 0.000009s : 60: predicate.reduce_eliminate 2.65% : 0.000018s : 142: predicate.redundant_stop_gradient_eliminater 0.33% : 0.000002s : 23: predicate.remove_not_recompute_node 1.87% : 0.000013s : 131: predicate.replace_applicator 0.62% : 0.000004s : 46: predicate.replace_old_param 0.11% : 0.000001s : 4: predicate.reset_defer_inline 1.20% : 0.000008s : 60: predicate.reshape_eliminate 1.14% : 0.000008s : 55: predicate.row_tensor_add_zeros_like 0.11% : 0.000001s : 4: predicate.row_tensor_eliminate 1.25% : 0.000009s : 55: predicate.same_eliminate 0.37% : 0.000003s : 23: predicate.set_cell_output_no_recompute 0.81% : 0.000006s : 23: predicate.shard_identity_eliminate 0.24% : 0.000002s : 8: predicate.special_op_eliminate 0.53% : 0.000004s : 23: predicate.specialize_transform 1.30% : 0.000009s : 55: predicate.split_environ_get_set_with_tuple_value 1.19% : 0.000008s : 46: predicate.stack_unstack_eliminate 0.14% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.96% : 0.000013s : 92: predicate.switch_defer_inline 2.90% : 0.000020s : 147: predicate.switch_layer_defer_inline 5.10% : 0.000035s : 247: predicate.switch_simplify 1.12% : 0.000008s : 60: predicate.tile_eliminate 1.08% : 0.000007s : 60: predicate.transpose_eliminate 1.38% : 0.000009s : 68: predicate.tuple_list_convert_item_index_to_positive 1.52% : 0.000010s : 68: predicate.tuple_list_get_item_const_eliminator 1.30% : 0.000009s : 68: predicate.tuple_list_get_item_depend_reorder 3.04% : 0.000021s : 107: predicate.tuple_list_get_item_eliminator 1.39% : 0.000010s : 68: predicate.tuple_list_get_set_item_eliminator 1.81% : 0.000012s : 91: predicate.tuple_list_set_item_eliminator 1.60% : 0.000011s : 84: predicate.tuple_to_list_eliminator_ 2.64% : 0.000018s : 142: predicate.updatestate_pure_node_eliminater 3.01% : 0.000021s : 165: predicate.updatestate_useless_node_eliminater 0.11% : 0.000001s : 4: predicate.value_based_eliminate 0.52% : 0.000004s : 23: predicate.virtual_dataset_eliminate 0.47% : 0.000003s : 23: predicate.virtual_output_eliminate 0.07% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.15% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002151 37 60.96% : 0.001311s : 17: func_graph_cloner_run.FuncGraphClonerGraph 39.04% : 0.000840s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.381340 233 0.00% : 0.000004s : 1: ForceFp32Comm 41.95% : 0.159984s : 1: add_attr 41.95% : 0.159968s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.01% : 0.000055s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.03% : 0.000132s : 1: auto_monad 0.01% : 0.000022s : 1: auto_monad_reorder 0.00% : 0.000003s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 0.12% : 0.000469s : 1: bootstrap 0.01% : 0.000038s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000019s : 1: control_data_broadcast_order 0.00% : 0.000011s : 1: convert_after_rewriter 0.01% : 0.000025s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000013s : 1: environ_conv 0.02% : 0.000066s : 1: event_method 0.00% : 0.000006s : 1: full_micro_interleaved_order_control 0.00% : 0.000006s : 1: get_jit_bprop_graph 0.00% : 0.000013s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.12% : 0.000474s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.19% : 0.000737s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.00% : 0.000014s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000020s : 1: opt.transform.mutable_eliminate 1.30% : 0.004950s : 117: opt.transform.opt_a 0.01% : 0.000030s : 1: opt.transform.opt_after_cconv 0.01% : 0.000026s : 1: opt.transform.opt_after_jit_grad 0.02% : 0.000092s : 28: opt.transform.opt_b 0.01% : 0.000048s : 2: opt.transform.opt_trans_graph 0.01% : 0.000036s : 4: opt.transform.symbol_engine_opt 3.85% : 0.014698s : 1: opt_a 0.03% : 0.000116s : 1: opt_after_cconv 0.14% : 0.000542s : 1: opt_after_jit_grad 0.06% : 0.000220s : 1: opt_b 4.51% : 0.017186s : 1: optimize 0.01% : 0.000024s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000008s : 1: order_py_execute_after_rewriter 0.01% : 0.000026s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000004s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000007s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000006s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.02% : 0.000074s : 1: pre_auto_parallel 0.01% : 0.000055s : 1: py_interpret_to_execute 0.00% : 0.000019s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.01% : 0.000019s : 1: remove_dup_value 0.58% : 0.002197s : 2: renormalize.infer 0.48% : 0.001829s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000009s : 1: rewriter_after_jit_bprop_graph 0.01% : 0.000048s : 1: rewriter_after_opt_a 0.05% : 0.000185s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.02% : 0.000092s : 1: symbol_engine_optimizer 0.02% : 0.000079s : 1: tuple_transform 4.31% : 0.016454s : 1: type_inference TotalTime = 0.734865, [21] [bootstrap]: 0.00074611 [type_inference]: 0.382925 [event_method]: 6.366e-05 [auto_monad]: 0.00012805 [graph_reusing]: 8.72e-06 [inline]: 2.99999e-06 [add_attr]: 0.00459269, [1] [add_attr_with_inline]: 0.00458091, [1] [Cycle 1]: 0.00010755, [2] [tag_attr]: 5.025e-05 [meta_addattr_fg_expand]: 9.94001e-06 [parallel-infer-symbol]: 4.03999e-06 [pre_auto_parallel]: 6.531e-05 [insert-virtual-dataset]: 2.66e-06 [parallel-infer-symbol-second]: 9.80013e-07 [dataset_repeat_opt]: 1.86e-06 [pipeline_split]: 1.79e-06 [optimize]: 0.345547, [53] [py_interpret_to_execute]: 4.498e-05 [rewriter_before_opt_a]: 0.00019067 [opt_a]: 0.342855, [3] [Cycle 1]: 0.00874643, [45] [expand_dump_flag]: 3.76999e-06 [switch_simplify]: 8.467e-05 [loop_unroll]: 6.718e-05 [a_1]: 0.00164761 [with_stream_mark]: 3.251e-05 [recompute_prepare]: 2.436e-05 [updatestate_depend_eliminate]: 1.026e-05 [updatestate_assign_eliminate]: 8.02e-06 [updatestate_loads_eliminate]: 7.63001e-06 [parameter_eliminate]: 3.14999e-06 [a_2]: 0.00028443 [accelerated_algorithm]: 3.489e-05 [shard]: 1.99e-06 [meta_shard_fg_expand]: 4.51002e-06 [shard_inline]: 1.586e-05 [merge_send_recv]: 1.79e-05 [auto_parallel]: 1.404e-05 [parallel]: 3.023e-05 [flash_sp]: 1.075e-05 [merge_comm]: 1.034e-05 [allreduce_fusion]: 9.36e-06 [matmul_add_comm_reduction]: 3.208e-05 [allreduce_slice_to_reducescatter]: 8.2e-07 [virtual_shard_identity]: 1.899e-05 [virtual_dataset]: 1.638e-05 [get_grad_eliminate_]: 1.649e-05 [virtual_output]: 1.67e-05 [merge_forward]: 1.027e-05 [cell_reuse_recompute_pass]: 1.79e-06 [offload_activation]: 1.931e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.235e-05 [merge_recompute_call_nodes]: 1.62001e-06 [before_grad]: 2.87e-05 [set_forward_comm_id_for_comm_node_pass]: 1.017e-05 [meta_fg_expand]: 0.00186101 [flash_sp_send_recv_attached]: 6.59999e-06 [receive_attached]: 2.46e-06 [after_resolve]: 6.475e-05 [a_after_grad]: 8.401e-05 [renormalize]: 0.00324572 [add_forward_monad_depend]: 1.078e-05 [auto_monad_grad]: 6.84001e-06 [auto_monad_eliminator]: 5.82e-05 [cse]: 0.00021801 [a_3]: 0.00033266 [Cycle 2]: 0.00313797, [45] [expand_dump_flag]: 2.44999e-06 [switch_simplify]: 4.799e-05 [loop_unroll]: 4.157e-05 [a_1]: 0.0015754 [with_stream_mark]: 1.607e-05 [recompute_prepare]: 9.27001e-06 [updatestate_depend_eliminate]: 3.8e-06 [updatestate_assign_eliminate]: 2.91e-06 [updatestate_loads_eliminate]: 2.54999e-06 [parameter_eliminate]: 2.01e-06 [a_2]: 6.877e-05 [accelerated_algorithm]: 1.089e-05 [shard]: 1.77999e-06 [meta_shard_fg_expand]: 1.86e-06 [shard_inline]: 5.75001e-06 [merge_send_recv]: 6.92002e-06 [auto_parallel]: 8.95999e-06 [parallel]: 7.73999e-06 [flash_sp]: 3.73999e-06 [merge_comm]: 3.38999e-06 [allreduce_fusion]: 3.89002e-06 [matmul_add_comm_reduction]: 7.76001e-06 [allreduce_slice_to_reducescatter]: 5.90022e-07 [virtual_shard_identity]: 7.34002e-06 [virtual_dataset]: 5.49998e-06 [get_grad_eliminate_]: 5.47001e-06 [virtual_output]: 5.62001e-06 [merge_forward]: 3.01001e-06 [cell_reuse_recompute_pass]: 9.79984e-07 [offload_activation]: 9.04e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.142e-05 [merge_recompute_call_nodes]: 1.14e-06 [before_grad]: 1.019e-05 [set_forward_comm_id_for_comm_node_pass]: 3.93001e-06 [meta_fg_expand]: 0.00013526 [flash_sp_send_recv_attached]: 1.68002e-06 [receive_attached]: 2.20002e-06 [after_resolve]: 1.297e-05 [a_after_grad]: 8.93002e-06 [renormalize]: 0.0007437 [add_forward_monad_depend]: 5.10999e-06 [auto_monad_grad]: 1.86e-06 [auto_monad_eliminator]: 1.253e-05 [cse]: 2.093e-05 [a_3]: 4.403e-05 [Cycle 3]: 0.330953, [45] [expand_dump_flag]: 1.54e-06 [switch_simplify]: 6.95998e-06 [loop_unroll]: 6.95998e-06 [a_1]: 0.00012143 [with_stream_mark]: 9.72001e-06 [recompute_prepare]: 5.99e-06 [updatestate_depend_eliminate]: 3.38e-06 [updatestate_assign_eliminate]: 2.44999e-06 [updatestate_loads_eliminate]: 2.19001e-06 [parameter_eliminate]: 1.22e-06 [a_2]: 0.330145 [accelerated_algorithm]: 5.527e-05 [shard]: 3.86001e-06 [meta_shard_fg_expand]: 6.79001e-06 [shard_inline]: 1.18e-05 [merge_send_recv]: 2.078e-05 [auto_parallel]: 1.592e-05 [parallel]: 1.28e-05 [flash_sp]: 2.66e-06 [merge_comm]: 3.21001e-06 [allreduce_fusion]: 3.56001e-06 [matmul_add_comm_reduction]: 1.354e-05 [allreduce_slice_to_reducescatter]: 8.50006e-07 [virtual_shard_identity]: 1.023e-05 [virtual_dataset]: 7.43e-06 [get_grad_eliminate_]: 5.80002e-06 [virtual_output]: 6.59999e-06 [merge_forward]: 4.62998e-06 [cell_reuse_recompute_pass]: 4.08001e-06 [offload_activation]: 1.274e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.593e-05 [merge_recompute_call_nodes]: 1.70001e-06 [before_grad]: 1.064e-05 [set_forward_comm_id_for_comm_node_pass]: 4.36002e-06 [meta_fg_expand]: 2.79001e-06 [flash_sp_send_recv_attached]: 2.69001e-06 [receive_attached]: 2.80002e-06 [after_resolve]: 1.328e-05 [a_after_grad]: 9.44998e-06 [renormalize]: 9.00181e-08 [add_forward_monad_depend]: 7.22002e-06 [auto_monad_grad]: 3.51001e-06 [auto_monad_eliminator]: 1.918e-05 [cse]: 3.829e-05 [a_3]: 4.051e-05 [py_interpret_to_execute_after_opt_a]: 2.068e-05 [slice_cell_reuse_recomputed_activation]: 2.26e-06 [rewriter_after_opt_a]: 4.677e-05 [convert_after_rewriter]: 6.74001e-06 [order_py_execute_after_rewriter]: 5.27001e-06 [mutable_eliminate]: 0.00077517 [opt_b]: 0.00020598, [1] [Cycle 1]: 0.00019594, [7] [b_1]: 0.00011915 [b_2]: 8.39998e-06 [updatestate_depend_eliminate]: 6.58e-06 [updatestate_assign_eliminate]: 3.34001e-06 [updatestate_loads_eliminate]: 2.70002e-06 [renormalize]: 5.8001e-07 [cse]: 2.005e-05 [optimize_parallel_all_gather_comm]: 1.88e-05 [overlap_param_gather]: 9.39e-06 [cconv]: 5.023e-05 [loop_unroll]: 0.00052656 [opt_after_cconv]: 0.00012178, [1] [Cycle 1]: 0.00011524, [7] [c_1]: 4.394e-05 [parameter_eliminate]: 4.01001e-06 [updatestate_depend_eliminate]: 8.55001e-06 [updatestate_assign_eliminate]: 2.58e-06 [updatestate_loads_eliminate]: 2.34001e-06 [cse]: 1.808e-05 [renormalize]: 5.50004e-07 [remove_dup_value]: 1.56e-05 [tuple_transform]: 9.911e-05, [1] [Cycle 1]: 9.396e-05, [4] [d_1]: 6.358e-05 [none_parameter_eliminate]: 1.79e-06 [renormalize]: 2.3999e-07 [switch_simplify]: 7.7e-06 [partial_unused_args_eliminate]: 2.05002e-06 [add_recomputation]: 5.328e-05 [cse_after_recomputation]: 2.425e-05, [1] [Cycle 1]: 1.928e-05, [1] [cse]: 1.381e-05 [environ_conv]: 9.00999e-06 [swap_dp_allreduce_reducescatter]: 5.34e-06 [bias_add_comm_swap]: 2.99999e-06 [label_micro_interleaved_index]: 5.37001e-06 [label_fine_grained_interleaved_index]: 2.59999e-06 [merge_cast_opt]: 1.57001e-06 [slice_recompute_activation]: 2.39001e-06 [micro_interleaved_order_control]: 2.24999e-06 [assign_add_opt]: 1.74e-06 [ForceFp32Comm]: 8.60018e-07 [remove_cast_before_assign_add]: 1.03001e-06 [full_micro_interleaved_order_control]: 2.22999e-06 [reorder_send_recv_between_fp_bp]: 2.86e-06 [comm_op_add_attrs]: 1.17e-06 [add_comm_op_reuse_tag]: 1.05001e-06 [interleave_split_concat_branches]: 1.38002e-06 [interleave_parallel_branches]: 1.09e-06 [overlap_opt_shard_in_pipeline]: 3.72998e-06 [overlap_opt_shard_grad_in_pipeline]: 1.87999e-06 [control_data_broadcast_order]: 1.246e-05 [grouped_pairwise_exchange_alltoall]: 1.64e-06 [offloading_packed_experts]: 4.03999e-06 [overlap_recompute_and_grad_model_parallel]: 5.17999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.55999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.40999e-06 [overlap_recompute_comm]: 2.59999e-06 [overlap_grad_ring_attention]: 4.28001e-06 [overlap_grad_flash_sp]: 2.232e-05 [begin_end_overlap_inline]: 5.3001e-07 [split_matmul_comm_elemetwise]: 2.98998e-06 [split_layernorm_comm]: 2.38998e-06 [handle_group_info]: 1.09003e-06 [symbol_engine_optimizer]: 9.226e-05, [1] [Cycle 1]: 8.665e-05, [6] [build]: 1.294e-05 [elim_shapecalc]: 1.061e-05 [elim_not_effective]: 1.38e-05 [opt_reshape]: 7.33e-06 [fold_const_symbol]: 1.154e-05 [renormalize]: 1.39989e-07 [detach_backward]: 2.44999e-06 [pipeline_parallel_scheduler]: 1.74e-06 [auto_monad_reorder]: 1.789e-05 [get_jit_bprop_graph]: 2.38002e-06 [rewriter_after_jit_bprop_graph]: 4.60001e-06 [opt_after_jit_grad]: 0.00050289 [validate]: 5.105e-05 Sums bootstrap : 0.000746s : 0.10% type_inference : 0.382925s : 52.54% event_method : 0.000064s : 0.01% auto_monad : 0.000128s : 0.02% graph_reusing : 0.000009s : 0.00% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000050s : 0.01% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000010s : 0.00% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000065s : 0.01% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000045s : 0.01% optimize.rewriter_before_opt_a : 0.000191s : 0.03% optimize.opt_a.expand_dump_flag : 0.000008s : 0.00% optimize.opt_a.switch_simplify : 0.000140s : 0.02% optimize.opt_a.loop_unroll : 0.000116s : 0.02% optimize.opt_a.a_1 : 0.003344s : 0.46% optimize.opt_a.with_stream_mark : 0.000058s : 0.01% optimize.opt_a.recompute_prepare : 0.000040s : 0.01% optimize.opt_a.updatestate_depend_eliminate : 0.000017s : 0.00% optimize.opt_a.updatestate_assign_eliminate : 0.000013s : 0.00% optimize.opt_a.updatestate_loads_eliminate : 0.000012s : 0.00% optimize.opt_a.parameter_eliminate : 0.000006s : 0.00% optimize.opt_a.a_2 : 0.330498s : 45.35% optimize.opt_a.accelerated_algorithm : 0.000101s : 0.01% optimize.opt_a.shard : 0.000008s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000013s : 0.00% optimize.opt_a.shard_inline : 0.000033s : 0.00% optimize.opt_a.merge_send_recv : 0.000046s : 0.01% optimize.opt_a.auto_parallel : 0.000039s : 0.01% optimize.opt_a.parallel : 0.000051s : 0.01% optimize.opt_a.flash_sp : 0.000017s : 0.00% optimize.opt_a.merge_comm : 0.000017s : 0.00% optimize.opt_a.allreduce_fusion : 0.000017s : 0.00% optimize.opt_a.matmul_add_comm_reduction : 0.000053s : 0.01% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000037s : 0.01% optimize.opt_a.virtual_dataset : 0.000029s : 0.00% optimize.opt_a.get_grad_eliminate_ : 0.000028s : 0.00% optimize.opt_a.virtual_output : 0.000029s : 0.00% optimize.opt_a.merge_forward : 0.000018s : 0.00% optimize.opt_a.cell_reuse_recompute_pass : 0.000007s : 0.00% optimize.opt_a.offload_activation : 0.000041s : 0.01% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000060s : 0.01% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.00% optimize.opt_a.before_grad : 0.000050s : 0.01% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000018s : 0.00% optimize.opt_a.meta_fg_expand : 0.001999s : 0.27% optimize.opt_a.flash_sp_send_recv_attached : 0.000011s : 0.00% optimize.opt_a.receive_attached : 0.000007s : 0.00% optimize.opt_a.after_resolve : 0.000091s : 0.01% optimize.opt_a.a_after_grad : 0.000102s : 0.01% optimize.opt_a.renormalize : 0.003990s : 0.55% optimize.opt_a.add_forward_monad_depend : 0.000023s : 0.00% optimize.opt_a.auto_monad_grad : 0.000012s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000090s : 0.01% optimize.opt_a.cse : 0.000277s : 0.04% optimize.opt_a.a_3 : 0.000417s : 0.06% optimize.py_interpret_to_execute_after_opt_a : 0.000021s : 0.00% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000047s : 0.01% optimize.convert_after_rewriter : 0.000007s : 0.00% optimize.order_py_execute_after_rewriter : 0.000005s : 0.00% optimize.mutable_eliminate : 0.000775s : 0.11% optimize.opt_b.b_1 : 0.000119s : 0.02% optimize.opt_b.b_2 : 0.000008s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000007s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000020s : 0.00% optimize.optimize_parallel_all_gather_comm : 0.000019s : 0.00% optimize.overlap_param_gather : 0.000009s : 0.00% optimize.cconv : 0.000050s : 0.01% optimize.loop_unroll : 0.000527s : 0.07% optimize.opt_after_cconv.c_1 : 0.000044s : 0.01% optimize.opt_after_cconv.parameter_eliminate : 0.000004s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000009s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000018s : 0.00% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000016s : 0.00% optimize.tuple_transform.d_1 : 0.000064s : 0.01% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000008s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000053s : 0.01% optimize.cse_after_recomputation.cse : 0.000014s : 0.00% optimize.environ_conv : 0.000009s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.00% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000002s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000004s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000012s : 0.00% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000002s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000003s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000022s : 0.00% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000003s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000013s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000011s : 0.00% optimize.symbol_engine_optimizer.elim_not_effective : 0.000014s : 0.00% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000012s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000018s : 0.00% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000005s : 0.00% opt_after_jit_grad : 0.000503s : 0.07% validate : 0.000051s : 0.01% Time group info: ------[substitution.] 0.000981 190 8.48% : 0.000083s : 6: substitution.arithmetic_simplify 0.23% : 0.000002s : 2: substitution.elim_not_effective 0.54% : 0.000005s : 6: substitution.float_depend_g_call 0.49% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.17% : 0.000002s : 2: substitution.fold_const_symbol 0.71% : 0.000007s : 4: substitution.graph_param_transform 0.34% : 0.000003s : 2: substitution.incorporate_call 0.25% : 0.000002s : 2: substitution.incorporate_call_switch 55.19% : 0.000541s : 16: substitution.inline 1.93% : 0.000019s : 2: substitution.inline_without_move 1.03% : 0.000010s : 14: substitution.j_node_and_user_rematch 2.41% : 0.000024s : 3: substitution.less_batch_normalization 1.54% : 0.000015s : 11: substitution.minmaximum_grad 0.70% : 0.000007s : 6: substitution.partial_eliminate 1.38% : 0.000014s : 14: substitution.remove_not_recompute_node 2.97% : 0.000029s : 10: substitution.replace_applicator 1.39% : 0.000014s : 12: substitution.replace_old_param 0.29% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.09% : 0.000030s : 11: substitution.tuple_list_convert_item_index_to_positive 1.48% : 0.000014s : 11: substitution.tuple_list_get_item_const_eliminator 2.00% : 0.000020s : 11: substitution.tuple_list_get_item_depend_reorder 11.44% : 0.000112s : 30: substitution.tuple_list_get_item_eliminator 1.94% : 0.000019s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.382822 2 99.49% : 0.380869s : 1: type_inference.infer 0.51% : 0.001953s : 1: type_inference.specialize ------[replace.] 0.000258 34 4.18% : 0.000011s : 2: replace.arithmetic_simplify 55.31% : 0.000143s : 16: replace.inline 40.51% : 0.000105s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.000621 34 8.04% : 0.000050s : 2: match.arithmetic_simplify 85.72% : 0.000533s : 16: match.inline 6.24% : 0.000039s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.000711 4766 1.15% : 0.000008s : 58: predicate.accumulaten_eliminater 0.21% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.43% : 0.000003s : 23: predicate.addn_check_dump 1.07% : 0.000008s : 58: predicate.addn_zero_filter 0.99% : 0.000007s : 58: predicate.adjust_all_reduce_mul_add 1.88% : 0.000013s : 83: predicate.arithmetic_simplify 1.09% : 0.000008s : 60: predicate.cast_eliminate 1.05% : 0.000007s : 55: predicate.check_bprop_eliminate 0.44% : 0.000003s : 23: predicate.compare_switch_simplify 0.06% : 0.000000s : 4: predicate.const_output_eliminate 0.43% : 0.000003s : 23: predicate.depend_value_elim 1.16% : 0.000008s : 60: predicate.dict_get_item_const_eliminator 1.22% : 0.000009s : 60: predicate.dict_get_item_eliminator 1.11% : 0.000008s : 60: predicate.dict_set_item_eliminator 0.26% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.09% : 0.000001s : 4: predicate.elim_not_effective 0.13% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.14% : 0.000008s : 64: predicate.environ_add_const_eliminate 1.11% : 0.000008s : 64: predicate.environ_get_add_eliminate 1.11% : 0.000008s : 64: predicate.environ_get_depend_swap 1.63% : 0.000012s : 87: predicate.environ_get_eliminate 1.12% : 0.000008s : 64: predicate.environ_get_set_eliminate 1.75% : 0.000012s : 92: predicate.exchange_switch_depend_value 2.50% : 0.000018s : 92: predicate.float_depend_g_call 0.41% : 0.000003s : 23: predicate.float_environ_get_switch 0.55% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.12% : 0.000001s : 4: predicate.fold_const_symbol 0.51% : 0.000004s : 23: predicate.get_grad_eliminate 0.07% : 0.000001s : 4: predicate.graph_param_transform 0.43% : 0.000003s : 23: predicate.incorporate_call 0.38% : 0.000003s : 23: predicate.incorporate_call_switch 5.50% : 0.000039s : 204: predicate.inline 1.15% : 0.000008s : 46: predicate.inline_without_move 0.24% : 0.000002s : 23: predicate.j_node_and_user_rematch 1.26% : 0.000009s : 23: predicate.less_batch_normalization 1.68% : 0.000012s : 84: predicate.list_to_tuple_eliminator_ 2.67% : 0.000019s : 142: predicate.load_eliminater 2.18% : 0.000016s : 4: predicate.loop_unroll_after_grad 2.47% : 0.000018s : 128: predicate.loop_unroll_before_grad 1.38% : 0.000010s : 68: predicate.make_slice_get_slice_eliminator 0.43% : 0.000003s : 23: predicate.merge_addn 1.01% : 0.000007s : 55: predicate.micro_step_allgather_replace 1.03% : 0.000007s : 55: predicate.mini_step_allgather_replace 1.10% : 0.000008s : 60: predicate.minmaximum_grad 0.32% : 0.000002s : 4: predicate.mutable_eliminate 0.16% : 0.000001s : 4: predicate.opt_reshape 0.11% : 0.000001s : 4: predicate.parallel_virtual_node 2.16% : 0.000015s : 92: predicate.partial_defer_inline 1.66% : 0.000012s : 80: predicate.partial_eliminate 1.02% : 0.000007s : 58: predicate.print_const_string_wrapper 0.44% : 0.000003s : 23: predicate.reduce_all_const_elim 1.34% : 0.000010s : 60: predicate.reduce_eliminate 2.55% : 0.000018s : 142: predicate.redundant_stop_gradient_eliminater 0.32% : 0.000002s : 23: predicate.remove_not_recompute_node 1.75% : 0.000012s : 131: predicate.replace_applicator 0.54% : 0.000004s : 46: predicate.replace_old_param 0.08% : 0.000001s : 4: predicate.reset_defer_inline 1.14% : 0.000008s : 60: predicate.reshape_eliminate 1.13% : 0.000008s : 55: predicate.row_tensor_add_zeros_like 0.11% : 0.000001s : 4: predicate.row_tensor_eliminate 1.21% : 0.000009s : 55: predicate.same_eliminate 0.32% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.62% : 0.000004s : 23: predicate.shard_identity_eliminate 0.20% : 0.000001s : 8: predicate.special_op_eliminate 0.50% : 0.000004s : 23: predicate.specialize_transform 1.20% : 0.000009s : 55: predicate.split_environ_get_set_with_tuple_value 1.13% : 0.000008s : 46: predicate.stack_unstack_eliminate 1.97% : 0.000014s : 4: predicate.switch_call_monad_eliminater 1.92% : 0.000014s : 92: predicate.switch_defer_inline 2.99% : 0.000021s : 147: predicate.switch_layer_defer_inline 4.98% : 0.000035s : 247: predicate.switch_simplify 1.11% : 0.000008s : 60: predicate.tile_eliminate 1.06% : 0.000008s : 60: predicate.transpose_eliminate 1.31% : 0.000009s : 68: predicate.tuple_list_convert_item_index_to_positive 1.37% : 0.000010s : 68: predicate.tuple_list_get_item_const_eliminator 1.26% : 0.000009s : 68: predicate.tuple_list_get_item_depend_reorder 2.77% : 0.000020s : 107: predicate.tuple_list_get_item_eliminator 1.31% : 0.000009s : 68: predicate.tuple_list_get_set_item_eliminator 1.79% : 0.000013s : 91: predicate.tuple_list_set_item_eliminator 1.50% : 0.000011s : 84: predicate.tuple_to_list_eliminator_ 2.51% : 0.000018s : 142: predicate.updatestate_pure_node_eliminater 3.00% : 0.000021s : 165: predicate.updatestate_useless_node_eliminater 0.11% : 0.000001s : 4: predicate.value_based_eliminate 0.54% : 0.000004s : 23: predicate.virtual_dataset_eliminate 0.59% : 0.000004s : 23: predicate.virtual_output_eliminate 0.07% : 0.000000s : 4: predicate.virtual_view_grad_eliminate 0.12% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.201027 37 99.57% : 0.200162s : 17: func_graph_cloner_run.FuncGraphClonerGraph 0.43% : 0.000865s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 1.424077 233 0.00% : 0.000004s : 1: ForceFp32Comm 0.32% : 0.004599s : 1: add_attr 0.32% : 0.004585s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.00% : 0.000057s : 1: add_recomputation 0.00% : 0.000005s : 1: assign_add_opt 0.01% : 0.000136s : 1: auto_monad 0.00% : 0.000022s : 1: auto_monad_reorder 0.00% : 0.000003s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 0.05% : 0.000780s : 1: bootstrap 0.00% : 0.000054s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000017s : 1: control_data_broadcast_order 0.00% : 0.000010s : 1: convert_after_rewriter 0.00% : 0.000027s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000012s : 1: environ_conv 0.01% : 0.000072s : 1: event_method 0.00% : 0.000006s : 1: full_micro_interleaved_order_control 0.00% : 0.000006s : 1: get_jit_bprop_graph 0.00% : 0.000013s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.04% : 0.000535s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.06% : 0.000784s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.00% : 0.000027s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000015s : 1: opt.transform.mutable_eliminate 23.52% : 0.334940s : 117: opt.transform.opt_a 0.00% : 0.000043s : 1: opt.transform.opt_after_cconv 0.00% : 0.000023s : 1: opt.transform.opt_after_jit_grad 0.01% : 0.000097s : 28: opt.transform.opt_b 0.00% : 0.000069s : 2: opt.transform.opt_trans_graph 0.00% : 0.000040s : 4: opt.transform.symbol_engine_opt 24.08% : 0.342859s : 1: opt_a 0.01% : 0.000151s : 1: opt_after_cconv 0.04% : 0.000512s : 1: opt_after_jit_grad 0.01% : 0.000211s : 1: opt_b 24.27% : 0.345553s : 1: optimize 0.00% : 0.000023s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000008s : 1: order_py_execute_after_rewriter 0.00% : 0.000026s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000007s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000013s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000006s : 1: overlap_recompute_comm 0.00% : 0.000009s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000006s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.00% : 0.000071s : 1: pre_auto_parallel 0.00% : 0.000049s : 1: py_interpret_to_execute 0.00% : 0.000024s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000019s : 1: remove_dup_value 0.15% : 0.002158s : 2: renormalize.infer 0.13% : 0.001813s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000008s : 1: rewriter_after_jit_bprop_graph 0.00% : 0.000051s : 1: rewriter_after_opt_a 0.01% : 0.000197s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000006s : 1: split_matmul_comm_elemetwise 0.00% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.01% : 0.000095s : 1: symbol_engine_optimizer 0.01% : 0.000102s : 1: tuple_transform 26.89% : 0.382957s : 1: type_inference TotalTime = 0.101763, [21] [bootstrap]: 0.00048346 [type_inference]: 0.0197688 [event_method]: 6.295e-05 [auto_monad]: 0.00012663 [graph_reusing]: 8.29002e-06 [inline]: 2.61e-06 [add_attr]: 0.00381841, [1] [add_attr_with_inline]: 0.00380646, [1] [Cycle 1]: 0.00010241, [2] [tag_attr]: 4.544e-05 [meta_addattr_fg_expand]: 9.87999e-06 [parallel-infer-symbol]: 4.32e-06 [pre_auto_parallel]: 6.388e-05 [insert-virtual-dataset]: 2.91e-06 [parallel-infer-symbol-second]: 7.09988e-07 [dataset_repeat_opt]: 2.10002e-06 [pipeline_split]: 1.62001e-06 [optimize]: 0.0763302, [53] [py_interpret_to_execute]: 4.634e-05 [rewriter_before_opt_a]: 0.00018921 [opt_a]: 0.0616524, [3] [Cycle 1]: 0.0574522, [45] [expand_dump_flag]: 4.08001e-06 [switch_simplify]: 7.83e-05 [loop_unroll]: 6.695e-05 [a_1]: 0.00155318 [with_stream_mark]: 3.45e-05 [recompute_prepare]: 2.848e-05 [updatestate_depend_eliminate]: 1.007e-05 [updatestate_assign_eliminate]: 8.96002e-06 [updatestate_loads_eliminate]: 7.35e-06 [parameter_eliminate]: 3.17002e-06 [a_2]: 0.00024755 [accelerated_algorithm]: 3.744e-05 [shard]: 2.04e-06 [meta_shard_fg_expand]: 4.85999e-06 [shard_inline]: 1.631e-05 [merge_send_recv]: 2.219e-05 [auto_parallel]: 1.575e-05 [parallel]: 2.13e-05 [flash_sp]: 1.243e-05 [merge_comm]: 1.047e-05 [allreduce_fusion]: 8.82999e-06 [matmul_add_comm_reduction]: 3.529e-05 [allreduce_slice_to_reducescatter]: 1.03001e-06 [virtual_shard_identity]: 2.035e-05 [virtual_dataset]: 1.591e-05 [get_grad_eliminate_]: 1.539e-05 [virtual_output]: 1.551e-05 [merge_forward]: 1.082e-05 [cell_reuse_recompute_pass]: 1.71e-06 [offload_activation]: 1.907e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.281e-05 [merge_recompute_call_nodes]: 1.45001e-06 [before_grad]: 3.052e-05 [set_forward_comm_id_for_comm_node_pass]: 1.191e-05 [meta_fg_expand]: 0.0503528 [flash_sp_send_recv_attached]: 9.20001e-06 [receive_attached]: 2.34999e-06 [after_resolve]: 8.99e-05 [a_after_grad]: 0.0001019 [renormalize]: 0.00341943 [add_forward_monad_depend]: 1.774e-05 [auto_monad_grad]: 7.01001e-06 [auto_monad_eliminator]: 6.427e-05 [cse]: 0.00023503 [a_3]: 0.00035476 [Cycle 2]: 0.00353406, [45] [expand_dump_flag]: 3.25e-06 [switch_simplify]: 4.914e-05 [loop_unroll]: 4.173e-05 [a_1]: 0.00169036 [with_stream_mark]: 2.89e-05 [recompute_prepare]: 1.489e-05 [updatestate_depend_eliminate]: 5.29998e-06 [updatestate_assign_eliminate]: 3.14001e-06 [updatestate_loads_eliminate]: 2.90998e-06 [parameter_eliminate]: 2.07999e-06 [a_2]: 6.973e-05 [accelerated_algorithm]: 1.483e-05 [shard]: 2.27001e-06 [meta_shard_fg_expand]: 3.36001e-06 [shard_inline]: 6.83e-06 [merge_send_recv]: 1.093e-05 [auto_parallel]: 1.109e-05 [parallel]: 1.063e-05 [flash_sp]: 4.08999e-06 [merge_comm]: 3.86999e-06 [allreduce_fusion]: 4.73001e-06 [matmul_add_comm_reduction]: 1.083e-05 [allreduce_slice_to_reducescatter]: 1.11997e-06 [virtual_shard_identity]: 9.31e-06 [virtual_dataset]: 6.07999e-06 [get_grad_eliminate_]: 6.09001e-06 [virtual_output]: 5.12e-06 [merge_forward]: 4.14002e-06 [cell_reuse_recompute_pass]: 2.59001e-06 [offload_activation]: 1.103e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.388e-05 [merge_recompute_call_nodes]: 1.35999e-06 [before_grad]: 9.57999e-06 [set_forward_comm_id_for_comm_node_pass]: 4.33001e-06 [meta_fg_expand]: 0.00014582 [flash_sp_send_recv_attached]: 2.11e-06 [receive_attached]: 2.57001e-06 [after_resolve]: 2.247e-05 [a_after_grad]: 1.008e-05 [renormalize]: 0.00085724 [add_forward_monad_depend]: 7.65e-06 [auto_monad_grad]: 2.49999e-06 [auto_monad_eliminator]: 1.92e-05 [cse]: 2.989e-05 [a_3]: 4.866e-05 [Cycle 3]: 0.00064478, [45] [expand_dump_flag]: 2.48e-06 [switch_simplify]: 8.15e-06 [loop_unroll]: 6.14001e-06 [a_1]: 0.00011837 [with_stream_mark]: 1.517e-05 [recompute_prepare]: 6.04001e-06 [updatestate_depend_eliminate]: 3.7e-06 [updatestate_assign_eliminate]: 2.62001e-06 [updatestate_loads_eliminate]: 2.64001e-06 [parameter_eliminate]: 1.13001e-06 [a_2]: 6.424e-05 [accelerated_algorithm]: 1.424e-05 [shard]: 1.29998e-06 [meta_shard_fg_expand]: 1.47001e-06 [shard_inline]: 5.82999e-06 [merge_send_recv]: 6.92002e-06 [auto_parallel]: 8.57e-06 [parallel]: 7.38e-06 [flash_sp]: 1.17e-06 [merge_comm]: 3.40998e-06 [allreduce_fusion]: 3.08e-06 [matmul_add_comm_reduction]: 7.18998e-06 [allreduce_slice_to_reducescatter]: 4.10015e-07 [virtual_shard_identity]: 8.80999e-06 [virtual_dataset]: 5.59e-06 [get_grad_eliminate_]: 4.96002e-06 [virtual_output]: 5.46998e-06 [merge_forward]: 4.07e-06 [cell_reuse_recompute_pass]: 1.82001e-06 [offload_activation]: 9.45001e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.321e-05 [merge_recompute_call_nodes]: 9.50007e-07 [before_grad]: 9.51e-06 [set_forward_comm_id_for_comm_node_pass]: 3.66001e-06 [meta_fg_expand]: 2.27999e-06 [flash_sp_send_recv_attached]: 9.00007e-07 [receive_attached]: 1.63002e-06 [after_resolve]: 1.034e-05 [a_after_grad]: 8.12e-06 [renormalize]: 8.00064e-08 [add_forward_monad_depend]: 2.40002e-06 [auto_monad_grad]: 1.45999e-06 [auto_monad_eliminator]: 8.18001e-06 [cse]: 1.638e-05 [a_3]: 3.379e-05 [py_interpret_to_execute_after_opt_a]: 1.477e-05 [slice_cell_reuse_recomputed_activation]: 1.94e-06 [rewriter_after_opt_a]: 4.238e-05 [convert_after_rewriter]: 8.53001e-06 [order_py_execute_after_rewriter]: 4.63999e-06 [mutable_eliminate]: 0.00077409 [opt_b]: 0.00019952, [1] [Cycle 1]: 0.00019077, [7] [b_1]: 0.00011053 [b_2]: 8.33999e-06 [updatestate_depend_eliminate]: 7.97e-06 [updatestate_assign_eliminate]: 3.04001e-06 [updatestate_loads_eliminate]: 2.53003e-06 [renormalize]: 5.00004e-07 [cse]: 2.335e-05 [optimize_parallel_all_gather_comm]: 1.735e-05 [overlap_param_gather]: 1.72001e-06 [cconv]: 2.939e-05 [loop_unroll]: 0.00047687 [opt_after_cconv]: 0.00011032, [1] [Cycle 1]: 0.00010307, [7] [c_1]: 3.075e-05 [parameter_eliminate]: 4.33001e-06 [updatestate_depend_eliminate]: 6.64999e-06 [updatestate_assign_eliminate]: 2.88e-06 [updatestate_loads_eliminate]: 2.82002e-06 [cse]: 2.011e-05 [renormalize]: 7.00005e-07 [remove_dup_value]: 1.363e-05 [tuple_transform]: 7.646e-05, [1] [Cycle 1]: 7.18e-05, [4] [d_1]: 4.392e-05 [none_parameter_eliminate]: 1.60001e-06 [renormalize]: 1.8999e-07 [switch_simplify]: 6.73e-06 [partial_unused_args_eliminate]: 1.59e-06 [add_recomputation]: 5.129e-05 [cse_after_recomputation]: 2.308e-05, [1] [Cycle 1]: 1.807e-05, [1] [cse]: 1.238e-05 [environ_conv]: 8.77e-06 [swap_dp_allreduce_reducescatter]: 5.29998e-06 [bias_add_comm_swap]: 2.85002e-06 [label_micro_interleaved_index]: 4.87998e-06 [label_fine_grained_interleaved_index]: 2.57001e-06 [merge_cast_opt]: 1.34998e-06 [slice_recompute_activation]: 2.32999e-06 [micro_interleaved_order_control]: 2.76e-06 [assign_add_opt]: 1.29e-06 [ForceFp32Comm]: 1.20999e-06 [remove_cast_before_assign_add]: 1.07998e-06 [full_micro_interleaved_order_control]: 2.26998e-06 [reorder_send_recv_between_fp_bp]: 2.96001e-06 [comm_op_add_attrs]: 1.35999e-06 [add_comm_op_reuse_tag]: 1.00001e-06 [interleave_split_concat_branches]: 1.15001e-06 [interleave_parallel_branches]: 1.07998e-06 [overlap_opt_shard_in_pipeline]: 1.20001e-06 [overlap_opt_shard_grad_in_pipeline]: 1.87999e-06 [control_data_broadcast_order]: 1.283e-05 [grouped_pairwise_exchange_alltoall]: 2.11e-06 [offloading_packed_experts]: 4.08001e-06 [overlap_recompute_and_grad_model_parallel]: 4.50999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.24003e-06 [overlap_recompute_allgather_and_fa_grad]: 1.35001e-06 [overlap_recompute_comm]: 2.37001e-06 [overlap_grad_ring_attention]: 3.93001e-06 [overlap_grad_flash_sp]: 2.263e-05 [begin_end_overlap_inline]: 7.00005e-07 [split_matmul_comm_elemetwise]: 2.39001e-06 [split_layernorm_comm]: 1.76e-06 [handle_group_info]: 1.10999e-06 [symbol_engine_optimizer]: 0.012249, [1] [Cycle 1]: 0.0122404, [6] [build]: 1.302e-05 [elim_shapecalc]: 1.349e-05 [elim_not_effective]: 1.406e-05 [opt_reshape]: 0.0120773 [fold_const_symbol]: 3.462e-05 [renormalize]: 7.89994e-07 [detach_backward]: 4.13999e-06 [pipeline_parallel_scheduler]: 2.06003e-06 [auto_monad_reorder]: 3.697e-05 [get_jit_bprop_graph]: 2.66e-06 [rewriter_after_jit_bprop_graph]: 1.212e-05 [opt_after_jit_grad]: 0.00077651 [validate]: 5.961e-05 Sums bootstrap : 0.000483s : 0.50% type_inference : 0.019769s : 20.50% event_method : 0.000063s : 0.07% auto_monad : 0.000127s : 0.13% graph_reusing : 0.000008s : 0.01% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000045s : 0.05% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000010s : 0.01% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000064s : 0.07% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000046s : 0.05% optimize.rewriter_before_opt_a : 0.000189s : 0.20% optimize.opt_a.expand_dump_flag : 0.000010s : 0.01% optimize.opt_a.switch_simplify : 0.000136s : 0.14% optimize.opt_a.loop_unroll : 0.000115s : 0.12% optimize.opt_a.a_1 : 0.003362s : 3.49% optimize.opt_a.with_stream_mark : 0.000079s : 0.08% optimize.opt_a.recompute_prepare : 0.000049s : 0.05% optimize.opt_a.updatestate_depend_eliminate : 0.000019s : 0.02% optimize.opt_a.updatestate_assign_eliminate : 0.000015s : 0.02% optimize.opt_a.updatestate_loads_eliminate : 0.000013s : 0.01% optimize.opt_a.parameter_eliminate : 0.000006s : 0.01% optimize.opt_a.a_2 : 0.000382s : 0.40% optimize.opt_a.accelerated_algorithm : 0.000067s : 0.07% optimize.opt_a.shard : 0.000006s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000010s : 0.01% optimize.opt_a.shard_inline : 0.000029s : 0.03% optimize.opt_a.merge_send_recv : 0.000040s : 0.04% optimize.opt_a.auto_parallel : 0.000035s : 0.04% optimize.opt_a.parallel : 0.000039s : 0.04% optimize.opt_a.flash_sp : 0.000018s : 0.02% optimize.opt_a.merge_comm : 0.000018s : 0.02% optimize.opt_a.allreduce_fusion : 0.000017s : 0.02% optimize.opt_a.matmul_add_comm_reduction : 0.000053s : 0.06% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000003s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000038s : 0.04% optimize.opt_a.virtual_dataset : 0.000028s : 0.03% optimize.opt_a.get_grad_eliminate_ : 0.000026s : 0.03% optimize.opt_a.virtual_output : 0.000026s : 0.03% optimize.opt_a.merge_forward : 0.000019s : 0.02% optimize.opt_a.cell_reuse_recompute_pass : 0.000006s : 0.01% optimize.opt_a.offload_activation : 0.000040s : 0.04% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000060s : 0.06% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.00% optimize.opt_a.before_grad : 0.000050s : 0.05% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000020s : 0.02% optimize.opt_a.meta_fg_expand : 0.050501s : 52.37% optimize.opt_a.flash_sp_send_recv_attached : 0.000012s : 0.01% optimize.opt_a.receive_attached : 0.000007s : 0.01% optimize.opt_a.after_resolve : 0.000123s : 0.13% optimize.opt_a.a_after_grad : 0.000120s : 0.12% optimize.opt_a.renormalize : 0.004277s : 4.44% optimize.opt_a.add_forward_monad_depend : 0.000028s : 0.03% optimize.opt_a.auto_monad_grad : 0.000011s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000092s : 0.10% optimize.opt_a.cse : 0.000281s : 0.29% optimize.opt_a.a_3 : 0.000437s : 0.45% optimize.py_interpret_to_execute_after_opt_a : 0.000015s : 0.02% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000042s : 0.04% optimize.convert_after_rewriter : 0.000009s : 0.01% optimize.order_py_execute_after_rewriter : 0.000005s : 0.00% optimize.mutable_eliminate : 0.000774s : 0.80% optimize.opt_b.b_1 : 0.000111s : 0.11% optimize.opt_b.b_2 : 0.000008s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000008s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000023s : 0.02% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.02% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000029s : 0.03% optimize.loop_unroll : 0.000477s : 0.49% optimize.opt_after_cconv.c_1 : 0.000031s : 0.03% optimize.opt_after_cconv.parameter_eliminate : 0.000004s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000007s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.cse : 0.000020s : 0.02% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000014s : 0.01% optimize.tuple_transform.d_1 : 0.000044s : 0.05% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000051s : 0.05% optimize.cse_after_recomputation.cse : 0.000012s : 0.01% optimize.environ_conv : 0.000009s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.01% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.01% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000013s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000023s : 0.02% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000013s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000013s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000014s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.012077s : 12.53% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000035s : 0.04% optimize.symbol_engine_optimizer.renormalize : 0.000001s : 0.00% detach_backward : 0.000004s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000037s : 0.04% get_jit_bprop_graph : 0.000003s : 0.00% rewriter_after_jit_bprop_graph : 0.000012s : 0.01% opt_after_jit_grad : 0.000777s : 0.81% validate : 0.000060s : 0.06% Time group info: ------[substitution.] 0.001039 190 9.24% : 0.000096s : 6: substitution.arithmetic_simplify 0.20% : 0.000002s : 2: substitution.elim_not_effective 0.52% : 0.000005s : 6: substitution.float_depend_g_call 0.51% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.40% : 0.000004s : 2: substitution.fold_const_symbol 0.62% : 0.000006s : 4: substitution.graph_param_transform 0.32% : 0.000003s : 2: substitution.incorporate_call 0.23% : 0.000002s : 2: substitution.incorporate_call_switch 56.45% : 0.000587s : 16: substitution.inline 2.68% : 0.000028s : 2: substitution.inline_without_move 1.10% : 0.000011s : 14: substitution.j_node_and_user_rematch 2.12% : 0.000022s : 3: substitution.less_batch_normalization 1.55% : 0.000016s : 11: substitution.minmaximum_grad 0.74% : 0.000008s : 6: substitution.partial_eliminate 1.30% : 0.000014s : 14: substitution.remove_not_recompute_node 3.44% : 0.000036s : 10: substitution.replace_applicator 1.27% : 0.000013s : 12: substitution.replace_old_param 0.27% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.26% : 0.000034s : 11: substitution.tuple_list_convert_item_index_to_positive 1.47% : 0.000015s : 11: substitution.tuple_list_get_item_const_eliminator 2.03% : 0.000021s : 11: substitution.tuple_list_get_item_depend_reorder 8.21% : 0.000085s : 30: substitution.tuple_list_get_item_eliminator 2.07% : 0.000021s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.019656 2 90.34% : 0.017756s : 1: type_inference.infer 9.66% : 0.001899s : 1: type_inference.specialize ------[replace.] 0.000282 34 4.81% : 0.000014s : 2: replace.arithmetic_simplify 55.89% : 0.000158s : 16: replace.inline 39.30% : 0.000111s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.000675 34 8.43% : 0.000057s : 2: match.arithmetic_simplify 85.44% : 0.000577s : 16: match.inline 6.13% : 0.000041s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.000683 4766 1.05% : 0.000007s : 58: predicate.accumulaten_eliminater 0.37% : 0.000003s : 4: predicate.ad_related_special_op_eliminate 0.42% : 0.000003s : 23: predicate.addn_check_dump 1.09% : 0.000007s : 58: predicate.addn_zero_filter 1.01% : 0.000007s : 58: predicate.adjust_all_reduce_mul_add 1.98% : 0.000014s : 83: predicate.arithmetic_simplify 1.12% : 0.000008s : 60: predicate.cast_eliminate 1.06% : 0.000007s : 55: predicate.check_bprop_eliminate 0.45% : 0.000003s : 23: predicate.compare_switch_simplify 0.06% : 0.000000s : 4: predicate.const_output_eliminate 0.42% : 0.000003s : 23: predicate.depend_value_elim 1.18% : 0.000008s : 60: predicate.dict_get_item_const_eliminator 1.25% : 0.000009s : 60: predicate.dict_get_item_eliminator 1.14% : 0.000008s : 60: predicate.dict_set_item_eliminator 0.41% : 0.000003s : 8: predicate.dumpgradient_eliminate 0.08% : 0.000001s : 4: predicate.elim_not_effective 0.15% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.22% : 0.000008s : 64: predicate.environ_add_const_eliminate 1.14% : 0.000008s : 64: predicate.environ_get_add_eliminate 1.14% : 0.000008s : 64: predicate.environ_get_depend_swap 1.62% : 0.000011s : 87: predicate.environ_get_eliminate 1.16% : 0.000008s : 64: predicate.environ_get_set_eliminate 1.80% : 0.000012s : 92: predicate.exchange_switch_depend_value 2.53% : 0.000017s : 92: predicate.float_depend_g_call 0.48% : 0.000003s : 23: predicate.float_environ_get_switch 0.54% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.16% : 0.000001s : 4: predicate.fold_const_symbol 0.50% : 0.000003s : 23: predicate.get_grad_eliminate 0.06% : 0.000000s : 4: predicate.graph_param_transform 0.43% : 0.000003s : 23: predicate.incorporate_call 0.43% : 0.000003s : 23: predicate.incorporate_call_switch 5.46% : 0.000037s : 204: predicate.inline 1.32% : 0.000009s : 46: predicate.inline_without_move 0.23% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.58% : 0.000004s : 23: predicate.less_batch_normalization 1.60% : 0.000011s : 84: predicate.list_to_tuple_eliminator_ 2.54% : 0.000017s : 142: predicate.load_eliminater 0.37% : 0.000003s : 4: predicate.loop_unroll_after_grad 2.45% : 0.000017s : 128: predicate.loop_unroll_before_grad 1.59% : 0.000011s : 68: predicate.make_slice_get_slice_eliminator 0.49% : 0.000003s : 23: predicate.merge_addn 1.03% : 0.000007s : 55: predicate.micro_step_allgather_replace 1.12% : 0.000008s : 55: predicate.mini_step_allgather_replace 1.14% : 0.000008s : 60: predicate.minmaximum_grad 0.39% : 0.000003s : 4: predicate.mutable_eliminate 1.28% : 0.000009s : 4: predicate.opt_reshape 0.17% : 0.000001s : 4: predicate.parallel_virtual_node 2.35% : 0.000016s : 92: predicate.partial_defer_inline 1.68% : 0.000011s : 80: predicate.partial_eliminate 1.08% : 0.000007s : 58: predicate.print_const_string_wrapper 0.44% : 0.000003s : 23: predicate.reduce_all_const_elim 1.41% : 0.000010s : 60: predicate.reduce_eliminate 2.62% : 0.000018s : 142: predicate.redundant_stop_gradient_eliminater 0.39% : 0.000003s : 23: predicate.remove_not_recompute_node 1.90% : 0.000013s : 131: predicate.replace_applicator 0.73% : 0.000005s : 46: predicate.replace_old_param 0.07% : 0.000001s : 4: predicate.reset_defer_inline 1.13% : 0.000008s : 60: predicate.reshape_eliminate 1.16% : 0.000008s : 55: predicate.row_tensor_add_zeros_like 0.11% : 0.000001s : 4: predicate.row_tensor_eliminate 1.23% : 0.000008s : 55: predicate.same_eliminate 0.32% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.63% : 0.000004s : 23: predicate.shard_identity_eliminate 0.22% : 0.000002s : 8: predicate.special_op_eliminate 0.55% : 0.000004s : 23: predicate.specialize_transform 1.55% : 0.000011s : 55: predicate.split_environ_get_set_with_tuple_value 1.35% : 0.000009s : 46: predicate.stack_unstack_eliminate 0.11% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.98% : 0.000014s : 92: predicate.switch_defer_inline 2.91% : 0.000020s : 147: predicate.switch_layer_defer_inline 5.11% : 0.000035s : 247: predicate.switch_simplify 1.11% : 0.000008s : 60: predicate.tile_eliminate 1.13% : 0.000008s : 60: predicate.transpose_eliminate 1.33% : 0.000009s : 68: predicate.tuple_list_convert_item_index_to_positive 1.43% : 0.000010s : 68: predicate.tuple_list_get_item_const_eliminator 1.29% : 0.000009s : 68: predicate.tuple_list_get_item_depend_reorder 2.99% : 0.000020s : 107: predicate.tuple_list_get_item_eliminator 1.32% : 0.000009s : 68: predicate.tuple_list_get_set_item_eliminator 1.78% : 0.000012s : 91: predicate.tuple_list_set_item_eliminator 1.56% : 0.000011s : 84: predicate.tuple_to_list_eliminator_ 2.53% : 0.000017s : 142: predicate.updatestate_pure_node_eliminater 3.00% : 0.000021s : 165: predicate.updatestate_useless_node_eliminater 0.09% : 0.000001s : 4: predicate.value_based_eliminate 0.54% : 0.000004s : 23: predicate.virtual_dataset_eliminate 0.48% : 0.000003s : 23: predicate.virtual_output_eliminate 0.08% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.11% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002423 37 64.31% : 0.001558s : 17: func_graph_cloner_run.FuncGraphClonerGraph 35.69% : 0.000865s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.203257 233 0.00% : 0.000004s : 1: ForceFp32Comm 1.88% : 0.003825s : 1: add_attr 1.87% : 0.003811s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.03% : 0.000056s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.07% : 0.000134s : 1: auto_monad 0.02% : 0.000041s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 0.26% : 0.000520s : 1: bootstrap 0.02% : 0.000033s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.01% : 0.000017s : 1: control_data_broadcast_order 0.01% : 0.000012s : 1: convert_after_rewriter 0.01% : 0.000026s : 1: cse_after_recomputation 0.00% : 0.000006s : 1: dataset_repeat_opt 0.00% : 0.000008s : 1: detach_backward 0.01% : 0.000012s : 1: environ_conv 0.03% : 0.000071s : 1: event_method 0.00% : 0.000006s : 1: full_micro_interleaved_order_control 0.00% : 0.000007s : 1: get_jit_bprop_graph 0.01% : 0.000012s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.24% : 0.000485s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.39% : 0.000785s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.01% : 0.000016s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000018s : 1: opt.transform.mutable_eliminate 2.41% : 0.004889s : 117: opt.transform.opt_a 0.01% : 0.000029s : 1: opt.transform.opt_after_cconv 0.01% : 0.000030s : 1: opt.transform.opt_after_jit_grad 0.05% : 0.000092s : 28: opt.transform.opt_b 0.02% : 0.000049s : 2: opt.transform.opt_trans_graph 5.97% : 0.012128s : 4: opt.transform.symbol_engine_opt 30.33% : 0.061656s : 1: opt_a 0.06% : 0.000114s : 1: opt_after_cconv 0.39% : 0.000786s : 1: opt_after_jit_grad 0.10% : 0.000203s : 1: opt_b 37.56% : 0.076337s : 1: optimize 0.01% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000008s : 1: order_py_execute_after_rewriter 0.01% : 0.000027s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000007s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000006s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.03% : 0.000069s : 1: pre_auto_parallel 0.02% : 0.000050s : 1: py_interpret_to_execute 0.01% : 0.000019s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.01% : 0.000018s : 1: remove_dup_value 1.12% : 0.002272s : 2: renormalize.infer 0.98% : 0.001982s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000016s : 1: rewriter_after_jit_bprop_graph 0.02% : 0.000047s : 1: rewriter_after_opt_a 0.10% : 0.000195s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000008s : 1: swap_dp_allreduce_reducescatter 6.03% : 0.012253s : 1: symbol_engine_optimizer 0.04% : 0.000079s : 1: tuple_transform 9.74% : 0.019797s : 1: type_inference [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:43:28.687.46 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:43:28.692.14 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:43:28.348.706 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:43:28.349.001 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. TotalTime = 0.193361, [21] [bootstrap]: 0.00044251 [type_inference]: 0.156045 [event_method]: 4.18e-05 [auto_monad]: 7.682e-05 [graph_reusing]: 7.4e-06 [inline]: 3.39001e-06 [add_attr]: 0.00363224, [1] [add_attr_with_inline]: 0.00361897, [1] [Cycle 1]: 7.763e-05, [2] [tag_attr]: 3.159e-05 [meta_addattr_fg_expand]: 6.53e-06 [parallel-infer-symbol]: 3.58999e-06 [pre_auto_parallel]: 5.89e-05 [insert-virtual-dataset]: 2.65002e-06 [parallel-infer-symbol-second]: 9.50007e-07 [dataset_repeat_opt]: 2.19001e-06 [pipeline_split]: 1.60999e-06 [optimize]: 0.00626945, [53] [py_interpret_to_execute]: 3.369e-05 [rewriter_before_opt_a]: 0.00011083 [opt_a]: 0.00367546, [2] [Cycle 1]: 0.00297973, [45] [expand_dump_flag]: 4.40999e-06 [switch_simplify]: 5.179e-05 [loop_unroll]: 0.00018144 [a_1]: 0.00096774 [with_stream_mark]: 3.178e-05 [recompute_prepare]: 1.603e-05 [updatestate_depend_eliminate]: 5.69999e-06 [updatestate_assign_eliminate]: 3.43999e-06 [updatestate_loads_eliminate]: 2.98e-06 [parameter_eliminate]: 2.91e-06 [a_2]: 8.623e-05 [accelerated_algorithm]: 7.83999e-06 [shard]: 3.23998e-06 [meta_shard_fg_expand]: 2.51998e-06 [shard_inline]: 5.96998e-06 [merge_send_recv]: 9.57001e-06 [auto_parallel]: 9.82999e-06 [parallel]: 2.01e-05 [flash_sp]: 9.50001e-06 [merge_comm]: 4.45999e-06 [allreduce_fusion]: 3.58999e-06 [matmul_add_comm_reduction]: 1.078e-05 [allreduce_slice_to_reducescatter]: 1.10001e-06 [virtual_shard_identity]: 1.092e-05 [virtual_dataset]: 6.26e-06 [get_grad_eliminate_]: 5.81e-06 [virtual_output]: 6.39001e-06 [merge_forward]: 4.53999e-06 [cell_reuse_recompute_pass]: 2.09999e-06 [offload_activation]: 1.055e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.534e-05 [merge_recompute_call_nodes]: 1.35999e-06 [before_grad]: 1.052e-05 [set_forward_comm_id_for_comm_node_pass]: 4.27e-06 [meta_fg_expand]: 2.57001e-06 [flash_sp_send_recv_attached]: 2.97002e-06 [receive_attached]: 2.17999e-06 [after_resolve]: 1.325e-05 [a_after_grad]: 9.96e-06 [renormalize]: 0.00101816 [add_forward_monad_depend]: 8.97e-06 [auto_monad_grad]: 2.84999e-06 [auto_monad_eliminator]: 2.004e-05 [cse]: 3.22e-05 [a_3]: 5.254e-05 [Cycle 2]: 0.00068283, [45] [expand_dump_flag]: 2.52001e-06 [switch_simplify]: 8.54e-06 [loop_unroll]: 5.98998e-06 [a_1]: 0.00013761 [with_stream_mark]: 1.744e-05 [recompute_prepare]: 7.05e-06 [updatestate_depend_eliminate]: 4.27998e-06 [updatestate_assign_eliminate]: 2.73998e-06 [updatestate_loads_eliminate]: 2.79999e-06 [parameter_eliminate]: 9.89996e-07 [a_2]: 7.069e-05 [accelerated_algorithm]: 6.28e-06 [shard]: 1.67999e-06 [meta_shard_fg_expand]: 1.57999e-06 [shard_inline]: 5.73002e-06 [merge_send_recv]: 7.24001e-06 [auto_parallel]: 1e-05 [parallel]: 7.51001e-06 [flash_sp]: 3.56999e-06 [merge_comm]: 3.8e-06 [allreduce_fusion]: 3.31999e-06 [matmul_add_comm_reduction]: 8.37e-06 [allreduce_slice_to_reducescatter]: 3.7998e-07 [virtual_shard_identity]: 8e-06 [virtual_dataset]: 6.34999e-06 [get_grad_eliminate_]: 5.12e-06 [virtual_output]: 5.10001e-06 [merge_forward]: 4.38999e-06 [cell_reuse_recompute_pass]: 2.25002e-06 [offload_activation]: 9.19998e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.246e-05 [merge_recompute_call_nodes]: 1.32e-06 [before_grad]: 8.95001e-06 [set_forward_comm_id_for_comm_node_pass]: 4.20999e-06 [meta_fg_expand]: 2.74001e-06 [flash_sp_send_recv_attached]: 1.50001e-06 [receive_attached]: 1.92999e-06 [after_resolve]: 1.089e-05 [a_after_grad]: 7.91001e-06 [renormalize]: 7.99773e-08 [add_forward_monad_depend]: 2.36e-06 [auto_monad_grad]: 2.35002e-06 [auto_monad_eliminator]: 1.121e-05 [cse]: 1.719e-05 [a_3]: 3.347e-05 [py_interpret_to_execute_after_opt_a]: 1.585e-05 [slice_cell_reuse_recomputed_activation]: 1.92999e-06 [rewriter_after_opt_a]: 5.034e-05 [convert_after_rewriter]: 4.136e-05 [order_py_execute_after_rewriter]: 6.56e-06 [mutable_eliminate]: 0.00078856 [opt_b]: 0.00021596, [1] [Cycle 1]: 0.00020599, [7] [b_1]: 0.00011647 [b_2]: 9.42001e-06 [updatestate_depend_eliminate]: 1.003e-05 [updatestate_assign_eliminate]: 3.00998e-06 [updatestate_loads_eliminate]: 2.73e-06 [renormalize]: 7.49977e-07 [cse]: 2.819e-05 [optimize_parallel_all_gather_comm]: 2.575e-05 [overlap_param_gather]: 2.10002e-06 [cconv]: 3.459e-05 [loop_unroll]: 0.00050187 [opt_after_cconv]: 0.00011273, [1] [Cycle 1]: 0.00010521, [7] [c_1]: 2.981e-05 [parameter_eliminate]: 6.35002e-06 [updatestate_depend_eliminate]: 6.49001e-06 [updatestate_assign_eliminate]: 2.29999e-06 [updatestate_loads_eliminate]: 2.50002e-06 [cse]: 2.429e-05 [renormalize]: 5.90022e-07 [remove_dup_value]: 1.586e-05 [tuple_transform]: 0.00010521, [1] [Cycle 1]: 0.00010057, [4] [d_1]: 4.625e-05 [none_parameter_eliminate]: 2.16e-06 [renormalize]: 1.8999e-07 [switch_simplify]: 7.5e-06 [partial_unused_args_eliminate]: 2.04e-06 [add_recomputation]: 6.082e-05 [cse_after_recomputation]: 2.355e-05, [1] [Cycle 1]: 1.766e-05, [1] [cse]: 1.217e-05 [environ_conv]: 6.24001e-06 [swap_dp_allreduce_reducescatter]: 5.20001e-06 [bias_add_comm_swap]: 3.23998e-06 [label_micro_interleaved_index]: 4.70001e-06 [label_fine_grained_interleaved_index]: 3.09999e-06 [merge_cast_opt]: 1.35001e-06 [slice_recompute_activation]: 2.01998e-06 [micro_interleaved_order_control]: 2.36e-06 [assign_add_opt]: 1.29e-06 [ForceFp32Comm]: 8.89995e-07 [remove_cast_before_assign_add]: 1.19e-06 [full_micro_interleaved_order_control]: 2.51e-06 [reorder_send_recv_between_fp_bp]: 2.59001e-06 [comm_op_add_attrs]: 1.14e-06 [add_comm_op_reuse_tag]: 1.54e-06 [interleave_split_concat_branches]: 1.12e-06 [interleave_parallel_branches]: 1.25001e-06 [overlap_opt_shard_in_pipeline]: 1.18001e-06 [overlap_opt_shard_grad_in_pipeline]: 1.72001e-06 [control_data_broadcast_order]: 1.424e-05 [grouped_pairwise_exchange_alltoall]: 1.59998e-06 [offloading_packed_experts]: 4e-06 [overlap_recompute_and_grad_model_parallel]: 4.90001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.42e-06 [overlap_recompute_allgather_and_fa_grad]: 1.57001e-06 [overlap_recompute_comm]: 2.29001e-06 [overlap_grad_ring_attention]: 4.42998e-06 [overlap_grad_flash_sp]: 2.469e-05 [begin_end_overlap_inline]: 5.39992e-07 [split_matmul_comm_elemetwise]: 2.37001e-06 [split_layernorm_comm]: 1.62999e-06 [handle_group_info]: 1.07998e-06 [symbol_engine_optimizer]: 8.825e-05, [1] [Cycle 1]: 8.311e-05, [6] [build]: 3.91001e-06 [elim_shapecalc]: 1.432e-05 [elim_not_effective]: 1.524e-05 [opt_reshape]: 6.81999e-06 [fold_const_symbol]: 9.54999e-06 [renormalize]: 1.59984e-07 [detach_backward]: 2.37001e-06 [pipeline_parallel_scheduler]: 1.81e-06 [auto_monad_reorder]: 2.03e-05 [get_jit_bprop_graph]: 1.79e-06 [rewriter_after_jit_bprop_graph]: 0.00018415 [opt_after_jit_grad]: 0.00056952 [validate]: 4.696e-05 Sums bootstrap : 0.000443s : 0.27% type_inference : 0.156045s : 95.80% event_method : 0.000042s : 0.03% auto_monad : 0.000077s : 0.05% graph_reusing : 0.000007s : 0.00% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000032s : 0.02% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.00% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000059s : 0.04% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000034s : 0.02% optimize.rewriter_before_opt_a : 0.000111s : 0.07% optimize.opt_a.expand_dump_flag : 0.000007s : 0.00% optimize.opt_a.switch_simplify : 0.000060s : 0.04% optimize.opt_a.loop_unroll : 0.000187s : 0.12% optimize.opt_a.a_1 : 0.001105s : 0.68% optimize.opt_a.with_stream_mark : 0.000049s : 0.03% optimize.opt_a.recompute_prepare : 0.000023s : 0.01% optimize.opt_a.updatestate_depend_eliminate : 0.000010s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.00% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.00% optimize.opt_a.parameter_eliminate : 0.000004s : 0.00% optimize.opt_a.a_2 : 0.000157s : 0.10% optimize.opt_a.accelerated_algorithm : 0.000014s : 0.01% optimize.opt_a.shard : 0.000005s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000004s : 0.00% optimize.opt_a.shard_inline : 0.000012s : 0.01% optimize.opt_a.merge_send_recv : 0.000017s : 0.01% optimize.opt_a.auto_parallel : 0.000020s : 0.01% optimize.opt_a.parallel : 0.000028s : 0.02% optimize.opt_a.flash_sp : 0.000013s : 0.01% optimize.opt_a.merge_comm : 0.000008s : 0.01% optimize.opt_a.allreduce_fusion : 0.000007s : 0.00% optimize.opt_a.matmul_add_comm_reduction : 0.000019s : 0.01% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000019s : 0.01% optimize.opt_a.virtual_dataset : 0.000013s : 0.01% optimize.opt_a.get_grad_eliminate_ : 0.000011s : 0.01% optimize.opt_a.virtual_output : 0.000011s : 0.01% optimize.opt_a.merge_forward : 0.000009s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000004s : 0.00% optimize.opt_a.offload_activation : 0.000020s : 0.01% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000028s : 0.02% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.00% optimize.opt_a.before_grad : 0.000019s : 0.01% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000008s : 0.01% optimize.opt_a.meta_fg_expand : 0.000005s : 0.00% optimize.opt_a.flash_sp_send_recv_attached : 0.000004s : 0.00% optimize.opt_a.receive_attached : 0.000004s : 0.00% optimize.opt_a.after_resolve : 0.000024s : 0.01% optimize.opt_a.a_after_grad : 0.000018s : 0.01% optimize.opt_a.renormalize : 0.001018s : 0.63% optimize.opt_a.add_forward_monad_depend : 0.000011s : 0.01% optimize.opt_a.auto_monad_grad : 0.000005s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000031s : 0.02% optimize.opt_a.cse : 0.000049s : 0.03% optimize.opt_a.a_3 : 0.000086s : 0.05% optimize.py_interpret_to_execute_after_opt_a : 0.000016s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000050s : 0.03% optimize.convert_after_rewriter : 0.000041s : 0.03% optimize.order_py_execute_after_rewriter : 0.000007s : 0.00% optimize.mutable_eliminate : 0.000789s : 0.48% optimize.opt_b.b_1 : 0.000116s : 0.07% optimize.opt_b.b_2 : 0.000009s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000010s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000028s : 0.02% optimize.optimize_parallel_all_gather_comm : 0.000026s : 0.02% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000035s : 0.02% optimize.loop_unroll : 0.000502s : 0.31% optimize.opt_after_cconv.c_1 : 0.000030s : 0.02% optimize.opt_after_cconv.parameter_eliminate : 0.000006s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.cse : 0.000024s : 0.01% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000016s : 0.01% optimize.tuple_transform.d_1 : 0.000046s : 0.03% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000061s : 0.04% optimize.cse_after_recomputation.cse : 0.000012s : 0.01% optimize.environ_conv : 0.000006s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.00% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000003s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000002s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000014s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000002s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000025s : 0.02% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000004s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000014s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000015s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000020s : 0.01% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000184s : 0.11% opt_after_jit_grad : 0.000570s : 0.35% validate : 0.000047s : 0.03% Time group info: ------[substitution.] 0.000326 40 14.10% : 0.000046s : 8: substitution.arithmetic_simplify 0.74% : 0.000002s : 2: substitution.elim_not_effective 0.38% : 0.000001s : 2: substitution.fold_const_symbol 2.12% : 0.000007s : 4: substitution.graph_param_transform 71.45% : 0.000233s : 6: substitution.inline 1.25% : 0.000004s : 4: substitution.j_node_and_user_rematch 1.87% : 0.000006s : 4: substitution.remove_not_recompute_node 1.81% : 0.000006s : 4: substitution.replace_old_param 6.29% : 0.000020s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.155966 2 99.32% : 0.154903s : 1: type_inference.infer 0.68% : 0.001063s : 1: type_inference.specialize ------[replace.] 0.000098 12 59.09% : 0.000058s : 6: replace.inline 40.91% : 0.000040s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000247 12 92.68% : 0.000229s : 6: match.inline 7.32% : 0.000018s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000242 1534 0.99% : 0.000002s : 17: predicate.accumulaten_eliminater 0.86% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.41% : 0.000001s : 8: predicate.addn_check_dump 0.96% : 0.000002s : 17: predicate.addn_zero_filter 0.91% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.18% : 0.000005s : 25: predicate.arithmetic_simplify 0.95% : 0.000002s : 17: predicate.cast_eliminate 0.57% : 0.000001s : 8: predicate.check_bprop_eliminate 0.48% : 0.000001s : 8: predicate.compare_switch_simplify 0.17% : 0.000000s : 4: predicate.const_output_eliminate 0.55% : 0.000001s : 8: predicate.depend_value_elim 1.00% : 0.000002s : 17: predicate.dict_get_item_const_eliminator 1.05% : 0.000003s : 17: predicate.dict_get_item_eliminator 1.05% : 0.000003s : 17: predicate.dict_set_item_eliminator 1.04% : 0.000003s : 8: predicate.dumpgradient_eliminate 0.25% : 0.000001s : 4: predicate.elim_not_effective 0.48% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.06% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.11% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.06% : 0.000003s : 21: predicate.environ_get_depend_swap 1.54% : 0.000004s : 29: predicate.environ_get_eliminate 1.04% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.62% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.57% : 0.000006s : 29: predicate.float_depend_g_call 0.49% : 0.000001s : 8: predicate.float_environ_get_switch 0.59% : 0.000001s : 12: predicate.float_tuple_getitem_switch 0.15% : 0.000000s : 4: predicate.fold_const_symbol 0.54% : 0.000001s : 8: predicate.get_grad_eliminate 0.33% : 0.000001s : 4: predicate.graph_param_transform 0.51% : 0.000001s : 8: predicate.incorporate_call 0.37% : 0.000001s : 8: predicate.incorporate_call_switch 6.70% : 0.000016s : 70: predicate.inline 0.80% : 0.000002s : 8: predicate.inline_without_move 0.25% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.83% : 0.000002s : 8: predicate.less_batch_normalization 1.74% : 0.000004s : 31: predicate.list_to_tuple_eliminator_ 2.44% : 0.000006s : 48: predicate.load_eliminater 1.06% : 0.000003s : 4: predicate.loop_unroll_after_grad 3.08% : 0.000007s : 47: predicate.loop_unroll_before_grad 1.69% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.40% : 0.000001s : 8: predicate.merge_addn 0.44% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.49% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.83% : 0.000002s : 17: predicate.minmaximum_grad 1.95% : 0.000005s : 4: predicate.mutable_eliminate 0.26% : 0.000001s : 4: predicate.opt_reshape 0.34% : 0.000001s : 4: predicate.parallel_virtual_node 2.51% : 0.000006s : 29: predicate.partial_defer_inline 1.52% : 0.000004s : 27: predicate.partial_eliminate 0.93% : 0.000002s : 17: predicate.print_const_string_wrapper 0.45% : 0.000001s : 8: predicate.reduce_all_const_elim 1.55% : 0.000004s : 17: predicate.reduce_eliminate 2.54% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.48% : 0.000001s : 8: predicate.remove_not_recompute_node 1.36% : 0.000003s : 31: predicate.replace_applicator 0.53% : 0.000001s : 8: predicate.replace_old_param 0.37% : 0.000001s : 4: predicate.reset_defer_inline 1.29% : 0.000003s : 17: predicate.reshape_eliminate 0.55% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.35% : 0.000001s : 4: predicate.row_tensor_eliminate 0.74% : 0.000002s : 8: predicate.same_eliminate 0.37% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.59% : 0.000001s : 8: predicate.shard_identity_eliminate 0.58% : 0.000001s : 8: predicate.special_op_eliminate 0.59% : 0.000001s : 8: predicate.specialize_transform 1.06% : 0.000003s : 8: predicate.split_environ_get_set_with_tuple_value 0.54% : 0.000001s : 8: predicate.stack_unstack_eliminate 0.31% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.71% : 0.000004s : 29: predicate.switch_defer_inline 2.08% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.33% : 0.000013s : 88: predicate.switch_simplify 0.94% : 0.000002s : 17: predicate.tile_eliminate 0.94% : 0.000002s : 17: predicate.transpose_eliminate 1.38% : 0.000003s : 25: predicate.tuple_list_convert_item_index_to_positive 1.31% : 0.000003s : 25: predicate.tuple_list_get_item_const_eliminator 1.19% : 0.000003s : 25: predicate.tuple_list_get_item_depend_reorder 3.52% : 0.000009s : 39: predicate.tuple_list_get_item_eliminator 1.25% : 0.000003s : 25: predicate.tuple_list_get_set_item_eliminator 1.81% : 0.000004s : 33: predicate.tuple_list_set_item_eliminator 1.74% : 0.000004s : 31: predicate.tuple_to_list_eliminator_ 2.40% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.10% : 0.000007s : 56: predicate.updatestate_useless_node_eliminater 0.33% : 0.000001s : 4: predicate.value_based_eliminate 0.55% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.47% : 0.000001s : 8: predicate.virtual_output_eliminate 0.21% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.39% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000907 15 59.57% : 0.000540s : 7: func_graph_cloner_run.FuncGraphClonerGraph 40.43% : 0.000367s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.206099 192 0.00% : 0.000004s : 1: ForceFp32Comm 1.77% : 0.003639s : 1: add_attr 1.76% : 0.003623s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.03% : 0.000066s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.04% : 0.000083s : 1: auto_monad 0.01% : 0.000024s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 12.73% : 0.026233s : 1: bootstrap 0.02% : 0.000038s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.01% : 0.000018s : 1: control_data_broadcast_order 0.02% : 0.000046s : 1: convert_after_rewriter 0.01% : 0.000026s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000009s : 1: environ_conv 0.02% : 0.000050s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000005s : 1: get_jit_bprop_graph 0.01% : 0.000012s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000007s : 1: inline 0.00% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.25% : 0.000512s : 1: loop_unroll 0.00% : 0.000005s : 1: merge_cast_opt 0.00% : 0.000006s : 1: micro_interleaved_order_control 0.39% : 0.000801s : 1: mutable_eliminate 0.00% : 0.000008s : 1: offloading_packed_experts 0.01% : 0.000018s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000025s : 1: opt.transform.mutable_eliminate 0.82% : 0.001694s : 78: opt.transform.opt_a 0.01% : 0.000029s : 1: opt.transform.opt_after_cconv 0.01% : 0.000030s : 1: opt.transform.opt_after_jit_grad 0.05% : 0.000097s : 28: opt.transform.opt_b 0.02% : 0.000051s : 2: opt.transform.opt_trans_graph 0.02% : 0.000041s : 4: opt.transform.symbol_engine_opt 1.78% : 0.003679s : 1: opt_a 0.06% : 0.000117s : 1: opt_after_cconv 0.28% : 0.000581s : 1: opt_after_jit_grad 0.11% : 0.000220s : 1: opt_b 3.05% : 0.006276s : 1: optimize 0.01% : 0.000030s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000010s : 1: order_py_execute_after_rewriter 0.01% : 0.000029s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000005s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.03% : 0.000064s : 1: pre_auto_parallel 0.02% : 0.000038s : 1: py_interpret_to_execute 0.01% : 0.000019s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.01% : 0.000019s : 1: remove_dup_value 0.26% : 0.000530s : 1: renormalize.infer 0.23% : 0.000477s : 1: renormalize.specialize 0.00% : 0.000005s : 1: reorder_send_recv_between_fp_bp 0.09% : 0.000191s : 1: rewriter_after_jit_bprop_graph 0.03% : 0.000055s : 1: rewriter_after_opt_a 0.06% : 0.000115s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000004s : 1: split_layernorm_comm 0.00% : 0.000006s : 1: split_matmul_comm_elemetwise 0.00% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.04% : 0.000091s : 1: symbol_engine_optimizer 0.05% : 0.000108s : 1: tuple_transform 75.72% : 0.156068s : 1: type_inference TotalTime = 0.113501, [21] [bootstrap]: 0.00043836 [type_inference]: 0.0622204 [event_method]: 8.071e-05 [auto_monad]: 0.00012798 [graph_reusing]: 9.61e-06 [inline]: 2.72001e-06 [add_attr]: 0.00451368, [1] [add_attr_with_inline]: 0.00449914, [1] [Cycle 1]: 0.00011405, [2] [tag_attr]: 5.334e-05 [meta_addattr_fg_expand]: 1.065e-05 [parallel-infer-symbol]: 4.25999e-06 [pre_auto_parallel]: 8.414e-05 [insert-virtual-dataset]: 2.41e-06 [parallel-infer-symbol-second]: 7.2e-07 [dataset_repeat_opt]: 2.39999e-06 [pipeline_split]: 1.72999e-06 [optimize]: 0.0451565, [53] [py_interpret_to_execute]: 0.00011948 [rewriter_before_opt_a]: 0.00021272 [opt_a]: 0.0422608, [3] [Cycle 1]: 0.03634, [45] [expand_dump_flag]: 5.92001e-06 [switch_simplify]: 9.275e-05 [loop_unroll]: 7.411e-05 [a_1]: 0.00196372 [with_stream_mark]: 4.182e-05 [recompute_prepare]: 2.999e-05 [updatestate_depend_eliminate]: 1.097e-05 [updatestate_assign_eliminate]: 8.07e-06 [updatestate_loads_eliminate]: 7.27002e-06 [parameter_eliminate]: 4.02002e-06 [a_2]: 0.00024878 [accelerated_algorithm]: 3.657e-05 [shard]: 1.92999e-06 [meta_shard_fg_expand]: 5.24e-06 [shard_inline]: 1.573e-05 [merge_send_recv]: 1.987e-05 [auto_parallel]: 1.654e-05 [parallel]: 2.106e-05 [flash_sp]: 1.365e-05 [merge_comm]: 1.025e-05 [allreduce_fusion]: 9.01002e-06 [matmul_add_comm_reduction]: 3.379e-05 [allreduce_slice_to_reducescatter]: 9.10019e-07 [virtual_shard_identity]: 2.013e-05 [virtual_dataset]: 1.599e-05 [get_grad_eliminate_]: 1.454e-05 [virtual_output]: 7.436e-05 [merge_forward]: 1.214e-05 [cell_reuse_recompute_pass]: 1.40999e-06 [offload_activation]: 2.274e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.473e-05 [merge_recompute_call_nodes]: 1.62001e-06 [before_grad]: 2.906e-05 [set_forward_comm_id_for_comm_node_pass]: 1.098e-05 [meta_fg_expand]: 0.0155041 [flash_sp_send_recv_attached]: 8.55999e-06 [receive_attached]: 2.61e-06 [after_resolve]: 0.00011569 [a_after_grad]: 0.00010297 [renormalize]: 0.00389943 [add_forward_monad_depend]: 1.762e-05 [auto_monad_grad]: 6.96999e-06 [auto_monad_eliminator]: 0.0125595 [cse]: 0.00034941 [a_3]: 0.00038029 [Cycle 2]: 0.00487833, [45] [expand_dump_flag]: 3.40998e-06 [switch_simplify]: 4.908e-05 [loop_unroll]: 0.00013255 [a_1]: 0.00253655 [with_stream_mark]: 3.142e-05 [recompute_prepare]: 1.394e-05 [updatestate_depend_eliminate]: 4.46002e-06 [updatestate_assign_eliminate]: 3.45e-06 [updatestate_loads_eliminate]: 2.83998e-06 [parameter_eliminate]: 2.33998e-06 [a_2]: 9.482e-05 [accelerated_algorithm]: 1.84e-05 [shard]: 2.22999e-06 [meta_shard_fg_expand]: 2.68e-06 [shard_inline]: 7.04001e-06 [merge_send_recv]: 9.64e-06 [auto_parallel]: 1.139e-05 [parallel]: 1.08e-05 [flash_sp]: 4.70001e-06 [merge_comm]: 4.08001e-06 [allreduce_fusion]: 3.41001e-06 [matmul_add_comm_reduction]: 1.128e-05 [allreduce_slice_to_reducescatter]: 1.09e-06 [virtual_shard_identity]: 8.97999e-06 [virtual_dataset]: 6.34999e-06 [get_grad_eliminate_]: 6.11e-06 [virtual_output]: 6.34001e-06 [merge_forward]: 4.65999e-06 [cell_reuse_recompute_pass]: 2.16e-06 [offload_activation]: 1.065e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.38e-05 [merge_recompute_call_nodes]: 1.75001e-06 [before_grad]: 1.08e-05 [set_forward_comm_id_for_comm_node_pass]: 4.15e-06 [meta_fg_expand]: 0.00032815 [flash_sp_send_recv_attached]: 3.78001e-06 [receive_attached]: 2.94001e-06 [after_resolve]: 2.582e-05 [a_after_grad]: 1.139e-05 [renormalize]: 0.0010255 [add_forward_monad_depend]: 8.03001e-06 [auto_monad_grad]: 2.78998e-06 [auto_monad_eliminator]: 1.814e-05 [cse]: 3.244e-05 [a_3]: 5.178e-05 [Cycle 3]: 0.00102036, [45] [expand_dump_flag]: 3.36001e-06 [switch_simplify]: 7.37002e-06 [loop_unroll]: 6.06e-06 [a_1]: 0.00012223 [with_stream_mark]: 1.635e-05 [recompute_prepare]: 5.90002e-06 [updatestate_depend_eliminate]: 4.76997e-06 [updatestate_assign_eliminate]: 3.06001e-06 [updatestate_loads_eliminate]: 2.56998e-06 [parameter_eliminate]: 1.42e-06 [a_2]: 6.604e-05 [accelerated_algorithm]: 1.5e-05 [shard]: 1.89e-06 [meta_shard_fg_expand]: 1.55999e-06 [shard_inline]: 6.14001e-06 [merge_send_recv]: 6.86001e-06 [auto_parallel]: 9.58002e-06 [parallel]: 8.70001e-06 [flash_sp]: 1.60999e-06 [merge_comm]: 3.46999e-06 [allreduce_fusion]: 3.06999e-06 [matmul_add_comm_reduction]: 9.98998e-06 [allreduce_slice_to_reducescatter]: 4.90021e-07 [virtual_shard_identity]: 7.87998e-06 [virtual_dataset]: 5.55001e-06 [get_grad_eliminate_]: 5.66e-06 [virtual_output]: 5.37999e-06 [merge_forward]: 3.98999e-06 [cell_reuse_recompute_pass]: 2.32001e-06 [offload_activation]: 1.082e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.201e-05 [merge_recompute_call_nodes]: 1.49e-06 [before_grad]: 9.37999e-06 [set_forward_comm_id_for_comm_node_pass]: 3.46001e-06 [meta_fg_expand]: 2.34001e-06 [flash_sp_send_recv_attached]: 1.54998e-06 [receive_attached]: 2.19999e-06 [after_resolve]: 6.547e-05 [a_after_grad]: 1.105e-05 [renormalize]: 5.00004e-08 [add_forward_monad_depend]: 6.21998e-06 [auto_monad_grad]: 2.18002e-06 [auto_monad_eliminator]: 1.716e-05 [cse]: 2.142e-05 [a_3]: 4.261e-05 [py_interpret_to_execute_after_opt_a]: 1.766e-05 [slice_cell_reuse_recomputed_activation]: 2.41e-06 [rewriter_after_opt_a]: 4.877e-05 [convert_after_rewriter]: 7.46001e-06 [order_py_execute_after_rewriter]: 4.84003e-06 [mutable_eliminate]: 0.00083891 [opt_b]: 0.0002467, [1] [Cycle 1]: 0.00023709, [7] [b_1]: 0.00014543 [b_2]: 8.95001e-06 [updatestate_depend_eliminate]: 1.03e-05 [updatestate_assign_eliminate]: 3.21999e-06 [updatestate_loads_eliminate]: 2.48e-06 [renormalize]: 6.19999e-07 [cse]: 2.787e-05 [optimize_parallel_all_gather_comm]: 2.106e-05 [overlap_param_gather]: 2.09e-06 [cconv]: 3.53e-05 [loop_unroll]: 0.00059577 [opt_after_cconv]: 0.00012765, [1] [Cycle 1]: 0.00012041, [7] [c_1]: 3.026e-05 [parameter_eliminate]: 4.62e-06 [updatestate_depend_eliminate]: 6.34001e-06 [updatestate_assign_eliminate]: 2.094e-05 [updatestate_loads_eliminate]: 2.63003e-06 [cse]: 2.062e-05 [renormalize]: 6.30011e-07 [remove_dup_value]: 1.429e-05 [tuple_transform]: 8.024e-05, [1] [Cycle 1]: 7.529e-05, [4] [d_1]: 4.723e-05 [none_parameter_eliminate]: 2.14e-06 [renormalize]: 3.30008e-07 [switch_simplify]: 7.25e-06 [partial_unused_args_eliminate]: 2.04e-06 [add_recomputation]: 5.277e-05 [cse_after_recomputation]: 2.181e-05, [1] [Cycle 1]: 1.684e-05, [1] [cse]: 1.146e-05 [environ_conv]: 1.055e-05 [swap_dp_allreduce_reducescatter]: 5.44e-06 [bias_add_comm_swap]: 2.94999e-06 [label_micro_interleaved_index]: 4.99e-06 [label_fine_grained_interleaved_index]: 3.11999e-06 [merge_cast_opt]: 1.69e-06 [slice_recompute_activation]: 2.05002e-06 [micro_interleaved_order_control]: 2.74999e-06 [assign_add_opt]: 1.27e-06 [ForceFp32Comm]: 8.59989e-07 [remove_cast_before_assign_add]: 1.20999e-06 [full_micro_interleaved_order_control]: 2.71e-06 [reorder_send_recv_between_fp_bp]: 2.56998e-06 [comm_op_add_attrs]: 1.29e-06 [add_comm_op_reuse_tag]: 9.79984e-07 [interleave_split_concat_branches]: 1.14998e-06 [interleave_parallel_branches]: 1.60001e-06 [overlap_opt_shard_in_pipeline]: 1.17e-06 [overlap_opt_shard_grad_in_pipeline]: 1.67001e-06 [control_data_broadcast_order]: 1.254e-05 [grouped_pairwise_exchange_alltoall]: 1.84e-06 [offloading_packed_experts]: 4.34997e-06 [overlap_recompute_and_grad_model_parallel]: 5.40999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.24e-06 [overlap_recompute_allgather_and_fa_grad]: 1.84e-06 [overlap_recompute_comm]: 2.13998e-06 [overlap_grad_ring_attention]: 4.02998e-06 [overlap_grad_flash_sp]: 2.064e-05 [begin_end_overlap_inline]: 5.00004e-07 [split_matmul_comm_elemetwise]: 2.49001e-06 [split_layernorm_comm]: 1.60001e-06 [handle_group_info]: 1.17999e-06 [symbol_engine_optimizer]: 8.683e-05, [1] [Cycle 1]: 8.21e-05, [6] [build]: 1.23e-05 [elim_shapecalc]: 1.066e-05 [elim_not_effective]: 1.343e-05 [opt_reshape]: 6.78998e-06 [fold_const_symbol]: 9.47001e-06 [renormalize]: 2.19996e-07 [detach_backward]: 2.19001e-06 [pipeline_parallel_scheduler]: 1.66e-06 [auto_monad_reorder]: 1.707e-05 [get_jit_bprop_graph]: 2.24001e-06 [rewriter_after_jit_bprop_graph]: 6.83e-06 [opt_after_jit_grad]: 0.00058913 [validate]: 5.141e-05 Sums bootstrap : 0.000438s : 0.41% type_inference : 0.062220s : 58.07% event_method : 0.000081s : 0.08% auto_monad : 0.000128s : 0.12% graph_reusing : 0.000010s : 0.01% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000053s : 0.05% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000011s : 0.01% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000084s : 0.08% insert-virtual-dataset : 0.000002s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000119s : 0.11% optimize.rewriter_before_opt_a : 0.000213s : 0.20% optimize.opt_a.expand_dump_flag : 0.000013s : 0.01% optimize.opt_a.switch_simplify : 0.000149s : 0.14% optimize.opt_a.loop_unroll : 0.000213s : 0.20% optimize.opt_a.a_1 : 0.004622s : 4.31% optimize.opt_a.with_stream_mark : 0.000090s : 0.08% optimize.opt_a.recompute_prepare : 0.000050s : 0.05% optimize.opt_a.updatestate_depend_eliminate : 0.000020s : 0.02% optimize.opt_a.updatestate_assign_eliminate : 0.000015s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000013s : 0.01% optimize.opt_a.parameter_eliminate : 0.000008s : 0.01% optimize.opt_a.a_2 : 0.000410s : 0.38% optimize.opt_a.accelerated_algorithm : 0.000070s : 0.07% optimize.opt_a.shard : 0.000006s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000009s : 0.01% optimize.opt_a.shard_inline : 0.000029s : 0.03% optimize.opt_a.merge_send_recv : 0.000036s : 0.03% optimize.opt_a.auto_parallel : 0.000038s : 0.04% optimize.opt_a.parallel : 0.000041s : 0.04% optimize.opt_a.flash_sp : 0.000020s : 0.02% optimize.opt_a.merge_comm : 0.000018s : 0.02% optimize.opt_a.allreduce_fusion : 0.000015s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000055s : 0.05% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000037s : 0.03% optimize.opt_a.virtual_dataset : 0.000028s : 0.03% optimize.opt_a.get_grad_eliminate_ : 0.000026s : 0.02% optimize.opt_a.virtual_output : 0.000086s : 0.08% optimize.opt_a.merge_forward : 0.000021s : 0.02% optimize.opt_a.cell_reuse_recompute_pass : 0.000006s : 0.01% optimize.opt_a.offload_activation : 0.000044s : 0.04% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000061s : 0.06% optimize.opt_a.merge_recompute_call_nodes : 0.000005s : 0.00% optimize.opt_a.before_grad : 0.000049s : 0.05% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000019s : 0.02% optimize.opt_a.meta_fg_expand : 0.015835s : 14.78% optimize.opt_a.flash_sp_send_recv_attached : 0.000014s : 0.01% optimize.opt_a.receive_attached : 0.000008s : 0.01% optimize.opt_a.after_resolve : 0.000207s : 0.19% optimize.opt_a.a_after_grad : 0.000125s : 0.12% optimize.opt_a.renormalize : 0.004925s : 4.60% optimize.opt_a.add_forward_monad_depend : 0.000032s : 0.03% optimize.opt_a.auto_monad_grad : 0.000012s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.012595s : 11.75% optimize.opt_a.cse : 0.000403s : 0.38% optimize.opt_a.a_3 : 0.000475s : 0.44% optimize.py_interpret_to_execute_after_opt_a : 0.000018s : 0.02% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000049s : 0.05% optimize.convert_after_rewriter : 0.000007s : 0.01% optimize.order_py_execute_after_rewriter : 0.000005s : 0.00% optimize.mutable_eliminate : 0.000839s : 0.78% optimize.opt_b.b_1 : 0.000145s : 0.14% optimize.opt_b.b_2 : 0.000009s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000010s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000028s : 0.03% optimize.optimize_parallel_all_gather_comm : 0.000021s : 0.02% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000035s : 0.03% optimize.loop_unroll : 0.000596s : 0.56% optimize.opt_after_cconv.c_1 : 0.000030s : 0.03% optimize.opt_after_cconv.parameter_eliminate : 0.000005s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000021s : 0.02% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.cse : 0.000021s : 0.02% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000014s : 0.01% optimize.tuple_transform.d_1 : 0.000047s : 0.04% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000053s : 0.05% optimize.cse_after_recomputation.cse : 0.000011s : 0.01% optimize.environ_conv : 0.000011s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.01% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000003s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000002s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000013s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000002s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000021s : 0.02% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000012s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000011s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000013s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.01% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000009s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000017s : 0.02% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000007s : 0.01% opt_after_jit_grad : 0.000589s : 0.55% validate : 0.000051s : 0.05% Time group info: ------[substitution.] 0.001953 195 5.11% : 0.000100s : 7: substitution.arithmetic_simplify 0.13% : 0.000003s : 2: substitution.elim_not_effective 0.31% : 0.000006s : 6: substitution.float_depend_g_call 0.23% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.07% : 0.000001s : 2: substitution.fold_const_symbol 0.33% : 0.000006s : 4: substitution.graph_param_transform 0.17% : 0.000003s : 2: substitution.incorporate_call 0.11% : 0.000002s : 2: substitution.incorporate_call_switch 72.11% : 0.001408s : 18: substitution.inline 1.44% : 0.000028s : 2: substitution.inline_without_move 0.51% : 0.000010s : 14: substitution.j_node_and_user_rematch 1.08% : 0.000021s : 3: substitution.less_batch_normalization 0.88% : 0.000017s : 11: substitution.minmaximum_grad 0.45% : 0.000009s : 6: substitution.partial_eliminate 0.63% : 0.000012s : 14: substitution.remove_not_recompute_node 2.24% : 0.000044s : 10: substitution.replace_applicator 0.85% : 0.000017s : 12: substitution.replace_old_param 0.22% : 0.000004s : 1: substitution.set_cell_output_no_recompute 1.83% : 0.000036s : 11: substitution.tuple_list_convert_item_index_to_positive 0.84% : 0.000016s : 11: substitution.tuple_list_get_item_const_eliminator 1.22% : 0.000024s : 11: substitution.tuple_list_get_item_depend_reorder 5.66% : 0.000111s : 32: substitution.tuple_list_get_item_eliminator 3.57% : 0.000070s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.062102 2 93.63% : 0.058146s : 1: type_inference.infer 6.37% : 0.003956s : 1: type_inference.specialize ------[replace.] 0.000346 38 4.34% : 0.000015s : 2: replace.arithmetic_simplify 54.75% : 0.000190s : 18: replace.inline 40.91% : 0.000142s : 18: replace.tuple_list_get_item_eliminator ------[match.] 0.001507 38 3.98% : 0.000060s : 2: match.arithmetic_simplify 92.63% : 0.001396s : 18: match.inline 3.39% : 0.000051s : 18: match.tuple_list_get_item_eliminator ------[predicate.] 0.000739 4976 1.08% : 0.000008s : 61: predicate.accumulaten_eliminater 0.23% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.41% : 0.000003s : 23: predicate.addn_check_dump 1.05% : 0.000008s : 61: predicate.addn_zero_filter 0.99% : 0.000007s : 61: predicate.adjust_all_reduce_mul_add 2.09% : 0.000015s : 86: predicate.arithmetic_simplify 1.18% : 0.000009s : 63: predicate.cast_eliminate 1.05% : 0.000008s : 55: predicate.check_bprop_eliminate 0.37% : 0.000003s : 23: predicate.compare_switch_simplify 0.05% : 0.000000s : 4: predicate.const_output_eliminate 0.42% : 0.000003s : 23: predicate.depend_value_elim 1.20% : 0.000009s : 63: predicate.dict_get_item_const_eliminator 1.28% : 0.000009s : 63: predicate.dict_get_item_eliminator 1.15% : 0.000009s : 63: predicate.dict_set_item_eliminator 0.30% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.06% : 0.000000s : 4: predicate.elim_not_effective 0.11% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.21% : 0.000009s : 67: predicate.environ_add_const_eliminate 1.18% : 0.000009s : 67: predicate.environ_get_add_eliminate 1.12% : 0.000008s : 67: predicate.environ_get_depend_swap 1.53% : 0.000011s : 90: predicate.environ_get_eliminate 1.23% : 0.000009s : 67: predicate.environ_get_set_eliminate 1.76% : 0.000013s : 99: predicate.exchange_switch_depend_value 2.73% : 0.000020s : 99: predicate.float_depend_g_call 0.40% : 0.000003s : 23: predicate.float_environ_get_switch 0.59% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.04% : 0.000000s : 4: predicate.fold_const_symbol 0.53% : 0.000004s : 23: predicate.get_grad_eliminate 0.06% : 0.000000s : 4: predicate.graph_param_transform 0.44% : 0.000003s : 23: predicate.incorporate_call 0.39% : 0.000003s : 23: predicate.incorporate_call_switch 5.89% : 0.000044s : 214: predicate.inline 1.48% : 0.000011s : 46: predicate.inline_without_move 0.21% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.87% : 0.000006s : 23: predicate.less_batch_normalization 1.62% : 0.000012s : 89: predicate.list_to_tuple_eliminator_ 2.52% : 0.000019s : 150: predicate.load_eliminater 0.32% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.73% : 0.000020s : 141: predicate.loop_unroll_before_grad 1.42% : 0.000010s : 71: predicate.make_slice_get_slice_eliminator 0.42% : 0.000003s : 23: predicate.merge_addn 0.97% : 0.000007s : 55: predicate.micro_step_allgather_replace 1.05% : 0.000008s : 55: predicate.mini_step_allgather_replace 1.23% : 0.000009s : 63: predicate.minmaximum_grad 0.52% : 0.000004s : 4: predicate.mutable_eliminate 0.08% : 0.000001s : 4: predicate.opt_reshape 0.10% : 0.000001s : 4: predicate.parallel_virtual_node 2.63% : 0.000019s : 99: predicate.partial_defer_inline 1.65% : 0.000012s : 85: predicate.partial_eliminate 1.07% : 0.000008s : 61: predicate.print_const_string_wrapper 0.41% : 0.000003s : 23: predicate.reduce_all_const_elim 1.38% : 0.000010s : 63: predicate.reduce_eliminate 2.54% : 0.000019s : 150: predicate.redundant_stop_gradient_eliminater 0.33% : 0.000002s : 23: predicate.remove_not_recompute_node 1.94% : 0.000014s : 136: predicate.replace_applicator 0.78% : 0.000006s : 46: predicate.replace_old_param 0.10% : 0.000001s : 4: predicate.reset_defer_inline 1.13% : 0.000008s : 63: predicate.reshape_eliminate 1.08% : 0.000008s : 55: predicate.row_tensor_add_zeros_like 0.13% : 0.000001s : 4: predicate.row_tensor_eliminate 1.38% : 0.000010s : 55: predicate.same_eliminate 0.28% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.51% : 0.000004s : 23: predicate.shard_identity_eliminate 0.19% : 0.000001s : 8: predicate.special_op_eliminate 0.53% : 0.000004s : 23: predicate.specialize_transform 1.36% : 0.000010s : 55: predicate.split_environ_get_set_with_tuple_value 1.29% : 0.000010s : 46: predicate.stack_unstack_eliminate 0.11% : 0.000001s : 4: predicate.switch_call_monad_eliminater 2.01% : 0.000015s : 99: predicate.switch_defer_inline 2.81% : 0.000021s : 154: predicate.switch_layer_defer_inline 5.14% : 0.000038s : 267: predicate.switch_simplify 1.11% : 0.000008s : 63: predicate.tile_eliminate 1.06% : 0.000008s : 63: predicate.transpose_eliminate 1.39% : 0.000010s : 71: predicate.tuple_list_convert_item_index_to_positive 1.44% : 0.000011s : 71: predicate.tuple_list_get_item_const_eliminator 1.27% : 0.000009s : 71: predicate.tuple_list_get_item_depend_reorder 3.37% : 0.000025s : 112: predicate.tuple_list_get_item_eliminator 1.35% : 0.000010s : 71: predicate.tuple_list_get_set_item_eliminator 1.85% : 0.000014s : 94: predicate.tuple_list_set_item_eliminator 1.64% : 0.000012s : 89: predicate.tuple_to_list_eliminator_ 2.48% : 0.000018s : 150: predicate.updatestate_pure_node_eliminater 3.02% : 0.000022s : 173: predicate.updatestate_useless_node_eliminater 0.11% : 0.000001s : 4: predicate.value_based_eliminate 0.53% : 0.000004s : 23: predicate.virtual_dataset_eliminate 0.78% : 0.000006s : 23: predicate.virtual_output_eliminate 0.07% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.09% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002603 41 60.52% : 0.001576s : 19: func_graph_cloner_run.FuncGraphClonerGraph 39.48% : 0.001028s : 22: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.174636 233 0.00% : 0.000004s : 1: ForceFp32Comm 2.59% : 0.004521s : 1: add_attr 2.58% : 0.004504s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.03% : 0.000057s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.08% : 0.000135s : 1: auto_monad 0.01% : 0.000021s : 1: auto_monad_reorder 0.00% : 0.000003s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 0.27% : 0.000470s : 1: bootstrap 0.02% : 0.000039s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.01% : 0.000017s : 1: control_data_broadcast_order 0.01% : 0.000011s : 1: convert_after_rewriter 0.01% : 0.000025s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.01% : 0.000014s : 1: environ_conv 0.05% : 0.000089s : 1: event_method 0.00% : 0.000007s : 1: full_micro_interleaved_order_control 0.00% : 0.000007s : 1: get_jit_bprop_graph 0.01% : 0.000014s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.35% : 0.000605s : 1: loop_unroll 0.00% : 0.000005s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.49% : 0.000850s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.01% : 0.000015s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000021s : 1: opt.transform.mutable_eliminate 3.67% : 0.006413s : 117: opt.transform.opt_a 0.02% : 0.000029s : 1: opt.transform.opt_after_cconv 0.01% : 0.000025s : 1: opt.transform.opt_after_jit_grad 0.07% : 0.000124s : 28: opt.transform.opt_b 0.03% : 0.000052s : 2: opt.transform.opt_trans_graph 0.02% : 0.000037s : 4: opt.transform.symbol_engine_opt 24.20% : 0.042264s : 1: opt_a 0.08% : 0.000131s : 1: opt_after_cconv 0.34% : 0.000600s : 1: opt_after_jit_grad 0.14% : 0.000250s : 1: opt_b 25.86% : 0.045163s : 1: optimize 0.01% : 0.000025s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000008s : 1: order_py_execute_after_rewriter 0.01% : 0.000025s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000004s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000005s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000006s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.05% : 0.000091s : 1: pre_auto_parallel 0.07% : 0.000130s : 1: py_interpret_to_execute 0.01% : 0.000021s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.01% : 0.000018s : 1: remove_dup_value 1.54% : 0.002691s : 2: renormalize.infer 1.27% : 0.002210s : 2: renormalize.specialize 0.00% : 0.000005s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000010s : 1: rewriter_after_jit_bprop_graph 0.03% : 0.000053s : 1: rewriter_after_opt_a 0.13% : 0.000221s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000004s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.05% : 0.000090s : 1: symbol_engine_optimizer 0.05% : 0.000083s : 1: tuple_transform 35.65% : 0.062249s : 1: type_inference TotalTime = 0.0718861, [21] [bootstrap]: 0.00069601 [type_inference]: 0.0590198 [event_method]: 4.771e-05 [auto_monad]: 7.591e-05 [graph_reusing]: 6.51999e-06 [inline]: 2.60002e-06 [add_attr]: 0.00476814, [1] [add_attr_with_inline]: 0.00475553, [1] [Cycle 1]: 0.00019016, [2] [tag_attr]: 3.479e-05 [meta_addattr_fg_expand]: 7.87e-06 [parallel-infer-symbol]: 4.18001e-06 [pre_auto_parallel]: 5.157e-05 [insert-virtual-dataset]: 2.43e-06 [parallel-infer-symbol-second]: 8.79983e-07 [dataset_repeat_opt]: 2.74999e-06 [pipeline_split]: 1.57999e-06 [optimize]: 0.00635429, [53] [py_interpret_to_execute]: 3.538e-05 [rewriter_before_opt_a]: 0.00010978 [opt_a]: 0.00361597, [2] [Cycle 1]: 0.002932, [45] [expand_dump_flag]: 3.4e-06 [switch_simplify]: 5.232e-05 [loop_unroll]: 3.685e-05 [a_1]: 0.00098138 [with_stream_mark]: 2.49e-05 [recompute_prepare]: 1.179e-05 [updatestate_depend_eliminate]: 4.22998e-06 [updatestate_assign_eliminate]: 3.38e-06 [updatestate_loads_eliminate]: 2.71e-06 [parameter_eliminate]: 2.39001e-06 [a_2]: 8.581e-05 [accelerated_algorithm]: 7.29001e-06 [shard]: 2.29999e-06 [meta_shard_fg_expand]: 2.57001e-06 [shard_inline]: 6.56999e-06 [merge_send_recv]: 8.82999e-06 [auto_parallel]: 8.53001e-06 [parallel]: 2e-05 [flash_sp]: 9.49999e-06 [merge_comm]: 3.81001e-06 [allreduce_fusion]: 3.28e-06 [matmul_add_comm_reduction]: 1.02e-05 [allreduce_slice_to_reducescatter]: 1.09998e-06 [virtual_shard_identity]: 9.20001e-06 [virtual_dataset]: 6.38003e-06 [get_grad_eliminate_]: 5.82001e-06 [virtual_output]: 6.29001e-06 [merge_forward]: 4.45999e-06 [cell_reuse_recompute_pass]: 2.34999e-06 [offload_activation]: 1.121e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.254e-05 [merge_recompute_call_nodes]: 1.47999e-06 [before_grad]: 1.053e-05 [set_forward_comm_id_for_comm_node_pass]: 3.56999e-06 [meta_fg_expand]: 2.78e-06 [flash_sp_send_recv_attached]: 2.66e-06 [receive_attached]: 2.49001e-06 [after_resolve]: 1.186e-05 [a_after_grad]: 8.64e-06 [renormalize]: 0.00104756 [add_forward_monad_depend]: 7.3e-06 [auto_monad_grad]: 2.76e-06 [auto_monad_eliminator]: 1.758e-05 [cse]: 3.288e-05 [a_3]: 0.00011604 [Cycle 2]: 0.00067058, [45] [expand_dump_flag]: 2.86999e-06 [switch_simplify]: 9.09e-06 [loop_unroll]: 5.56e-06 [a_1]: 0.00014013 [with_stream_mark]: 1.901e-05 [recompute_prepare]: 6.33e-06 [updatestate_depend_eliminate]: 3.5e-06 [updatestate_assign_eliminate]: 2.68e-06 [updatestate_loads_eliminate]: 2.98e-06 [parameter_eliminate]: 1.34e-06 [a_2]: 7.214e-05 [accelerated_algorithm]: 5.62001e-06 [shard]: 1.64e-06 [meta_shard_fg_expand]: 1.54998e-06 [shard_inline]: 5.59e-06 [merge_send_recv]: 5.62001e-06 [auto_parallel]: 6.73e-06 [parallel]: 6.16e-06 [flash_sp]: 3.43e-06 [merge_comm]: 3.43999e-06 [allreduce_fusion]: 2.97002e-06 [matmul_add_comm_reduction]: 7.98001e-06 [allreduce_slice_to_reducescatter]: 6.29982e-07 [virtual_shard_identity]: 6.31998e-06 [virtual_dataset]: 5.14e-06 [get_grad_eliminate_]: 5.19e-06 [virtual_output]: 4.99e-06 [merge_forward]: 3.10998e-06 [cell_reuse_recompute_pass]: 2.48002e-06 [offload_activation]: 8.36002e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.116e-05 [merge_recompute_call_nodes]: 1.15001e-06 [before_grad]: 8.92e-06 [set_forward_comm_id_for_comm_node_pass]: 4.14002e-06 [meta_fg_expand]: 2.27999e-06 [flash_sp_send_recv_attached]: 1.33002e-06 [receive_attached]: 2.31e-06 [after_resolve]: 1.137e-05 [a_after_grad]: 8.87e-06 [renormalize]: 8.00064e-08 [add_forward_monad_depend]: 1.57999e-06 [auto_monad_grad]: 1.20001e-06 [auto_monad_eliminator]: 8.48001e-06 [cse]: 1.754e-05 [a_3]: 3.351e-05 [py_interpret_to_execute_after_opt_a]: 1.262e-05 [slice_cell_reuse_recomputed_activation]: 1.96e-06 [rewriter_after_opt_a]: 3.857e-05 [convert_after_rewriter]: 6.86001e-06 [order_py_execute_after_rewriter]: 5.30999e-06 [mutable_eliminate]: 0.00080495 [opt_b]: 0.00019642, [1] [Cycle 1]: 0.00018885, [7] [b_1]: 0.00011191 [b_2]: 8.29002e-06 [updatestate_depend_eliminate]: 7.6e-06 [updatestate_assign_eliminate]: 2.99001e-06 [updatestate_loads_eliminate]: 2.16e-06 [renormalize]: 7.2e-07 [cse]: 2.092e-05 [optimize_parallel_all_gather_comm]: 1.864e-05 [overlap_param_gather]: 2.01e-06 [cconv]: 2.836e-05 [loop_unroll]: 0.0005986 [opt_after_cconv]: 0.00010998, [1] [Cycle 1]: 0.0001015, [7] [c_1]: 2.984e-05 [parameter_eliminate]: 4.4e-06 [updatestate_depend_eliminate]: 7.25998e-06 [updatestate_assign_eliminate]: 2.54999e-06 [updatestate_loads_eliminate]: 2.36e-06 [cse]: 2.022e-05 [renormalize]: 2.50002e-07 [remove_dup_value]: 1.4e-05 [tuple_transform]: 7.423e-05, [1] [Cycle 1]: 6.921e-05, [4] [d_1]: 4.275e-05 [none_parameter_eliminate]: 1.42e-06 [renormalize]: 1.69995e-07 [switch_simplify]: 6.17001e-06 [partial_unused_args_eliminate]: 3.10002e-06 [add_recomputation]: 5.799e-05 [cse_after_recomputation]: 2.801e-05, [1] [Cycle 1]: 2.242e-05, [1] [cse]: 1.533e-05 [environ_conv]: 5.64998e-06 [swap_dp_allreduce_reducescatter]: 5.25999e-06 [bias_add_comm_swap]: 2.64001e-06 [label_micro_interleaved_index]: 4.50999e-06 [label_fine_grained_interleaved_index]: 2.75002e-06 [merge_cast_opt]: 1.43002e-06 [slice_recompute_activation]: 2.36998e-06 [micro_interleaved_order_control]: 2.63e-06 [assign_add_opt]: 1.44e-06 [ForceFp32Comm]: 1.08001e-06 [remove_cast_before_assign_add]: 1.33002e-06 [full_micro_interleaved_order_control]: 2.14999e-06 [reorder_send_recv_between_fp_bp]: 3.01001e-06 [comm_op_add_attrs]: 1.15999e-06 [add_comm_op_reuse_tag]: 1.01002e-06 [interleave_split_concat_branches]: 1.77001e-06 [interleave_parallel_branches]: 1.05999e-06 [overlap_opt_shard_in_pipeline]: 1.59998e-06 [overlap_opt_shard_grad_in_pipeline]: 1.93002e-06 [control_data_broadcast_order]: 1.446e-05 [grouped_pairwise_exchange_alltoall]: 1.99999e-06 [offloading_packed_experts]: 3.88001e-06 [overlap_recompute_and_grad_model_parallel]: 5.67001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.50999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.45001e-06 [overlap_recompute_comm]: 2.44001e-06 [overlap_grad_ring_attention]: 4.43999e-06 [overlap_grad_flash_sp]: 2.113e-05 [begin_end_overlap_inline]: 7.30011e-07 [split_matmul_comm_elemetwise]: 2.39001e-06 [split_layernorm_comm]: 1.70001e-06 [handle_group_info]: 1.11002e-06 [symbol_engine_optimizer]: 0.00015399, [1] [Cycle 1]: 0.00014846, [6] [build]: 3.8e-06 [elim_shapecalc]: 7.41e-05 [elim_not_effective]: 1.729e-05 [opt_reshape]: 6.95002e-06 [fold_const_symbol]: 1.012e-05 [renormalize]: 2.80008e-07 [detach_backward]: 2.48e-06 [pipeline_parallel_scheduler]: 2.14e-06 [auto_monad_reorder]: 2.183e-05 [get_jit_bprop_graph]: 1.92001e-06 [rewriter_after_jit_bprop_graph]: 5.52999e-06 [opt_after_jit_grad]: 0.00059998 [validate]: 4.563e-05 Sums bootstrap : 0.000696s : 1.06% type_inference : 0.059020s : 89.47% event_method : 0.000048s : 0.07% auto_monad : 0.000076s : 0.12% graph_reusing : 0.000007s : 0.01% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000035s : 0.05% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000008s : 0.01% parallel-infer-symbol : 0.000004s : 0.01% pre_auto_parallel : 0.000052s : 0.08% insert-virtual-dataset : 0.000002s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000003s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000035s : 0.05% optimize.rewriter_before_opt_a : 0.000110s : 0.17% optimize.opt_a.expand_dump_flag : 0.000006s : 0.01% optimize.opt_a.switch_simplify : 0.000061s : 0.09% optimize.opt_a.loop_unroll : 0.000042s : 0.06% optimize.opt_a.a_1 : 0.001122s : 1.70% optimize.opt_a.with_stream_mark : 0.000044s : 0.07% optimize.opt_a.recompute_prepare : 0.000018s : 0.03% optimize.opt_a.updatestate_depend_eliminate : 0.000008s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.01% optimize.opt_a.parameter_eliminate : 0.000004s : 0.01% optimize.opt_a.a_2 : 0.000158s : 0.24% optimize.opt_a.accelerated_algorithm : 0.000013s : 0.02% optimize.opt_a.shard : 0.000004s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000004s : 0.01% optimize.opt_a.shard_inline : 0.000012s : 0.02% optimize.opt_a.merge_send_recv : 0.000014s : 0.02% optimize.opt_a.auto_parallel : 0.000015s : 0.02% optimize.opt_a.parallel : 0.000026s : 0.04% optimize.opt_a.flash_sp : 0.000013s : 0.02% optimize.opt_a.merge_comm : 0.000007s : 0.01% optimize.opt_a.allreduce_fusion : 0.000006s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000018s : 0.03% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000016s : 0.02% optimize.opt_a.virtual_dataset : 0.000012s : 0.02% optimize.opt_a.get_grad_eliminate_ : 0.000011s : 0.02% optimize.opt_a.virtual_output : 0.000011s : 0.02% optimize.opt_a.merge_forward : 0.000008s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000005s : 0.01% optimize.opt_a.offload_activation : 0.000020s : 0.03% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000024s : 0.04% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.00% optimize.opt_a.before_grad : 0.000019s : 0.03% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000008s : 0.01% optimize.opt_a.meta_fg_expand : 0.000005s : 0.01% optimize.opt_a.flash_sp_send_recv_attached : 0.000004s : 0.01% optimize.opt_a.receive_attached : 0.000005s : 0.01% optimize.opt_a.after_resolve : 0.000023s : 0.04% optimize.opt_a.a_after_grad : 0.000018s : 0.03% optimize.opt_a.renormalize : 0.001048s : 1.59% optimize.opt_a.add_forward_monad_depend : 0.000009s : 0.01% optimize.opt_a.auto_monad_grad : 0.000004s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000026s : 0.04% optimize.opt_a.cse : 0.000050s : 0.08% optimize.opt_a.a_3 : 0.000150s : 0.23% optimize.py_interpret_to_execute_after_opt_a : 0.000013s : 0.02% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000039s : 0.06% optimize.convert_after_rewriter : 0.000007s : 0.01% optimize.order_py_execute_after_rewriter : 0.000005s : 0.01% optimize.mutable_eliminate : 0.000805s : 1.22% optimize.opt_b.b_1 : 0.000112s : 0.17% optimize.opt_b.b_2 : 0.000008s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000008s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000021s : 0.03% optimize.optimize_parallel_all_gather_comm : 0.000019s : 0.03% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000028s : 0.04% optimize.loop_unroll : 0.000599s : 0.91% optimize.opt_after_cconv.c_1 : 0.000030s : 0.05% optimize.opt_after_cconv.parameter_eliminate : 0.000004s : 0.01% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000007s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000020s : 0.03% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000014s : 0.02% optimize.tuple_transform.d_1 : 0.000043s : 0.06% optimize.tuple_transform.none_parameter_eliminate : 0.000001s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000006s : 0.01% optimize.partial_unused_args_eliminate : 0.000003s : 0.00% optimize.add_recomputation : 0.000058s : 0.09% optimize.cse_after_recomputation.cse : 0.000015s : 0.02% optimize.environ_conv : 0.000006s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.01% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.01% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000002s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000002s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000014s : 0.02% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.01% optimize.overlap_recompute_and_grad_model_parallel : 0.000006s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000002s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.01% optimize.overlap_grad_flash_sp : 0.000021s : 0.03% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000004s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000074s : 0.11% optimize.symbol_engine_optimizer.elim_not_effective : 0.000017s : 0.03% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.01% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.02% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000022s : 0.03% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000006s : 0.01% opt_after_jit_grad : 0.000600s : 0.91% validate : 0.000046s : 0.07% Time group info: ------[substitution.] 0.000312 40 13.76% : 0.000043s : 8: substitution.arithmetic_simplify 0.81% : 0.000003s : 2: substitution.elim_not_effective 0.45% : 0.000001s : 2: substitution.fold_const_symbol 1.99% : 0.000006s : 4: substitution.graph_param_transform 71.38% : 0.000223s : 6: substitution.inline 1.25% : 0.000004s : 4: substitution.j_node_and_user_rematch 1.48% : 0.000005s : 4: substitution.remove_not_recompute_node 1.89% : 0.000006s : 4: substitution.replace_old_param 6.98% : 0.000022s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.058935 2 97.73% : 0.057598s : 1: type_inference.infer 2.27% : 0.001337s : 1: type_inference.specialize ------[replace.] 0.000096 12 56.91% : 0.000055s : 6: replace.inline 43.09% : 0.000042s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000238 12 91.91% : 0.000219s : 6: match.inline 8.09% : 0.000019s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000286 1534 0.79% : 0.000002s : 17: predicate.accumulaten_eliminater 0.89% : 0.000003s : 4: predicate.ad_related_special_op_eliminate 0.31% : 0.000001s : 8: predicate.addn_check_dump 0.91% : 0.000003s : 17: predicate.addn_zero_filter 0.73% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.23% : 0.000006s : 25: predicate.arithmetic_simplify 0.88% : 0.000003s : 17: predicate.cast_eliminate 0.40% : 0.000001s : 8: predicate.check_bprop_eliminate 0.37% : 0.000001s : 8: predicate.compare_switch_simplify 0.13% : 0.000000s : 4: predicate.const_output_eliminate 0.41% : 0.000001s : 8: predicate.depend_value_elim 0.81% : 0.000002s : 17: predicate.dict_get_item_const_eliminator 0.96% : 0.000003s : 17: predicate.dict_get_item_eliminator 0.78% : 0.000002s : 17: predicate.dict_set_item_eliminator 0.95% : 0.000003s : 8: predicate.dumpgradient_eliminate 0.22% : 0.000001s : 4: predicate.elim_not_effective 0.63% : 0.000002s : 4: predicate.elim_shapecalc_of_broadcastargs 0.91% : 0.000003s : 21: predicate.environ_add_const_eliminate 0.97% : 0.000003s : 21: predicate.environ_get_add_eliminate 0.94% : 0.000003s : 21: predicate.environ_get_depend_swap 1.32% : 0.000004s : 29: predicate.environ_get_eliminate 1.00% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.32% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.21% : 0.000006s : 29: predicate.float_depend_g_call 0.37% : 0.000001s : 8: predicate.float_environ_get_switch 0.48% : 0.000001s : 12: predicate.float_tuple_getitem_switch 0.11% : 0.000000s : 4: predicate.fold_const_symbol 0.46% : 0.000001s : 8: predicate.get_grad_eliminate 0.21% : 0.000001s : 4: predicate.graph_param_transform 0.46% : 0.000001s : 8: predicate.incorporate_call 0.33% : 0.000001s : 8: predicate.incorporate_call_switch 5.36% : 0.000015s : 70: predicate.inline 0.45% : 0.000001s : 8: predicate.inline_without_move 0.22% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.55% : 0.000002s : 8: predicate.less_batch_normalization 1.57% : 0.000004s : 31: predicate.list_to_tuple_eliminator_ 2.12% : 0.000006s : 48: predicate.load_eliminater 1.21% : 0.000003s : 4: predicate.loop_unroll_after_grad 2.12% : 0.000006s : 47: predicate.loop_unroll_before_grad 1.35% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.37% : 0.000001s : 8: predicate.merge_addn 0.38% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.41% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.74% : 0.000002s : 17: predicate.minmaximum_grad 1.26% : 0.000004s : 4: predicate.mutable_eliminate 0.23% : 0.000001s : 4: predicate.opt_reshape 0.36% : 0.000001s : 4: predicate.parallel_virtual_node 1.74% : 0.000005s : 29: predicate.partial_defer_inline 1.32% : 0.000004s : 27: predicate.partial_eliminate 0.79% : 0.000002s : 17: predicate.print_const_string_wrapper 0.36% : 0.000001s : 8: predicate.reduce_all_const_elim 1.06% : 0.000003s : 17: predicate.reduce_eliminate 2.16% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.43% : 0.000001s : 8: predicate.remove_not_recompute_node 1.25% : 0.000004s : 31: predicate.replace_applicator 0.33% : 0.000001s : 8: predicate.replace_old_param 0.25% : 0.000001s : 4: predicate.reset_defer_inline 0.90% : 0.000003s : 17: predicate.reshape_eliminate 0.46% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.24% : 0.000001s : 4: predicate.row_tensor_eliminate 0.76% : 0.000002s : 8: predicate.same_eliminate 0.41% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.56% : 0.000002s : 8: predicate.shard_identity_eliminate 0.52% : 0.000001s : 8: predicate.special_op_eliminate 0.57% : 0.000002s : 8: predicate.specialize_transform 0.67% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.50% : 0.000001s : 8: predicate.stack_unstack_eliminate 0.28% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.42% : 0.000004s : 29: predicate.switch_defer_inline 1.80% : 0.000005s : 37: predicate.switch_layer_defer_inline 4.41% : 0.000013s : 88: predicate.switch_simplify 16.46% : 0.000047s : 17: predicate.tile_eliminate 1.18% : 0.000003s : 17: predicate.transpose_eliminate 1.36% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.17% : 0.000003s : 25: predicate.tuple_list_get_item_const_eliminator 1.07% : 0.000003s : 25: predicate.tuple_list_get_item_depend_reorder 2.78% : 0.000008s : 39: predicate.tuple_list_get_item_eliminator 1.09% : 0.000003s : 25: predicate.tuple_list_get_set_item_eliminator 1.59% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 1.69% : 0.000005s : 31: predicate.tuple_to_list_eliminator_ 2.09% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 2.53% : 0.000007s : 56: predicate.updatestate_useless_node_eliminater 0.25% : 0.000001s : 4: predicate.value_based_eliminate 0.40% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.48% : 0.000001s : 8: predicate.virtual_output_eliminate 0.20% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.24% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000948 15 54.75% : 0.000519s : 7: func_graph_cloner_run.FuncGraphClonerGraph 45.25% : 0.000429s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.085800 192 0.00% : 0.000004s : 1: ForceFp32Comm 5.57% : 0.004775s : 1: add_attr 5.55% : 0.004760s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.07% : 0.000062s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.10% : 0.000082s : 1: auto_monad 0.03% : 0.000026s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.01% : 0.000006s : 1: bias_add_comm_swap 0.85% : 0.000729s : 1: bootstrap 0.04% : 0.000032s : 1: cconv 0.01% : 0.000005s : 1: comm_op_add_attrs 0.02% : 0.000017s : 1: control_data_broadcast_order 0.01% : 0.000010s : 1: convert_after_rewriter 0.04% : 0.000032s : 1: cse_after_recomputation 0.01% : 0.000006s : 1: dataset_repeat_opt 0.01% : 0.000006s : 1: detach_backward 0.01% : 0.000009s : 1: environ_conv 0.06% : 0.000056s : 1: event_method 0.01% : 0.000005s : 1: full_micro_interleaved_order_control 0.01% : 0.000005s : 1: get_jit_bprop_graph 0.01% : 0.000010s : 1: graph_reusing 0.01% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.01% : 0.000006s : 1: inline 0.01% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.01% : 0.000005s : 1: interleave_split_concat_branches 0.01% : 0.000006s : 1: label_fine_grained_interleaved_index 0.01% : 0.000007s : 1: label_micro_interleaved_index 0.71% : 0.000611s : 1: loop_unroll 0.01% : 0.000005s : 1: merge_cast_opt 0.01% : 0.000005s : 1: micro_interleaved_order_control 0.95% : 0.000816s : 1: mutable_eliminate 0.01% : 0.000009s : 1: offloading_packed_experts 0.02% : 0.000021s : 1: opt.transform.loop_unroll_optimizer 0.02% : 0.000018s : 1: opt.transform.mutable_eliminate 1.88% : 0.001616s : 78: opt.transform.opt_a 0.03% : 0.000028s : 1: opt.transform.opt_after_cconv 0.03% : 0.000027s : 1: opt.transform.opt_after_jit_grad 0.11% : 0.000092s : 28: opt.transform.opt_b 0.05% : 0.000047s : 2: opt.transform.opt_trans_graph 0.05% : 0.000046s : 4: opt.transform.symbol_engine_opt 4.22% : 0.003619s : 1: opt_a 0.13% : 0.000114s : 1: opt_after_cconv 0.71% : 0.000609s : 1: opt_after_jit_grad 0.23% : 0.000200s : 1: opt_b 7.41% : 0.006360s : 1: optimize 0.03% : 0.000022s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000008s : 1: order_py_execute_after_rewriter 0.03% : 0.000025s : 1: overlap_grad_flash_sp 0.01% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000007s : 1: overlap_grad_ring_attention 0.01% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.01% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.01% : 0.000005s : 1: overlap_param_gather 0.01% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000009s : 1: overlap_recompute_and_grad_model_parallel 0.01% : 0.000006s : 1: overlap_recompute_comm 0.01% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.01% : 0.000008s : 1: partial_unused_args_eliminate 0.01% : 0.000005s : 1: pipeline_parallel_scheduler 0.01% : 0.000005s : 1: pipeline_split 0.07% : 0.000056s : 1: pre_auto_parallel 0.05% : 0.000040s : 1: py_interpret_to_execute 0.02% : 0.000016s : 1: py_interpret_to_execute_after_opt_a 0.01% : 0.000004s : 1: remove_cast_before_assign_add 0.02% : 0.000017s : 1: remove_dup_value 0.71% : 0.000605s : 1: renormalize.infer 0.50% : 0.000430s : 1: renormalize.specialize 0.01% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000009s : 1: rewriter_after_jit_bprop_graph 0.05% : 0.000043s : 1: rewriter_after_opt_a 0.13% : 0.000114s : 1: rewriter_before_opt_a 0.01% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.01% : 0.000005s : 1: slice_recompute_activation 0.01% : 0.000005s : 1: split_layernorm_comm 0.01% : 0.000006s : 1: split_matmul_comm_elemetwise 0.01% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.18% : 0.000157s : 1: symbol_engine_optimizer 0.19% : 0.000166s : 1: tuple_transform 68.82% : 0.059045s : 1: type_inference TotalTime = 0.284885, [21] [bootstrap]: 0.0607917 [type_inference]: 0.136852 [event_method]: 9.859e-05 [auto_monad]: 0.00014018 [graph_reusing]: 1.034e-05 [inline]: 3.42997e-06 [add_attr]: 0.0162561, [1] [add_attr_with_inline]: 0.0162418, [1] [Cycle 1]: 0.00200197, [2] [tag_attr]: 0.0019075 [meta_addattr_fg_expand]: 2.872e-05 [parallel-infer-symbol]: 4.89998e-06 [pre_auto_parallel]: 8.905e-05 [insert-virtual-dataset]: 2.887e-05 [parallel-infer-symbol-second]: 9.19972e-07 [dataset_repeat_opt]: 3.33998e-06 [pipeline_split]: 1.87999e-06 [optimize]: 0.0682848, [53] [py_interpret_to_execute]: 0.00120345 [rewriter_before_opt_a]: 0.00022467 [opt_a]: 0.0565765, [3] [Cycle 1]: 0.0398231, [45] [expand_dump_flag]: 6.08998e-06 [switch_simplify]: 0.00066146 [loop_unroll]: 0.00024661 [a_1]: 0.00655594 [with_stream_mark]: 5.104e-05 [recompute_prepare]: 3.972e-05 [updatestate_depend_eliminate]: 1.242e-05 [updatestate_assign_eliminate]: 8.32e-06 [updatestate_loads_eliminate]: 7.64002e-06 [parameter_eliminate]: 4.42e-06 [a_2]: 0.00026139 [accelerated_algorithm]: 3.834e-05 [shard]: 2.22001e-06 [meta_shard_fg_expand]: 7.16001e-06 [shard_inline]: 1.761e-05 [merge_send_recv]: 2.079e-05 [auto_parallel]: 1.733e-05 [parallel]: 2.375e-05 [flash_sp]: 1.329e-05 [merge_comm]: 1.02e-05 [allreduce_fusion]: 9.14e-06 [matmul_add_comm_reduction]: 0.00177021 [allreduce_slice_to_reducescatter]: 1.57999e-06 [virtual_shard_identity]: 3.82e-05 [virtual_dataset]: 1.772e-05 [get_grad_eliminate_]: 1.607e-05 [virtual_output]: 1.594e-05 [merge_forward]: 1.724e-05 [cell_reuse_recompute_pass]: 2.73998e-06 [offload_activation]: 2.277e-05 [cell_reuse_handle_not_recompute_node_pass]: 4.107e-05 [merge_recompute_call_nodes]: 2.20002e-06 [before_grad]: 2.994e-05 [set_forward_comm_id_for_comm_node_pass]: 1.663e-05 [meta_fg_expand]: 0.00939207 [flash_sp_send_recv_attached]: 1.227e-05 [receive_attached]: 2.46e-06 [after_resolve]: 9.581e-05 [a_after_grad]: 0.00010001 [renormalize]: 0.0156493 [add_forward_monad_depend]: 2.006e-05 [auto_monad_grad]: 8.45999e-06 [auto_monad_eliminator]: 7.217e-05 [cse]: 0.00025463 [a_3]: 0.00372529 [Cycle 2]: 0.0126185, [45] [expand_dump_flag]: 4.99e-06 [switch_simplify]: 5.127e-05 [loop_unroll]: 4.195e-05 [a_1]: 0.00583128 [with_stream_mark]: 2.925e-05 [recompute_prepare]: 1.206e-05 [updatestate_depend_eliminate]: 4.32003e-06 [updatestate_assign_eliminate]: 3.31999e-06 [updatestate_loads_eliminate]: 2.56998e-06 [parameter_eliminate]: 5.56e-06 [a_2]: 9.055e-05 [accelerated_algorithm]: 1.798e-05 [shard]: 2.94999e-06 [meta_shard_fg_expand]: 3.32997e-06 [shard_inline]: 7.12997e-06 [merge_send_recv]: 1.246e-05 [auto_parallel]: 1.376e-05 [parallel]: 1.022e-05 [flash_sp]: 5.33002e-06 [merge_comm]: 3.33e-06 [allreduce_fusion]: 3.26001e-06 [matmul_add_comm_reduction]: 1.154e-05 [allreduce_slice_to_reducescatter]: 6.39993e-07 [virtual_shard_identity]: 8.62e-06 [virtual_dataset]: 5.92999e-06 [get_grad_eliminate_]: 5.84999e-06 [virtual_output]: 5.19e-06 [merge_forward]: 4.22e-06 [cell_reuse_recompute_pass]: 2.34001e-06 [offload_activation]: 1.067e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.287e-05 [merge_recompute_call_nodes]: 1.55999e-06 [before_grad]: 1.082e-05 [set_forward_comm_id_for_comm_node_pass]: 3.58e-06 [meta_fg_expand]: 0.00146182 [flash_sp_send_recv_attached]: 2.82002e-06 [receive_attached]: 2.90002e-06 [after_resolve]: 2.831e-05 [a_after_grad]: 1.155e-05 [renormalize]: 0.00262422 [add_forward_monad_depend]: 1.072e-05 [auto_monad_grad]: 2.32001e-06 [auto_monad_eliminator]: 2.108e-05 [cse]: 3.319e-05 [a_3]: 5.617e-05 [Cycle 3]: 0.00411029, [45] [expand_dump_flag]: 2.46998e-06 [switch_simplify]: 8.27998e-06 [loop_unroll]: 6.16e-06 [a_1]: 0.00156535 [with_stream_mark]: 2.307e-05 [recompute_prepare]: 1.127e-05 [updatestate_depend_eliminate]: 5.12e-06 [updatestate_assign_eliminate]: 2.96999e-06 [updatestate_loads_eliminate]: 2.74999e-06 [parameter_eliminate]: 1.76e-06 [a_2]: 7.048e-05 [accelerated_algorithm]: 1.916e-05 [shard]: 2.22001e-06 [meta_shard_fg_expand]: 2.16e-06 [shard_inline]: 6.76999e-06 [merge_send_recv]: 8.25999e-06 [auto_parallel]: 1.01e-05 [parallel]: 9.76e-06 [flash_sp]: 1.82999e-06 [merge_comm]: 3.13e-06 [allreduce_fusion]: 3.38e-06 [matmul_add_comm_reduction]: 1.026e-05 [allreduce_slice_to_reducescatter]: 7.80012e-07 [virtual_shard_identity]: 4.109e-05 [virtual_dataset]: 7.82e-06 [get_grad_eliminate_]: 6.28e-06 [virtual_output]: 6.87002e-06 [merge_forward]: 1.428e-05 [cell_reuse_recompute_pass]: 3.68e-06 [offload_activation]: 1.933e-05 [cell_reuse_handle_not_recompute_node_pass]: 2.426e-05 [merge_recompute_call_nodes]: 2.48e-06 [before_grad]: 1.01e-05 [set_forward_comm_id_for_comm_node_pass]: 8.69998e-06 [meta_fg_expand]: 3.33998e-06 [flash_sp_send_recv_attached]: 2.74001e-06 [receive_attached]: 2.16998e-06 [after_resolve]: 1.434e-05 [a_after_grad]: 9.97999e-06 [renormalize]: 1.09983e-07 [add_forward_monad_depend]: 6.46e-06 [auto_monad_grad]: 3.21001e-06 [auto_monad_eliminator]: 1.716e-05 [cse]: 3.248e-05 [a_3]: 4.021e-05 [py_interpret_to_execute_after_opt_a]: 2.033e-05 [slice_cell_reuse_recomputed_activation]: 2.68998e-06 [rewriter_after_opt_a]: 4.927e-05 [convert_after_rewriter]: 7.06001e-06 [order_py_execute_after_rewriter]: 6.51e-06 [mutable_eliminate]: 0.00224007 [opt_b]: 0.00265323, [1] [Cycle 1]: 0.00263933, [7] [b_1]: 0.00247699 [b_2]: 1.45e-05 [updatestate_depend_eliminate]: 1.717e-05 [updatestate_assign_eliminate]: 3.14001e-06 [updatestate_loads_eliminate]: 3.43999e-06 [renormalize]: 1.15001e-06 [cse]: 4.311e-05 [optimize_parallel_all_gather_comm]: 2.776e-05 [overlap_param_gather]: 1.99e-06 [cconv]: 3.972e-05 [loop_unroll]: 0.00278786 [opt_after_cconv]: 0.00162316, [1] [Cycle 1]: 0.00160863, [7] [c_1]: 4.222e-05 [parameter_eliminate]: 0.00142179 [updatestate_depend_eliminate]: 1.87e-05 [updatestate_assign_eliminate]: 3.63e-06 [updatestate_loads_eliminate]: 3.19001e-06 [cse]: 4.476e-05 [renormalize]: 7.2e-07 [remove_dup_value]: 1.836e-05 [tuple_transform]: 0.00014293, [1] [Cycle 1]: 0.00013504, [4] [d_1]: 9.59e-05 [none_parameter_eliminate]: 2.93e-06 [renormalize]: 2.3999e-07 [switch_simplify]: 1.006e-05 [partial_unused_args_eliminate]: 2.21998e-06 [add_recomputation]: 6.727e-05 [cse_after_recomputation]: 3.029e-05, [1] [Cycle 1]: 2.403e-05, [1] [cse]: 1.707e-05 [environ_conv]: 1.292e-05 [swap_dp_allreduce_reducescatter]: 6.56e-06 [bias_add_comm_swap]: 3.68e-06 [label_micro_interleaved_index]: 8.42998e-06 [label_fine_grained_interleaved_index]: 3.36999e-06 [merge_cast_opt]: 1.94e-06 [slice_recompute_activation]: 2.76e-06 [micro_interleaved_order_control]: 2.81999e-06 [assign_add_opt]: 1.59e-06 [ForceFp32Comm]: 9.20001e-07 [remove_cast_before_assign_add]: 1.13001e-06 [full_micro_interleaved_order_control]: 2.24001e-06 [reorder_send_recv_between_fp_bp]: 2.86999e-06 [comm_op_add_attrs]: 1.22999e-06 [add_comm_op_reuse_tag]: 1.19e-06 [interleave_split_concat_branches]: 1.51002e-06 [interleave_parallel_branches]: 1.14e-06 [overlap_opt_shard_in_pipeline]: 1.47999e-06 [overlap_opt_shard_grad_in_pipeline]: 1.95001e-06 [control_data_broadcast_order]: 1.856e-05 [grouped_pairwise_exchange_alltoall]: 1.66998e-06 [offloading_packed_experts]: 4.57e-06 [overlap_recompute_and_grad_model_parallel]: 5.99999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.34e-06 [overlap_recompute_allgather_and_fa_grad]: 1.53002e-06 [overlap_recompute_comm]: 2.69999e-06 [overlap_grad_ring_attention]: 4.57e-06 [overlap_grad_flash_sp]: 2.418e-05 [begin_end_overlap_inline]: 6.10016e-07 [split_matmul_comm_elemetwise]: 2.04e-06 [split_layernorm_comm]: 1.74e-06 [handle_group_info]: 1.14e-06 [symbol_engine_optimizer]: 0.00010565, [1] [Cycle 1]: 9.968e-05, [6] [build]: 1.272e-05 [elim_shapecalc]: 1.586e-05 [elim_not_effective]: 1.717e-05 [opt_reshape]: 7.45998e-06 [fold_const_symbol]: 1.101e-05 [renormalize]: 2.10013e-07 [detach_backward]: 2.71e-06 [pipeline_parallel_scheduler]: 1.76e-06 [auto_monad_reorder]: 0.00120901 [get_jit_bprop_graph]: 2.98e-06 [rewriter_after_jit_bprop_graph]: 1.108e-05 [opt_after_jit_grad]: 0.0007939 [validate]: 6.175e-05 Sums bootstrap : 0.060792s : 22.94% type_inference : 0.136852s : 51.63% event_method : 0.000099s : 0.04% auto_monad : 0.000140s : 0.05% graph_reusing : 0.000010s : 0.00% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.001908s : 0.72% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000029s : 0.01% parallel-infer-symbol : 0.000005s : 0.00% pre_auto_parallel : 0.000089s : 0.03% insert-virtual-dataset : 0.000029s : 0.01% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000003s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.001203s : 0.45% optimize.rewriter_before_opt_a : 0.000225s : 0.08% optimize.opt_a.expand_dump_flag : 0.000014s : 0.01% optimize.opt_a.switch_simplify : 0.000721s : 0.27% optimize.opt_a.loop_unroll : 0.000295s : 0.11% optimize.opt_a.a_1 : 0.013953s : 5.26% optimize.opt_a.with_stream_mark : 0.000103s : 0.04% optimize.opt_a.recompute_prepare : 0.000063s : 0.02% optimize.opt_a.updatestate_depend_eliminate : 0.000022s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000015s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000013s : 0.00% optimize.opt_a.parameter_eliminate : 0.000012s : 0.00% optimize.opt_a.a_2 : 0.000422s : 0.16% optimize.opt_a.accelerated_algorithm : 0.000075s : 0.03% optimize.opt_a.shard : 0.000007s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000013s : 0.00% optimize.opt_a.shard_inline : 0.000032s : 0.01% optimize.opt_a.merge_send_recv : 0.000042s : 0.02% optimize.opt_a.auto_parallel : 0.000041s : 0.02% optimize.opt_a.parallel : 0.000044s : 0.02% optimize.opt_a.flash_sp : 0.000020s : 0.01% optimize.opt_a.merge_comm : 0.000017s : 0.01% optimize.opt_a.allreduce_fusion : 0.000016s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.001792s : 0.68% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000003s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000088s : 0.03% optimize.opt_a.virtual_dataset : 0.000031s : 0.01% optimize.opt_a.get_grad_eliminate_ : 0.000028s : 0.01% optimize.opt_a.virtual_output : 0.000028s : 0.01% optimize.opt_a.merge_forward : 0.000036s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000009s : 0.00% optimize.opt_a.offload_activation : 0.000053s : 0.02% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000078s : 0.03% optimize.opt_a.merge_recompute_call_nodes : 0.000006s : 0.00% optimize.opt_a.before_grad : 0.000051s : 0.02% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000029s : 0.01% optimize.opt_a.meta_fg_expand : 0.010857s : 4.10% optimize.opt_a.flash_sp_send_recv_attached : 0.000018s : 0.01% optimize.opt_a.receive_attached : 0.000008s : 0.00% optimize.opt_a.after_resolve : 0.000138s : 0.05% optimize.opt_a.a_after_grad : 0.000122s : 0.05% optimize.opt_a.renormalize : 0.018274s : 6.89% optimize.opt_a.add_forward_monad_depend : 0.000037s : 0.01% optimize.opt_a.auto_monad_grad : 0.000014s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000110s : 0.04% optimize.opt_a.cse : 0.000320s : 0.12% optimize.opt_a.a_3 : 0.003822s : 1.44% optimize.py_interpret_to_execute_after_opt_a : 0.000020s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000003s : 0.00% optimize.rewriter_after_opt_a : 0.000049s : 0.02% optimize.convert_after_rewriter : 0.000007s : 0.00% optimize.order_py_execute_after_rewriter : 0.000007s : 0.00% optimize.mutable_eliminate : 0.002240s : 0.85% optimize.opt_b.b_1 : 0.002477s : 0.93% optimize.opt_b.b_2 : 0.000014s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000017s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000043s : 0.02% optimize.optimize_parallel_all_gather_comm : 0.000028s : 0.01% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000040s : 0.01% optimize.loop_unroll : 0.002788s : 1.05% optimize.opt_after_cconv.c_1 : 0.000042s : 0.02% optimize.opt_after_cconv.parameter_eliminate : 0.001422s : 0.54% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000019s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000004s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.cse : 0.000045s : 0.02% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000018s : 0.01% optimize.tuple_transform.d_1 : 0.000096s : 0.04% optimize.tuple_transform.none_parameter_eliminate : 0.000003s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000010s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000067s : 0.03% optimize.cse_after_recomputation.cse : 0.000017s : 0.01% optimize.environ_conv : 0.000013s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000007s : 0.00% optimize.bias_add_comm_swap : 0.000004s : 0.00% optimize.label_micro_interleaved_index : 0.000008s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000003s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.00% optimize.assign_add_opt : 0.000002s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000002s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000019s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000005s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000006s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000002s : 0.00% optimize.overlap_recompute_comm : 0.000003s : 0.00% optimize.overlap_grad_ring_attention : 0.000005s : 0.00% optimize.overlap_grad_flash_sp : 0.000024s : 0.01% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000013s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000016s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000017s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000011s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.001209s : 0.46% get_jit_bprop_graph : 0.000003s : 0.00% rewriter_after_jit_bprop_graph : 0.000011s : 0.00% opt_after_jit_grad : 0.000794s : 0.30% validate : 0.000062s : 0.02% Time group info: ------[substitution.] 0.003667 195 2.51% : 0.000092s : 7: substitution.arithmetic_simplify 0.07% : 0.000003s : 2: substitution.elim_not_effective 0.25% : 0.000009s : 6: substitution.float_depend_g_call 0.14% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.05% : 0.000002s : 2: substitution.fold_const_symbol 0.22% : 0.000008s : 4: substitution.graph_param_transform 0.09% : 0.000003s : 2: substitution.incorporate_call 0.05% : 0.000002s : 2: substitution.incorporate_call_switch 21.69% : 0.000795s : 18: substitution.inline 0.88% : 0.000032s : 2: substitution.inline_without_move 0.30% : 0.000011s : 14: substitution.j_node_and_user_rematch 0.66% : 0.000024s : 3: substitution.less_batch_normalization 0.54% : 0.000020s : 11: substitution.minmaximum_grad 0.26% : 0.000010s : 6: substitution.partial_eliminate 0.36% : 0.000013s : 14: substitution.remove_not_recompute_node 1.27% : 0.000046s : 10: substitution.replace_applicator 0.49% : 0.000018s : 12: substitution.replace_old_param 0.14% : 0.000005s : 1: substitution.set_cell_output_no_recompute 38.89% : 0.001426s : 11: substitution.tuple_list_convert_item_index_to_positive 0.68% : 0.000025s : 11: substitution.tuple_list_get_item_const_eliminator 0.72% : 0.000026s : 11: substitution.tuple_list_get_item_depend_reorder 28.99% : 0.001063s : 32: substitution.tuple_list_get_item_eliminator 0.77% : 0.000028s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.136674 2 92.90% : 0.126977s : 1: type_inference.infer 7.10% : 0.009697s : 1: type_inference.specialize ------[replace.] 0.000380 38 3.24% : 0.000012s : 2: replace.arithmetic_simplify 51.98% : 0.000198s : 18: replace.inline 44.78% : 0.000170s : 18: replace.tuple_list_get_item_eliminator ------[match.] 0.000891 38 5.37% : 0.000048s : 2: match.arithmetic_simplify 88.03% : 0.000784s : 18: match.inline 6.61% : 0.000059s : 18: match.tuple_list_get_item_eliminator ------[predicate.] 0.007086 4976 0.13% : 0.000010s : 61: predicate.accumulaten_eliminater 0.03% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.05% : 0.000003s : 23: predicate.addn_check_dump 0.12% : 0.000009s : 61: predicate.addn_zero_filter 0.10% : 0.000007s : 61: predicate.adjust_all_reduce_mul_add 0.28% : 0.000020s : 86: predicate.arithmetic_simplify 0.13% : 0.000009s : 63: predicate.cast_eliminate 0.15% : 0.000011s : 55: predicate.check_bprop_eliminate 0.04% : 0.000003s : 23: predicate.compare_switch_simplify 0.01% : 0.000001s : 4: predicate.const_output_eliminate 0.05% : 0.000003s : 23: predicate.depend_value_elim 0.13% : 0.000009s : 63: predicate.dict_get_item_const_eliminator 0.22% : 0.000016s : 63: predicate.dict_get_item_eliminator 0.12% : 0.000008s : 63: predicate.dict_set_item_eliminator 0.04% : 0.000003s : 8: predicate.dumpgradient_eliminate 0.01% : 0.000000s : 4: predicate.elim_not_effective 0.01% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 0.14% : 0.000010s : 67: predicate.environ_add_const_eliminate 20.13% : 0.001427s : 67: predicate.environ_get_add_eliminate 0.12% : 0.000009s : 67: predicate.environ_get_depend_swap 0.18% : 0.000013s : 90: predicate.environ_get_eliminate 0.13% : 0.000009s : 67: predicate.environ_get_set_eliminate 0.19% : 0.000013s : 99: predicate.exchange_switch_depend_value 0.31% : 0.000022s : 99: predicate.float_depend_g_call 0.04% : 0.000003s : 23: predicate.float_environ_get_switch 0.05% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.00% : 0.000000s : 4: predicate.fold_const_symbol 0.05% : 0.000004s : 23: predicate.get_grad_eliminate 0.01% : 0.000001s : 4: predicate.graph_param_transform 0.05% : 0.000003s : 23: predicate.incorporate_call 0.05% : 0.000003s : 23: predicate.incorporate_call_switch 0.68% : 0.000048s : 214: predicate.inline 0.14% : 0.000010s : 46: predicate.inline_without_move 0.02% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.07% : 0.000005s : 23: predicate.less_batch_normalization 0.19% : 0.000013s : 89: predicate.list_to_tuple_eliminator_ 0.31% : 0.000022s : 150: predicate.load_eliminater 28.60% : 0.002026s : 4: predicate.loop_unroll_after_grad 0.29% : 0.000021s : 141: predicate.loop_unroll_before_grad 0.18% : 0.000013s : 71: predicate.make_slice_get_slice_eliminator 0.05% : 0.000003s : 23: predicate.merge_addn 0.10% : 0.000007s : 55: predicate.micro_step_allgather_replace 0.11% : 0.000008s : 55: predicate.mini_step_allgather_replace 0.12% : 0.000009s : 63: predicate.minmaximum_grad 0.08% : 0.000006s : 4: predicate.mutable_eliminate 0.01% : 0.000001s : 4: predicate.opt_reshape 0.02% : 0.000001s : 4: predicate.parallel_virtual_node 0.30% : 0.000021s : 99: predicate.partial_defer_inline 0.18% : 0.000012s : 85: predicate.partial_eliminate 0.11% : 0.000008s : 61: predicate.print_const_string_wrapper 0.05% : 0.000004s : 23: predicate.reduce_all_const_elim 0.18% : 0.000013s : 63: predicate.reduce_eliminate 0.28% : 0.000020s : 150: predicate.redundant_stop_gradient_eliminater 0.05% : 0.000003s : 23: predicate.remove_not_recompute_node 0.21% : 0.000015s : 136: predicate.replace_applicator 0.07% : 0.000005s : 46: predicate.replace_old_param 0.02% : 0.000001s : 4: predicate.reset_defer_inline 0.13% : 0.000010s : 63: predicate.reshape_eliminate 0.12% : 0.000008s : 55: predicate.row_tensor_add_zeros_like 0.02% : 0.000002s : 4: predicate.row_tensor_eliminate 0.14% : 0.000010s : 55: predicate.same_eliminate 0.03% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.15% : 0.000010s : 23: predicate.shard_identity_eliminate 0.03% : 0.000002s : 8: predicate.special_op_eliminate 0.07% : 0.000005s : 23: predicate.specialize_transform 0.15% : 0.000010s : 55: predicate.split_environ_get_set_with_tuple_value 0.13% : 0.000009s : 46: predicate.stack_unstack_eliminate 0.02% : 0.000001s : 4: predicate.switch_call_monad_eliminater 0.20% : 0.000014s : 99: predicate.switch_defer_inline 17.37% : 0.001231s : 154: predicate.switch_layer_defer_inline 0.62% : 0.000044s : 267: predicate.switch_simplify 0.12% : 0.000009s : 63: predicate.tile_eliminate 0.11% : 0.000008s : 63: predicate.transpose_eliminate 0.16% : 0.000011s : 71: predicate.tuple_list_convert_item_index_to_positive 23.10% : 0.001636s : 71: predicate.tuple_list_get_item_const_eliminator 0.14% : 0.000010s : 71: predicate.tuple_list_get_item_depend_reorder 0.37% : 0.000026s : 112: predicate.tuple_list_get_item_eliminator 0.15% : 0.000011s : 71: predicate.tuple_list_get_set_item_eliminator 0.24% : 0.000017s : 94: predicate.tuple_list_set_item_eliminator 0.18% : 0.000012s : 89: predicate.tuple_to_list_eliminator_ 0.26% : 0.000018s : 150: predicate.updatestate_pure_node_eliminater 0.35% : 0.000025s : 173: predicate.updatestate_useless_node_eliminater 0.01% : 0.000001s : 4: predicate.value_based_eliminate 0.06% : 0.000004s : 23: predicate.virtual_dataset_eliminate 0.05% : 0.000004s : 23: predicate.virtual_output_eliminate 0.01% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.02% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.015919 41 91.67% : 0.014593s : 19: func_graph_cloner_run.FuncGraphClonerGraph 8.33% : 0.001327s : 22: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.411833 233 0.00% : 0.000004s : 1: ForceFp32Comm 3.95% : 0.016264s : 1: add_attr 3.94% : 0.016247s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.02% : 0.000074s : 1: add_recomputation 0.00% : 0.000005s : 1: assign_add_opt 0.04% : 0.000148s : 1: auto_monad 0.30% : 0.001235s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000007s : 1: bias_add_comm_swap 14.77% : 0.060829s : 1: bootstrap 0.01% : 0.000043s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.01% : 0.000023s : 1: control_data_broadcast_order 0.00% : 0.000010s : 1: convert_after_rewriter 0.01% : 0.000033s : 1: cse_after_recomputation 0.00% : 0.000006s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000017s : 1: environ_conv 0.03% : 0.000110s : 1: event_method 0.00% : 0.000007s : 1: full_micro_interleaved_order_control 0.00% : 0.000010s : 1: get_jit_bprop_graph 0.00% : 0.000015s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000007s : 1: inline 0.01% : 0.000034s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.01% : 0.000022s : 1: label_fine_grained_interleaved_index 0.00% : 0.000012s : 1: label_micro_interleaved_index 0.68% : 0.002808s : 1: loop_unroll 0.00% : 0.000007s : 1: merge_cast_opt 0.00% : 0.000006s : 1: micro_interleaved_order_control 0.55% : 0.002259s : 1: mutable_eliminate 0.00% : 0.000008s : 1: offloading_packed_experts 0.50% : 0.002063s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000032s : 1: opt.transform.mutable_eliminate 4.79% : 0.019724s : 117: opt.transform.opt_a 0.01% : 0.000038s : 1: opt.transform.opt_after_cconv 0.01% : 0.000033s : 1: opt.transform.opt_after_jit_grad 0.57% : 0.002341s : 28: opt.transform.opt_b 0.02% : 0.000076s : 2: opt.transform.opt_trans_graph 0.01% : 0.000047s : 4: opt.transform.symbol_engine_opt 13.74% : 0.056580s : 1: opt_a 0.40% : 0.001629s : 1: opt_after_cconv 0.20% : 0.000804s : 1: opt_after_jit_grad 0.65% : 0.002659s : 1: opt_b 16.58% : 0.068291s : 1: optimize 0.01% : 0.000031s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000009s : 1: order_py_execute_after_rewriter 0.01% : 0.000029s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000008s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000005s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000006s : 1: overlap_param_gather 0.00% : 0.000005s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000009s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000006s : 1: overlap_recompute_comm 0.00% : 0.000010s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000007s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.02% : 0.000094s : 1: pre_auto_parallel 0.30% : 0.001220s : 1: py_interpret_to_execute 0.01% : 0.000024s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.01% : 0.000022s : 1: remove_dup_value 2.97% : 0.012221s : 2: renormalize.infer 1.46% : 0.006020s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000015s : 1: rewriter_after_jit_bprop_graph 0.01% : 0.000055s : 1: rewriter_after_opt_a 0.06% : 0.000233s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000006s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000010s : 1: swap_dp_allreduce_reducescatter 0.03% : 0.000108s : 1: symbol_engine_optimizer 0.04% : 0.000146s : 1: tuple_transform 33.24% : 0.136886s : 1: type_inference [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:43:33.662.365 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:43:33.662.691 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. TotalTime = 1.05405, [21] [bootstrap]: 0.00048216 [type_inference]: 0.982779 [event_method]: 0.00011629 [auto_monad]: 0.00028744 [graph_reusing]: 2.162e-05 [inline]: 7.07002e-06 [add_attr]: 0.00454323, [1] [add_attr_with_inline]: 0.00452894, [1] [Cycle 1]: 0.00017824, [2] [tag_attr]: 0.00011291 [meta_addattr_fg_expand]: 2.361e-05 [parallel-infer-symbol]: 3.8e-06 [pre_auto_parallel]: 0.0001885 [insert-virtual-dataset]: 2.79999e-06 [parallel-infer-symbol-second]: 2.33998e-06 [dataset_repeat_opt]: 2.58e-06 [pipeline_split]: 1.72001e-06 [optimize]: 0.0648191, [53] [py_interpret_to_execute]: 0.00012667 [rewriter_before_opt_a]: 0.00040506 [opt_a]: 0.0619037, [2] [Cycle 1]: 0.0611849, [45] [expand_dump_flag]: 1.042e-05 [switch_simplify]: 0.0486198 [loop_unroll]: 0.00021492 [a_1]: 0.00501243 [with_stream_mark]: 3.103e-05 [recompute_prepare]: 2.019e-05 [updatestate_depend_eliminate]: 6.90998e-06 [updatestate_assign_eliminate]: 5.03002e-06 [updatestate_loads_eliminate]: 4.92e-06 [parameter_eliminate]: 2.15002e-06 [a_2]: 0.00026983 [accelerated_algorithm]: 9.67999e-06 [shard]: 2.59999e-06 [meta_shard_fg_expand]: 7.8e-06 [shard_inline]: 7.95e-06 [merge_send_recv]: 9.97001e-06 [auto_parallel]: 1.217e-05 [parallel]: 2.661e-05 [flash_sp]: 1.21e-05 [merge_comm]: 4.56002e-06 [allreduce_fusion]: 3.29001e-06 [matmul_add_comm_reduction]: 8.70001e-06 [allreduce_slice_to_reducescatter]: 7.59988e-07 [virtual_shard_identity]: 1.046e-05 [virtual_dataset]: 6.07999e-06 [get_grad_eliminate_]: 8.08001e-06 [virtual_output]: 6.48e-06 [merge_forward]: 4.72e-06 [cell_reuse_recompute_pass]: 1.88002e-06 [offload_activation]: 1.089e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.456e-05 [merge_recompute_call_nodes]: 1.62999e-06 [before_grad]: 1.093e-05 [set_forward_comm_id_for_comm_node_pass]: 4.50001e-06 [meta_fg_expand]: 6.41e-06 [flash_sp_send_recv_attached]: 3.45e-06 [receive_attached]: 2.58e-06 [after_resolve]: 1.334e-05 [a_after_grad]: 9.99001e-06 [renormalize]: 0.00627129 [add_forward_monad_depend]: 1.377e-05 [auto_monad_grad]: 3.40998e-06 [auto_monad_eliminator]: 2.224e-05 [cse]: 3.112e-05 [a_3]: 6.121e-05 [Cycle 2]: 0.00070368, [45] [expand_dump_flag]: 2.24001e-06 [switch_simplify]: 8.91002e-06 [loop_unroll]: 6.02999e-06 [a_1]: 0.00015816 [with_stream_mark]: 2.088e-05 [recompute_prepare]: 6.09999e-06 [updatestate_depend_eliminate]: 4.2e-06 [updatestate_assign_eliminate]: 2.76e-06 [updatestate_loads_eliminate]: 3.28998e-06 [parameter_eliminate]: 1.73002e-06 [a_2]: 7.298e-05 [accelerated_algorithm]: 7.33999e-06 [shard]: 2.54001e-06 [meta_shard_fg_expand]: 2.02999e-06 [shard_inline]: 6.08998e-06 [merge_send_recv]: 7.61001e-06 [auto_parallel]: 9.19e-06 [parallel]: 9.01998e-06 [flash_sp]: 3.98001e-06 [merge_comm]: 3.6e-06 [allreduce_fusion]: 3.35e-06 [matmul_add_comm_reduction]: 9.54999e-06 [allreduce_slice_to_reducescatter]: 6.60017e-07 [virtual_shard_identity]: 6.98e-06 [virtual_dataset]: 5.86998e-06 [get_grad_eliminate_]: 5.75001e-06 [virtual_output]: 6.01998e-06 [merge_forward]: 3.85e-06 [cell_reuse_recompute_pass]: 3.26001e-06 [offload_activation]: 9.74999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.153e-05 [merge_recompute_call_nodes]: 1.36998e-06 [before_grad]: 9.37999e-06 [set_forward_comm_id_for_comm_node_pass]: 3.98001e-06 [meta_fg_expand]: 2.74001e-06 [flash_sp_send_recv_attached]: 1.62999e-06 [receive_attached]: 2.28002e-06 [after_resolve]: 1.169e-05 [a_after_grad]: 9.02999e-06 [renormalize]: 8.9989e-08 [add_forward_monad_depend]: 1.55001e-06 [auto_monad_grad]: 1.17e-06 [auto_monad_eliminator]: 7.12002e-06 [cse]: 1.408e-05 [a_3]: 3.372e-05 [py_interpret_to_execute_after_opt_a]: 1.682e-05 [slice_cell_reuse_recomputed_activation]: 2.14999e-06 [rewriter_after_opt_a]: 4.29e-05 [convert_after_rewriter]: 7.30003e-06 [order_py_execute_after_rewriter]: 5.57001e-06 [mutable_eliminate]: 0.00074479 [opt_b]: 0.00020182, [1] [Cycle 1]: 0.00019365, [7] [b_1]: 0.00011766 [b_2]: 7.41999e-06 [updatestate_depend_eliminate]: 7.18998e-06 [updatestate_assign_eliminate]: 2.52001e-06 [updatestate_loads_eliminate]: 2.39001e-06 [renormalize]: 5.59987e-07 [cse]: 2.023e-05 [optimize_parallel_all_gather_comm]: 1.736e-05 [overlap_param_gather]: 1.79998e-06 [cconv]: 3.279e-05 [loop_unroll]: 0.00049927 [opt_after_cconv]: 0.00010437, [1] [Cycle 1]: 9.754e-05, [7] [c_1]: 3.176e-05 [parameter_eliminate]: 3.88999e-06 [updatestate_depend_eliminate]: 6.14001e-06 [updatestate_assign_eliminate]: 2.30002e-06 [updatestate_loads_eliminate]: 2.54999e-06 [cse]: 1.719e-05 [renormalize]: 5.10016e-07 [remove_dup_value]: 1.424e-05 [tuple_transform]: 7.929e-05, [1] [Cycle 1]: 7.483e-05, [4] [d_1]: 4.661e-05 [none_parameter_eliminate]: 1.49e-06 [renormalize]: 1.59984e-07 [switch_simplify]: 7.47002e-06 [partial_unused_args_eliminate]: 1.64e-06 [add_recomputation]: 5.633e-05 [cse_after_recomputation]: 2.212e-05, [1] [Cycle 1]: 1.755e-05, [1] [cse]: 1.214e-05 [environ_conv]: 5.18002e-06 [swap_dp_allreduce_reducescatter]: 5.52999e-06 [bias_add_comm_swap]: 2.63003e-06 [label_micro_interleaved_index]: 4.79998e-06 [label_fine_grained_interleaved_index]: 2.50002e-06 [merge_cast_opt]: 1.42999e-06 [slice_recompute_activation]: 2.32999e-06 [micro_interleaved_order_control]: 2.39999e-06 [assign_add_opt]: 1.52999e-06 [ForceFp32Comm]: 1.20001e-06 [remove_cast_before_assign_add]: 1.17e-06 [full_micro_interleaved_order_control]: 2.37999e-06 [reorder_send_recv_between_fp_bp]: 2.62001e-06 [comm_op_add_attrs]: 1.44e-06 [add_comm_op_reuse_tag]: 1.30999e-06 [interleave_split_concat_branches]: 1.14e-06 [interleave_parallel_branches]: 1.00001e-06 [overlap_opt_shard_in_pipeline]: 1.42999e-06 [overlap_opt_shard_grad_in_pipeline]: 2.04e-06 [control_data_broadcast_order]: 1.189e-05 [grouped_pairwise_exchange_alltoall]: 1.67001e-06 [offloading_packed_experts]: 3.65e-06 [overlap_recompute_and_grad_model_parallel]: 4.28001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.38002e-06 [overlap_recompute_allgather_and_fa_grad]: 1.35999e-06 [overlap_recompute_comm]: 2.22999e-06 [overlap_grad_ring_attention]: 4.35e-06 [overlap_grad_flash_sp]: 2.127e-05 [begin_end_overlap_inline]: 5.00004e-07 [split_matmul_comm_elemetwise]: 2.24001e-06 [split_layernorm_comm]: 2.24001e-06 [handle_group_info]: 1.10999e-06 [symbol_engine_optimizer]: 0.00019474, [1] [Cycle 1]: 0.00018977, [6] [build]: 9.708e-05 [elim_shapecalc]: 1.17e-05 [elim_not_effective]: 2.434e-05 [opt_reshape]: 6.71e-06 [fold_const_symbol]: 1.591e-05 [renormalize]: 2.09984e-07 [detach_backward]: 2.69999e-06 [pipeline_parallel_scheduler]: 1.73002e-06 [auto_monad_reorder]: 1.714e-05 [get_jit_bprop_graph]: 2.04e-06 [rewriter_after_jit_bprop_graph]: 4.62e-06 [opt_after_jit_grad]: 0.0005156 [validate]: 4.679e-05 Sums bootstrap : 0.000482s : 0.05% type_inference : 0.982779s : 93.73% event_method : 0.000116s : 0.01% auto_monad : 0.000287s : 0.03% graph_reusing : 0.000022s : 0.00% inline : 0.000007s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000113s : 0.01% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000024s : 0.00% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000189s : 0.02% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000002s : 0.00% dataset_repeat_opt : 0.000003s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000127s : 0.01% optimize.rewriter_before_opt_a : 0.000405s : 0.04% optimize.opt_a.expand_dump_flag : 0.000013s : 0.00% optimize.opt_a.switch_simplify : 0.048629s : 4.64% optimize.opt_a.loop_unroll : 0.000221s : 0.02% optimize.opt_a.a_1 : 0.005171s : 0.49% optimize.opt_a.with_stream_mark : 0.000052s : 0.00% optimize.opt_a.recompute_prepare : 0.000026s : 0.00% optimize.opt_a.updatestate_depend_eliminate : 0.000011s : 0.00% optimize.opt_a.updatestate_assign_eliminate : 0.000008s : 0.00% optimize.opt_a.updatestate_loads_eliminate : 0.000008s : 0.00% optimize.opt_a.parameter_eliminate : 0.000004s : 0.00% optimize.opt_a.a_2 : 0.000343s : 0.03% optimize.opt_a.accelerated_algorithm : 0.000017s : 0.00% optimize.opt_a.shard : 0.000005s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000010s : 0.00% optimize.opt_a.shard_inline : 0.000014s : 0.00% optimize.opt_a.merge_send_recv : 0.000018s : 0.00% optimize.opt_a.auto_parallel : 0.000021s : 0.00% optimize.opt_a.parallel : 0.000036s : 0.00% optimize.opt_a.flash_sp : 0.000016s : 0.00% optimize.opt_a.merge_comm : 0.000008s : 0.00% optimize.opt_a.allreduce_fusion : 0.000007s : 0.00% optimize.opt_a.matmul_add_comm_reduction : 0.000018s : 0.00% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000017s : 0.00% optimize.opt_a.virtual_dataset : 0.000012s : 0.00% optimize.opt_a.get_grad_eliminate_ : 0.000014s : 0.00% optimize.opt_a.virtual_output : 0.000012s : 0.00% optimize.opt_a.merge_forward : 0.000009s : 0.00% optimize.opt_a.cell_reuse_recompute_pass : 0.000005s : 0.00% optimize.opt_a.offload_activation : 0.000021s : 0.00% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000026s : 0.00% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.00% optimize.opt_a.before_grad : 0.000020s : 0.00% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000008s : 0.00% optimize.opt_a.meta_fg_expand : 0.000009s : 0.00% optimize.opt_a.flash_sp_send_recv_attached : 0.000005s : 0.00% optimize.opt_a.receive_attached : 0.000005s : 0.00% optimize.opt_a.after_resolve : 0.000025s : 0.00% optimize.opt_a.a_after_grad : 0.000019s : 0.00% optimize.opt_a.renormalize : 0.006271s : 0.60% optimize.opt_a.add_forward_monad_depend : 0.000015s : 0.00% optimize.opt_a.auto_monad_grad : 0.000005s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000029s : 0.00% optimize.opt_a.cse : 0.000045s : 0.00% optimize.opt_a.a_3 : 0.000095s : 0.01% optimize.py_interpret_to_execute_after_opt_a : 0.000017s : 0.00% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000043s : 0.00% optimize.convert_after_rewriter : 0.000007s : 0.00% optimize.order_py_execute_after_rewriter : 0.000006s : 0.00% optimize.mutable_eliminate : 0.000745s : 0.07% optimize.opt_b.b_1 : 0.000118s : 0.01% optimize.opt_b.b_2 : 0.000007s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000007s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000020s : 0.00% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.00% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000033s : 0.00% optimize.loop_unroll : 0.000499s : 0.05% optimize.opt_after_cconv.c_1 : 0.000032s : 0.00% optimize.opt_after_cconv.parameter_eliminate : 0.000004s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.cse : 0.000017s : 0.00% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000014s : 0.00% optimize.tuple_transform.d_1 : 0.000047s : 0.00% optimize.tuple_transform.none_parameter_eliminate : 0.000001s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000056s : 0.01% optimize.cse_after_recomputation.cse : 0.000012s : 0.00% optimize.environ_conv : 0.000005s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.00% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000002s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000012s : 0.00% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000004s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000021s : 0.00% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000097s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000012s : 0.00% optimize.symbol_engine_optimizer.elim_not_effective : 0.000024s : 0.00% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000016s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000017s : 0.00% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000005s : 0.00% opt_after_jit_grad : 0.000516s : 0.05% validate : 0.000047s : 0.00% Time group info: ------[substitution.] 0.001657 325 3.38% : 0.000056s : 7: substitution.arithmetic_simplify 0.35% : 0.000006s : 4: substitution.depend_value_elim 0.72% : 0.000012s : 2: substitution.elim_not_effective 0.44% : 0.000007s : 2: substitution.fold_const_symbol 0.38% : 0.000006s : 4: substitution.graph_param_transform 61.40% : 0.001018s : 37: substitution.inline 0.28% : 0.000005s : 4: substitution.j_node_and_user_rematch 0.61% : 0.000010s : 8: substitution.minmaximum_grad 0.38% : 0.000006s : 4: substitution.remove_not_recompute_node 0.44% : 0.000007s : 4: substitution.replace_old_param 2.98% : 0.000049s : 10: substitution.switch_simplify 6.33% : 0.000105s : 42: substitution.tuple_list_convert_item_index_to_positive 3.15% : 0.000052s : 46: substitution.tuple_list_get_item_const_eliminator 6.91% : 0.000115s : 46: substitution.tuple_list_get_item_depend_reorder 7.92% : 0.000131s : 59: substitution.tuple_list_get_item_eliminator 4.34% : 0.000072s : 46: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.982548 2 98.67% : 0.969466s : 1: type_inference.infer 1.33% : 0.013082s : 1: type_inference.specialize ------[replace.] 0.000732 68 4.15% : 0.000030s : 4: replace.depend_value_elim 47.69% : 0.000349s : 37: replace.inline 24.92% : 0.000182s : 10: replace.switch_simplify 9.95% : 0.000073s : 4: replace.tuple_list_get_item_depend_reorder 13.29% : 0.000097s : 13: replace.tuple_list_get_item_eliminator ------[match.] 0.001126 68 0.34% : 0.000004s : 4: match.depend_value_elim 88.73% : 0.000999s : 37: match.inline 3.43% : 0.000039s : 10: match.switch_simplify 4.42% : 0.000050s : 4: match.tuple_list_get_item_depend_reorder 3.08% : 0.000035s : 13: match.tuple_list_get_item_eliminator ------[predicate.] 0.000813 4995 1.24% : 0.000010s : 74: predicate.accumulaten_eliminater 0.20% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.25% : 0.000002s : 16: predicate.addn_check_dump 1.32% : 0.000011s : 74: predicate.addn_zero_filter 1.22% : 0.000010s : 74: predicate.adjust_all_reduce_mul_add 2.15% : 0.000018s : 90: predicate.arithmetic_simplify 1.33% : 0.000011s : 74: predicate.cast_eliminate 0.14% : 0.000001s : 8: predicate.check_bprop_eliminate 0.26% : 0.000002s : 16: predicate.compare_switch_simplify 0.05% : 0.000000s : 4: predicate.const_output_eliminate 0.34% : 0.000003s : 18: predicate.depend_value_elim 1.35% : 0.000011s : 74: predicate.dict_get_item_const_eliminator 1.42% : 0.000012s : 74: predicate.dict_get_item_eliminator 1.30% : 0.000011s : 74: predicate.dict_set_item_eliminator 0.36% : 0.000003s : 8: predicate.dumpgradient_eliminate 0.08% : 0.000001s : 4: predicate.elim_not_effective 0.11% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.31% : 0.000011s : 78: predicate.environ_add_const_eliminate 1.27% : 0.000010s : 78: predicate.environ_get_add_eliminate 1.29% : 0.000011s : 78: predicate.environ_get_depend_swap 1.57% : 0.000013s : 94: predicate.environ_get_eliminate 1.33% : 0.000011s : 78: predicate.environ_get_set_eliminate 2.21% : 0.000018s : 128: predicate.exchange_switch_depend_value 3.01% : 0.000025s : 128: predicate.float_depend_g_call 0.31% : 0.000003s : 16: predicate.float_environ_get_switch 0.31% : 0.000003s : 20: predicate.float_tuple_getitem_switch 0.05% : 0.000000s : 4: predicate.fold_const_symbol 0.21% : 0.000002s : 8: predicate.get_grad_eliminate 0.07% : 0.000001s : 4: predicate.graph_param_transform 0.31% : 0.000003s : 16: predicate.incorporate_call 0.25% : 0.000002s : 16: predicate.incorporate_call_switch 6.13% : 0.000050s : 234: predicate.inline 0.20% : 0.000002s : 8: predicate.inline_without_move 0.08% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.32% : 0.000003s : 8: predicate.less_batch_normalization 1.74% : 0.000014s : 99: predicate.list_to_tuple_eliminator_ 2.81% : 0.000023s : 173: predicate.load_eliminater 0.30% : 0.000002s : 4: predicate.loop_unroll_after_grad 6.48% : 0.000053s : 181: predicate.loop_unroll_before_grad 1.53% : 0.000012s : 86: predicate.make_slice_get_slice_eliminator 0.26% : 0.000002s : 16: predicate.merge_addn 0.13% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.15% : 0.000001s : 8: predicate.mini_step_allgather_replace 1.28% : 0.000010s : 74: predicate.minmaximum_grad 0.38% : 0.000003s : 4: predicate.mutable_eliminate 0.08% : 0.000001s : 4: predicate.opt_reshape 0.09% : 0.000001s : 4: predicate.parallel_virtual_node 3.23% : 0.000026s : 128: predicate.partial_defer_inline 1.81% : 0.000015s : 95: predicate.partial_eliminate 1.25% : 0.000010s : 74: predicate.print_const_string_wrapper 0.22% : 0.000002s : 12: predicate.reduce_all_const_elim 1.64% : 0.000013s : 74: predicate.reduce_eliminate 3.00% : 0.000024s : 173: predicate.redundant_stop_gradient_eliminater 0.17% : 0.000001s : 8: predicate.remove_not_recompute_node 1.34% : 0.000011s : 99: predicate.replace_applicator 0.11% : 0.000001s : 8: predicate.replace_old_param 0.09% : 0.000001s : 4: predicate.reset_defer_inline 1.30% : 0.000011s : 74: predicate.reshape_eliminate 0.16% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.09% : 0.000001s : 4: predicate.row_tensor_eliminate 0.23% : 0.000002s : 8: predicate.same_eliminate 0.18% : 0.000001s : 12: predicate.set_cell_output_no_recompute 0.20% : 0.000002s : 8: predicate.shard_identity_eliminate 0.21% : 0.000002s : 8: predicate.special_op_eliminate 0.33% : 0.000003s : 16: predicate.specialize_transform 0.27% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.19% : 0.000002s : 8: predicate.stack_unstack_eliminate 0.09% : 0.000001s : 4: predicate.switch_call_monad_eliminater 2.40% : 0.000020s : 128: predicate.switch_defer_inline 2.58% : 0.000021s : 136: predicate.switch_layer_defer_inline 7.15% : 0.000058s : 349: predicate.switch_simplify 1.28% : 0.000010s : 74: predicate.tile_eliminate 1.27% : 0.000010s : 74: predicate.transpose_eliminate 1.71% : 0.000014s : 82: predicate.tuple_list_convert_item_index_to_positive 1.98% : 0.000016s : 86: predicate.tuple_list_get_item_const_eliminator 1.64% : 0.000013s : 86: predicate.tuple_list_get_item_depend_reorder 3.23% : 0.000026s : 115: predicate.tuple_list_get_item_eliminator 1.66% : 0.000014s : 86: predicate.tuple_list_get_set_item_eliminator 2.12% : 0.000017s : 102: predicate.tuple_list_set_item_eliminator 1.64% : 0.000013s : 99: predicate.tuple_to_list_eliminator_ 2.86% : 0.000023s : 173: predicate.updatestate_pure_node_eliminater 3.15% : 0.000026s : 189: predicate.updatestate_useless_node_eliminater 0.09% : 0.000001s : 4: predicate.value_based_eliminate 0.18% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.19% : 0.000002s : 8: predicate.virtual_output_eliminate 0.07% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.11% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.178332 87 98.36% : 0.175403s : 46: func_graph_cloner_run.FuncGraphClonerGraph 1.64% : 0.002929s : 41: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 1.184344 209 0.00% : 0.000004s : 1: ForceFp32Comm 0.38% : 0.004550s : 1: add_attr 0.38% : 0.004534s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.01% : 0.000061s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.03% : 0.000297s : 1: auto_monad 0.00% : 0.000021s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000005s : 1: bias_add_comm_swap 0.04% : 0.000512s : 1: bootstrap 0.00% : 0.000036s : 1: cconv 0.00% : 0.000005s : 1: comm_op_add_attrs 0.00% : 0.000015s : 1: control_data_broadcast_order 0.00% : 0.000010s : 1: convert_after_rewriter 0.00% : 0.000025s : 1: cse_after_recomputation 0.00% : 0.000008s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000009s : 1: environ_conv 0.01% : 0.000125s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000005s : 1: get_jit_bprop_graph 0.00% : 0.000026s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000011s : 1: inline 0.00% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.04% : 0.000508s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.06% : 0.000754s : 1: mutable_eliminate 0.00% : 0.000006s : 1: offloading_packed_experts 0.00% : 0.000017s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000018s : 1: opt.transform.mutable_eliminate 4.61% : 0.054541s : 95: opt.transform.opt_a 0.00% : 0.000030s : 1: opt.transform.opt_after_cconv 0.00% : 0.000028s : 1: opt.transform.opt_after_jit_grad 0.01% : 0.000097s : 28: opt.transform.opt_b 0.00% : 0.000052s : 2: opt.transform.opt_trans_graph 0.00% : 0.000055s : 4: opt.transform.symbol_engine_opt 5.23% : 0.061908s : 1: opt_a 0.01% : 0.000109s : 1: opt_after_cconv 0.04% : 0.000525s : 1: opt_after_jit_grad 0.02% : 0.000205s : 1: opt_b 5.47% : 0.064826s : 1: optimize 0.00% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000009s : 1: order_py_execute_after_rewriter 0.00% : 0.000025s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000006s : 1: overlap_param_gather 0.00% : 0.000005s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000007s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000006s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.02% : 0.000199s : 1: pre_auto_parallel 0.01% : 0.000132s : 1: py_interpret_to_execute 0.00% : 0.000020s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000018s : 1: remove_dup_value 0.32% : 0.003792s : 1: renormalize.infer 0.21% : 0.002465s : 1: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000008s : 1: rewriter_after_jit_bprop_graph 0.00% : 0.000048s : 1: rewriter_after_opt_a 0.03% : 0.000411s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.02% : 0.000197s : 1: symbol_engine_optimizer 0.01% : 0.000082s : 1: tuple_transform 82.98% : 0.982801s : 1: type_inference TotalTime = 0.0567188, [21] [bootstrap]: 0.00058642 [type_inference]: 0.0125923 [event_method]: 5.764e-05 [auto_monad]: 7.543e-05 [graph_reusing]: 7.06001e-06 [inline]: 3.08e-06 [add_attr]: 0.00387258, [1] [add_attr_with_inline]: 0.0038613, [1] [Cycle 1]: 8.822e-05, [2] [tag_attr]: 4.079e-05 [meta_addattr_fg_expand]: 6.41e-06 [parallel-infer-symbol]: 3.75e-06 [pre_auto_parallel]: 4.584e-05 [insert-virtual-dataset]: 2.46e-06 [parallel-infer-symbol-second]: 8.90024e-07 [dataset_repeat_opt]: 2.26e-06 [pipeline_split]: 1.55999e-06 [optimize]: 0.0385305, [53] [py_interpret_to_execute]: 3.163e-05 [rewriter_before_opt_a]: 0.00016788 [opt_a]: 0.003512, [2] [Cycle 1]: 0.00279646, [45] [expand_dump_flag]: 3.79002e-06 [switch_simplify]: 5.44e-05 [loop_unroll]: 3.895e-05 [a_1]: 0.00104958 [with_stream_mark]: 2.181e-05 [recompute_prepare]: 9.42001e-06 [updatestate_depend_eliminate]: 4.19002e-06 [updatestate_assign_eliminate]: 3.22997e-06 [updatestate_loads_eliminate]: 3.03e-06 [parameter_eliminate]: 1.97999e-06 [a_2]: 8.002e-05 [accelerated_algorithm]: 6.91001e-06 [shard]: 2.78998e-06 [meta_shard_fg_expand]: 1.96e-06 [shard_inline]: 6.01e-06 [merge_send_recv]: 8.47e-06 [auto_parallel]: 9.02e-06 [parallel]: 1.868e-05 [flash_sp]: 9.10999e-06 [merge_comm]: 3.93001e-06 [allreduce_fusion]: 3.33e-06 [matmul_add_comm_reduction]: 9.86998e-06 [allreduce_slice_to_reducescatter]: 6.89994e-07 [virtual_shard_identity]: 7.75e-06 [virtual_dataset]: 6.21e-06 [get_grad_eliminate_]: 5.50001e-06 [virtual_output]: 5.85002e-06 [merge_forward]: 4.15999e-06 [cell_reuse_recompute_pass]: 1.49e-06 [offload_activation]: 1.012e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.202e-05 [merge_recompute_call_nodes]: 1.35001e-06 [before_grad]: 1.082e-05 [set_forward_comm_id_for_comm_node_pass]: 3.58999e-06 [meta_fg_expand]: 2.82002e-06 [flash_sp_send_recv_attached]: 2.54001e-06 [receive_attached]: 2.29001e-06 [after_resolve]: 1.144e-05 [a_after_grad]: 9.43002e-06 [renormalize]: 0.00099543 [add_forward_monad_depend]: 5.79e-06 [auto_monad_grad]: 2.45002e-06 [auto_monad_eliminator]: 1.616e-05 [cse]: 2.913e-05 [a_3]: 4.558e-05 [Cycle 2]: 0.00070408, [45] [expand_dump_flag]: 1.79e-06 [switch_simplify]: 7.68001e-06 [loop_unroll]: 5.82999e-06 [a_1]: 0.00013338 [with_stream_mark]: 1.332e-05 [recompute_prepare]: 5.94e-06 [updatestate_depend_eliminate]: 3.22002e-06 [updatestate_assign_eliminate]: 2.56e-06 [updatestate_loads_eliminate]: 2.74999e-06 [parameter_eliminate]: 1.30001e-06 [a_2]: 0.0001302 [accelerated_algorithm]: 7.62002e-06 [shard]: 1.68002e-06 [meta_shard_fg_expand]: 1.40001e-06 [shard_inline]: 6.45002e-06 [merge_send_recv]: 6.70998e-06 [auto_parallel]: 7.43e-06 [parallel]: 6.58e-06 [flash_sp]: 3.65e-06 [merge_comm]: 2.98998e-06 [allreduce_fusion]: 3.01999e-06 [matmul_add_comm_reduction]: 7.07002e-06 [allreduce_slice_to_reducescatter]: 6.69999e-07 [virtual_shard_identity]: 6.31998e-06 [virtual_dataset]: 5.64998e-06 [get_grad_eliminate_]: 5.46e-06 [virtual_output]: 5.13002e-06 [merge_forward]: 3.16001e-06 [cell_reuse_recompute_pass]: 1.74998e-06 [offload_activation]: 6.44999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.099e-05 [merge_recompute_call_nodes]: 9.20001e-07 [before_grad]: 9.44e-06 [set_forward_comm_id_for_comm_node_pass]: 3.36999e-06 [meta_fg_expand]: 1.95001e-06 [flash_sp_send_recv_attached]: 1.10999e-06 [receive_attached]: 1.57999e-06 [after_resolve]: 1.003e-05 [a_after_grad]: 9.01998e-06 [renormalize]: 1.00001e-07 [add_forward_monad_depend]: 1.76e-06 [auto_monad_grad]: 1.48002e-06 [auto_monad_eliminator]: 7.28999e-06 [cse]: 1.517e-05 [a_3]: 3.324e-05 [py_interpret_to_execute_after_opt_a]: 1.16e-05 [slice_cell_reuse_recomputed_activation]: 1.95001e-06 [rewriter_after_opt_a]: 3.445e-05 [convert_after_rewriter]: 6.46e-06 [order_py_execute_after_rewriter]: 5.30001e-06 [mutable_eliminate]: 0.032584 [opt_b]: 0.00026202, [1] [Cycle 1]: 0.00025084, [7] [b_1]: 0.00013749 [b_2]: 9.69999e-06 [updatestate_depend_eliminate]: 1.406e-05 [updatestate_assign_eliminate]: 3.10002e-06 [updatestate_loads_eliminate]: 2.75002e-06 [renormalize]: 1.47001e-06 [cse]: 4.321e-05 [optimize_parallel_all_gather_comm]: 2.699e-05 [overlap_param_gather]: 1.95001e-06 [cconv]: 4.072e-05 [loop_unroll]: 0.00078826 [opt_after_cconv]: 0.00012301, [1] [Cycle 1]: 0.00011586, [7] [c_1]: 3.15e-05 [parameter_eliminate]: 7.15e-06 [updatestate_depend_eliminate]: 8.80999e-06 [updatestate_assign_eliminate]: 2.66999e-06 [updatestate_loads_eliminate]: 3.25e-06 [cse]: 2.757e-05 [renormalize]: 7.60017e-07 [remove_dup_value]: 1.49e-05 [tuple_transform]: 8.609e-05, [1] [Cycle 1]: 8.048e-05, [4] [d_1]: 5.13e-05 [none_parameter_eliminate]: 2.00002e-06 [renormalize]: 1.90019e-07 [switch_simplify]: 7.05e-06 [partial_unused_args_eliminate]: 2.15002e-06 [add_recomputation]: 5.878e-05 [cse_after_recomputation]: 2.184e-05, [1] [Cycle 1]: 1.675e-05, [1] [cse]: 1.152e-05 [environ_conv]: 6.78998e-06 [swap_dp_allreduce_reducescatter]: 5.80002e-06 [bias_add_comm_swap]: 3.48e-06 [label_micro_interleaved_index]: 5.90002e-06 [label_fine_grained_interleaved_index]: 2.93e-06 [merge_cast_opt]: 1.73002e-06 [slice_recompute_activation]: 2.09e-06 [micro_interleaved_order_control]: 2.74999e-06 [assign_add_opt]: 1.42e-06 [ForceFp32Comm]: 9.29984e-07 [remove_cast_before_assign_add]: 1.12999e-06 [full_micro_interleaved_order_control]: 2.53e-06 [reorder_send_recv_between_fp_bp]: 2.61999e-06 [comm_op_add_attrs]: 1.25001e-06 [add_comm_op_reuse_tag]: 1.20999e-06 [interleave_split_concat_branches]: 1.19e-06 [interleave_parallel_branches]: 1.20999e-06 [overlap_opt_shard_in_pipeline]: 1.72999e-06 [overlap_opt_shard_grad_in_pipeline]: 1.73002e-06 [control_data_broadcast_order]: 1.592e-05 [grouped_pairwise_exchange_alltoall]: 1.79e-06 [offloading_packed_experts]: 4.06001e-06 [overlap_recompute_and_grad_model_parallel]: 5.27001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.64998e-06 [overlap_recompute_allgather_and_fa_grad]: 1.35001e-06 [overlap_recompute_comm]: 2.51e-06 [overlap_grad_ring_attention]: 4.58001e-06 [overlap_grad_flash_sp]: 2.495e-05 [begin_end_overlap_inline]: 5.19998e-07 [split_matmul_comm_elemetwise]: 2.16e-06 [split_layernorm_comm]: 2.15002e-06 [handle_group_info]: 1.13001e-06 [symbol_engine_optimizer]: 0.0003079, [1] [Cycle 1]: 0.00030201, [6] [build]: 0.00019451 [elim_shapecalc]: 2.11e-05 [elim_not_effective]: 1.763e-05 [opt_reshape]: 7.35998e-06 [fold_const_symbol]: 1.946e-05 [renormalize]: 2.3999e-07 [detach_backward]: 2.80997e-06 [pipeline_parallel_scheduler]: 1.79998e-06 [auto_monad_reorder]: 2.111e-05 [get_jit_bprop_graph]: 2.27999e-06 [rewriter_after_jit_bprop_graph]: 6.71e-06 [opt_after_jit_grad]: 0.00061955 [validate]: 4.864e-05 Sums bootstrap : 0.000586s : 1.13% type_inference : 0.012592s : 24.36% event_method : 0.000058s : 0.11% auto_monad : 0.000075s : 0.15% graph_reusing : 0.000007s : 0.01% inline : 0.000003s : 0.01% add_attr.add_attr_with_inline.tag_attr : 0.000041s : 0.08% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000006s : 0.01% parallel-infer-symbol : 0.000004s : 0.01% pre_auto_parallel : 0.000046s : 0.09% insert-virtual-dataset : 0.000002s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000032s : 0.06% optimize.rewriter_before_opt_a : 0.000168s : 0.32% optimize.opt_a.expand_dump_flag : 0.000006s : 0.01% optimize.opt_a.switch_simplify : 0.000062s : 0.12% optimize.opt_a.loop_unroll : 0.000045s : 0.09% optimize.opt_a.a_1 : 0.001183s : 2.29% optimize.opt_a.with_stream_mark : 0.000035s : 0.07% optimize.opt_a.recompute_prepare : 0.000015s : 0.03% optimize.opt_a.updatestate_depend_eliminate : 0.000007s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.01% optimize.opt_a.parameter_eliminate : 0.000003s : 0.01% optimize.opt_a.a_2 : 0.000210s : 0.41% optimize.opt_a.accelerated_algorithm : 0.000015s : 0.03% optimize.opt_a.shard : 0.000004s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000003s : 0.01% optimize.opt_a.shard_inline : 0.000012s : 0.02% optimize.opt_a.merge_send_recv : 0.000015s : 0.03% optimize.opt_a.auto_parallel : 0.000016s : 0.03% optimize.opt_a.parallel : 0.000025s : 0.05% optimize.opt_a.flash_sp : 0.000013s : 0.02% optimize.opt_a.merge_comm : 0.000007s : 0.01% optimize.opt_a.allreduce_fusion : 0.000006s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000017s : 0.03% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000014s : 0.03% optimize.opt_a.virtual_dataset : 0.000012s : 0.02% optimize.opt_a.get_grad_eliminate_ : 0.000011s : 0.02% optimize.opt_a.virtual_output : 0.000011s : 0.02% optimize.opt_a.merge_forward : 0.000007s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000003s : 0.01% optimize.opt_a.offload_activation : 0.000017s : 0.03% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000023s : 0.04% optimize.opt_a.merge_recompute_call_nodes : 0.000002s : 0.00% optimize.opt_a.before_grad : 0.000020s : 0.04% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000007s : 0.01% optimize.opt_a.meta_fg_expand : 0.000005s : 0.01% optimize.opt_a.flash_sp_send_recv_attached : 0.000004s : 0.01% optimize.opt_a.receive_attached : 0.000004s : 0.01% optimize.opt_a.after_resolve : 0.000021s : 0.04% optimize.opt_a.a_after_grad : 0.000018s : 0.04% optimize.opt_a.renormalize : 0.000996s : 1.93% optimize.opt_a.add_forward_monad_depend : 0.000008s : 0.01% optimize.opt_a.auto_monad_grad : 0.000004s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000023s : 0.05% optimize.opt_a.cse : 0.000044s : 0.09% optimize.opt_a.a_3 : 0.000079s : 0.15% optimize.py_interpret_to_execute_after_opt_a : 0.000012s : 0.02% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000034s : 0.07% optimize.convert_after_rewriter : 0.000006s : 0.01% optimize.order_py_execute_after_rewriter : 0.000005s : 0.01% optimize.mutable_eliminate : 0.032584s : 63.03% optimize.opt_b.b_1 : 0.000137s : 0.27% optimize.opt_b.b_2 : 0.000010s : 0.02% optimize.opt_b.updatestate_depend_eliminate : 0.000014s : 0.03% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.01% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000043s : 0.08% optimize.optimize_parallel_all_gather_comm : 0.000027s : 0.05% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000041s : 0.08% optimize.loop_unroll : 0.000788s : 1.52% optimize.opt_after_cconv.c_1 : 0.000032s : 0.06% optimize.opt_after_cconv.parameter_eliminate : 0.000007s : 0.01% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000009s : 0.02% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.cse : 0.000028s : 0.05% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000015s : 0.03% optimize.tuple_transform.d_1 : 0.000051s : 0.10% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000059s : 0.11% optimize.cse_after_recomputation.cse : 0.000012s : 0.02% optimize.environ_conv : 0.000007s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.01% optimize.bias_add_comm_swap : 0.000003s : 0.01% optimize.label_micro_interleaved_index : 0.000006s : 0.01% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.01% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.01% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000003s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.01% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000002s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000016s : 0.03% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.01% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000002s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000003s : 0.00% optimize.overlap_grad_ring_attention : 0.000005s : 0.01% optimize.overlap_grad_flash_sp : 0.000025s : 0.05% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000195s : 0.38% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000021s : 0.04% optimize.symbol_engine_optimizer.elim_not_effective : 0.000018s : 0.03% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.01% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000019s : 0.04% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.01% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000021s : 0.04% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000007s : 0.01% opt_after_jit_grad : 0.000620s : 1.20% validate : 0.000049s : 0.09% Time group info: ------[substitution.] 0.000344 40 15.22% : 0.000052s : 8: substitution.arithmetic_simplify 1.16% : 0.000004s : 2: substitution.elim_not_effective 2.83% : 0.000010s : 2: substitution.fold_const_symbol 2.11% : 0.000007s : 4: substitution.graph_param_transform 68.05% : 0.000234s : 6: substitution.inline 1.20% : 0.000004s : 4: substitution.j_node_and_user_rematch 1.42% : 0.000005s : 4: substitution.remove_not_recompute_node 1.48% : 0.000005s : 4: substitution.replace_old_param 6.52% : 0.000022s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.012522 2 90.72% : 0.011360s : 1: type_inference.infer 9.28% : 0.001163s : 1: type_inference.specialize ------[replace.] 0.000100 12 57.91% : 0.000058s : 6: replace.inline 42.09% : 0.000042s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000250 12 92.19% : 0.000230s : 6: match.inline 7.81% : 0.000020s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000249 1534 0.91% : 0.000002s : 17: predicate.accumulaten_eliminater 1.16% : 0.000003s : 4: predicate.ad_related_special_op_eliminate 0.54% : 0.000001s : 8: predicate.addn_check_dump 0.95% : 0.000002s : 17: predicate.addn_zero_filter 0.88% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.01% : 0.000005s : 25: predicate.arithmetic_simplify 0.89% : 0.000002s : 17: predicate.cast_eliminate 0.45% : 0.000001s : 8: predicate.check_bprop_eliminate 0.37% : 0.000001s : 8: predicate.compare_switch_simplify 0.14% : 0.000000s : 4: predicate.const_output_eliminate 0.43% : 0.000001s : 8: predicate.depend_value_elim 0.93% : 0.000002s : 17: predicate.dict_get_item_const_eliminator 1.01% : 0.000003s : 17: predicate.dict_get_item_eliminator 0.93% : 0.000002s : 17: predicate.dict_set_item_eliminator 1.06% : 0.000003s : 8: predicate.dumpgradient_eliminate 0.28% : 0.000001s : 4: predicate.elim_not_effective 0.75% : 0.000002s : 4: predicate.elim_shapecalc_of_broadcastargs 1.19% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.08% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.10% : 0.000003s : 21: predicate.environ_get_depend_swap 1.64% : 0.000004s : 29: predicate.environ_get_eliminate 1.23% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.54% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.45% : 0.000006s : 29: predicate.float_depend_g_call 0.42% : 0.000001s : 8: predicate.float_environ_get_switch 0.64% : 0.000002s : 12: predicate.float_tuple_getitem_switch 0.13% : 0.000000s : 4: predicate.fold_const_symbol 0.55% : 0.000001s : 8: predicate.get_grad_eliminate 0.24% : 0.000001s : 4: predicate.graph_param_transform 0.58% : 0.000001s : 8: predicate.incorporate_call 0.38% : 0.000001s : 8: predicate.incorporate_call_switch 6.34% : 0.000016s : 70: predicate.inline 0.69% : 0.000002s : 8: predicate.inline_without_move 0.25% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.85% : 0.000002s : 8: predicate.less_batch_normalization 1.81% : 0.000005s : 31: predicate.list_to_tuple_eliminator_ 2.84% : 0.000007s : 48: predicate.load_eliminater 1.32% : 0.000003s : 4: predicate.loop_unroll_after_grad 2.41% : 0.000006s : 47: predicate.loop_unroll_before_grad 1.56% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.41% : 0.000001s : 8: predicate.merge_addn 0.43% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.40% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.82% : 0.000002s : 17: predicate.minmaximum_grad 2.79% : 0.000007s : 4: predicate.mutable_eliminate 0.27% : 0.000001s : 4: predicate.opt_reshape 0.44% : 0.000001s : 4: predicate.parallel_virtual_node 2.09% : 0.000005s : 29: predicate.partial_defer_inline 1.55% : 0.000004s : 27: predicate.partial_eliminate 0.87% : 0.000002s : 17: predicate.print_const_string_wrapper 0.43% : 0.000001s : 8: predicate.reduce_all_const_elim 1.19% : 0.000003s : 17: predicate.reduce_eliminate 2.50% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.33% : 0.000001s : 8: predicate.remove_not_recompute_node 1.34% : 0.000003s : 31: predicate.replace_applicator 0.42% : 0.000001s : 8: predicate.replace_old_param 0.48% : 0.000001s : 4: predicate.reset_defer_inline 0.99% : 0.000002s : 17: predicate.reshape_eliminate 0.53% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.36% : 0.000001s : 4: predicate.row_tensor_eliminate 0.59% : 0.000001s : 8: predicate.same_eliminate 0.37% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.57% : 0.000001s : 8: predicate.shard_identity_eliminate 0.71% : 0.000002s : 8: predicate.special_op_eliminate 0.54% : 0.000001s : 8: predicate.specialize_transform 0.71% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.51% : 0.000001s : 8: predicate.stack_unstack_eliminate 0.31% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.65% : 0.000004s : 29: predicate.switch_defer_inline 1.98% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.01% : 0.000012s : 88: predicate.switch_simplify 0.90% : 0.000002s : 17: predicate.tile_eliminate 0.91% : 0.000002s : 17: predicate.transpose_eliminate 1.60% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.44% : 0.000004s : 25: predicate.tuple_list_get_item_const_eliminator 1.56% : 0.000004s : 25: predicate.tuple_list_get_item_depend_reorder 3.81% : 0.000010s : 39: predicate.tuple_list_get_item_eliminator 1.42% : 0.000004s : 25: predicate.tuple_list_get_set_item_eliminator 2.32% : 0.000006s : 33: predicate.tuple_list_set_item_eliminator 1.78% : 0.000004s : 31: predicate.tuple_to_list_eliminator_ 2.36% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.14% : 0.000008s : 56: predicate.updatestate_useless_node_eliminater 0.34% : 0.000001s : 4: predicate.value_based_eliminate 0.47% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.48% : 0.000001s : 8: predicate.virtual_output_eliminate 0.24% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.70% : 0.000002s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000916 15 57.41% : 0.000526s : 7: func_graph_cloner_run.FuncGraphClonerGraph 42.59% : 0.000390s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.101959 192 0.00% : 0.000004s : 1: ForceFp32Comm 3.80% : 0.003878s : 1: add_attr 3.79% : 0.003866s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.06% : 0.000063s : 1: add_recomputation 0.00% : 0.000005s : 1: assign_add_opt 0.08% : 0.000081s : 1: auto_monad 0.03% : 0.000026s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.01% : 0.000007s : 1: bias_add_comm_swap 0.61% : 0.000619s : 1: bootstrap 0.04% : 0.000044s : 1: cconv 0.01% : 0.000005s : 1: comm_op_add_attrs 0.02% : 0.000019s : 1: control_data_broadcast_order 0.01% : 0.000010s : 1: convert_after_rewriter 0.03% : 0.000026s : 1: cse_after_recomputation 0.01% : 0.000005s : 1: dataset_repeat_opt 0.01% : 0.000008s : 1: detach_backward 0.01% : 0.000010s : 1: environ_conv 0.07% : 0.000067s : 1: event_method 0.01% : 0.000005s : 1: full_micro_interleaved_order_control 0.01% : 0.000005s : 1: get_jit_bprop_graph 0.01% : 0.000011s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.01% : 0.000006s : 1: inline 0.01% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.01% : 0.000007s : 1: label_fine_grained_interleaved_index 0.01% : 0.000009s : 1: label_micro_interleaved_index 0.78% : 0.000800s : 1: loop_unroll 0.01% : 0.000005s : 1: merge_cast_opt 0.05% : 0.000048s : 1: micro_interleaved_order_control 31.98% : 0.032609s : 1: mutable_eliminate 0.01% : 0.000008s : 1: offloading_packed_experts 0.02% : 0.000022s : 1: opt.transform.loop_unroll_optimizer 0.04% : 0.000042s : 1: opt.transform.mutable_eliminate 1.63% : 0.001667s : 78: opt.transform.opt_a 0.03% : 0.000030s : 1: opt.transform.opt_after_cconv 0.03% : 0.000029s : 1: opt.transform.opt_after_jit_grad 0.11% : 0.000110s : 28: opt.transform.opt_b 0.06% : 0.000056s : 2: opt.transform.opt_trans_graph 0.06% : 0.000059s : 4: opt.transform.symbol_engine_opt 3.45% : 0.003515s : 1: opt_a 0.12% : 0.000127s : 1: opt_after_cconv 0.62% : 0.000630s : 1: opt_after_jit_grad 0.26% : 0.000267s : 1: opt_b 37.80% : 0.038537s : 1: optimize 0.03% : 0.000032s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000008s : 1: order_py_execute_after_rewriter 0.03% : 0.000029s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000008s : 1: overlap_grad_ring_attention 0.01% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000005s : 1: overlap_opt_shard_in_pipeline 0.01% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.01% : 0.000006s : 1: overlap_recompute_comm 0.01% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.01% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000004s : 1: pipeline_split 0.05% : 0.000051s : 1: pre_auto_parallel 0.03% : 0.000035s : 1: py_interpret_to_execute 0.01% : 0.000015s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.02% : 0.000020s : 1: remove_dup_value 0.54% : 0.000546s : 1: renormalize.infer 0.43% : 0.000439s : 1: renormalize.specialize 0.01% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000011s : 1: rewriter_after_jit_bprop_graph 0.04% : 0.000038s : 1: rewriter_after_opt_a 0.17% : 0.000176s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.01% : 0.000006s : 1: split_matmul_comm_elemetwise 0.01% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.31% : 0.000311s : 1: symbol_engine_optimizer 0.09% : 0.000089s : 1: tuple_transform 12.43% : 0.012670s : 1: type_inference TotalTime = 0.0563911, [21] [bootstrap]: 0.00052465 [type_inference]: 0.0245669 [event_method]: 6.662e-05 [auto_monad]: 0.00015053 [graph_reusing]: 1.029e-05 [inline]: 2.66e-06 [add_attr]: 0.00414877, [1] [add_attr_with_inline]: 0.00413713, [1] [Cycle 1]: 0.00011389, [2] [tag_attr]: 5.816e-05 [meta_addattr_fg_expand]: 1.183e-05 [parallel-infer-symbol]: 3.51001e-06 [pre_auto_parallel]: 7.745e-05 [insert-virtual-dataset]: 2.58e-06 [parallel-infer-symbol-second]: 7.89994e-07 [dataset_repeat_opt]: 2.15002e-06 [pipeline_split]: 1.80001e-06 [optimize]: 0.0256893, [53] [py_interpret_to_execute]: 5.638e-05 [rewriter_before_opt_a]: 0.000211 [opt_a]: 0.0198469, [3] [Cycle 1]: 0.0109858, [45] [expand_dump_flag]: 4.68001e-06 [switch_simplify]: 9.925e-05 [loop_unroll]: 8.216e-05 [a_1]: 0.0019684 [with_stream_mark]: 3.887e-05 [recompute_prepare]: 2.871e-05 [updatestate_depend_eliminate]: 1.199e-05 [updatestate_assign_eliminate]: 9.35001e-06 [updatestate_loads_eliminate]: 8.43001e-06 [parameter_eliminate]: 3.02002e-06 [a_2]: 0.00028419 [accelerated_algorithm]: 3.951e-05 [shard]: 2.44999e-06 [meta_shard_fg_expand]: 5.84999e-06 [shard_inline]: 1.817e-05 [merge_send_recv]: 2.12e-05 [auto_parallel]: 1.689e-05 [parallel]: 2.097e-05 [flash_sp]: 1.27e-05 [merge_comm]: 1.193e-05 [allreduce_fusion]: 1.036e-05 [matmul_add_comm_reduction]: 3.601e-05 [allreduce_slice_to_reducescatter]: 1.14e-06 [virtual_shard_identity]: 2.203e-05 [virtual_dataset]: 1.72e-05 [get_grad_eliminate_]: 3.638e-05 [virtual_output]: 2.044e-05 [merge_forward]: 1.3e-05 [cell_reuse_recompute_pass]: 2.31e-06 [offload_activation]: 2.235e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.911e-05 [merge_recompute_call_nodes]: 1.78002e-06 [before_grad]: 3.291e-05 [set_forward_comm_id_for_comm_node_pass]: 1.272e-05 [meta_fg_expand]: 0.00184625 [flash_sp_send_recv_attached]: 6.39999e-06 [receive_attached]: 2.31e-06 [after_resolve]: 6.865e-05 [a_after_grad]: 8.967e-05 [renormalize]: 0.00485328 [add_forward_monad_depend]: 1.331e-05 [auto_monad_grad]: 6.88e-06 [auto_monad_eliminator]: 9.101e-05 [cse]: 0.00024623 [a_3]: 0.00044759 [Cycle 2]: 0.00662044, [45] [expand_dump_flag]: 2.99999e-06 [switch_simplify]: 5.954e-05 [loop_unroll]: 5.578e-05 [a_1]: 0.00265918 [with_stream_mark]: 3.939e-05 [recompute_prepare]: 2.515e-05 [updatestate_depend_eliminate]: 1.311e-05 [updatestate_assign_eliminate]: 1.154e-05 [updatestate_loads_eliminate]: 1.006e-05 [parameter_eliminate]: 2.48e-06 [a_2]: 0.00040635 [accelerated_algorithm]: 2.67e-05 [shard]: 2.31e-06 [meta_shard_fg_expand]: 7.16999e-06 [shard_inline]: 1.982e-05 [merge_send_recv]: 1.873e-05 [auto_parallel]: 1.977e-05 [parallel]: 1.016e-05 [flash_sp]: 4.93001e-06 [merge_comm]: 1.255e-05 [allreduce_fusion]: 1.165e-05 [matmul_add_comm_reduction]: 2.177e-05 [allreduce_slice_to_reducescatter]: 8.99978e-07 [virtual_shard_identity]: 2.164e-05 [virtual_dataset]: 1.858e-05 [get_grad_eliminate_]: 1.904e-05 [virtual_output]: 1.898e-05 [merge_forward]: 5.987e-05 [cell_reuse_recompute_pass]: 2.81999e-06 [offload_activation]: 2.644e-05 [cell_reuse_handle_not_recompute_node_pass]: 4.716e-05 [merge_recompute_call_nodes]: 1.71e-06 [before_grad]: 3.565e-05 [set_forward_comm_id_for_comm_node_pass]: 1.363e-05 [meta_fg_expand]: 0.00024524 [flash_sp_send_recv_attached]: 2.11e-06 [receive_attached]: 2.20002e-06 [after_resolve]: 3.013e-05 [a_after_grad]: 3.101e-05 [renormalize]: 0.00195332 [add_forward_monad_depend]: 1.068e-05 [auto_monad_grad]: 2.96001e-06 [auto_monad_eliminator]: 3.877e-05 [cse]: 0.00011542 [a_3]: 0.00014471 [Cycle 3]: 0.00221974, [45] [expand_dump_flag]: 3.08e-06 [switch_simplify]: 2.159e-05 [loop_unroll]: 1.92e-05 [a_1]: 0.00070103 [with_stream_mark]: 2.935e-05 [recompute_prepare]: 0.00012552 [updatestate_depend_eliminate]: 1.548e-05 [updatestate_assign_eliminate]: 1.131e-05 [updatestate_loads_eliminate]: 1.056e-05 [parameter_eliminate]: 2.92002e-06 [a_2]: 0.00028701 [accelerated_algorithm]: 0.00010375 [shard]: 2.29001e-06 [meta_shard_fg_expand]: 5.74e-06 [shard_inline]: 2.037e-05 [merge_send_recv]: 1.986e-05 [auto_parallel]: 2.032e-05 [parallel]: 8.86002e-06 [flash_sp]: 1.70001e-06 [merge_comm]: 1.245e-05 [allreduce_fusion]: 1.179e-05 [matmul_add_comm_reduction]: 2.121e-05 [allreduce_slice_to_reducescatter]: 1.02e-06 [virtual_shard_identity]: 2.094e-05 [virtual_dataset]: 1.867e-05 [get_grad_eliminate_]: 1.75e-05 [virtual_output]: 1.765e-05 [merge_forward]: 1.174e-05 [cell_reuse_recompute_pass]: 3.27002e-06 [offload_activation]: 2.356e-05 [cell_reuse_handle_not_recompute_node_pass]: 4.204e-05 [merge_recompute_call_nodes]: 1.52001e-06 [before_grad]: 3.447e-05 [set_forward_comm_id_for_comm_node_pass]: 1.292e-05 [meta_fg_expand]: 8.45999e-06 [flash_sp_send_recv_attached]: 1.76003e-06 [receive_attached]: 2.32999e-06 [after_resolve]: 2.374e-05 [a_after_grad]: 2.97e-05 [renormalize]: 9.00181e-08 [add_forward_monad_depend]: 3.31999e-06 [auto_monad_grad]: 2.59001e-06 [auto_monad_eliminator]: 3.344e-05 [cse]: 7.551e-05 [a_3]: 0.00012769 [py_interpret_to_execute_after_opt_a]: 3.08e-05 [slice_cell_reuse_recomputed_activation]: 2.19999e-06 [rewriter_after_opt_a]: 0.0002517 [convert_after_rewriter]: 2.222e-05 [order_py_execute_after_rewriter]: 1.291e-05 [mutable_eliminate]: 0.00076426 [opt_b]: 0.00097853, [1] [Cycle 1]: 0.00096808, [7] [b_1]: 0.00075547 [b_2]: 2.502e-05 [updatestate_depend_eliminate]: 1.963e-05 [updatestate_assign_eliminate]: 1.039e-05 [updatestate_loads_eliminate]: 1.01e-05 [renormalize]: 1.05001e-06 [cse]: 9.265e-05 [optimize_parallel_all_gather_comm]: 4.065e-05 [overlap_param_gather]: 2.11e-06 [cconv]: 4.118e-05 [loop_unroll]: 0.00073549 [opt_after_cconv]: 0.00029305, [1] [Cycle 1]: 0.00028408, [7] [c_1]: 0.00010388 [parameter_eliminate]: 5.49998e-06 [updatestate_depend_eliminate]: 3.227e-05 [updatestate_assign_eliminate]: 1.031e-05 [updatestate_loads_eliminate]: 1.034e-05 [cse]: 8.11e-05 [renormalize]: 7.39994e-07 [remove_dup_value]: 0.00010291 [tuple_transform]: 0.00039449, [1] [Cycle 1]: 0.00038622, [4] [d_1]: 0.00032758 [none_parameter_eliminate]: 3.18e-06 [renormalize]: 4.59986e-07 [switch_simplify]: 2.286e-05 [partial_unused_args_eliminate]: 2.02999e-06 [add_recomputation]: 0.0001333 [cse_after_recomputation]: 7.052e-05, [1] [Cycle 1]: 6.442e-05, [1] [cse]: 5.62e-05 [environ_conv]: 2.053e-05 [swap_dp_allreduce_reducescatter]: 1.634e-05 [bias_add_comm_swap]: 2.93e-06 [label_micro_interleaved_index]: 8.25999e-06 [label_fine_grained_interleaved_index]: 2.61e-06 [merge_cast_opt]: 1.50001e-06 [slice_recompute_activation]: 2.09e-06 [micro_interleaved_order_control]: 2.72001e-06 [assign_add_opt]: 1.40001e-06 [ForceFp32Comm]: 9.20001e-07 [remove_cast_before_assign_add]: 1.09e-06 [full_micro_interleaved_order_control]: 2.27999e-06 [reorder_send_recv_between_fp_bp]: 2.53998e-06 [comm_op_add_attrs]: 1.20999e-06 [add_comm_op_reuse_tag]: 1.25001e-06 [interleave_split_concat_branches]: 1.49e-06 [interleave_parallel_branches]: 1.04e-06 [overlap_opt_shard_in_pipeline]: 1.25999e-06 [overlap_opt_shard_grad_in_pipeline]: 1.70001e-06 [control_data_broadcast_order]: 3.693e-05 [grouped_pairwise_exchange_alltoall]: 2.12999e-06 [offloading_packed_experts]: 1.073e-05 [overlap_recompute_and_grad_model_parallel]: 9.92999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.27999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.45999e-06 [overlap_recompute_comm]: 2.46e-06 [overlap_grad_ring_attention]: 8.93002e-06 [overlap_grad_flash_sp]: 5.049e-05 [begin_end_overlap_inline]: 6.19999e-07 [split_matmul_comm_elemetwise]: 2.33998e-06 [split_layernorm_comm]: 1.76003e-06 [handle_group_info]: 1.18001e-06 [symbol_engine_optimizer]: 0.00113958, [1] [Cycle 1]: 0.00113268, [6] [build]: 0.00086259 [elim_shapecalc]: 8.096e-05 [elim_not_effective]: 4.531e-05 [opt_reshape]: 2.787e-05 [fold_const_symbol]: 5.278e-05 [renormalize]: 4.60015e-07 [detach_backward]: 3.33998e-06 [pipeline_parallel_scheduler]: 1.99e-06 [auto_monad_reorder]: 5.255e-05 [get_jit_bprop_graph]: 2.24999e-06 [rewriter_after_jit_bprop_graph]: 6.21e-06 [opt_after_jit_grad]: 0.00079162 [validate]: 8.692e-05 Sums bootstrap : 0.000525s : 1.04% type_inference : 0.024567s : 48.59% event_method : 0.000067s : 0.13% auto_monad : 0.000151s : 0.30% graph_reusing : 0.000010s : 0.02% inline : 0.000003s : 0.01% add_attr.add_attr_with_inline.tag_attr : 0.000058s : 0.12% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000012s : 0.02% parallel-infer-symbol : 0.000004s : 0.01% pre_auto_parallel : 0.000077s : 0.15% insert-virtual-dataset : 0.000003s : 0.01% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000056s : 0.11% optimize.rewriter_before_opt_a : 0.000211s : 0.42% optimize.opt_a.expand_dump_flag : 0.000011s : 0.02% optimize.opt_a.switch_simplify : 0.000180s : 0.36% optimize.opt_a.loop_unroll : 0.000157s : 0.31% optimize.opt_a.a_1 : 0.005329s : 10.54% optimize.opt_a.with_stream_mark : 0.000108s : 0.21% optimize.opt_a.recompute_prepare : 0.000179s : 0.35% optimize.opt_a.updatestate_depend_eliminate : 0.000041s : 0.08% optimize.opt_a.updatestate_assign_eliminate : 0.000032s : 0.06% optimize.opt_a.updatestate_loads_eliminate : 0.000029s : 0.06% optimize.opt_a.parameter_eliminate : 0.000008s : 0.02% optimize.opt_a.a_2 : 0.000978s : 1.93% optimize.opt_a.accelerated_algorithm : 0.000170s : 0.34% optimize.opt_a.shard : 0.000007s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000019s : 0.04% optimize.opt_a.shard_inline : 0.000058s : 0.12% optimize.opt_a.merge_send_recv : 0.000060s : 0.12% optimize.opt_a.auto_parallel : 0.000057s : 0.11% optimize.opt_a.parallel : 0.000040s : 0.08% optimize.opt_a.flash_sp : 0.000019s : 0.04% optimize.opt_a.merge_comm : 0.000037s : 0.07% optimize.opt_a.allreduce_fusion : 0.000034s : 0.07% optimize.opt_a.matmul_add_comm_reduction : 0.000079s : 0.16% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000003s : 0.01% optimize.opt_a.virtual_shard_identity : 0.000065s : 0.13% optimize.opt_a.virtual_dataset : 0.000054s : 0.11% optimize.opt_a.get_grad_eliminate_ : 0.000073s : 0.14% optimize.opt_a.virtual_output : 0.000057s : 0.11% optimize.opt_a.merge_forward : 0.000085s : 0.17% optimize.opt_a.cell_reuse_recompute_pass : 0.000008s : 0.02% optimize.opt_a.offload_activation : 0.000072s : 0.14% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000128s : 0.25% optimize.opt_a.merge_recompute_call_nodes : 0.000005s : 0.01% optimize.opt_a.before_grad : 0.000103s : 0.20% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000039s : 0.08% optimize.opt_a.meta_fg_expand : 0.002100s : 4.15% optimize.opt_a.flash_sp_send_recv_attached : 0.000010s : 0.02% optimize.opt_a.receive_attached : 0.000007s : 0.01% optimize.opt_a.after_resolve : 0.000123s : 0.24% optimize.opt_a.a_after_grad : 0.000150s : 0.30% optimize.opt_a.renormalize : 0.006807s : 13.46% optimize.opt_a.add_forward_monad_depend : 0.000027s : 0.05% optimize.opt_a.auto_monad_grad : 0.000012s : 0.02% optimize.opt_a.auto_monad_eliminator : 0.000163s : 0.32% optimize.opt_a.cse : 0.000437s : 0.86% optimize.opt_a.a_3 : 0.000720s : 1.42% optimize.py_interpret_to_execute_after_opt_a : 0.000031s : 0.06% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000252s : 0.50% optimize.convert_after_rewriter : 0.000022s : 0.04% optimize.order_py_execute_after_rewriter : 0.000013s : 0.03% optimize.mutable_eliminate : 0.000764s : 1.51% optimize.opt_b.b_1 : 0.000755s : 1.49% optimize.opt_b.b_2 : 0.000025s : 0.05% optimize.opt_b.updatestate_depend_eliminate : 0.000020s : 0.04% optimize.opt_b.updatestate_assign_eliminate : 0.000010s : 0.02% optimize.opt_b.updatestate_loads_eliminate : 0.000010s : 0.02% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000093s : 0.18% optimize.optimize_parallel_all_gather_comm : 0.000041s : 0.08% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000041s : 0.08% optimize.loop_unroll : 0.000735s : 1.45% optimize.opt_after_cconv.c_1 : 0.000104s : 0.21% optimize.opt_after_cconv.parameter_eliminate : 0.000005s : 0.01% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000032s : 0.06% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000010s : 0.02% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000010s : 0.02% optimize.opt_after_cconv.cse : 0.000081s : 0.16% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000103s : 0.20% optimize.tuple_transform.d_1 : 0.000328s : 0.65% optimize.tuple_transform.none_parameter_eliminate : 0.000003s : 0.01% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000023s : 0.05% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000133s : 0.26% optimize.cse_after_recomputation.cse : 0.000056s : 0.11% optimize.environ_conv : 0.000021s : 0.04% optimize.swap_dp_allreduce_reducescatter : 0.000016s : 0.03% optimize.bias_add_comm_swap : 0.000003s : 0.01% optimize.label_micro_interleaved_index : 0.000008s : 0.02% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.01% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.01% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.01% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000037s : 0.07% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000011s : 0.02% optimize.overlap_recompute_and_grad_model_parallel : 0.000010s : 0.02% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000009s : 0.02% optimize.overlap_grad_flash_sp : 0.000050s : 0.10% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000863s : 1.71% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000081s : 0.16% optimize.symbol_engine_optimizer.elim_not_effective : 0.000045s : 0.09% optimize.symbol_engine_optimizer.opt_reshape : 0.000028s : 0.06% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000053s : 0.10% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.01% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000053s : 0.10% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000006s : 0.01% opt_after_jit_grad : 0.000792s : 1.57% validate : 0.000087s : 0.17% Time group info: ------[substitution.] 0.001653 367 4.87% : 0.000081s : 16: substitution.arithmetic_simplify 0.79% : 0.000013s : 15: substitution.elim_not_effective 1.90% : 0.000031s : 2: substitution.elim_shapecalc_of_broadcastargs 0.33% : 0.000005s : 6: substitution.float_depend_g_call 1.06% : 0.000017s : 10: substitution.float_tuple_getitem_switch 1.49% : 0.000025s : 15: substitution.fold_const_symbol 1.02% : 0.000017s : 17: substitution.graph_param_transform 0.20% : 0.000003s : 2: substitution.incorporate_call 0.12% : 0.000002s : 2: substitution.incorporate_call_switch 43.71% : 0.000723s : 20: substitution.inline 1.24% : 0.000020s : 2: substitution.inline_without_move 1.24% : 0.000020s : 42: substitution.j_node_and_user_rematch 1.39% : 0.000023s : 3: substitution.less_batch_normalization 1.65% : 0.000027s : 14: substitution.minmaximum_grad 0.33% : 0.000005s : 2: substitution.opt_reshape 0.41% : 0.000007s : 6: substitution.partial_eliminate 0.65% : 0.000011s : 4: substitution.reduce_eliminate 1.86% : 0.000031s : 42: substitution.remove_not_recompute_node 4.47% : 0.000074s : 10: substitution.replace_applicator 0.80% : 0.000013s : 12: substitution.replace_old_param 2.60% : 0.000043s : 6: substitution.reshape_eliminate 0.21% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.66% : 0.000061s : 18: substitution.tuple_list_convert_item_index_to_positive 1.92% : 0.000032s : 18: substitution.tuple_list_get_item_const_eliminator 2.55% : 0.000042s : 18: substitution.tuple_list_get_item_depend_reorder 7.94% : 0.000131s : 46: substitution.tuple_list_get_item_eliminator 11.58% : 0.000191s : 18: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.024432 2 88.10% : 0.021524s : 1: type_inference.infer 11.90% : 0.002908s : 1: type_inference.specialize ------[replace.] 0.000310 40 58.80% : 0.000182s : 20: replace.inline 41.20% : 0.000128s : 20: replace.tuple_list_get_item_eliminator ------[match.] 0.000756 40 93.90% : 0.000710s : 20: match.inline 6.10% : 0.000046s : 20: match.tuple_list_get_item_eliminator ------[predicate.] 0.001480 8180 0.81% : 0.000012s : 89: predicate.accumulaten_eliminater 0.35% : 0.000005s : 17: predicate.ad_related_special_op_eliminate 0.46% : 0.000007s : 51: predicate.addn_check_dump 0.81% : 0.000012s : 89: predicate.addn_zero_filter 0.75% : 0.000011s : 89: predicate.adjust_all_reduce_mul_add 1.75% : 0.000026s : 140: predicate.arithmetic_simplify 0.81% : 0.000012s : 89: predicate.cast_eliminate 0.88% : 0.000013s : 96: predicate.check_bprop_eliminate 0.54% : 0.000008s : 51: predicate.compare_switch_simplify 0.09% : 0.000001s : 17: predicate.const_output_eliminate 0.48% : 0.000007s : 51: predicate.depend_value_elim 1.03% : 0.000015s : 89: predicate.dict_get_item_const_eliminator 0.90% : 0.000013s : 89: predicate.dict_get_item_eliminator 0.85% : 0.000013s : 89: predicate.dict_set_item_eliminator 0.45% : 0.000007s : 34: predicate.dumpgradient_eliminate 0.10% : 0.000002s : 17: predicate.elim_not_effective 0.26% : 0.000004s : 17: predicate.elim_shapecalc_of_broadcastargs 0.95% : 0.000014s : 106: predicate.environ_add_const_eliminate 0.94% : 0.000014s : 106: predicate.environ_get_add_eliminate 0.95% : 0.000014s : 106: predicate.environ_get_depend_swap 1.49% : 0.000022s : 157: predicate.environ_get_eliminate 0.97% : 0.000014s : 106: predicate.environ_get_set_eliminate 1.19% : 0.000018s : 129: predicate.exchange_switch_depend_value 1.69% : 0.000025s : 129: predicate.float_depend_g_call 0.50% : 0.000007s : 51: predicate.float_environ_get_switch 0.69% : 0.000010s : 68: predicate.float_tuple_getitem_switch 0.08% : 0.000001s : 17: predicate.fold_const_symbol 0.52% : 0.000008s : 51: predicate.get_grad_eliminate 0.10% : 0.000001s : 17: predicate.graph_param_transform 0.48% : 0.000007s : 51: predicate.incorporate_call 0.48% : 0.000007s : 51: predicate.incorporate_call_switch 4.60% : 0.000068s : 354: predicate.inline 0.93% : 0.000014s : 74: predicate.inline_without_move 0.24% : 0.000004s : 51: predicate.j_node_and_user_rematch 0.67% : 0.000010s : 51: predicate.less_batch_normalization 1.43% : 0.000021s : 143: predicate.list_to_tuple_eliminator_ 2.05% : 0.000030s : 232: predicate.load_eliminater 0.47% : 0.000007s : 17: predicate.loop_unroll_after_grad 1.63% : 0.000024s : 179: predicate.loop_unroll_before_grad 1.27% : 0.000019s : 123: predicate.make_slice_get_slice_eliminator 0.49% : 0.000007s : 51: predicate.merge_addn 0.87% : 0.000013s : 96: predicate.micro_step_allgather_replace 0.96% : 0.000014s : 96: predicate.mini_step_allgather_replace 0.82% : 0.000012s : 89: predicate.minmaximum_grad 0.44% : 0.000007s : 17: predicate.mutable_eliminate 0.21% : 0.000003s : 17: predicate.opt_reshape 0.19% : 0.000003s : 17: predicate.parallel_virtual_node 1.59% : 0.000023s : 129: predicate.partial_defer_inline 1.30% : 0.000019s : 126: predicate.partial_eliminate 0.78% : 0.000012s : 89: predicate.print_const_string_wrapper 0.49% : 0.000007s : 51: predicate.reduce_all_const_elim 1.06% : 0.000016s : 89: predicate.reduce_eliminate 2.07% : 0.000031s : 232: predicate.redundant_stop_gradient_eliminater 0.29% : 0.000004s : 51: predicate.remove_not_recompute_node 1.31% : 0.000019s : 205: predicate.replace_applicator 0.47% : 0.000007s : 74: predicate.replace_old_param 0.10% : 0.000002s : 17: predicate.reset_defer_inline 0.86% : 0.000013s : 89: predicate.reshape_eliminate 0.89% : 0.000013s : 96: predicate.row_tensor_add_zeros_like 0.20% : 0.000003s : 17: predicate.row_tensor_eliminate 1.10% : 0.000016s : 96: predicate.same_eliminate 0.31% : 0.000005s : 51: predicate.set_cell_output_no_recompute 0.57% : 0.000008s : 51: predicate.shard_identity_eliminate 0.35% : 0.000005s : 34: predicate.special_op_eliminate 0.56% : 0.000008s : 51: predicate.specialize_transform 1.07% : 0.000016s : 96: predicate.split_environ_get_set_with_tuple_value 0.85% : 0.000013s : 74: predicate.stack_unstack_eliminate 0.19% : 0.000003s : 17: predicate.switch_call_monad_eliminater 1.28% : 0.000019s : 129: predicate.switch_defer_inline 2.13% : 0.000032s : 225: predicate.switch_layer_defer_inline 3.63% : 0.000054s : 376: predicate.switch_simplify 0.85% : 0.000013s : 89: predicate.tile_eliminate 0.80% : 0.000012s : 89: predicate.transpose_eliminate 1.26% : 0.000019s : 123: predicate.tuple_list_convert_item_index_to_positive 1.38% : 0.000020s : 123: predicate.tuple_list_get_item_const_eliminator 1.17% : 0.000017s : 123: predicate.tuple_list_get_item_depend_reorder 20.86% : 0.000309s : 194: predicate.tuple_list_get_item_eliminator 1.22% : 0.000018s : 123: predicate.tuple_list_get_set_item_eliminator 1.79% : 0.000027s : 174: predicate.tuple_list_set_item_eliminator 1.36% : 0.000020s : 143: predicate.tuple_to_list_eliminator_ 2.01% : 0.000030s : 232: predicate.updatestate_pure_node_eliminater 2.55% : 0.000038s : 283: predicate.updatestate_useless_node_eliminater 0.17% : 0.000003s : 17: predicate.value_based_eliminate 0.51% : 0.000007s : 51: predicate.virtual_dataset_eliminate 0.59% : 0.000009s : 51: predicate.virtual_output_eliminate 0.17% : 0.000003s : 17: predicate.virtual_view_grad_eliminate 0.20% : 0.000003s : 17: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.003022 43 60.18% : 0.001819s : 19: func_graph_cloner_run.FuncGraphClonerGraph 39.82% : 0.001203s : 24: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.102647 233 0.00% : 0.000004s : 1: ForceFp32Comm 4.05% : 0.004155s : 1: add_attr 4.03% : 0.004141s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.14% : 0.000140s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.15% : 0.000158s : 1: auto_monad 0.06% : 0.000057s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.01% : 0.000006s : 1: bias_add_comm_swap 0.54% : 0.000557s : 1: bootstrap 0.04% : 0.000045s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.04% : 0.000040s : 1: control_data_broadcast_order 0.03% : 0.000027s : 1: convert_after_rewriter 0.07% : 0.000074s : 1: cse_after_recomputation 0.01% : 0.000005s : 1: dataset_repeat_opt 0.01% : 0.000009s : 1: detach_backward 0.02% : 0.000024s : 1: environ_conv 0.07% : 0.000075s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.01% : 0.000006s : 1: get_jit_bprop_graph 0.01% : 0.000015s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.01% : 0.000006s : 1: inline 0.01% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.01% : 0.000005s : 1: label_fine_grained_interleaved_index 0.01% : 0.000011s : 1: label_micro_interleaved_index 0.73% : 0.000748s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.01% : 0.000006s : 1: micro_interleaved_order_control 0.75% : 0.000775s : 1: mutable_eliminate 0.01% : 0.000014s : 1: offloading_packed_experts 0.04% : 0.000037s : 1: opt.transform.loop_unroll_optimizer 0.03% : 0.000036s : 1: opt.transform.mutable_eliminate 8.13% : 0.008345s : 117: opt.transform.opt_a 0.10% : 0.000102s : 1: opt.transform.opt_after_cconv 0.07% : 0.000068s : 1: opt.transform.opt_after_jit_grad 0.72% : 0.000740s : 28: opt.transform.opt_b 0.34% : 0.000345s : 2: opt.transform.opt_trans_graph 0.19% : 0.000200s : 4: opt.transform.symbol_engine_opt 19.34% : 0.019851s : 1: opt_a 0.29% : 0.000297s : 1: opt_after_cconv 0.78% : 0.000804s : 1: opt_after_jit_grad 0.96% : 0.000983s : 1: opt_b 25.04% : 0.025701s : 1: optimize 0.04% : 0.000045s : 1: optimize_parallel_all_gather_comm 0.02% : 0.000016s : 1: order_py_execute_after_rewriter 0.05% : 0.000055s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000012s : 1: overlap_grad_ring_attention 0.00% : 0.000004s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.01% : 0.000006s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000014s : 1: overlap_recompute_and_grad_model_parallel 0.01% : 0.000005s : 1: overlap_recompute_comm 0.01% : 0.000007s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.01% : 0.000006s : 1: partial_unused_args_eliminate 0.01% : 0.000006s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.08% : 0.000082s : 1: pre_auto_parallel 0.06% : 0.000061s : 1: py_interpret_to_execute 0.03% : 0.000035s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.15% : 0.000152s : 1: remove_dup_value 4.08% : 0.004192s : 2: renormalize.infer 2.52% : 0.002589s : 2: renormalize.specialize 0.01% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000010s : 1: rewriter_after_jit_bprop_graph 0.26% : 0.000263s : 1: rewriter_after_opt_a 0.21% : 0.000216s : 1: rewriter_before_opt_a 0.01% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.01% : 0.000006s : 1: split_matmul_comm_elemetwise 0.02% : 0.000019s : 1: swap_dp_allreduce_reducescatter 1.11% : 0.001143s : 1: symbol_engine_optimizer 0.39% : 0.000398s : 1: tuple_transform 23.96% : 0.024597s : 1: type_inference Start to compare view out with no view out Compare ['Success'] for 0'th output. *****************End Generalize test GRAPH_MODE_O0 ViewTensor case***************** *****************Start Generalize test GRAPH_MODE_O0 EmptyTensor case***************** [WARNING]: TEST_OP_GENERALIZE GRAPH_MODE_O0 'EmptyTensor' in 'disable_case', EmptyTensor case is skipped. *****************End Generalize test GRAPH_MODE_O0 EmptyTensor case***************** *****************Start Generalize test GRAPH_MODE_O0 ScalarTensor case***************** [INFO]: TEST_OP_GENERALIZE Start to test scalar tensor [INFO]: TEST_OP_GENERALIZE Start to test 1d tensor [INFO]: TEST_OP_GENERALIZE Start to compare scalar tensor out with 1d tensor out Compare ['Success'] for 0'th output. *****************End Generalize test GRAPH_MODE_O0 ScalarTensor case***************** *****************Start Generalize test GRAPH_MODE_O0 GradByRequirement case***************** [INFO]: TEST_OP_GENERALIZE Start to test all grads [INFO]: TEST_OP_GENERALIZE Start to test grad by requirement on position 0 [INFO]: TEST_OP_GENERALIZE Start to test grad by requirement on position 1 [INFO]: TEST_OP_GENERALIZE Start to compare grad by requirement with all grads Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. *****************End Generalize test GRAPH_MODE_O0 GradByRequirement case***************** *****************Start Generalize test GRAPH_MODE_O0 Deterministic case***************** [INFO]: TEST_OP_GENERALIZE test deterministic case loop 0 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 1 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 2 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 3 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 4 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. *****************End Generalize test GRAPH_MODE_O0 Deterministic case***************** [WARNING]: TEST_OP_GENERALIZE GRAPH_MODE_GE is skipped. ************************************************************************************************** **************************************TEST_OP_GENERALIZE End************************************** ************************************************************************************************** ************************************************************************************************* **************************************TEST_OP_DYNAMIC Start************************************** ************************************************************************************************* [WARNING]: TEST_DYNAMIC GRAPH_MODE_GE is skipped, Ge backend don't support dynamic situations any longer. *********************Begin DynamicTest for with PYNATIVE_MODE ********************* Start DynamicTest running on PYNATIVE_MODE with static shape with first inputs... End Start DynamicTest running on PYNATIVE_MODE with static shape with second inputs... End Start DynamicTest testing with [PYNATIVE_MODE] [Resize]... Compare ['Success'] for 0'th output of Resize. End Start DynamicTest testing with [PYNATIVE_MODE] [DYNAMIC_SHAPE] [None]... Compare ['Success'] for 0'th output of Forward. Compare ['Success'] for 0'th output of Grad. Compare ['Success'] for 1'th output of Grad. End Start DynamicTest testing with [PYNATIVE_MODE] [DYNAMIC_RANK] [None]... Compare ['Success'] for 0'th output of Forward. Compare ['Success'] for 0'th output of Grad. Compare ['Success'] for 1'th output of Grad. End *********************End DynamicTest for with PYNATIVE_MODE ********************* *********************Begin DynamicTest for with GRAPH_MODE_O0 ********************* Start DynamicTest running on GRAPH_MODE_O0 with static shape with first inputs... End Start DynamicTest running on GRAPH_MODE_O0 with static shape with second inputs... End Start DynamicTest testing with [GRAPH_MODE_O0] [Resize]... Compare ['Success'] for 0'th output of Resize. End Start DynamicTest testing with [GRAPH_MODE_O0] [DYNAMIC_SHAPE] [None]... [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:44:53.967.728 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:44:53.968.835 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. TotalTime = 0.233575, [21] [bootstrap]: 0.00106748 [type_inference]: 0.118868 [event_method]: 0.00011383 [auto_monad]: 0.00011446 [graph_reusing]: 7.92e-06 [inline]: 2.67001e-06 [add_attr]: 0.00865443, [1] [add_attr_with_inline]: 0.00861361, [1] [Cycle 1]: 9.723e-05, [2] [tag_attr]: 3.708e-05 [meta_addattr_fg_expand]: 7.1e-06 [parallel-infer-symbol]: 4.13999e-06 [pre_auto_parallel]: 8.112e-05 [insert-virtual-dataset]: 2.54001e-06 [parallel-infer-symbol-second]: 7.00005e-07 [dataset_repeat_opt]: 2.16998e-06 [pipeline_split]: 1.64e-06 [optimize]: 0.103523, [53] [py_interpret_to_execute]: 3.781e-05 [rewriter_before_opt_a]: 0.00011887 [opt_a]: 0.00444271, [2] [Cycle 1]: 0.00369185, [45] [expand_dump_flag]: 3.46999e-06 [switch_simplify]: 5.514e-05 [loop_unroll]: 3.997e-05 [a_1]: 0.00108987 [with_stream_mark]: 8.119e-05 [recompute_prepare]: 1.901e-05 [updatestate_depend_eliminate]: 7.27997e-06 [updatestate_assign_eliminate]: 3.39001e-06 [updatestate_loads_eliminate]: 3.23998e-06 [parameter_eliminate]: 3.46001e-06 [a_2]: 8.917e-05 [accelerated_algorithm]: 7.91001e-06 [shard]: 2.96001e-06 [meta_shard_fg_expand]: 2.44001e-06 [shard_inline]: 6.63e-06 [merge_send_recv]: 9.56e-06 [auto_parallel]: 0.00024463 [parallel]: 0.0001015 [flash_sp]: 1.298e-05 [merge_comm]: 6.15002e-06 [allreduce_fusion]: 3.76999e-06 [matmul_add_comm_reduction]: 1.209e-05 [allreduce_slice_to_reducescatter]: 1.20001e-06 [virtual_shard_identity]: 1.977e-05 [virtual_dataset]: 7.9e-06 [get_grad_eliminate_]: 6.78e-06 [virtual_output]: 7.31999e-06 [merge_forward]: 4.54002e-06 [cell_reuse_recompute_pass]: 2.53e-06 [offload_activation]: 1.118e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.799e-05 [merge_recompute_call_nodes]: 1.62001e-06 [before_grad]: 1.329e-05 [set_forward_comm_id_for_comm_node_pass]: 5.44e-06 [meta_fg_expand]: 4.60001e-06 [flash_sp_send_recv_attached]: 3.30003e-06 [receive_attached]: 2.76e-06 [after_resolve]: 1.454e-05 [a_after_grad]: 1.064e-05 [renormalize]: 0.0012678 [add_forward_monad_depend]: 9.12999e-06 [auto_monad_grad]: 3.26001e-06 [auto_monad_eliminator]: 2.165e-05 [cse]: 3.203e-05 [a_3]: 5.489e-05 [Cycle 2]: 0.00073716, [45] [expand_dump_flag]: 2.37999e-06 [switch_simplify]: 8.43999e-06 [loop_unroll]: 6.56e-06 [a_1]: 0.00015261 [with_stream_mark]: 1.962e-05 [recompute_prepare]: 6.88998e-06 [updatestate_depend_eliminate]: 3.88001e-06 [updatestate_assign_eliminate]: 2.78e-06 [updatestate_loads_eliminate]: 3.13e-06 [parameter_eliminate]: 1.65001e-06 [a_2]: 7.361e-05 [accelerated_algorithm]: 6.59001e-06 [shard]: 2.18002e-06 [meta_shard_fg_expand]: 3.28e-06 [shard_inline]: 7.34002e-06 [merge_send_recv]: 7.81001e-06 [auto_parallel]: 9.76e-06 [parallel]: 8.59998e-06 [flash_sp]: 3.31999e-06 [merge_comm]: 4.33001e-06 [allreduce_fusion]: 3.32002e-06 [matmul_add_comm_reduction]: 1.016e-05 [allreduce_slice_to_reducescatter]: 7.2e-07 [virtual_shard_identity]: 9.15001e-06 [virtual_dataset]: 6.35002e-06 [get_grad_eliminate_]: 5.93002e-06 [virtual_output]: 6.02001e-06 [merge_forward]: 4.05e-06 [cell_reuse_recompute_pass]: 3.03e-06 [offload_activation]: 9.67001e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.325e-05 [merge_recompute_call_nodes]: 1.25001e-06 [before_grad]: 9.39e-06 [set_forward_comm_id_for_comm_node_pass]: 4.05998e-06 [meta_fg_expand]: 2.63998e-06 [flash_sp_send_recv_attached]: 1.62999e-06 [receive_attached]: 2.76e-06 [after_resolve]: 1.213e-05 [a_after_grad]: 9.35001e-06 [renormalize]: 9.00181e-08 [add_forward_monad_depend]: 2.44001e-06 [auto_monad_grad]: 2.56e-06 [auto_monad_eliminator]: 1.244e-05 [cse]: 1.719e-05 [a_3]: 3.661e-05 [py_interpret_to_execute_after_opt_a]: 1.702e-05 [slice_cell_reuse_recomputed_activation]: 2.22999e-06 [rewriter_after_opt_a]: 0.095718 [convert_after_rewriter]: 2.804e-05 [order_py_execute_after_rewriter]: 7.25998e-06 [mutable_eliminate]: 0.00091446 [opt_b]: 0.00026942, [1] [Cycle 1]: 0.00025854, [7] [b_1]: 0.0001461 [b_2]: 8.98002e-06 [updatestate_depend_eliminate]: 9.96998e-06 [updatestate_assign_eliminate]: 3.43999e-06 [updatestate_loads_eliminate]: 3.28998e-06 [renormalize]: 8.29983e-07 [cse]: 4.348e-05 [optimize_parallel_all_gather_comm]: 2.599e-05 [overlap_param_gather]: 2.19999e-06 [cconv]: 4.012e-05 [loop_unroll]: 0.00068522 [opt_after_cconv]: 0.00013732, [1] [Cycle 1]: 0.00012818, [7] [c_1]: 3.527e-05 [parameter_eliminate]: 6.16998e-06 [updatestate_depend_eliminate]: 9.53002e-06 [updatestate_assign_eliminate]: 2.78998e-06 [updatestate_loads_eliminate]: 2.35997e-06 [cse]: 3.222e-05 [renormalize]: 1.37999e-06 [remove_dup_value]: 1.708e-05 [tuple_transform]: 9.705e-05, [1] [Cycle 1]: 9.164e-05, [4] [d_1]: 5.819e-05 [none_parameter_eliminate]: 2.12001e-06 [renormalize]: 2.10013e-07 [switch_simplify]: 8.13001e-06 [partial_unused_args_eliminate]: 1.71002e-06 [add_recomputation]: 0.00012737 [cse_after_recomputation]: 2.793e-05, [1] [Cycle 1]: 2.215e-05, [1] [cse]: 1.504e-05 [environ_conv]: 7.81001e-06 [swap_dp_allreduce_reducescatter]: 6.94001e-06 [bias_add_comm_swap]: 4.02002e-06 [label_micro_interleaved_index]: 7.60998e-06 [label_fine_grained_interleaved_index]: 3.01001e-06 [merge_cast_opt]: 1.55001e-06 [slice_recompute_activation]: 2.37999e-06 [micro_interleaved_order_control]: 2.99999e-06 [assign_add_opt]: 1.79e-06 [ForceFp32Comm]: 1.30999e-06 [remove_cast_before_assign_add]: 1.18001e-06 [full_micro_interleaved_order_control]: 2.76e-06 [reorder_send_recv_between_fp_bp]: 2.74999e-06 [comm_op_add_attrs]: 1.13001e-06 [add_comm_op_reuse_tag]: 1.02998e-06 [interleave_split_concat_branches]: 1.35001e-06 [interleave_parallel_branches]: 1.50001e-06 [overlap_opt_shard_in_pipeline]: 1.25999e-06 [overlap_opt_shard_grad_in_pipeline]: 1.99999e-06 [control_data_broadcast_order]: 1.737e-05 [grouped_pairwise_exchange_alltoall]: 1.60001e-06 [offloading_packed_experts]: 4.32e-06 [overlap_recompute_and_grad_model_parallel]: 4.95999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.37e-06 [overlap_recompute_allgather_and_fa_grad]: 1.35999e-06 [overlap_recompute_comm]: 2.40002e-06 [overlap_grad_ring_attention]: 4.13001e-06 [overlap_grad_flash_sp]: 2.352e-05 [begin_end_overlap_inline]: 5.50004e-07 [split_matmul_comm_elemetwise]: 2.41e-06 [split_layernorm_comm]: 1.96998e-06 [handle_group_info]: 1.42e-06 [symbol_engine_optimizer]: 0.00024378, [1] [Cycle 1]: 0.00023699, [6] [build]: 0.00012678 [elim_shapecalc]: 1.997e-05 [elim_not_effective]: 1.865e-05 [opt_reshape]: 7.99002e-06 [fold_const_symbol]: 1.961e-05 [renormalize]: 2.00002e-07 [detach_backward]: 2.78e-06 [pipeline_parallel_scheduler]: 1.67001e-06 [auto_monad_reorder]: 2.112e-05 [get_jit_bprop_graph]: 2.36e-06 [rewriter_after_jit_bprop_graph]: 7.48e-06 [opt_after_jit_grad]: 0.00076504 [validate]: 0.0001148 Sums bootstrap : 0.001067s : 0.48% type_inference : 0.118868s : 53.18% event_method : 0.000114s : 0.05% auto_monad : 0.000114s : 0.05% graph_reusing : 0.000008s : 0.00% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000037s : 0.02% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.00% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000081s : 0.04% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000038s : 0.02% optimize.rewriter_before_opt_a : 0.000119s : 0.05% optimize.opt_a.expand_dump_flag : 0.000006s : 0.00% optimize.opt_a.switch_simplify : 0.000064s : 0.03% optimize.opt_a.loop_unroll : 0.000047s : 0.02% optimize.opt_a.a_1 : 0.001242s : 0.56% optimize.opt_a.with_stream_mark : 0.000101s : 0.05% optimize.opt_a.recompute_prepare : 0.000026s : 0.01% optimize.opt_a.updatestate_depend_eliminate : 0.000011s : 0.00% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.00% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.00% optimize.opt_a.parameter_eliminate : 0.000005s : 0.00% optimize.opt_a.a_2 : 0.000163s : 0.07% optimize.opt_a.accelerated_algorithm : 0.000015s : 0.01% optimize.opt_a.shard : 0.000005s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000006s : 0.00% optimize.opt_a.shard_inline : 0.000014s : 0.01% optimize.opt_a.merge_send_recv : 0.000017s : 0.01% optimize.opt_a.auto_parallel : 0.000254s : 0.11% optimize.opt_a.parallel : 0.000110s : 0.05% optimize.opt_a.flash_sp : 0.000016s : 0.01% optimize.opt_a.merge_comm : 0.000010s : 0.00% optimize.opt_a.allreduce_fusion : 0.000007s : 0.00% optimize.opt_a.matmul_add_comm_reduction : 0.000022s : 0.01% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000029s : 0.01% optimize.opt_a.virtual_dataset : 0.000014s : 0.01% optimize.opt_a.get_grad_eliminate_ : 0.000013s : 0.01% optimize.opt_a.virtual_output : 0.000013s : 0.01% optimize.opt_a.merge_forward : 0.000009s : 0.00% optimize.opt_a.cell_reuse_recompute_pass : 0.000006s : 0.00% optimize.opt_a.offload_activation : 0.000021s : 0.01% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000031s : 0.01% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.00% optimize.opt_a.before_grad : 0.000023s : 0.01% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000009s : 0.00% optimize.opt_a.meta_fg_expand : 0.000007s : 0.00% optimize.opt_a.flash_sp_send_recv_attached : 0.000005s : 0.00% optimize.opt_a.receive_attached : 0.000006s : 0.00% optimize.opt_a.after_resolve : 0.000027s : 0.01% optimize.opt_a.a_after_grad : 0.000020s : 0.01% optimize.opt_a.renormalize : 0.001268s : 0.57% optimize.opt_a.add_forward_monad_depend : 0.000012s : 0.01% optimize.opt_a.auto_monad_grad : 0.000006s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000034s : 0.02% optimize.opt_a.cse : 0.000049s : 0.02% optimize.opt_a.a_3 : 0.000091s : 0.04% optimize.py_interpret_to_execute_after_opt_a : 0.000017s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.095718s : 42.82% optimize.convert_after_rewriter : 0.000028s : 0.01% optimize.order_py_execute_after_rewriter : 0.000007s : 0.00% optimize.mutable_eliminate : 0.000914s : 0.41% optimize.opt_b.b_1 : 0.000146s : 0.07% optimize.opt_b.b_2 : 0.000009s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000010s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000043s : 0.02% optimize.optimize_parallel_all_gather_comm : 0.000026s : 0.01% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000040s : 0.02% optimize.loop_unroll : 0.000685s : 0.31% optimize.opt_after_cconv.c_1 : 0.000035s : 0.02% optimize.opt_after_cconv.parameter_eliminate : 0.000006s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000010s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000032s : 0.01% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000017s : 0.01% optimize.tuple_transform.d_1 : 0.000058s : 0.03% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000008s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000127s : 0.06% optimize.cse_after_recomputation.cse : 0.000015s : 0.01% optimize.environ_conv : 0.000008s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000007s : 0.00% optimize.bias_add_comm_swap : 0.000004s : 0.00% optimize.label_micro_interleaved_index : 0.000008s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.00% optimize.assign_add_opt : 0.000002s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000003s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000002s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000017s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000024s : 0.01% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000127s : 0.06% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000020s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000019s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000008s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000020s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000021s : 0.01% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000007s : 0.00% opt_after_jit_grad : 0.000765s : 0.34% validate : 0.000115s : 0.05% Time group info: ------[substitution.] 0.000430 40 16.00% : 0.000069s : 8: substitution.arithmetic_simplify 0.98% : 0.000004s : 2: substitution.elim_not_effective 2.24% : 0.000010s : 2: substitution.fold_const_symbol 1.76% : 0.000008s : 4: substitution.graph_param_transform 70.03% : 0.000301s : 6: substitution.inline 1.01% : 0.000004s : 4: substitution.j_node_and_user_rematch 1.21% : 0.000005s : 4: substitution.remove_not_recompute_node 1.46% : 0.000006s : 4: substitution.replace_old_param 5.30% : 0.000023s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.118596 2 98.82% : 0.117198s : 1: type_inference.infer 1.18% : 0.001398s : 1: type_inference.specialize ------[replace.] 0.000109 12 57.72% : 0.000063s : 6: replace.inline 42.28% : 0.000046s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000315 12 93.70% : 0.000295s : 6: match.inline 6.30% : 0.000020s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000295 1534 0.83% : 0.000002s : 17: predicate.accumulaten_eliminater 0.81% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.37% : 0.000001s : 8: predicate.addn_check_dump 0.86% : 0.000003s : 17: predicate.addn_zero_filter 0.70% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 1.99% : 0.000006s : 25: predicate.arithmetic_simplify 0.92% : 0.000003s : 17: predicate.cast_eliminate 0.45% : 0.000001s : 8: predicate.check_bprop_eliminate 0.31% : 0.000001s : 8: predicate.compare_switch_simplify 0.13% : 0.000000s : 4: predicate.const_output_eliminate 0.42% : 0.000001s : 8: predicate.depend_value_elim 0.89% : 0.000003s : 17: predicate.dict_get_item_const_eliminator 1.10% : 0.000003s : 17: predicate.dict_get_item_eliminator 0.87% : 0.000003s : 17: predicate.dict_set_item_eliminator 0.76% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.30% : 0.000001s : 4: predicate.elim_not_effective 0.55% : 0.000002s : 4: predicate.elim_shapecalc_of_broadcastargs 1.14% : 0.000003s : 21: predicate.environ_add_const_eliminate 0.94% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.04% : 0.000003s : 21: predicate.environ_get_depend_swap 1.38% : 0.000004s : 29: predicate.environ_get_eliminate 0.91% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.40% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.32% : 0.000007s : 29: predicate.float_depend_g_call 0.35% : 0.000001s : 8: predicate.float_environ_get_switch 0.60% : 0.000002s : 12: predicate.float_tuple_getitem_switch 0.12% : 0.000000s : 4: predicate.fold_const_symbol 0.58% : 0.000002s : 8: predicate.get_grad_eliminate 0.30% : 0.000001s : 4: predicate.graph_param_transform 0.45% : 0.000001s : 8: predicate.incorporate_call 0.32% : 0.000001s : 8: predicate.incorporate_call_switch 15.79% : 0.000047s : 70: predicate.inline 0.86% : 0.000003s : 8: predicate.inline_without_move 0.22% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.63% : 0.000002s : 8: predicate.less_batch_normalization 1.72% : 0.000005s : 31: predicate.list_to_tuple_eliminator_ 2.22% : 0.000007s : 48: predicate.load_eliminater 1.07% : 0.000003s : 4: predicate.loop_unroll_after_grad 2.42% : 0.000007s : 47: predicate.loop_unroll_before_grad 1.52% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.40% : 0.000001s : 8: predicate.merge_addn 0.36% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.47% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.74% : 0.000002s : 17: predicate.minmaximum_grad 1.20% : 0.000004s : 4: predicate.mutable_eliminate 0.21% : 0.000001s : 4: predicate.opt_reshape 0.31% : 0.000001s : 4: predicate.parallel_virtual_node 1.99% : 0.000006s : 29: predicate.partial_defer_inline 1.32% : 0.000004s : 27: predicate.partial_eliminate 0.78% : 0.000002s : 17: predicate.print_const_string_wrapper 0.40% : 0.000001s : 8: predicate.reduce_all_const_elim 1.05% : 0.000003s : 17: predicate.reduce_eliminate 2.17% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.40% : 0.000001s : 8: predicate.remove_not_recompute_node 1.29% : 0.000004s : 31: predicate.replace_applicator 0.31% : 0.000001s : 8: predicate.replace_old_param 0.31% : 0.000001s : 4: predicate.reset_defer_inline 0.90% : 0.000003s : 17: predicate.reshape_eliminate 0.47% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.40% : 0.000001s : 4: predicate.row_tensor_eliminate 0.74% : 0.000002s : 8: predicate.same_eliminate 0.40% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.79% : 0.000002s : 8: predicate.shard_identity_eliminate 0.70% : 0.000002s : 8: predicate.special_op_eliminate 0.64% : 0.000002s : 8: predicate.specialize_transform 0.85% : 0.000003s : 8: predicate.split_environ_get_set_with_tuple_value 0.53% : 0.000002s : 8: predicate.stack_unstack_eliminate 0.27% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.45% : 0.000004s : 29: predicate.switch_defer_inline 1.79% : 0.000005s : 37: predicate.switch_layer_defer_inline 4.69% : 0.000014s : 88: predicate.switch_simplify 0.74% : 0.000002s : 17: predicate.tile_eliminate 0.78% : 0.000002s : 17: predicate.transpose_eliminate 1.50% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.39% : 0.000004s : 25: predicate.tuple_list_get_item_const_eliminator 1.31% : 0.000004s : 25: predicate.tuple_list_get_item_depend_reorder 3.17% : 0.000009s : 39: predicate.tuple_list_get_item_eliminator 1.25% : 0.000004s : 25: predicate.tuple_list_get_set_item_eliminator 1.82% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 2.24% : 0.000007s : 31: predicate.tuple_to_list_eliminator_ 2.09% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 2.58% : 0.000008s : 56: predicate.updatestate_useless_node_eliminater 0.33% : 0.000001s : 4: predicate.value_based_eliminate 0.63% : 0.000002s : 8: predicate.virtual_dataset_eliminate 0.53% : 0.000002s : 8: predicate.virtual_output_eliminate 0.20% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.54% : 0.000002s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.001940 15 72.40% : 0.001404s : 7: func_graph_cloner_run.FuncGraphClonerGraph 27.60% : 0.000535s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.348773 192 0.00% : 0.000004s : 1: ForceFp32Comm 2.48% : 0.008662s : 1: add_attr 2.47% : 0.008618s : 1: add_attr_with_inline 0.01% : 0.000046s : 1: add_comm_op_reuse_tag 0.04% : 0.000135s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.03% : 0.000121s : 1: auto_monad 0.01% : 0.000025s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000007s : 1: bias_add_comm_swap 0.32% : 0.001101s : 1: bootstrap 0.01% : 0.000045s : 1: cconv 0.00% : 0.000005s : 1: comm_op_add_attrs 0.01% : 0.000022s : 1: control_data_broadcast_order 0.01% : 0.000037s : 1: convert_after_rewriter 0.01% : 0.000031s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000007s : 1: detach_backward 0.00% : 0.000011s : 1: environ_conv 0.04% : 0.000124s : 1: event_method 0.00% : 0.000006s : 1: full_micro_interleaved_order_control 0.00% : 0.000008s : 1: get_jit_bprop_graph 0.00% : 0.000012s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000005s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000008s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000011s : 1: label_micro_interleaved_index 0.20% : 0.000700s : 1: loop_unroll 0.00% : 0.000006s : 1: merge_cast_opt 0.00% : 0.000006s : 1: micro_interleaved_order_control 0.27% : 0.000940s : 1: mutable_eliminate 0.00% : 0.000008s : 1: offloading_packed_experts 0.01% : 0.000023s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000029s : 1: opt.transform.mutable_eliminate 0.50% : 0.001728s : 78: opt.transform.opt_a 0.01% : 0.000033s : 1: opt.transform.opt_after_cconv 0.01% : 0.000038s : 1: opt.transform.opt_after_jit_grad 0.03% : 0.000118s : 28: opt.transform.opt_b 0.02% : 0.000064s : 2: opt.transform.opt_trans_graph 0.02% : 0.000061s : 4: opt.transform.symbol_engine_opt 1.27% : 0.004446s : 1: opt_a 0.04% : 0.000144s : 1: opt_after_cconv 0.22% : 0.000784s : 1: opt_after_jit_grad 0.08% : 0.000274s : 1: opt_b 29.68% : 0.103530s : 1: optimize 0.01% : 0.000030s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000011s : 1: order_py_execute_after_rewriter 0.01% : 0.000027s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000006s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000005s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000009s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.03% : 0.000087s : 1: pre_auto_parallel 0.01% : 0.000044s : 1: py_interpret_to_execute 0.01% : 0.000021s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.01% : 0.000021s : 1: remove_dup_value 0.17% : 0.000602s : 1: renormalize.infer 0.19% : 0.000651s : 1: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000011s : 1: rewriter_after_jit_bprop_graph 27.46% : 0.095759s : 1: rewriter_after_opt_a 0.04% : 0.000125s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000006s : 1: split_matmul_comm_elemetwise 0.00% : 0.000010s : 1: swap_dp_allreduce_reducescatter 0.07% : 0.000247s : 1: symbol_engine_optimizer 0.03% : 0.000100s : 1: tuple_transform 34.09% : 0.118904s : 1: type_inference TotalTime = 0.246977, [21] [bootstrap]: 0.00055236 [type_inference]: 0.0949067 [event_method]: 0.00012968 [auto_monad]: 0.00014925 [graph_reusing]: 1.136e-05 [inline]: 2.64001e-06 [add_attr]: 0.00450254, [1] [add_attr_with_inline]: 0.00448884, [1] [Cycle 1]: 0.00013414, [2] [tag_attr]: 7.878e-05 [meta_addattr_fg_expand]: 1.348e-05 [parallel-infer-symbol]: 4.32e-06 [pre_auto_parallel]: 7.82e-05 [insert-virtual-dataset]: 2.61999e-06 [parallel-infer-symbol-second]: 1.05999e-06 [dataset_repeat_opt]: 2.06998e-06 [pipeline_split]: 1.67001e-06 [optimize]: 0.145386, [53] [py_interpret_to_execute]: 5.383e-05 [rewriter_before_opt_a]: 0.00022032 [opt_a]: 0.12903, [3] [Cycle 1]: 0.120138, [45] [expand_dump_flag]: 5.07999e-06 [switch_simplify]: 0.00010156 [loop_unroll]: 0.00011816 [a_1]: 0.0128212 [with_stream_mark]: 4.556e-05 [recompute_prepare]: 3.536e-05 [updatestate_depend_eliminate]: 1.301e-05 [updatestate_assign_eliminate]: 9.64e-06 [updatestate_loads_eliminate]: 8.65999e-06 [parameter_eliminate]: 3.62002e-06 [a_2]: 0.00028904 [accelerated_algorithm]: 4.499e-05 [shard]: 2.93e-06 [meta_shard_fg_expand]: 3.166e-05 [shard_inline]: 2.542e-05 [merge_send_recv]: 2.308e-05 [auto_parallel]: 2.174e-05 [parallel]: 3.037e-05 [flash_sp]: 1.346e-05 [merge_comm]: 1.36e-05 [allreduce_fusion]: 1.125e-05 [matmul_add_comm_reduction]: 3.852e-05 [allreduce_slice_to_reducescatter]: 9.5999e-07 [virtual_shard_identity]: 2.575e-05 [virtual_dataset]: 1.881e-05 [get_grad_eliminate_]: 1.766e-05 [virtual_output]: 1.967e-05 [merge_forward]: 1.257e-05 [cell_reuse_recompute_pass]: 2.43e-06 [offload_activation]: 2.322e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.993e-05 [merge_recompute_call_nodes]: 1.64e-06 [before_grad]: 3.463e-05 [set_forward_comm_id_for_comm_node_pass]: 1.345e-05 [meta_fg_expand]: 0.00252049 [flash_sp_send_recv_attached]: 7.42002e-06 [receive_attached]: 2.78e-06 [after_resolve]: 0.0970011 [a_after_grad]: 0.00015477 [renormalize]: 0.00514749 [add_forward_monad_depend]: 1.676e-05 [auto_monad_grad]: 8.53001e-06 [auto_monad_eliminator]: 9.352e-05 [cse]: 0.00031324 [a_3]: 0.00045548 [Cycle 2]: 0.00673751, [45] [expand_dump_flag]: 3.14001e-06 [switch_simplify]: 6.434e-05 [loop_unroll]: 5.617e-05 [a_1]: 0.00222994 [with_stream_mark]: 4.057e-05 [recompute_prepare]: 2.863e-05 [updatestate_depend_eliminate]: 1.382e-05 [updatestate_assign_eliminate]: 1.134e-05 [updatestate_loads_eliminate]: 9.92999e-06 [parameter_eliminate]: 2.91e-06 [a_2]: 0.0002965 [accelerated_algorithm]: 2.906e-05 [shard]: 2.47001e-06 [meta_shard_fg_expand]: 7.24001e-06 [shard_inline]: 2.096e-05 [merge_send_recv]: 2.032e-05 [auto_parallel]: 2.125e-05 [parallel]: 1.14e-05 [flash_sp]: 4.72998e-06 [merge_comm]: 1.322e-05 [allreduce_fusion]: 1.253e-05 [matmul_add_comm_reduction]: 2.31e-05 [allreduce_slice_to_reducescatter]: 1.19e-06 [virtual_shard_identity]: 2.471e-05 [virtual_dataset]: 1.898e-05 [get_grad_eliminate_]: 1.832e-05 [virtual_output]: 1.825e-05 [merge_forward]: 1.392e-05 [cell_reuse_recompute_pass]: 1.98997e-06 [offload_activation]: 2.501e-05 [cell_reuse_handle_not_recompute_node_pass]: 4.37e-05 [merge_recompute_call_nodes]: 1.76003e-06 [before_grad]: 5.288e-05 [set_forward_comm_id_for_comm_node_pass]: 1.634e-05 [meta_fg_expand]: 0.00022585 [flash_sp_send_recv_attached]: 2.92002e-06 [receive_attached]: 2.84001e-06 [after_resolve]: 3.725e-05 [a_after_grad]: 3.203e-05 [renormalize]: 0.00258148 [add_forward_monad_depend]: 1.16e-05 [auto_monad_grad]: 2.83998e-06 [auto_monad_eliminator]: 4.586e-05 [cse]: 0.00012401 [a_3]: 0.00015077 [Cycle 3]: 0.00213194, [45] [expand_dump_flag]: 3.51999e-06 [switch_simplify]: 2.415e-05 [loop_unroll]: 1.967e-05 [a_1]: 0.00070002 [with_stream_mark]: 3.071e-05 [recompute_prepare]: 2.4e-05 [updatestate_depend_eliminate]: 1.308e-05 [updatestate_assign_eliminate]: 1.139e-05 [updatestate_loads_eliminate]: 9.89001e-06 [parameter_eliminate]: 2.83e-06 [a_2]: 0.00032347 [accelerated_algorithm]: 2.964e-05 [shard]: 2.28998e-06 [meta_shard_fg_expand]: 6.44999e-06 [shard_inline]: 2.163e-05 [merge_send_recv]: 2.15e-05 [auto_parallel]: 2.121e-05 [parallel]: 9.32001e-06 [flash_sp]: 2.12999e-06 [merge_comm]: 1.24e-05 [allreduce_fusion]: 1.195e-05 [matmul_add_comm_reduction]: 2.146e-05 [allreduce_slice_to_reducescatter]: 1.20001e-06 [virtual_shard_identity]: 2.078e-05 [virtual_dataset]: 1.851e-05 [get_grad_eliminate_]: 1.867e-05 [virtual_output]: 1.769e-05 [merge_forward]: 1.242e-05 [cell_reuse_recompute_pass]: 3.35998e-06 [offload_activation]: 2.315e-05 [cell_reuse_handle_not_recompute_node_pass]: 4.336e-05 [merge_recompute_call_nodes]: 1.44e-06 [before_grad]: 3.581e-05 [set_forward_comm_id_for_comm_node_pass]: 1.569e-05 [meta_fg_expand]: 9.12001e-06 [flash_sp_send_recv_attached]: 1.71002e-06 [receive_attached]: 2.43998e-06 [after_resolve]: 2.661e-05 [a_after_grad]: 3.059e-05 [renormalize]: 9.00181e-08 [add_forward_monad_depend]: 4.25999e-06 [auto_monad_grad]: 2.69001e-06 [auto_monad_eliminator]: 3.509e-05 [cse]: 7.861e-05 [a_3]: 0.00012991 [py_interpret_to_execute_after_opt_a]: 3.265e-05 [slice_cell_reuse_recomputed_activation]: 2.11e-06 [rewriter_after_opt_a]: 0.00020537 [convert_after_rewriter]: 2.157e-05 [order_py_execute_after_rewriter]: 1.346e-05 [mutable_eliminate]: 0.00080581 [opt_b]: 0.00074678, [1] [Cycle 1]: 0.00073578, [7] [b_1]: 0.00051597 [b_2]: 2.858e-05 [updatestate_depend_eliminate]: 2.102e-05 [updatestate_assign_eliminate]: 1.143e-05 [updatestate_loads_eliminate]: 1.028e-05 [renormalize]: 7.79983e-07 [cse]: 9.499e-05 [optimize_parallel_all_gather_comm]: 4.386e-05 [overlap_param_gather]: 2.28002e-06 [cconv]: 4.22e-05 [loop_unroll]: 0.00072759 [opt_after_cconv]: 0.00027798, [1] [Cycle 1]: 0.00027005, [7] [c_1]: 0.00010424 [parameter_eliminate]: 6.01e-06 [updatestate_depend_eliminate]: 1.923e-05 [updatestate_assign_eliminate]: 1.073e-05 [updatestate_loads_eliminate]: 1.023e-05 [cse]: 8.094e-05 [renormalize]: 9.30013e-07 [remove_dup_value]: 0.00010884 [tuple_transform]: 0.0116143, [1] [Cycle 1]: 0.0116058, [4] [d_1]: 0.011486 [none_parameter_eliminate]: 6.63998e-06 [renormalize]: 8.79983e-07 [switch_simplify]: 3.825e-05 [partial_unused_args_eliminate]: 3.80998e-06 [add_recomputation]: 0.00016207 [cse_after_recomputation]: 0.00010965, [1] [Cycle 1]: 9.876e-05, [1] [cse]: 9.145e-05 [environ_conv]: 2.196e-05 [swap_dp_allreduce_reducescatter]: 1.821e-05 [bias_add_comm_swap]: 3.76999e-06 [label_micro_interleaved_index]: 7.51001e-06 [label_fine_grained_interleaved_index]: 2.88e-06 [merge_cast_opt]: 1.76003e-06 [slice_recompute_activation]: 2.36e-06 [micro_interleaved_order_control]: 2.88e-06 [assign_add_opt]: 1.64e-06 [ForceFp32Comm]: 1.16997e-06 [remove_cast_before_assign_add]: 1.09003e-06 [full_micro_interleaved_order_control]: 2.06e-06 [reorder_send_recv_between_fp_bp]: 2.81e-06 [comm_op_add_attrs]: 1.19998e-06 [add_comm_op_reuse_tag]: 1.27e-06 [interleave_split_concat_branches]: 1.20999e-06 [interleave_parallel_branches]: 1.02e-06 [overlap_opt_shard_in_pipeline]: 1.20001e-06 [overlap_opt_shard_grad_in_pipeline]: 1.69e-06 [control_data_broadcast_order]: 4.044e-05 [grouped_pairwise_exchange_alltoall]: 1.75001e-06 [offloading_packed_experts]: 1.027e-05 [overlap_recompute_and_grad_model_parallel]: 1.012e-05 [overlap_grad_matmul_and_grad_allreduce]: 1.20999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.30001e-06 [overlap_recompute_comm]: 2.68998e-06 [overlap_grad_ring_attention]: 9.51998e-06 [overlap_grad_flash_sp]: 5.29e-05 [begin_end_overlap_inline]: 5.09986e-07 [split_matmul_comm_elemetwise]: 2.29999e-06 [split_layernorm_comm]: 1.68002e-06 [handle_group_info]: 1.19e-06 [symbol_engine_optimizer]: 0.00066467, [1] [Cycle 1]: 0.00065743, [6] [build]: 0.00033067 [elim_shapecalc]: 7.101e-05 [elim_not_effective]: 4.646e-05 [opt_reshape]: 9.156e-05 [fold_const_symbol]: 5.827e-05 [renormalize]: 5.19998e-07 [detach_backward]: 2.97002e-06 [pipeline_parallel_scheduler]: 1.94e-06 [auto_monad_reorder]: 5.407e-05 [get_jit_bprop_graph]: 3.04001e-06 [rewriter_after_jit_bprop_graph]: 8.37e-06 [opt_after_jit_grad]: 0.00088677 [validate]: 9.074e-05 Sums bootstrap : 0.000552s : 0.23% type_inference : 0.094907s : 39.44% event_method : 0.000130s : 0.05% auto_monad : 0.000149s : 0.06% graph_reusing : 0.000011s : 0.00% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000079s : 0.03% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000013s : 0.01% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000078s : 0.03% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000054s : 0.02% optimize.rewriter_before_opt_a : 0.000220s : 0.09% optimize.opt_a.expand_dump_flag : 0.000012s : 0.00% optimize.opt_a.switch_simplify : 0.000190s : 0.08% optimize.opt_a.loop_unroll : 0.000194s : 0.08% optimize.opt_a.a_1 : 0.015751s : 6.55% optimize.opt_a.with_stream_mark : 0.000117s : 0.05% optimize.opt_a.recompute_prepare : 0.000088s : 0.04% optimize.opt_a.updatestate_depend_eliminate : 0.000040s : 0.02% optimize.opt_a.updatestate_assign_eliminate : 0.000032s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000028s : 0.01% optimize.opt_a.parameter_eliminate : 0.000009s : 0.00% optimize.opt_a.a_2 : 0.000909s : 0.38% optimize.opt_a.accelerated_algorithm : 0.000104s : 0.04% optimize.opt_a.shard : 0.000008s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000045s : 0.02% optimize.opt_a.shard_inline : 0.000068s : 0.03% optimize.opt_a.merge_send_recv : 0.000065s : 0.03% optimize.opt_a.auto_parallel : 0.000064s : 0.03% optimize.opt_a.parallel : 0.000051s : 0.02% optimize.opt_a.flash_sp : 0.000020s : 0.01% optimize.opt_a.merge_comm : 0.000039s : 0.02% optimize.opt_a.allreduce_fusion : 0.000036s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000083s : 0.03% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000003s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000071s : 0.03% optimize.opt_a.virtual_dataset : 0.000056s : 0.02% optimize.opt_a.get_grad_eliminate_ : 0.000055s : 0.02% optimize.opt_a.virtual_output : 0.000056s : 0.02% optimize.opt_a.merge_forward : 0.000039s : 0.02% optimize.opt_a.cell_reuse_recompute_pass : 0.000008s : 0.00% optimize.opt_a.offload_activation : 0.000071s : 0.03% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000127s : 0.05% optimize.opt_a.merge_recompute_call_nodes : 0.000005s : 0.00% optimize.opt_a.before_grad : 0.000123s : 0.05% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000045s : 0.02% optimize.opt_a.meta_fg_expand : 0.002755s : 1.14% optimize.opt_a.flash_sp_send_recv_attached : 0.000012s : 0.01% optimize.opt_a.receive_attached : 0.000008s : 0.00% optimize.opt_a.after_resolve : 0.097065s : 40.33% optimize.opt_a.a_after_grad : 0.000217s : 0.09% optimize.opt_a.renormalize : 0.007729s : 3.21% optimize.opt_a.add_forward_monad_depend : 0.000033s : 0.01% optimize.opt_a.auto_monad_grad : 0.000014s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000174s : 0.07% optimize.opt_a.cse : 0.000516s : 0.21% optimize.opt_a.a_3 : 0.000736s : 0.31% optimize.py_interpret_to_execute_after_opt_a : 0.000033s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000205s : 0.09% optimize.convert_after_rewriter : 0.000022s : 0.01% optimize.order_py_execute_after_rewriter : 0.000013s : 0.01% optimize.mutable_eliminate : 0.000806s : 0.33% optimize.opt_b.b_1 : 0.000516s : 0.21% optimize.opt_b.b_2 : 0.000029s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000021s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000011s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000010s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000095s : 0.04% optimize.optimize_parallel_all_gather_comm : 0.000044s : 0.02% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000042s : 0.02% optimize.loop_unroll : 0.000728s : 0.30% optimize.opt_after_cconv.c_1 : 0.000104s : 0.04% optimize.opt_after_cconv.parameter_eliminate : 0.000006s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000019s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000011s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000010s : 0.00% optimize.opt_after_cconv.cse : 0.000081s : 0.03% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000109s : 0.05% optimize.tuple_transform.d_1 : 0.011486s : 4.77% optimize.tuple_transform.none_parameter_eliminate : 0.000007s : 0.00% optimize.tuple_transform.renormalize : 0.000001s : 0.00% optimize.tuple_transform.switch_simplify : 0.000038s : 0.02% optimize.partial_unused_args_eliminate : 0.000004s : 0.00% optimize.add_recomputation : 0.000162s : 0.07% optimize.cse_after_recomputation.cse : 0.000091s : 0.04% optimize.environ_conv : 0.000022s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000018s : 0.01% optimize.bias_add_comm_swap : 0.000004s : 0.00% optimize.label_micro_interleaved_index : 0.000008s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.00% optimize.assign_add_opt : 0.000002s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000040s : 0.02% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000010s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000010s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000003s : 0.00% optimize.overlap_grad_ring_attention : 0.000010s : 0.00% optimize.overlap_grad_flash_sp : 0.000053s : 0.02% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000331s : 0.14% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000071s : 0.03% optimize.symbol_engine_optimizer.elim_not_effective : 0.000046s : 0.02% optimize.symbol_engine_optimizer.opt_reshape : 0.000092s : 0.04% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000058s : 0.02% optimize.symbol_engine_optimizer.renormalize : 0.000001s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000054s : 0.02% get_jit_bprop_graph : 0.000003s : 0.00% rewriter_after_jit_bprop_graph : 0.000008s : 0.00% opt_after_jit_grad : 0.000887s : 0.37% validate : 0.000091s : 0.04% Time group info: ------[substitution.] 0.001814 367 5.15% : 0.000093s : 16: substitution.arithmetic_simplify 0.66% : 0.000012s : 15: substitution.elim_not_effective 1.56% : 0.000028s : 2: substitution.elim_shapecalc_of_broadcastargs 0.43% : 0.000008s : 6: substitution.float_depend_g_call 1.02% : 0.000018s : 10: substitution.float_tuple_getitem_switch 1.39% : 0.000025s : 15: substitution.fold_const_symbol 1.51% : 0.000027s : 17: substitution.graph_param_transform 0.17% : 0.000003s : 2: substitution.incorporate_call 0.12% : 0.000002s : 2: substitution.incorporate_call_switch 53.00% : 0.000961s : 20: substitution.inline 2.66% : 0.000048s : 2: substitution.inline_without_move 1.18% : 0.000021s : 42: substitution.j_node_and_user_rematch 1.40% : 0.000025s : 3: substitution.less_batch_normalization 1.31% : 0.000024s : 14: substitution.minmaximum_grad 0.23% : 0.000004s : 2: substitution.opt_reshape 0.57% : 0.000010s : 6: substitution.partial_eliminate 0.59% : 0.000011s : 4: substitution.reduce_eliminate 1.62% : 0.000029s : 42: substitution.remove_not_recompute_node 2.38% : 0.000043s : 10: substitution.replace_applicator 1.41% : 0.000026s : 12: substitution.replace_old_param 1.11% : 0.000020s : 6: substitution.reshape_eliminate 0.25% : 0.000005s : 1: substitution.set_cell_output_no_recompute 3.80% : 0.000069s : 18: substitution.tuple_list_convert_item_index_to_positive 1.62% : 0.000029s : 18: substitution.tuple_list_get_item_const_eliminator 2.35% : 0.000043s : 18: substitution.tuple_list_get_item_depend_reorder 10.36% : 0.000188s : 46: substitution.tuple_list_get_item_eliminator 2.15% : 0.000039s : 18: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.094787 2 96.89% : 0.091838s : 1: type_inference.infer 3.11% : 0.002949s : 1: type_inference.specialize ------[replace.] 0.000373 40 58.03% : 0.000216s : 20: replace.inline 41.97% : 0.000157s : 20: replace.tuple_list_get_item_eliminator ------[match.] 0.000996 40 94.94% : 0.000946s : 20: match.inline 5.06% : 0.000050s : 20: match.tuple_list_get_item_eliminator ------[predicate.] 0.001342 8180 0.92% : 0.000012s : 89: predicate.accumulaten_eliminater 0.49% : 0.000007s : 17: predicate.ad_related_special_op_eliminate 0.53% : 0.000007s : 51: predicate.addn_check_dump 0.97% : 0.000013s : 89: predicate.addn_zero_filter 0.91% : 0.000012s : 89: predicate.adjust_all_reduce_mul_add 2.13% : 0.000029s : 140: predicate.arithmetic_simplify 0.94% : 0.000013s : 89: predicate.cast_eliminate 1.01% : 0.000014s : 96: predicate.check_bprop_eliminate 0.52% : 0.000007s : 51: predicate.compare_switch_simplify 0.10% : 0.000001s : 17: predicate.const_output_eliminate 0.54% : 0.000007s : 51: predicate.depend_value_elim 1.01% : 0.000014s : 89: predicate.dict_get_item_const_eliminator 1.26% : 0.000017s : 89: predicate.dict_get_item_eliminator 0.93% : 0.000012s : 89: predicate.dict_set_item_eliminator 0.51% : 0.000007s : 34: predicate.dumpgradient_eliminate 0.12% : 0.000002s : 17: predicate.elim_not_effective 0.25% : 0.000003s : 17: predicate.elim_shapecalc_of_broadcastargs 1.07% : 0.000014s : 106: predicate.environ_add_const_eliminate 1.13% : 0.000015s : 106: predicate.environ_get_add_eliminate 1.04% : 0.000014s : 106: predicate.environ_get_depend_swap 1.76% : 0.000024s : 157: predicate.environ_get_eliminate 1.13% : 0.000015s : 106: predicate.environ_get_set_eliminate 1.37% : 0.000018s : 129: predicate.exchange_switch_depend_value 2.18% : 0.000029s : 129: predicate.float_depend_g_call 0.55% : 0.000007s : 51: predicate.float_environ_get_switch 0.79% : 0.000011s : 68: predicate.float_tuple_getitem_switch 0.11% : 0.000001s : 17: predicate.fold_const_symbol 0.53% : 0.000007s : 51: predicate.get_grad_eliminate 0.14% : 0.000002s : 17: predicate.graph_param_transform 0.55% : 0.000007s : 51: predicate.incorporate_call 0.50% : 0.000007s : 51: predicate.incorporate_call_switch 5.76% : 0.000077s : 354: predicate.inline 1.86% : 0.000025s : 74: predicate.inline_without_move 1.34% : 0.000018s : 51: predicate.j_node_and_user_rematch 0.76% : 0.000010s : 51: predicate.less_batch_normalization 1.62% : 0.000022s : 143: predicate.list_to_tuple_eliminator_ 2.30% : 0.000031s : 232: predicate.load_eliminater 0.48% : 0.000006s : 17: predicate.loop_unroll_after_grad 3.87% : 0.000052s : 179: predicate.loop_unroll_before_grad 1.48% : 0.000020s : 123: predicate.make_slice_get_slice_eliminator 0.54% : 0.000007s : 51: predicate.merge_addn 0.96% : 0.000013s : 96: predicate.micro_step_allgather_replace 0.98% : 0.000013s : 96: predicate.mini_step_allgather_replace 0.93% : 0.000012s : 89: predicate.minmaximum_grad 0.48% : 0.000006s : 17: predicate.mutable_eliminate 0.22% : 0.000003s : 17: predicate.opt_reshape 0.21% : 0.000003s : 17: predicate.parallel_virtual_node 2.30% : 0.000031s : 129: predicate.partial_defer_inline 1.47% : 0.000020s : 126: predicate.partial_eliminate 0.94% : 0.000013s : 89: predicate.print_const_string_wrapper 0.56% : 0.000007s : 51: predicate.reduce_all_const_elim 1.34% : 0.000018s : 89: predicate.reduce_eliminate 2.39% : 0.000032s : 232: predicate.redundant_stop_gradient_eliminater 0.32% : 0.000004s : 51: predicate.remove_not_recompute_node 1.54% : 0.000021s : 205: predicate.replace_applicator 0.59% : 0.000008s : 74: predicate.replace_old_param 0.12% : 0.000002s : 17: predicate.reset_defer_inline 1.11% : 0.000015s : 89: predicate.reshape_eliminate 1.00% : 0.000013s : 96: predicate.row_tensor_add_zeros_like 0.24% : 0.000003s : 17: predicate.row_tensor_eliminate 1.19% : 0.000016s : 96: predicate.same_eliminate 0.32% : 0.000004s : 51: predicate.set_cell_output_no_recompute 0.67% : 0.000009s : 51: predicate.shard_identity_eliminate 0.42% : 0.000006s : 34: predicate.special_op_eliminate 0.62% : 0.000008s : 51: predicate.specialize_transform 1.22% : 0.000016s : 96: predicate.split_environ_get_set_with_tuple_value 1.15% : 0.000015s : 74: predicate.stack_unstack_eliminate 0.20% : 0.000003s : 17: predicate.switch_call_monad_eliminater 1.61% : 0.000022s : 129: predicate.switch_defer_inline 2.39% : 0.000032s : 225: predicate.switch_layer_defer_inline 4.43% : 0.000059s : 376: predicate.switch_simplify 0.99% : 0.000013s : 89: predicate.tile_eliminate 0.95% : 0.000013s : 89: predicate.transpose_eliminate 1.59% : 0.000021s : 123: predicate.tuple_list_convert_item_index_to_positive 1.57% : 0.000021s : 123: predicate.tuple_list_get_item_const_eliminator 1.39% : 0.000019s : 123: predicate.tuple_list_get_item_depend_reorder 3.23% : 0.000043s : 194: predicate.tuple_list_get_item_eliminator 1.65% : 0.000022s : 123: predicate.tuple_list_get_set_item_eliminator 2.11% : 0.000028s : 174: predicate.tuple_list_set_item_eliminator 1.52% : 0.000020s : 143: predicate.tuple_to_list_eliminator_ 2.28% : 0.000031s : 232: predicate.updatestate_pure_node_eliminater 2.91% : 0.000039s : 283: predicate.updatestate_useless_node_eliminater 0.20% : 0.000003s : 17: predicate.value_based_eliminate 0.59% : 0.000008s : 51: predicate.virtual_dataset_eliminate 0.61% : 0.000008s : 51: predicate.virtual_output_eliminate 0.20% : 0.000003s : 17: predicate.virtual_view_grad_eliminate 0.29% : 0.000004s : 17: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.003491 43 53.95% : 0.001884s : 19: func_graph_cloner_run.FuncGraphClonerGraph 46.05% : 0.001608s : 24: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.532503 233 0.00% : 0.000004s : 1: ForceFp32Comm 0.85% : 0.004510s : 1: add_attr 0.84% : 0.004493s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.03% : 0.000169s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.03% : 0.000159s : 1: auto_monad 0.01% : 0.000059s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000007s : 1: bias_add_comm_swap 0.11% : 0.000584s : 1: bootstrap 0.01% : 0.000046s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.01% : 0.000044s : 1: control_data_broadcast_order 0.00% : 0.000026s : 1: convert_after_rewriter 0.02% : 0.000113s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000007s : 1: detach_backward 0.00% : 0.000026s : 1: environ_conv 0.03% : 0.000142s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000007s : 1: get_jit_bprop_graph 0.00% : 0.000016s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000005s : 1: handle_group_info 0.00% : 0.000007s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000011s : 1: label_micro_interleaved_index 0.14% : 0.000739s : 1: loop_unroll 0.00% : 0.000005s : 1: merge_cast_opt 0.00% : 0.000006s : 1: micro_interleaved_order_control 0.15% : 0.000818s : 1: mutable_eliminate 0.00% : 0.000013s : 1: offloading_packed_experts 0.01% : 0.000040s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000039s : 1: opt.transform.mutable_eliminate 21.71% : 0.115607s : 117: opt.transform.opt_a 0.02% : 0.000102s : 1: opt.transform.opt_after_cconv 0.01% : 0.000079s : 1: opt.transform.opt_after_jit_grad 0.10% : 0.000508s : 28: opt.transform.opt_b 2.16% : 0.011514s : 2: opt.transform.opt_trans_graph 0.05% : 0.000261s : 4: opt.transform.symbol_engine_opt 24.23% : 0.129035s : 1: opt_a 0.05% : 0.000282s : 1: opt_after_cconv 0.17% : 0.000900s : 1: opt_after_jit_grad 0.14% : 0.000752s : 1: opt_b 27.30% : 0.145393s : 1: optimize 0.01% : 0.000048s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000017s : 1: order_py_execute_after_rewriter 0.01% : 0.000056s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000013s : 1: overlap_grad_ring_attention 0.00% : 0.000004s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000006s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000013s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000006s : 1: overlap_recompute_comm 0.00% : 0.000009s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000009s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.02% : 0.000083s : 1: pre_auto_parallel 0.01% : 0.000059s : 1: py_interpret_to_execute 0.01% : 0.000037s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.02% : 0.000115s : 1: remove_dup_value 0.87% : 0.004619s : 2: renormalize.infer 0.58% : 0.003078s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000012s : 1: rewriter_after_jit_bprop_graph 0.04% : 0.000216s : 1: rewriter_after_opt_a 0.04% : 0.000226s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000022s : 1: swap_dp_allreduce_reducescatter 0.13% : 0.000668s : 1: symbol_engine_optimizer 2.18% : 0.011619s : 1: tuple_transform 17.83% : 0.094932s : 1: type_inference Compare ['Success'] for 0'th output of Forward. Compare ['Success'] for 0'th output of Grad. Compare ['Success'] for 1'th output of Grad. End Start DynamicTest testing with [GRAPH_MODE_O0] [DYNAMIC_RANK] [None]... Compare ['Success'] for 0'th output of Forward. Compare ['Success'] for 0'th output of Grad. Compare ['Success'] for 1'th output of Grad. End *********************End DynamicTest for with GRAPH_MODE_O0 ********************* ************************************************************************************************* ***************************************TEST_OP_DYNAMIC End*************************************** ************************************************************************************************* . [hook] pytest_runtest_teardown:test_mint_mul_dynamic_shape_test_op[pynative] tests/st/mint/test_functional_mul.py::test_mint_mul_dynamic_shape_test_op[pynative],max_mem:4.0M TotalTime = 0.0220634, [21] [bootstrap]: 0.00056096 [type_inference]: 0.0103797 [event_method]: 4.633e-05 [auto_monad]: 7.415e-05 [graph_reusing]: 6.86001e-06 [inline]: 2.16998e-06 [add_attr]: 0.00436977, [1] [add_attr_with_inline]: 0.00435728, [1] [Cycle 1]: 7.369e-05, [2] [tag_attr]: 2.981e-05 [meta_addattr_fg_expand]: 6.87002e-06 [parallel-infer-symbol]: 3.16001e-06 [pre_auto_parallel]: 4.936e-05 [insert-virtual-dataset]: 2.46e-06 [parallel-infer-symbol-second]: 8.50006e-07 [dataset_repeat_opt]: 1.90001e-06 [pipeline_split]: 1.65001e-06 [optimize]: 0.00563337, [53] [py_interpret_to_execute]: 3.539e-05 [rewriter_before_opt_a]: 0.00010963 [opt_a]: 0.00329534, [2] [Cycle 1]: 0.0026249, [45] [expand_dump_flag]: 3.58e-06 [switch_simplify]: 5.347e-05 [loop_unroll]: 3.93e-05 [a_1]: 0.00091106 [with_stream_mark]: 2.128e-05 [recompute_prepare]: 9.19e-06 [updatestate_depend_eliminate]: 4.18999e-06 [updatestate_assign_eliminate]: 3.36001e-06 [updatestate_loads_eliminate]: 3.21001e-06 [parameter_eliminate]: 2.17001e-06 [a_2]: 8.168e-05 [accelerated_algorithm]: 7.18e-06 [shard]: 1.97001e-06 [meta_shard_fg_expand]: 1.99e-06 [shard_inline]: 5.97999e-06 [merge_send_recv]: 9.41e-06 [auto_parallel]: 7.87e-06 [parallel]: 2.901e-05 [flash_sp]: 8.69e-06 [merge_comm]: 3.75e-06 [allreduce_fusion]: 3.5e-06 [matmul_add_comm_reduction]: 9.60001e-06 [allreduce_slice_to_reducescatter]: 6.60017e-07 [virtual_shard_identity]: 8.65999e-06 [virtual_dataset]: 6.66e-06 [get_grad_eliminate_]: 6.16998e-06 [virtual_output]: 5.86998e-06 [merge_forward]: 4.18001e-06 [cell_reuse_recompute_pass]: 1.23002e-06 [offload_activation]: 9.86998e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.204e-05 [merge_recompute_call_nodes]: 1.44e-06 [before_grad]: 1.023e-05 [set_forward_comm_id_for_comm_node_pass]: 3.76999e-06 [meta_fg_expand]: 2.97002e-06 [flash_sp_send_recv_attached]: 2.32999e-06 [receive_attached]: 2.54999e-06 [after_resolve]: 1.163e-05 [a_after_grad]: 1.041e-05 [renormalize]: 0.00094187 [add_forward_monad_depend]: 6.66e-06 [auto_monad_grad]: 2.52001e-06 [auto_monad_eliminator]: 1.609e-05 [cse]: 3.02e-05 [a_3]: 4.825e-05 [Cycle 2]: 0.00065984, [45] [expand_dump_flag]: 1.80001e-06 [switch_simplify]: 8e-06 [loop_unroll]: 6.21e-06 [a_1]: 0.00013739 [with_stream_mark]: 1.298e-05 [recompute_prepare]: 6.44999e-06 [updatestate_depend_eliminate]: 3.40998e-06 [updatestate_assign_eliminate]: 2.46998e-06 [updatestate_loads_eliminate]: 3.01001e-06 [parameter_eliminate]: 1.02998e-06 [a_2]: 7.221e-05 [accelerated_algorithm]: 6.79999e-06 [shard]: 2.26e-06 [meta_shard_fg_expand]: 1.91003e-06 [shard_inline]: 6.17999e-06 [merge_send_recv]: 6.33e-06 [auto_parallel]: 6.17001e-06 [parallel]: 6.43e-06 [flash_sp]: 3.5e-06 [merge_comm]: 3.25998e-06 [allreduce_fusion]: 3.31999e-06 [matmul_add_comm_reduction]: 7.82e-06 [allreduce_slice_to_reducescatter]: 5.39992e-07 [virtual_shard_identity]: 6.74999e-06 [virtual_dataset]: 5.84e-06 [get_grad_eliminate_]: 5.34e-06 [virtual_output]: 5.29e-06 [merge_forward]: 3.55e-06 [cell_reuse_recompute_pass]: 1.97999e-06 [offload_activation]: 7.78999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.088e-05 [merge_recompute_call_nodes]: 9.90025e-07 [before_grad]: 9.19e-06 [set_forward_comm_id_for_comm_node_pass]: 3.94002e-06 [meta_fg_expand]: 2.59999e-06 [flash_sp_send_recv_attached]: 1.25001e-06 [receive_attached]: 1.69e-06 [after_resolve]: 1.058e-05 [a_after_grad]: 8.54e-06 [renormalize]: 6.99947e-08 [add_forward_monad_depend]: 1.82001e-06 [auto_monad_grad]: 1.87999e-06 [auto_monad_eliminator]: 9.59999e-06 [cse]: 1.484e-05 [a_3]: 3.431e-05 [py_interpret_to_execute_after_opt_a]: 1.163e-05 [slice_cell_reuse_recomputed_activation]: 2.63e-06 [rewriter_after_opt_a]: 4.037e-05 [convert_after_rewriter]: 7.03998e-06 [order_py_execute_after_rewriter]: 5.74e-06 [mutable_eliminate]: 0.00064054 [opt_b]: 0.00020119, [1] [Cycle 1]: 0.00019487, [7] [b_1]: 0.00011778 [b_2]: 8.94998e-06 [updatestate_depend_eliminate]: 6.78e-06 [updatestate_assign_eliminate]: 2.65002e-06 [updatestate_loads_eliminate]: 2.09999e-06 [renormalize]: 3.29979e-07 [cse]: 2.025e-05 [optimize_parallel_all_gather_comm]: 1.967e-05 [overlap_param_gather]: 2.26e-06 [cconv]: 2.932e-05 [loop_unroll]: 0.00047124 [opt_after_cconv]: 0.0001078, [1] [Cycle 1]: 0.00010167, [7] [c_1]: 2.979e-05 [parameter_eliminate]: 4.08999e-06 [updatestate_depend_eliminate]: 6.16e-06 [updatestate_assign_eliminate]: 3.83999e-06 [updatestate_loads_eliminate]: 2.41e-06 [cse]: 1.989e-05 [renormalize]: 5.19998e-07 [remove_dup_value]: 1.45e-05 [tuple_transform]: 8.298e-05, [1] [Cycle 1]: 7.705e-05, [4] [d_1]: 4.833e-05 [none_parameter_eliminate]: 1.91998e-06 [renormalize]: 1.50001e-07 [switch_simplify]: 7.58999e-06 [partial_unused_args_eliminate]: 1.77999e-06 [add_recomputation]: 6.514e-05 [cse_after_recomputation]: 2.353e-05, [1] [Cycle 1]: 1.82e-05, [1] [cse]: 1.246e-05 [environ_conv]: 1.22e-05 [swap_dp_allreduce_reducescatter]: 5.56998e-06 [bias_add_comm_swap]: 2.84999e-06 [label_micro_interleaved_index]: 5.34e-06 [label_fine_grained_interleaved_index]: 3.15998e-06 [merge_cast_opt]: 1.42e-06 [slice_recompute_activation]: 2.07001e-06 [micro_interleaved_order_control]: 2.83e-06 [assign_add_opt]: 1.25001e-06 [ForceFp32Comm]: 1.25999e-06 [remove_cast_before_assign_add]: 1.12999e-06 [full_micro_interleaved_order_control]: 2.15002e-06 [reorder_send_recv_between_fp_bp]: 2.74001e-06 [comm_op_add_attrs]: 1.15999e-06 [add_comm_op_reuse_tag]: 1.02e-06 [interleave_split_concat_branches]: 2.04e-06 [interleave_parallel_branches]: 1.10999e-06 [overlap_opt_shard_in_pipeline]: 1.62999e-06 [overlap_opt_shard_grad_in_pipeline]: 1.72999e-06 [control_data_broadcast_order]: 1.494e-05 [grouped_pairwise_exchange_alltoall]: 1.66002e-06 [offloading_packed_experts]: 4.15999e-06 [overlap_recompute_and_grad_model_parallel]: 5.09e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.24e-06 [overlap_recompute_allgather_and_fa_grad]: 1.61998e-06 [overlap_recompute_comm]: 2.22001e-06 [overlap_grad_ring_attention]: 4.08001e-06 [overlap_grad_flash_sp]: 2.01e-05 [begin_end_overlap_inline]: 6.19999e-07 [split_matmul_comm_elemetwise]: 2.11998e-06 [split_layernorm_comm]: 1.94999e-06 [handle_group_info]: 1.07998e-06 [symbol_engine_optimizer]: 8.078e-05, [1] [Cycle 1]: 7.552e-05, [6] [build]: 3.16001e-06 [elim_shapecalc]: 1.094e-05 [elim_not_effective]: 1.374e-05 [opt_reshape]: 7.5e-06 [fold_const_symbol]: 9.82001e-06 [renormalize]: 2.09984e-07 [detach_backward]: 2.31998e-06 [pipeline_parallel_scheduler]: 1.47999e-06 [auto_monad_reorder]: 1.736e-05 [get_jit_bprop_graph]: 2.49001e-06 [rewriter_after_jit_bprop_graph]: 0.0001695 [opt_after_jit_grad]: 0.00051405 [validate]: 4.498e-05 Sums bootstrap : 0.000561s : 3.37% type_inference : 0.010380s : 62.32% event_method : 0.000046s : 0.28% auto_monad : 0.000074s : 0.45% graph_reusing : 0.000007s : 0.04% inline : 0.000002s : 0.01% add_attr.add_attr_with_inline.tag_attr : 0.000030s : 0.18% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.04% parallel-infer-symbol : 0.000003s : 0.02% pre_auto_parallel : 0.000049s : 0.30% insert-virtual-dataset : 0.000002s : 0.01% parallel-infer-symbol-second : 0.000001s : 0.01% dataset_repeat_opt : 0.000002s : 0.01% pipeline_split : 0.000002s : 0.01% optimize.py_interpret_to_execute : 0.000035s : 0.21% optimize.rewriter_before_opt_a : 0.000110s : 0.66% optimize.opt_a.expand_dump_flag : 0.000005s : 0.03% optimize.opt_a.switch_simplify : 0.000061s : 0.37% optimize.opt_a.loop_unroll : 0.000046s : 0.27% optimize.opt_a.a_1 : 0.001048s : 6.29% optimize.opt_a.with_stream_mark : 0.000034s : 0.21% optimize.opt_a.recompute_prepare : 0.000016s : 0.09% optimize.opt_a.updatestate_depend_eliminate : 0.000008s : 0.05% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.04% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.04% optimize.opt_a.parameter_eliminate : 0.000003s : 0.02% optimize.opt_a.a_2 : 0.000154s : 0.92% optimize.opt_a.accelerated_algorithm : 0.000014s : 0.08% optimize.opt_a.shard : 0.000004s : 0.03% optimize.opt_a.meta_shard_fg_expand : 0.000004s : 0.02% optimize.opt_a.shard_inline : 0.000012s : 0.07% optimize.opt_a.merge_send_recv : 0.000016s : 0.09% optimize.opt_a.auto_parallel : 0.000014s : 0.08% optimize.opt_a.parallel : 0.000035s : 0.21% optimize.opt_a.flash_sp : 0.000012s : 0.07% optimize.opt_a.merge_comm : 0.000007s : 0.04% optimize.opt_a.allreduce_fusion : 0.000007s : 0.04% optimize.opt_a.matmul_add_comm_reduction : 0.000017s : 0.10% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.01% optimize.opt_a.virtual_shard_identity : 0.000015s : 0.09% optimize.opt_a.virtual_dataset : 0.000013s : 0.08% optimize.opt_a.get_grad_eliminate_ : 0.000012s : 0.07% optimize.opt_a.virtual_output : 0.000011s : 0.07% optimize.opt_a.merge_forward : 0.000008s : 0.05% optimize.opt_a.cell_reuse_recompute_pass : 0.000003s : 0.02% optimize.opt_a.offload_activation : 0.000018s : 0.11% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000023s : 0.14% optimize.opt_a.merge_recompute_call_nodes : 0.000002s : 0.01% optimize.opt_a.before_grad : 0.000019s : 0.12% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000008s : 0.05% optimize.opt_a.meta_fg_expand : 0.000006s : 0.03% optimize.opt_a.flash_sp_send_recv_attached : 0.000004s : 0.02% optimize.opt_a.receive_attached : 0.000004s : 0.03% optimize.opt_a.after_resolve : 0.000022s : 0.13% optimize.opt_a.a_after_grad : 0.000019s : 0.11% optimize.opt_a.renormalize : 0.000942s : 5.66% optimize.opt_a.add_forward_monad_depend : 0.000008s : 0.05% optimize.opt_a.auto_monad_grad : 0.000004s : 0.03% optimize.opt_a.auto_monad_eliminator : 0.000026s : 0.15% optimize.opt_a.cse : 0.000045s : 0.27% optimize.opt_a.a_3 : 0.000083s : 0.50% optimize.py_interpret_to_execute_after_opt_a : 0.000012s : 0.07% optimize.slice_cell_reuse_recomputed_activation : 0.000003s : 0.02% optimize.rewriter_after_opt_a : 0.000040s : 0.24% optimize.convert_after_rewriter : 0.000007s : 0.04% optimize.order_py_execute_after_rewriter : 0.000006s : 0.03% optimize.mutable_eliminate : 0.000641s : 3.85% optimize.opt_b.b_1 : 0.000118s : 0.71% optimize.opt_b.b_2 : 0.000009s : 0.05% optimize.opt_b.updatestate_depend_eliminate : 0.000007s : 0.04% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.02% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.01% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000020s : 0.12% optimize.optimize_parallel_all_gather_comm : 0.000020s : 0.12% optimize.overlap_param_gather : 0.000002s : 0.01% optimize.cconv : 0.000029s : 0.18% optimize.loop_unroll : 0.000471s : 2.83% optimize.opt_after_cconv.c_1 : 0.000030s : 0.18% optimize.opt_after_cconv.parameter_eliminate : 0.000004s : 0.02% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.04% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000004s : 0.02% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.01% optimize.opt_after_cconv.cse : 0.000020s : 0.12% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000014s : 0.09% optimize.tuple_transform.d_1 : 0.000048s : 0.29% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.01% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000008s : 0.05% optimize.partial_unused_args_eliminate : 0.000002s : 0.01% optimize.add_recomputation : 0.000065s : 0.39% optimize.cse_after_recomputation.cse : 0.000012s : 0.07% optimize.environ_conv : 0.000012s : 0.07% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.03% optimize.bias_add_comm_swap : 0.000003s : 0.02% optimize.label_micro_interleaved_index : 0.000005s : 0.03% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.02% optimize.merge_cast_opt : 0.000001s : 0.01% optimize.slice_recompute_activation : 0.000002s : 0.01% optimize.micro_interleaved_order_control : 0.000003s : 0.02% optimize.assign_add_opt : 0.000001s : 0.01% optimize.ForceFp32Comm : 0.000001s : 0.01% optimize.remove_cast_before_assign_add : 0.000001s : 0.01% optimize.full_micro_interleaved_order_control : 0.000002s : 0.01% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.02% optimize.comm_op_add_attrs : 0.000001s : 0.01% optimize.add_comm_op_reuse_tag : 0.000001s : 0.01% optimize.interleave_split_concat_branches : 0.000002s : 0.01% optimize.interleave_parallel_branches : 0.000001s : 0.01% optimize.overlap_opt_shard_in_pipeline : 0.000002s : 0.01% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.01% optimize.control_data_broadcast_order : 0.000015s : 0.09% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.01% optimize.offloading_packed_experts : 0.000004s : 0.02% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.03% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.01% optimize.overlap_recompute_allgather_and_fa_grad : 0.000002s : 0.01% optimize.overlap_recompute_comm : 0.000002s : 0.01% optimize.overlap_grad_ring_attention : 0.000004s : 0.02% optimize.overlap_grad_flash_sp : 0.000020s : 0.12% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.01% optimize.split_layernorm_comm : 0.000002s : 0.01% optimize.handle_group_info : 0.000001s : 0.01% optimize.symbol_engine_optimizer.build : 0.000003s : 0.02% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000011s : 0.07% optimize.symbol_engine_optimizer.elim_not_effective : 0.000014s : 0.08% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.05% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.06% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.01% pipeline_parallel_scheduler : 0.000001s : 0.01% auto_monad_reorder : 0.000017s : 0.10% get_jit_bprop_graph : 0.000002s : 0.01% rewriter_after_jit_bprop_graph : 0.000170s : 1.02% opt_after_jit_grad : 0.000514s : 3.09% validate : 0.000045s : 0.27% Time group info: ------[substitution.] 0.000294 40 11.88% : 0.000035s : 8: substitution.arithmetic_simplify 0.67% : 0.000002s : 2: substitution.elim_not_effective 0.51% : 0.000002s : 2: substitution.fold_const_symbol 2.13% : 0.000006s : 4: substitution.graph_param_transform 73.23% : 0.000215s : 6: substitution.inline 1.25% : 0.000004s : 4: substitution.j_node_and_user_rematch 1.68% : 0.000005s : 4: substitution.remove_not_recompute_node 1.69% : 0.000005s : 4: substitution.replace_old_param 6.96% : 0.000020s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.010306 2 89.40% : 0.009214s : 1: type_inference.infer 10.60% : 0.001092s : 1: type_inference.specialize ------[replace.] 0.000092 12 61.14% : 0.000056s : 6: replace.inline 38.86% : 0.000036s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000229 12 92.23% : 0.000211s : 6: match.inline 7.77% : 0.000018s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000239 1534 0.98% : 0.000002s : 17: predicate.accumulaten_eliminater 0.70% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.39% : 0.000001s : 8: predicate.addn_check_dump 0.99% : 0.000002s : 17: predicate.addn_zero_filter 0.90% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.51% : 0.000006s : 25: predicate.arithmetic_simplify 0.95% : 0.000002s : 17: predicate.cast_eliminate 0.50% : 0.000001s : 8: predicate.check_bprop_eliminate 0.41% : 0.000001s : 8: predicate.compare_switch_simplify 0.16% : 0.000000s : 4: predicate.const_output_eliminate 0.47% : 0.000001s : 8: predicate.depend_value_elim 0.99% : 0.000002s : 17: predicate.dict_get_item_const_eliminator 1.07% : 0.000003s : 17: predicate.dict_get_item_eliminator 0.95% : 0.000002s : 17: predicate.dict_set_item_eliminator 0.87% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.21% : 0.000001s : 4: predicate.elim_not_effective 0.30% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.14% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.12% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.16% : 0.000003s : 21: predicate.environ_get_depend_swap 1.62% : 0.000004s : 29: predicate.environ_get_eliminate 1.21% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.64% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.58% : 0.000006s : 29: predicate.float_depend_g_call 0.44% : 0.000001s : 8: predicate.float_environ_get_switch 0.60% : 0.000001s : 12: predicate.float_tuple_getitem_switch 0.15% : 0.000000s : 4: predicate.fold_const_symbol 0.58% : 0.000001s : 8: predicate.get_grad_eliminate 0.20% : 0.000000s : 4: predicate.graph_param_transform 0.47% : 0.000001s : 8: predicate.incorporate_call 0.39% : 0.000001s : 8: predicate.incorporate_call_switch 6.38% : 0.000015s : 70: predicate.inline 0.82% : 0.000002s : 8: predicate.inline_without_move 0.27% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.70% : 0.000002s : 8: predicate.less_batch_normalization 1.99% : 0.000005s : 31: predicate.list_to_tuple_eliminator_ 2.67% : 0.000006s : 48: predicate.load_eliminater 0.97% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.66% : 0.000006s : 47: predicate.loop_unroll_before_grad 1.83% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.55% : 0.000001s : 8: predicate.merge_addn 0.46% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.49% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.89% : 0.000002s : 17: predicate.minmaximum_grad 0.99% : 0.000002s : 4: predicate.mutable_eliminate 0.26% : 0.000001s : 4: predicate.opt_reshape 0.31% : 0.000001s : 4: predicate.parallel_virtual_node 2.20% : 0.000005s : 29: predicate.partial_defer_inline 1.68% : 0.000004s : 27: predicate.partial_eliminate 1.13% : 0.000003s : 17: predicate.print_const_string_wrapper 0.47% : 0.000001s : 8: predicate.reduce_all_const_elim 1.31% : 0.000003s : 17: predicate.reduce_eliminate 2.60% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.36% : 0.000001s : 8: predicate.remove_not_recompute_node 1.48% : 0.000004s : 31: predicate.replace_applicator 0.45% : 0.000001s : 8: predicate.replace_old_param 0.28% : 0.000001s : 4: predicate.reset_defer_inline 0.97% : 0.000002s : 17: predicate.reshape_eliminate 0.56% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.47% : 0.000001s : 4: predicate.row_tensor_eliminate 0.60% : 0.000001s : 8: predicate.same_eliminate 0.39% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.68% : 0.000002s : 8: predicate.shard_identity_eliminate 0.68% : 0.000002s : 8: predicate.special_op_eliminate 0.59% : 0.000001s : 8: predicate.specialize_transform 0.75% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.55% : 0.000001s : 8: predicate.stack_unstack_eliminate 0.26% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.79% : 0.000004s : 29: predicate.switch_defer_inline 2.18% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.40% : 0.000013s : 88: predicate.switch_simplify 1.03% : 0.000002s : 17: predicate.tile_eliminate 1.05% : 0.000003s : 17: predicate.transpose_eliminate 1.62% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.55% : 0.000004s : 25: predicate.tuple_list_get_item_const_eliminator 1.44% : 0.000003s : 25: predicate.tuple_list_get_item_depend_reorder 3.68% : 0.000009s : 39: predicate.tuple_list_get_item_eliminator 1.48% : 0.000004s : 25: predicate.tuple_list_get_set_item_eliminator 2.12% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 1.79% : 0.000004s : 31: predicate.tuple_to_list_eliminator_ 2.48% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.06% : 0.000007s : 56: predicate.updatestate_useless_node_eliminater 0.31% : 0.000001s : 4: predicate.value_based_eliminate 0.61% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.51% : 0.000001s : 8: predicate.virtual_output_eliminate 0.21% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.36% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000898 15 59.37% : 0.000533s : 7: func_graph_cloner_run.FuncGraphClonerGraph 40.63% : 0.000365s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.034605 192 0.01% : 0.000004s : 1: ForceFp32Comm 12.65% : 0.004376s : 1: add_attr 12.60% : 0.004361s : 1: add_attr_with_inline 0.01% : 0.000004s : 1: add_comm_op_reuse_tag 0.20% : 0.000070s : 1: add_recomputation 0.07% : 0.000023s : 1: assign_add_opt 0.23% : 0.000081s : 1: auto_monad 0.06% : 0.000022s : 1: auto_monad_reorder 0.01% : 0.000004s : 1: begin_end_overlap_inline 0.02% : 0.000007s : 1: bias_add_comm_swap 1.72% : 0.000595s : 1: bootstrap 0.10% : 0.000034s : 1: cconv 0.02% : 0.000005s : 1: comm_op_add_attrs 0.06% : 0.000019s : 1: control_data_broadcast_order 0.03% : 0.000011s : 1: convert_after_rewriter 0.08% : 0.000027s : 1: cse_after_recomputation 0.01% : 0.000005s : 1: dataset_repeat_opt 0.02% : 0.000006s : 1: detach_backward 0.05% : 0.000016s : 1: environ_conv 0.16% : 0.000054s : 1: event_method 0.02% : 0.000005s : 1: full_micro_interleaved_order_control 0.02% : 0.000006s : 1: get_jit_bprop_graph 0.03% : 0.000011s : 1: graph_reusing 0.01% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.01% : 0.000004s : 1: handle_group_info 0.02% : 0.000005s : 1: inline 0.02% : 0.000007s : 1: insert-virtual-dataset 0.01% : 0.000004s : 1: interleave_parallel_branches 0.01% : 0.000005s : 1: interleave_split_concat_branches 0.02% : 0.000007s : 1: label_fine_grained_interleaved_index 0.02% : 0.000008s : 1: label_micro_interleaved_index 1.39% : 0.000480s : 1: loop_unroll 0.02% : 0.000006s : 1: merge_cast_opt 0.02% : 0.000006s : 1: micro_interleaved_order_control 1.88% : 0.000649s : 1: mutable_eliminate 0.02% : 0.000008s : 1: offloading_packed_experts 0.05% : 0.000016s : 1: opt.transform.loop_unroll_optimizer 0.05% : 0.000016s : 1: opt.transform.mutable_eliminate 4.28% : 0.001482s : 78: opt.transform.opt_a 0.08% : 0.000028s : 1: opt.transform.opt_after_cconv 0.07% : 0.000025s : 1: opt.transform.opt_after_jit_grad 0.28% : 0.000097s : 28: opt.transform.opt_b 0.15% : 0.000053s : 2: opt.transform.opt_trans_graph 0.11% : 0.000038s : 4: opt.transform.symbol_engine_opt 9.53% : 0.003299s : 1: opt_a 0.32% : 0.000112s : 1: opt_after_cconv 1.51% : 0.000524s : 1: opt_after_jit_grad 0.59% : 0.000205s : 1: opt_b 16.29% : 0.005639s : 1: optimize 0.07% : 0.000024s : 1: optimize_parallel_all_gather_comm 0.03% : 0.000009s : 1: order_py_execute_after_rewriter 0.07% : 0.000023s : 1: overlap_grad_flash_sp 0.01% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.02% : 0.000007s : 1: overlap_grad_ring_attention 0.01% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.01% : 0.000005s : 1: overlap_opt_shard_in_pipeline 0.02% : 0.000006s : 1: overlap_param_gather 0.01% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.02% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.02% : 0.000005s : 1: overlap_recompute_comm 0.02% : 0.000007s : 1: parallel-infer-symbol 0.01% : 0.000004s : 1: parallel-infer-symbol-second 0.01% : 0.000005s : 1: partial_unused_args_eliminate 0.01% : 0.000005s : 1: pipeline_parallel_scheduler 0.01% : 0.000005s : 1: pipeline_split 0.16% : 0.000054s : 1: pre_auto_parallel 0.12% : 0.000040s : 1: py_interpret_to_execute 0.04% : 0.000015s : 1: py_interpret_to_execute_after_opt_a 0.01% : 0.000004s : 1: remove_cast_before_assign_add 0.05% : 0.000018s : 1: remove_dup_value 1.33% : 0.000460s : 1: renormalize.infer 1.37% : 0.000473s : 1: renormalize.specialize 0.02% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.51% : 0.000177s : 1: rewriter_after_jit_bprop_graph 0.13% : 0.000045s : 1: rewriter_after_opt_a 0.33% : 0.000114s : 1: rewriter_before_opt_a 0.02% : 0.000006s : 1: slice_cell_reuse_recomputed_activation 0.01% : 0.000005s : 1: slice_recompute_activation 0.01% : 0.000005s : 1: split_layernorm_comm 0.02% : 0.000006s : 1: split_matmul_comm_elemetwise 0.03% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.24% : 0.000084s : 1: symbol_engine_optimizer 0.25% : 0.000086s : 1: tuple_transform 30.05% : 0.010400s : 1: type_inference ************************************************************************************************** *************************************TEST_OP_GENERALIZE Start************************************* ************************************************************************************************** *****************Start Generalize test PYNATIVE_MODE DiscontiguousInput case***************** [INFO]: TEST_OP_GENERALIZE generate discontiguous input with shape (10, 10, 8), is_contiguous False, origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE generate discontiguous input with shape (10, 10, 8), is_contiguous False, origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE Start to test discontiguous input [INFO]: TEST_OP_GENERALIZE Start to test contiguous input [INFO]: TEST_OP_GENERALIZE Start to compare discontiguous out with contiguous out Compare ['Success'] for 0'th output. *****************End Generalize test PYNATIVE_MODE DiscontiguousInput case***************** *****************Start Generalize test PYNATIVE_MODE ViewTensor case***************** [INFO]: TEST_OP_GENERALIZE generate view input with shape (10, 10, 8), origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE generate view input with shape (10, 10, 8), origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE Start to test no view input [INFO]: TEST_OP_GENERALIZE Start to test view input [INFO]: TEST_OP_GENERALIZE Start to compare view out with no view out Compare ['Success'] for 0'th output. *****************End Generalize test PYNATIVE_MODE ViewTensor case***************** *****************Start Generalize test PYNATIVE_MODE EmptyTensor case***************** [WARNING]: TEST_OP_GENERALIZE PYNATIVE_MODE 'EmptyTensor' in 'disable_case', EmptyTensor case is skipped. *****************End Generalize test PYNATIVE_MODE EmptyTensor case***************** *****************Start Generalize test PYNATIVE_MODE ScalarTensor case***************** [INFO]: TEST_OP_GENERALIZE Start to test scalar tensor [INFO]: TEST_OP_GENERALIZE Start to test 1d tensor [INFO]: TEST_OP_GENERALIZE Start to compare scalar tensor out with 1d tensor out Compare ['Success'] for 0'th output. *****************End Generalize test PYNATIVE_MODE ScalarTensor case***************** *****************Start Generalize test PYNATIVE_MODE GradByRequirement case***************** [INFO]: TEST_OP_GENERALIZE Start to test all grads [INFO]: TEST_OP_GENERALIZE Start to test grad by requirement on position 0 [INFO]: TEST_OP_GENERALIZE Start to test grad by requirement on position 1 [INFO]: TEST_OP_GENERALIZE Start to compare grad by requirement with all grads Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. *****************End Generalize test PYNATIVE_MODE GradByRequirement case***************** *****************Start Generalize test PYNATIVE_MODE Deterministic case***************** [INFO]: TEST_OP_GENERALIZE test deterministic case loop 0 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 1 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 2 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 3 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 4 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. *****************End Generalize test PYNATIVE_MODE Deterministic case***************** *****************Start Generalize test GRAPH_MODE_O0 DiscontiguousInput case***************** [WARNING]: TEST_OP_GENERALIZE GRAPH_MODE_O0 'DiscontiguousInput' is skipped temporarily with 'GRAPH_MODE_O0' and 'GRAPH_MODE_GE', it's will cause accuracy issue after 'Tensor storage refactor'. *****************End Generalize test GRAPH_MODE_O0 DiscontiguousInput case***************** *****************Start Generalize test GRAPH_MODE_O0 ViewTensor case***************** [INFO]: TEST_OP_GENERALIZE generate view input with shape (10, 10, 8), origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE generate view input with shape (10, 10, 8), origin input with shape (10, 10, 8) [INFO]: TEST_OP_GENERALIZE Start to test no view input [INFO]: TEST_OP_GENERALIZE Start to test view input [INFO]: TEST_OP_GENERALIZE TotalTime = 0.0380053, [21] [bootstrap]: 0.00050999 [type_inference]: 0.0267916 [event_method]: 5.001e-05 [auto_monad]: 8.098e-05 [graph_reusing]: 6.81999e-06 [inline]: 3.41999e-06 [add_attr]: 0.00380563, [1] [add_attr_with_inline]: 0.00379173, [1] [Cycle 1]: 8.464e-05, [2] [tag_attr]: 3.123e-05 [meta_addattr_fg_expand]: 7.04001e-06 [parallel-infer-symbol]: 3.88999e-06 [pre_auto_parallel]: 4.776e-05 [insert-virtual-dataset]: 3.26001e-06 [parallel-infer-symbol-second]: 9.00007e-07 [dataset_repeat_opt]: 1.99e-06 [pipeline_split]: 2.02001e-06 [optimize]: 0.00593637, [53] [py_interpret_to_execute]: 3.499e-05 [rewriter_before_opt_a]: 0.00010777 [opt_a]: 0.00359028, [2] [Cycle 1]: 0.00289818, [45] [expand_dump_flag]: 3.36001e-06 [switch_simplify]: 5.419e-05 [loop_unroll]: 3.988e-05 [a_1]: 0.00094476 [with_stream_mark]: 2.529e-05 [recompute_prepare]: 1.308e-05 [updatestate_depend_eliminate]: 5.62999e-06 [updatestate_assign_eliminate]: 3.3e-06 [updatestate_loads_eliminate]: 2.95002e-06 [parameter_eliminate]: 1.87001e-06 [a_2]: 8.708e-05 [accelerated_algorithm]: 7.16001e-06 [shard]: 2.04999e-06 [meta_shard_fg_expand]: 2.20002e-06 [shard_inline]: 6.69001e-06 [merge_send_recv]: 9.10001e-06 [auto_parallel]: 8.55999e-06 [parallel]: 1.963e-05 [flash_sp]: 9.77001e-06 [merge_comm]: 5.09e-06 [allreduce_fusion]: 3.5e-06 [matmul_add_comm_reduction]: 9.91e-06 [allreduce_slice_to_reducescatter]: 8.2e-07 [virtual_shard_identity]: 9.58002e-06 [virtual_dataset]: 6.66e-06 [get_grad_eliminate_]: 6.09001e-06 [virtual_output]: 6.09999e-06 [merge_forward]: 5.05001e-06 [cell_reuse_recompute_pass]: 1.71e-06 [offload_activation]: 1.045e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.445e-05 [merge_recompute_call_nodes]: 1.44998e-06 [before_grad]: 1.02e-05 [set_forward_comm_id_for_comm_node_pass]: 4.53001e-06 [meta_fg_expand]: 2.68e-06 [flash_sp_send_recv_attached]: 2.79001e-06 [receive_attached]: 2.07001e-06 [after_resolve]: 1.229e-05 [a_after_grad]: 1.03e-05 [renormalize]: 0.0011403 [add_forward_monad_depend]: 7.08e-06 [auto_monad_grad]: 2.74001e-06 [auto_monad_eliminator]: 1.953e-05 [cse]: 3.091e-05 [a_3]: 5.082e-05 [Cycle 2]: 0.00068009, [45] [expand_dump_flag]: 1.95001e-06 [switch_simplify]: 8.38999e-06 [loop_unroll]: 5.96998e-06 [a_1]: 0.00013919 [with_stream_mark]: 1.579e-05 [recompute_prepare]: 7.08e-06 [updatestate_depend_eliminate]: 3.47002e-06 [updatestate_assign_eliminate]: 2.52001e-06 [updatestate_loads_eliminate]: 2.83e-06 [parameter_eliminate]: 1.37e-06 [a_2]: 7.186e-05 [accelerated_algorithm]: 6.99001e-06 [shard]: 1.81e-06 [meta_shard_fg_expand]: 2.06e-06 [shard_inline]: 5.91003e-06 [merge_send_recv]: 6.26e-06 [auto_parallel]: 6.48e-06 [parallel]: 6.39999e-06 [flash_sp]: 3.3e-06 [merge_comm]: 3.86999e-06 [allreduce_fusion]: 3.19001e-06 [matmul_add_comm_reduction]: 8.22e-06 [allreduce_slice_to_reducescatter]: 4.00003e-07 [virtual_shard_identity]: 8.91002e-06 [virtual_dataset]: 5.67999e-06 [get_grad_eliminate_]: 5.36998e-06 [virtual_output]: 5.24e-06 [merge_forward]: 4.05998e-06 [cell_reuse_recompute_pass]: 1.96e-06 [offload_activation]: 8.66002e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.218e-05 [merge_recompute_call_nodes]: 8.29983e-07 [before_grad]: 9.71e-06 [set_forward_comm_id_for_comm_node_pass]: 3.91001e-06 [meta_fg_expand]: 2.36e-06 [flash_sp_send_recv_attached]: 1.34998e-06 [receive_attached]: 1.88002e-06 [after_resolve]: 1.202e-05 [a_after_grad]: 9.05999e-06 [renormalize]: 8.00064e-08 [add_forward_monad_depend]: 2.61e-06 [auto_monad_grad]: 1.59e-06 [auto_monad_eliminator]: 9.54e-06 [cse]: 1.632e-05 [a_3]: 3.395e-05 [py_interpret_to_execute_after_opt_a]: 1.361e-05 [slice_cell_reuse_recomputed_activation]: 2.14999e-06 [rewriter_after_opt_a]: 4.281e-05 [convert_after_rewriter]: 7.9e-06 [order_py_execute_after_rewriter]: 5.30001e-06 [mutable_eliminate]: 0.00066402 [opt_b]: 0.00020304, [1] [Cycle 1]: 0.00019532, [7] [b_1]: 0.00011619 [b_2]: 8.12e-06 [updatestate_depend_eliminate]: 7.43e-06 [updatestate_assign_eliminate]: 2.82002e-06 [updatestate_loads_eliminate]: 2.24999e-06 [renormalize]: 8.39995e-07 [cse]: 2.032e-05 [optimize_parallel_all_gather_comm]: 1.796e-05 [overlap_param_gather]: 1.92999e-06 [cconv]: 3.036e-05 [loop_unroll]: 0.00047595 [opt_after_cconv]: 0.00010865, [1] [Cycle 1]: 0.00010236, [7] [c_1]: 3.038e-05 [parameter_eliminate]: 3.26001e-06 [updatestate_depend_eliminate]: 7.97e-06 [updatestate_assign_eliminate]: 3.33e-06 [updatestate_loads_eliminate]: 2.17001e-06 [cse]: 1.957e-05 [renormalize]: 3.39991e-07 [remove_dup_value]: 1.46e-05 [tuple_transform]: 7.724e-05, [1] [Cycle 1]: 7.206e-05, [4] [d_1]: 4.475e-05 [none_parameter_eliminate]: 1.69998e-06 [renormalize]: 1.69995e-07 [switch_simplify]: 6.64001e-06 [partial_unused_args_eliminate]: 1.77999e-06 [add_recomputation]: 5.374e-05 [cse_after_recomputation]: 2.201e-05, [1] [Cycle 1]: 1.705e-05, [1] [cse]: 1.12e-05 [environ_conv]: 5.35001e-06 [swap_dp_allreduce_reducescatter]: 5.40999e-06 [bias_add_comm_swap]: 2.90998e-06 [label_micro_interleaved_index]: 5.05001e-06 [label_fine_grained_interleaved_index]: 2.79001e-06 [merge_cast_opt]: 1.47999e-06 [slice_recompute_activation]: 2.11e-06 [micro_interleaved_order_control]: 2.33998e-06 [assign_add_opt]: 1.55999e-06 [ForceFp32Comm]: 9.90025e-07 [remove_cast_before_assign_add]: 1.00001e-06 [full_micro_interleaved_order_control]: 2.07999e-06 [reorder_send_recv_between_fp_bp]: 2.69001e-06 [comm_op_add_attrs]: 1.10001e-06 [add_comm_op_reuse_tag]: 1.03001e-06 [interleave_split_concat_branches]: 1.30999e-06 [interleave_parallel_branches]: 1.04e-06 [overlap_opt_shard_in_pipeline]: 1.22e-06 [overlap_opt_shard_grad_in_pipeline]: 1.74e-06 [control_data_broadcast_order]: 1.615e-05 [grouped_pairwise_exchange_alltoall]: 1.73002e-06 [offloading_packed_experts]: 4.33001e-06 [overlap_recompute_and_grad_model_parallel]: 5.05001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.32e-06 [overlap_recompute_allgather_and_fa_grad]: 1.41002e-06 [overlap_recompute_comm]: 2.78e-06 [overlap_grad_ring_attention]: 5.12e-06 [overlap_grad_flash_sp]: 2.12e-05 [begin_end_overlap_inline]: 6.60017e-07 [split_matmul_comm_elemetwise]: 2.31e-06 [split_layernorm_comm]: 1.71e-06 [handle_group_info]: 1.34e-06 [symbol_engine_optimizer]: 8.701e-05, [1] [Cycle 1]: 8.136e-05, [6] [build]: 3.81999e-06 [elim_shapecalc]: 1.481e-05 [elim_not_effective]: 1.379e-05 [opt_reshape]: 6.64001e-06 [fold_const_symbol]: 1.051e-05 [renormalize]: 2.10013e-07 [detach_backward]: 2.53e-06 [pipeline_parallel_scheduler]: 1.57001e-06 [auto_monad_reorder]: 1.898e-05 [get_jit_bprop_graph]: 1.79e-06 [rewriter_after_jit_bprop_graph]: 5.73002e-06 [opt_after_jit_grad]: 0.00051103 [validate]: 4.23e-05 Sums bootstrap : 0.000510s : 1.54% type_inference : 0.026792s : 80.84% event_method : 0.000050s : 0.15% auto_monad : 0.000081s : 0.24% graph_reusing : 0.000007s : 0.02% inline : 0.000003s : 0.01% add_attr.add_attr_with_inline.tag_attr : 0.000031s : 0.09% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.02% parallel-infer-symbol : 0.000004s : 0.01% pre_auto_parallel : 0.000048s : 0.14% insert-virtual-dataset : 0.000003s : 0.01% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.01% pipeline_split : 0.000002s : 0.01% optimize.py_interpret_to_execute : 0.000035s : 0.11% optimize.rewriter_before_opt_a : 0.000108s : 0.33% optimize.opt_a.expand_dump_flag : 0.000005s : 0.02% optimize.opt_a.switch_simplify : 0.000063s : 0.19% optimize.opt_a.loop_unroll : 0.000046s : 0.14% optimize.opt_a.a_1 : 0.001084s : 3.27% optimize.opt_a.with_stream_mark : 0.000041s : 0.12% optimize.opt_a.recompute_prepare : 0.000020s : 0.06% optimize.opt_a.updatestate_depend_eliminate : 0.000009s : 0.03% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.02% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.02% optimize.opt_a.parameter_eliminate : 0.000003s : 0.01% optimize.opt_a.a_2 : 0.000159s : 0.48% optimize.opt_a.accelerated_algorithm : 0.000014s : 0.04% optimize.opt_a.shard : 0.000004s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000004s : 0.01% optimize.opt_a.shard_inline : 0.000013s : 0.04% optimize.opt_a.merge_send_recv : 0.000015s : 0.05% optimize.opt_a.auto_parallel : 0.000015s : 0.05% optimize.opt_a.parallel : 0.000026s : 0.08% optimize.opt_a.flash_sp : 0.000013s : 0.04% optimize.opt_a.merge_comm : 0.000009s : 0.03% optimize.opt_a.allreduce_fusion : 0.000007s : 0.02% optimize.opt_a.matmul_add_comm_reduction : 0.000018s : 0.05% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000018s : 0.06% optimize.opt_a.virtual_dataset : 0.000012s : 0.04% optimize.opt_a.get_grad_eliminate_ : 0.000011s : 0.03% optimize.opt_a.virtual_output : 0.000011s : 0.03% optimize.opt_a.merge_forward : 0.000009s : 0.03% optimize.opt_a.cell_reuse_recompute_pass : 0.000004s : 0.01% optimize.opt_a.offload_activation : 0.000019s : 0.06% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000027s : 0.08% optimize.opt_a.merge_recompute_call_nodes : 0.000002s : 0.01% optimize.opt_a.before_grad : 0.000020s : 0.06% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000008s : 0.03% optimize.opt_a.meta_fg_expand : 0.000005s : 0.02% optimize.opt_a.flash_sp_send_recv_attached : 0.000004s : 0.01% optimize.opt_a.receive_attached : 0.000004s : 0.01% optimize.opt_a.after_resolve : 0.000024s : 0.07% optimize.opt_a.a_after_grad : 0.000019s : 0.06% optimize.opt_a.renormalize : 0.001140s : 3.44% optimize.opt_a.add_forward_monad_depend : 0.000010s : 0.03% optimize.opt_a.auto_monad_grad : 0.000004s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000029s : 0.09% optimize.opt_a.cse : 0.000047s : 0.14% optimize.opt_a.a_3 : 0.000085s : 0.26% optimize.py_interpret_to_execute_after_opt_a : 0.000014s : 0.04% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.01% optimize.rewriter_after_opt_a : 0.000043s : 0.13% optimize.convert_after_rewriter : 0.000008s : 0.02% optimize.order_py_execute_after_rewriter : 0.000005s : 0.02% optimize.mutable_eliminate : 0.000664s : 2.00% optimize.opt_b.b_1 : 0.000116s : 0.35% optimize.opt_b.b_2 : 0.000008s : 0.02% optimize.opt_b.updatestate_depend_eliminate : 0.000007s : 0.02% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.01% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000020s : 0.06% optimize.optimize_parallel_all_gather_comm : 0.000018s : 0.05% optimize.overlap_param_gather : 0.000002s : 0.01% optimize.cconv : 0.000030s : 0.09% optimize.loop_unroll : 0.000476s : 1.44% optimize.opt_after_cconv.c_1 : 0.000030s : 0.09% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000008s : 0.02% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.01% optimize.opt_after_cconv.cse : 0.000020s : 0.06% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000015s : 0.04% optimize.tuple_transform.d_1 : 0.000045s : 0.14% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.01% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.02% optimize.partial_unused_args_eliminate : 0.000002s : 0.01% optimize.add_recomputation : 0.000054s : 0.16% optimize.cse_after_recomputation.cse : 0.000011s : 0.03% optimize.environ_conv : 0.000005s : 0.02% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.02% optimize.bias_add_comm_swap : 0.000003s : 0.01% optimize.label_micro_interleaved_index : 0.000005s : 0.02% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.01% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.01% optimize.micro_interleaved_order_control : 0.000002s : 0.01% optimize.assign_add_opt : 0.000002s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.01% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.01% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.01% optimize.control_data_broadcast_order : 0.000016s : 0.05% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.01% optimize.offloading_packed_experts : 0.000004s : 0.01% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.02% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000003s : 0.01% optimize.overlap_grad_ring_attention : 0.000005s : 0.02% optimize.overlap_grad_flash_sp : 0.000021s : 0.06% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.01% optimize.split_layernorm_comm : 0.000002s : 0.01% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000004s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000015s : 0.04% optimize.symbol_engine_optimizer.elim_not_effective : 0.000014s : 0.04% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.02% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000011s : 0.03% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.01% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000019s : 0.06% get_jit_bprop_graph : 0.000002s : 0.01% rewriter_after_jit_bprop_graph : 0.000006s : 0.02% opt_after_jit_grad : 0.000511s : 1.54% validate : 0.000042s : 0.13% Time group info: ------[substitution.] 0.000319 40 12.77% : 0.000041s : 8: substitution.arithmetic_simplify 0.63% : 0.000002s : 2: substitution.elim_not_effective 0.67% : 0.000002s : 2: substitution.fold_const_symbol 1.95% : 0.000006s : 4: substitution.graph_param_transform 73.18% : 0.000234s : 6: substitution.inline 1.23% : 0.000004s : 4: substitution.j_node_and_user_rematch 1.51% : 0.000005s : 4: substitution.remove_not_recompute_node 1.65% : 0.000005s : 4: substitution.replace_old_param 6.41% : 0.000020s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.026711 2 95.76% : 0.025578s : 1: type_inference.infer 4.24% : 0.001133s : 1: type_inference.specialize ------[replace.] 0.000104 12 59.22% : 0.000061s : 6: replace.inline 40.78% : 0.000042s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000248 12 92.76% : 0.000230s : 6: match.inline 7.24% : 0.000018s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000245 1534 1.02% : 0.000002s : 17: predicate.accumulaten_eliminater 0.88% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.49% : 0.000001s : 8: predicate.addn_check_dump 1.03% : 0.000003s : 17: predicate.addn_zero_filter 0.86% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.36% : 0.000006s : 25: predicate.arithmetic_simplify 1.04% : 0.000003s : 17: predicate.cast_eliminate 0.45% : 0.000001s : 8: predicate.check_bprop_eliminate 0.42% : 0.000001s : 8: predicate.compare_switch_simplify 0.15% : 0.000000s : 4: predicate.const_output_eliminate 0.45% : 0.000001s : 8: predicate.depend_value_elim 0.95% : 0.000002s : 17: predicate.dict_get_item_const_eliminator 1.02% : 0.000003s : 17: predicate.dict_get_item_eliminator 0.95% : 0.000002s : 17: predicate.dict_set_item_eliminator 0.93% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.27% : 0.000001s : 4: predicate.elim_not_effective 0.64% : 0.000002s : 4: predicate.elim_shapecalc_of_broadcastargs 1.14% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.06% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.08% : 0.000003s : 21: predicate.environ_get_depend_swap 1.63% : 0.000004s : 29: predicate.environ_get_eliminate 1.13% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.57% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.63% : 0.000006s : 29: predicate.float_depend_g_call 0.41% : 0.000001s : 8: predicate.float_environ_get_switch 0.64% : 0.000002s : 12: predicate.float_tuple_getitem_switch 0.15% : 0.000000s : 4: predicate.fold_const_symbol 0.60% : 0.000001s : 8: predicate.get_grad_eliminate 0.18% : 0.000000s : 4: predicate.graph_param_transform 0.48% : 0.000001s : 8: predicate.incorporate_call 0.40% : 0.000001s : 8: predicate.incorporate_call_switch 6.57% : 0.000016s : 70: predicate.inline 1.00% : 0.000002s : 8: predicate.inline_without_move 0.27% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.76% : 0.000002s : 8: predicate.less_batch_normalization 1.74% : 0.000004s : 31: predicate.list_to_tuple_eliminator_ 2.65% : 0.000006s : 48: predicate.load_eliminater 1.13% : 0.000003s : 4: predicate.loop_unroll_after_grad 2.62% : 0.000006s : 47: predicate.loop_unroll_before_grad 1.57% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.45% : 0.000001s : 8: predicate.merge_addn 0.45% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.44% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.86% : 0.000002s : 17: predicate.minmaximum_grad 1.17% : 0.000003s : 4: predicate.mutable_eliminate 0.26% : 0.000001s : 4: predicate.opt_reshape 0.43% : 0.000001s : 4: predicate.parallel_virtual_node 2.18% : 0.000005s : 29: predicate.partial_defer_inline 1.63% : 0.000004s : 27: predicate.partial_eliminate 1.00% : 0.000002s : 17: predicate.print_const_string_wrapper 0.45% : 0.000001s : 8: predicate.reduce_all_const_elim 1.35% : 0.000003s : 17: predicate.reduce_eliminate 2.59% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.82% : 0.000002s : 8: predicate.remove_not_recompute_node 1.37% : 0.000003s : 31: predicate.replace_applicator 0.38% : 0.000001s : 8: predicate.replace_old_param 0.40% : 0.000001s : 4: predicate.reset_defer_inline 0.98% : 0.000002s : 17: predicate.reshape_eliminate 0.54% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.32% : 0.000001s : 4: predicate.row_tensor_eliminate 0.75% : 0.000002s : 8: predicate.same_eliminate 0.36% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.82% : 0.000002s : 8: predicate.shard_identity_eliminate 0.53% : 0.000001s : 8: predicate.special_op_eliminate 0.60% : 0.000001s : 8: predicate.specialize_transform 0.95% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.64% : 0.000002s : 8: predicate.stack_unstack_eliminate 0.26% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.74% : 0.000004s : 29: predicate.switch_defer_inline 2.08% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.57% : 0.000014s : 88: predicate.switch_simplify 0.94% : 0.000002s : 17: predicate.tile_eliminate 1.15% : 0.000003s : 17: predicate.transpose_eliminate 1.52% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.44% : 0.000004s : 25: predicate.tuple_list_get_item_const_eliminator 1.30% : 0.000003s : 25: predicate.tuple_list_get_item_depend_reorder 3.23% : 0.000008s : 39: predicate.tuple_list_get_item_eliminator 1.36% : 0.000003s : 25: predicate.tuple_list_get_set_item_eliminator 1.93% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 1.82% : 0.000004s : 31: predicate.tuple_to_list_eliminator_ 2.47% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.15% : 0.000008s : 56: predicate.updatestate_useless_node_eliminater 0.40% : 0.000001s : 4: predicate.value_based_eliminate 0.54% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.52% : 0.000001s : 8: predicate.virtual_output_eliminate 0.22% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.33% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000961 15 59.33% : 0.000570s : 7: func_graph_cloner_run.FuncGraphClonerGraph 40.67% : 0.000391s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.050556 192 0.01% : 0.000004s : 1: ForceFp32Comm 7.54% : 0.003812s : 1: add_attr 7.51% : 0.003796s : 1: add_attr_with_inline 0.01% : 0.000004s : 1: add_comm_op_reuse_tag 0.12% : 0.000058s : 1: add_recomputation 0.01% : 0.000005s : 1: assign_add_opt 0.17% : 0.000087s : 1: auto_monad 0.05% : 0.000023s : 1: auto_monad_reorder 0.01% : 0.000005s : 1: begin_end_overlap_inline 0.01% : 0.000006s : 1: bias_add_comm_swap 1.08% : 0.000545s : 1: bootstrap 0.07% : 0.000034s : 1: cconv 0.03% : 0.000017s : 1: comm_op_add_attrs 0.04% : 0.000020s : 1: control_data_broadcast_order 0.02% : 0.000011s : 1: convert_after_rewriter 0.05% : 0.000025s : 1: cse_after_recomputation 0.01% : 0.000005s : 1: dataset_repeat_opt 0.01% : 0.000006s : 1: detach_backward 0.02% : 0.000008s : 1: environ_conv 0.11% : 0.000058s : 1: event_method 0.01% : 0.000005s : 1: full_micro_interleaved_order_control 0.01% : 0.000005s : 1: get_jit_bprop_graph 0.02% : 0.000011s : 1: graph_reusing 0.01% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.01% : 0.000004s : 1: handle_group_info 0.01% : 0.000007s : 1: inline 0.01% : 0.000007s : 1: insert-virtual-dataset 0.01% : 0.000004s : 1: interleave_parallel_branches 0.01% : 0.000004s : 1: interleave_split_concat_branches 0.01% : 0.000006s : 1: label_fine_grained_interleaved_index 0.02% : 0.000008s : 1: label_micro_interleaved_index 0.96% : 0.000485s : 1: loop_unroll 0.01% : 0.000005s : 1: merge_cast_opt 0.01% : 0.000005s : 1: micro_interleaved_order_control 1.33% : 0.000675s : 1: mutable_eliminate 0.01% : 0.000008s : 1: offloading_packed_experts 0.03% : 0.000017s : 1: opt.transform.loop_unroll_optimizer 0.04% : 0.000018s : 1: opt.transform.mutable_eliminate 3.05% : 0.001540s : 78: opt.transform.opt_a 0.06% : 0.000029s : 1: opt.transform.opt_after_cconv 0.05% : 0.000027s : 1: opt.transform.opt_after_jit_grad 0.19% : 0.000095s : 28: opt.transform.opt_b 0.10% : 0.000049s : 2: opt.transform.opt_trans_graph 0.08% : 0.000042s : 4: opt.transform.symbol_engine_opt 7.11% : 0.003593s : 1: opt_a 0.22% : 0.000112s : 1: opt_after_cconv 1.03% : 0.000521s : 1: opt_after_jit_grad 0.41% : 0.000207s : 1: opt_b 11.75% : 0.005943s : 1: optimize 0.04% : 0.000022s : 1: optimize_parallel_all_gather_comm 0.02% : 0.000008s : 1: order_py_execute_after_rewriter 0.07% : 0.000038s : 1: overlap_grad_flash_sp 0.01% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.02% : 0.000008s : 1: overlap_grad_ring_attention 0.01% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.01% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.01% : 0.000005s : 1: overlap_param_gather 0.01% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.02% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.01% : 0.000006s : 1: overlap_recompute_comm 0.02% : 0.000008s : 1: parallel-infer-symbol 0.01% : 0.000004s : 1: parallel-infer-symbol-second 0.01% : 0.000005s : 1: partial_unused_args_eliminate 0.01% : 0.000005s : 1: pipeline_parallel_scheduler 0.01% : 0.000005s : 1: pipeline_split 0.11% : 0.000053s : 1: pre_auto_parallel 0.08% : 0.000040s : 1: py_interpret_to_execute 0.03% : 0.000017s : 1: py_interpret_to_execute_after_opt_a 0.01% : 0.000004s : 1: remove_cast_before_assign_add 0.04% : 0.000018s : 1: remove_dup_value 1.35% : 0.000680s : 1: renormalize.infer 0.89% : 0.000449s : 1: renormalize.specialize 0.01% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.02% : 0.000009s : 1: rewriter_after_jit_bprop_graph 0.09% : 0.000047s : 1: rewriter_after_opt_a 0.23% : 0.000114s : 1: rewriter_before_opt_a 0.01% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.01% : 0.000005s : 1: slice_recompute_activation 0.01% : 0.000005s : 1: split_layernorm_comm 0.01% : 0.000005s : 1: split_matmul_comm_elemetwise 0.02% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.18% : 0.000090s : 1: symbol_engine_optimizer 0.16% : 0.000080s : 1: tuple_transform 53.04% : 0.026813s : 1: type_inference TotalTime = 0.0492903, [21] [bootstrap]: 0.00052531 [type_inference]: 0.00980005 [event_method]: 4.978e-05 [auto_monad]: 8.157e-05 [graph_reusing]: 6.98998e-06 [inline]: 2.89999e-06 [add_attr]: 0.0282127, [1] [add_attr_with_inline]: 0.028196, [1] [Cycle 1]: 0.00010025, [2] [tag_attr]: 4.389e-05 [meta_addattr_fg_expand]: 7.45998e-06 [parallel-infer-symbol]: 3.85998e-06 [pre_auto_parallel]: 0.00313363 [insert-virtual-dataset]: 3.13998e-06 [parallel-infer-symbol-second]: 3.55e-06 [dataset_repeat_opt]: 3.44001e-06 [pipeline_split]: 2.14e-06 [optimize]: 0.00652465, [53] [py_interpret_to_execute]: 6.911e-05 [rewriter_before_opt_a]: 0.0001254 [opt_a]: 0.00384897, [2] [Cycle 1]: 0.00310664, [45] [expand_dump_flag]: 5.81e-06 [switch_simplify]: 5.729e-05 [loop_unroll]: 4.117e-05 [a_1]: 0.00103079 [with_stream_mark]: 3.063e-05 [recompute_prepare]: 1.546e-05 [updatestate_depend_eliminate]: 5.20999e-06 [updatestate_assign_eliminate]: 3.4e-06 [updatestate_loads_eliminate]: 3.32997e-06 [parameter_eliminate]: 2.14e-06 [a_2]: 8.686e-05 [accelerated_algorithm]: 8.37e-06 [shard]: 2.61e-06 [meta_shard_fg_expand]: 2.72001e-06 [shard_inline]: 7.07997e-06 [merge_send_recv]: 9.20999e-06 [auto_parallel]: 1.002e-05 [parallel]: 2.239e-05 [flash_sp]: 1.34e-05 [merge_comm]: 4.12e-06 [allreduce_fusion]: 3.65e-06 [matmul_add_comm_reduction]: 1.101e-05 [allreduce_slice_to_reducescatter]: 8.89995e-07 [virtual_shard_identity]: 1.384e-05 [virtual_dataset]: 6.68998e-06 [get_grad_eliminate_]: 7.79997e-06 [virtual_output]: 7.2e-06 [merge_forward]: 4.67e-06 [cell_reuse_recompute_pass]: 1.77001e-06 [offload_activation]: 1.112e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.523e-05 [merge_recompute_call_nodes]: 1.86998e-06 [before_grad]: 1.238e-05 [set_forward_comm_id_for_comm_node_pass]: 4.51002e-06 [meta_fg_expand]: 3.42002e-06 [flash_sp_send_recv_attached]: 3.39001e-06 [receive_attached]: 2.32999e-06 [after_resolve]: 1.344e-05 [a_after_grad]: 1.009e-05 [renormalize]: 0.00118634 [add_forward_monad_depend]: 6.59001e-06 [auto_monad_grad]: 2.66999e-06 [auto_monad_eliminator]: 1.989e-05 [cse]: 3.091e-05 [a_3]: 5.694e-05 [Cycle 2]: 0.00072317, [45] [expand_dump_flag]: 3.2e-06 [switch_simplify]: 8.72e-06 [loop_unroll]: 6.38e-06 [a_1]: 0.00015049 [with_stream_mark]: 1.848e-05 [recompute_prepare]: 6.97002e-06 [updatestate_depend_eliminate]: 3.28998e-06 [updatestate_assign_eliminate]: 3.10002e-06 [updatestate_loads_eliminate]: 3.28e-06 [parameter_eliminate]: 1.44e-06 [a_2]: 7.279e-05 [accelerated_algorithm]: 6.89999e-06 [shard]: 3.04001e-06 [meta_shard_fg_expand]: 2.54999e-06 [shard_inline]: 6.29001e-06 [merge_send_recv]: 7.55e-06 [auto_parallel]: 8.03999e-06 [parallel]: 7.43999e-06 [flash_sp]: 3.39001e-06 [merge_comm]: 4.13999e-06 [allreduce_fusion]: 4.04997e-06 [matmul_add_comm_reduction]: 8.74e-06 [allreduce_slice_to_reducescatter]: 7.00005e-07 [virtual_shard_identity]: 8.77999e-06 [virtual_dataset]: 5.96e-06 [get_grad_eliminate_]: 5.64e-06 [virtual_output]: 5.74e-06 [merge_forward]: 4.56002e-06 [cell_reuse_recompute_pass]: 2.96001e-06 [offload_activation]: 1.077e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.336e-05 [merge_recompute_call_nodes]: 1.12999e-06 [before_grad]: 9.83998e-06 [set_forward_comm_id_for_comm_node_pass]: 4.01001e-06 [meta_fg_expand]: 2.36e-06 [flash_sp_send_recv_attached]: 1.64e-06 [receive_attached]: 2.11e-06 [after_resolve]: 1.153e-05 [a_after_grad]: 9.76e-06 [renormalize]: 8.9989e-08 [add_forward_monad_depend]: 2.21998e-06 [auto_monad_grad]: 1.66e-06 [auto_monad_eliminator]: 9.37001e-06 [cse]: 1.615e-05 [a_3]: 3.501e-05 [py_interpret_to_execute_after_opt_a]: 1.543e-05 [slice_cell_reuse_recomputed_activation]: 2.21e-06 [rewriter_after_opt_a]: 4.068e-05 [convert_after_rewriter]: 6.59001e-06 [order_py_execute_after_rewriter]: 5.45001e-06 [mutable_eliminate]: 0.00085706 [opt_b]: 0.00027015, [1] [Cycle 1]: 0.00026127, [7] [b_1]: 0.00017804 [b_2]: 8.60001e-06 [updatestate_depend_eliminate]: 6.30002e-06 [updatestate_assign_eliminate]: 2.73e-06 [updatestate_loads_eliminate]: 3.36001e-06 [renormalize]: 3.30008e-07 [cse]: 2.076e-05 [optimize_parallel_all_gather_comm]: 1.905e-05 [overlap_param_gather]: 2.02001e-06 [cconv]: 2.504e-05 [loop_unroll]: 0.00049818 [opt_after_cconv]: 0.00011298, [1] [Cycle 1]: 0.00010465, [7] [c_1]: 3.15e-05 [parameter_eliminate]: 3.68e-06 [updatestate_depend_eliminate]: 6.20002e-06 [updatestate_assign_eliminate]: 2.66e-06 [updatestate_loads_eliminate]: 2.67001e-06 [cse]: 2.101e-05 [renormalize]: 3.7998e-07 [remove_dup_value]: 1.497e-05 [tuple_transform]: 8.876e-05, [1] [Cycle 1]: 8.387e-05, [4] [d_1]: 5.493e-05 [none_parameter_eliminate]: 1.86e-06 [renormalize]: 2.29978e-07 [switch_simplify]: 7.2e-06 [partial_unused_args_eliminate]: 1.99999e-06 [add_recomputation]: 5.725e-05 [cse_after_recomputation]: 2.409e-05, [1] [Cycle 1]: 1.875e-05, [1] [cse]: 1.315e-05 [environ_conv]: 6.41e-06 [swap_dp_allreduce_reducescatter]: 5.19e-06 [bias_add_comm_swap]: 2.89999e-06 [label_micro_interleaved_index]: 6.26e-06 [label_fine_grained_interleaved_index]: 3.40003e-06 [merge_cast_opt]: 1.39998e-06 [slice_recompute_activation]: 2.09e-06 [micro_interleaved_order_control]: 2.69001e-06 [assign_add_opt]: 1.62001e-06 [ForceFp32Comm]: 9.20001e-07 [remove_cast_before_assign_add]: 1.09003e-06 [full_micro_interleaved_order_control]: 2.23002e-06 [reorder_send_recv_between_fp_bp]: 2.76e-06 [comm_op_add_attrs]: 1.32999e-06 [add_comm_op_reuse_tag]: 9.89996e-07 [interleave_split_concat_branches]: 1.25999e-06 [interleave_parallel_branches]: 1.14e-06 [overlap_opt_shard_in_pipeline]: 1.22e-06 [overlap_opt_shard_grad_in_pipeline]: 1.74998e-06 [control_data_broadcast_order]: 1.425e-05 [grouped_pairwise_exchange_alltoall]: 1.60001e-06 [offloading_packed_experts]: 4.75001e-06 [overlap_recompute_and_grad_model_parallel]: 4.88001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.30999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.37e-06 [overlap_recompute_comm]: 2.24999e-06 [overlap_grad_ring_attention]: 4.57998e-06 [overlap_grad_flash_sp]: 2.153e-05 [begin_end_overlap_inline]: 5.3001e-07 [split_matmul_comm_elemetwise]: 2.29001e-06 [split_layernorm_comm]: 2.21e-06 [handle_group_info]: 1.20001e-06 [symbol_engine_optimizer]: 8.644e-05, [1] [Cycle 1]: 8.044e-05, [6] [build]: 3.61999e-06 [elim_shapecalc]: 1.264e-05 [elim_not_effective]: 1.405e-05 [opt_reshape]: 7.14001e-06 [fold_const_symbol]: 9.84999e-06 [renormalize]: 2.30008e-07 [detach_backward]: 2.65002e-06 [pipeline_parallel_scheduler]: 1.50001e-06 [auto_monad_reorder]: 1.885e-05 [get_jit_bprop_graph]: 2.73998e-06 [rewriter_after_jit_bprop_graph]: 5.49e-06 [opt_after_jit_grad]: 0.0006277 [validate]: 5.085e-05 Sums bootstrap : 0.000525s : 2.63% type_inference : 0.009800s : 49.13% event_method : 0.000050s : 0.25% auto_monad : 0.000082s : 0.41% graph_reusing : 0.000007s : 0.04% inline : 0.000003s : 0.01% add_attr.add_attr_with_inline.tag_attr : 0.000044s : 0.22% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.04% parallel-infer-symbol : 0.000004s : 0.02% pre_auto_parallel : 0.003134s : 15.71% insert-virtual-dataset : 0.000003s : 0.02% parallel-infer-symbol-second : 0.000004s : 0.02% dataset_repeat_opt : 0.000003s : 0.02% pipeline_split : 0.000002s : 0.01% optimize.py_interpret_to_execute : 0.000069s : 0.35% optimize.rewriter_before_opt_a : 0.000125s : 0.63% optimize.opt_a.expand_dump_flag : 0.000009s : 0.05% optimize.opt_a.switch_simplify : 0.000066s : 0.33% optimize.opt_a.loop_unroll : 0.000048s : 0.24% optimize.opt_a.a_1 : 0.001181s : 5.92% optimize.opt_a.with_stream_mark : 0.000049s : 0.25% optimize.opt_a.recompute_prepare : 0.000022s : 0.11% optimize.opt_a.updatestate_depend_eliminate : 0.000008s : 0.04% optimize.opt_a.updatestate_assign_eliminate : 0.000007s : 0.03% optimize.opt_a.updatestate_loads_eliminate : 0.000007s : 0.03% optimize.opt_a.parameter_eliminate : 0.000004s : 0.02% optimize.opt_a.a_2 : 0.000160s : 0.80% optimize.opt_a.accelerated_algorithm : 0.000015s : 0.08% optimize.opt_a.shard : 0.000006s : 0.03% optimize.opt_a.meta_shard_fg_expand : 0.000005s : 0.03% optimize.opt_a.shard_inline : 0.000013s : 0.07% optimize.opt_a.merge_send_recv : 0.000017s : 0.08% optimize.opt_a.auto_parallel : 0.000018s : 0.09% optimize.opt_a.parallel : 0.000030s : 0.15% optimize.opt_a.flash_sp : 0.000017s : 0.08% optimize.opt_a.merge_comm : 0.000008s : 0.04% optimize.opt_a.allreduce_fusion : 0.000008s : 0.04% optimize.opt_a.matmul_add_comm_reduction : 0.000020s : 0.10% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.01% optimize.opt_a.virtual_shard_identity : 0.000023s : 0.11% optimize.opt_a.virtual_dataset : 0.000013s : 0.06% optimize.opt_a.get_grad_eliminate_ : 0.000013s : 0.07% optimize.opt_a.virtual_output : 0.000013s : 0.06% optimize.opt_a.merge_forward : 0.000009s : 0.05% optimize.opt_a.cell_reuse_recompute_pass : 0.000005s : 0.02% optimize.opt_a.offload_activation : 0.000022s : 0.11% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000029s : 0.14% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.02% optimize.opt_a.before_grad : 0.000022s : 0.11% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000009s : 0.04% optimize.opt_a.meta_fg_expand : 0.000006s : 0.03% optimize.opt_a.flash_sp_send_recv_attached : 0.000005s : 0.03% optimize.opt_a.receive_attached : 0.000004s : 0.02% optimize.opt_a.after_resolve : 0.000025s : 0.13% optimize.opt_a.a_after_grad : 0.000020s : 0.10% optimize.opt_a.renormalize : 0.001186s : 5.95% optimize.opt_a.add_forward_monad_depend : 0.000009s : 0.04% optimize.opt_a.auto_monad_grad : 0.000004s : 0.02% optimize.opt_a.auto_monad_eliminator : 0.000029s : 0.15% optimize.opt_a.cse : 0.000047s : 0.24% optimize.opt_a.a_3 : 0.000092s : 0.46% optimize.py_interpret_to_execute_after_opt_a : 0.000015s : 0.08% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.01% optimize.rewriter_after_opt_a : 0.000041s : 0.20% optimize.convert_after_rewriter : 0.000007s : 0.03% optimize.order_py_execute_after_rewriter : 0.000005s : 0.03% optimize.mutable_eliminate : 0.000857s : 4.30% optimize.opt_b.b_1 : 0.000178s : 0.89% optimize.opt_b.b_2 : 0.000009s : 0.04% optimize.opt_b.updatestate_depend_eliminate : 0.000006s : 0.03% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.02% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000021s : 0.10% optimize.optimize_parallel_all_gather_comm : 0.000019s : 0.10% optimize.overlap_param_gather : 0.000002s : 0.01% optimize.cconv : 0.000025s : 0.13% optimize.loop_unroll : 0.000498s : 2.50% optimize.opt_after_cconv.c_1 : 0.000031s : 0.16% optimize.opt_after_cconv.parameter_eliminate : 0.000004s : 0.02% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.03% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.cse : 0.000021s : 0.11% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000015s : 0.08% optimize.tuple_transform.d_1 : 0.000055s : 0.28% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.01% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.04% optimize.partial_unused_args_eliminate : 0.000002s : 0.01% optimize.add_recomputation : 0.000057s : 0.29% optimize.cse_after_recomputation.cse : 0.000013s : 0.07% optimize.environ_conv : 0.000006s : 0.03% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.03% optimize.bias_add_comm_swap : 0.000003s : 0.01% optimize.label_micro_interleaved_index : 0.000006s : 0.03% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.02% optimize.merge_cast_opt : 0.000001s : 0.01% optimize.slice_recompute_activation : 0.000002s : 0.01% optimize.micro_interleaved_order_control : 0.000003s : 0.01% optimize.assign_add_opt : 0.000002s : 0.01% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.01% optimize.full_micro_interleaved_order_control : 0.000002s : 0.01% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.01% optimize.comm_op_add_attrs : 0.000001s : 0.01% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.01% optimize.interleave_parallel_branches : 0.000001s : 0.01% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.01% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.01% optimize.control_data_broadcast_order : 0.000014s : 0.07% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.01% optimize.offloading_packed_experts : 0.000005s : 0.02% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.02% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.01% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.01% optimize.overlap_recompute_comm : 0.000002s : 0.01% optimize.overlap_grad_ring_attention : 0.000005s : 0.02% optimize.overlap_grad_flash_sp : 0.000022s : 0.11% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.01% optimize.split_layernorm_comm : 0.000002s : 0.01% optimize.handle_group_info : 0.000001s : 0.01% optimize.symbol_engine_optimizer.build : 0.000004s : 0.02% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000013s : 0.06% optimize.symbol_engine_optimizer.elim_not_effective : 0.000014s : 0.07% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.04% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.05% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.01% pipeline_parallel_scheduler : 0.000002s : 0.01% auto_monad_reorder : 0.000019s : 0.09% get_jit_bprop_graph : 0.000003s : 0.01% rewriter_after_jit_bprop_graph : 0.000005s : 0.03% opt_after_jit_grad : 0.000628s : 3.15% validate : 0.000051s : 0.25% Time group info: ------[substitution.] 0.000375 40 12.00% : 0.000045s : 8: substitution.arithmetic_simplify 0.56% : 0.000002s : 2: substitution.elim_not_effective 0.34% : 0.000001s : 2: substitution.fold_const_symbol 1.78% : 0.000007s : 4: substitution.graph_param_transform 74.07% : 0.000277s : 6: substitution.inline 1.24% : 0.000005s : 4: substitution.j_node_and_user_rematch 1.58% : 0.000006s : 4: substitution.remove_not_recompute_node 1.50% : 0.000006s : 4: substitution.replace_old_param 6.93% : 0.000026s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.009721 2 88.77% : 0.008630s : 1: type_inference.infer 11.23% : 0.001091s : 1: type_inference.specialize ------[replace.] 0.000106 12 60.69% : 0.000064s : 6: replace.inline 39.31% : 0.000041s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000295 12 92.11% : 0.000272s : 6: match.inline 7.89% : 0.000023s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000259 1534 1.02% : 0.000003s : 17: predicate.accumulaten_eliminater 0.63% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.43% : 0.000001s : 8: predicate.addn_check_dump 0.94% : 0.000002s : 17: predicate.addn_zero_filter 0.90% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.44% : 0.000006s : 25: predicate.arithmetic_simplify 1.19% : 0.000003s : 17: predicate.cast_eliminate 0.55% : 0.000001s : 8: predicate.check_bprop_eliminate 0.43% : 0.000001s : 8: predicate.compare_switch_simplify 0.16% : 0.000000s : 4: predicate.const_output_eliminate 0.51% : 0.000001s : 8: predicate.depend_value_elim 0.99% : 0.000003s : 17: predicate.dict_get_item_const_eliminator 1.13% : 0.000003s : 17: predicate.dict_get_item_eliminator 0.92% : 0.000002s : 17: predicate.dict_set_item_eliminator 0.96% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.20% : 0.000001s : 4: predicate.elim_not_effective 0.26% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.25% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.06% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.01% : 0.000003s : 21: predicate.environ_get_depend_swap 1.65% : 0.000004s : 29: predicate.environ_get_eliminate 1.10% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.60% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.73% : 0.000007s : 29: predicate.float_depend_g_call 0.37% : 0.000001s : 8: predicate.float_environ_get_switch 0.60% : 0.000002s : 12: predicate.float_tuple_getitem_switch 0.14% : 0.000000s : 4: predicate.fold_const_symbol 0.61% : 0.000002s : 8: predicate.get_grad_eliminate 0.17% : 0.000000s : 4: predicate.graph_param_transform 0.44% : 0.000001s : 8: predicate.incorporate_call 0.39% : 0.000001s : 8: predicate.incorporate_call_switch 6.40% : 0.000017s : 70: predicate.inline 0.72% : 0.000002s : 8: predicate.inline_without_move 0.26% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.75% : 0.000002s : 8: predicate.less_batch_normalization 1.83% : 0.000005s : 31: predicate.list_to_tuple_eliminator_ 2.58% : 0.000007s : 48: predicate.load_eliminater 0.93% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.47% : 0.000006s : 47: predicate.loop_unroll_before_grad 1.79% : 0.000005s : 25: predicate.make_slice_get_slice_eliminator 0.46% : 0.000001s : 8: predicate.merge_addn 0.57% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.43% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.81% : 0.000002s : 17: predicate.minmaximum_grad 1.06% : 0.000003s : 4: predicate.mutable_eliminate 0.28% : 0.000001s : 4: predicate.opt_reshape 0.32% : 0.000001s : 4: predicate.parallel_virtual_node 2.31% : 0.000006s : 29: predicate.partial_defer_inline 1.56% : 0.000004s : 27: predicate.partial_eliminate 0.89% : 0.000002s : 17: predicate.print_const_string_wrapper 0.49% : 0.000001s : 8: predicate.reduce_all_const_elim 1.56% : 0.000004s : 17: predicate.reduce_eliminate 2.46% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.38% : 0.000001s : 8: predicate.remove_not_recompute_node 1.50% : 0.000004s : 31: predicate.replace_applicator 0.45% : 0.000001s : 8: predicate.replace_old_param 0.38% : 0.000001s : 4: predicate.reset_defer_inline 1.02% : 0.000003s : 17: predicate.reshape_eliminate 0.65% : 0.000002s : 8: predicate.row_tensor_add_zeros_like 0.39% : 0.000001s : 4: predicate.row_tensor_eliminate 0.67% : 0.000002s : 8: predicate.same_eliminate 0.34% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.82% : 0.000002s : 8: predicate.shard_identity_eliminate 0.58% : 0.000002s : 8: predicate.special_op_eliminate 0.61% : 0.000002s : 8: predicate.specialize_transform 0.85% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.70% : 0.000002s : 8: predicate.stack_unstack_eliminate 0.31% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.71% : 0.000004s : 29: predicate.switch_defer_inline 2.05% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.30% : 0.000014s : 88: predicate.switch_simplify 1.04% : 0.000003s : 17: predicate.tile_eliminate 1.04% : 0.000003s : 17: predicate.transpose_eliminate 1.90% : 0.000005s : 25: predicate.tuple_list_convert_item_index_to_positive 1.85% : 0.000005s : 25: predicate.tuple_list_get_item_const_eliminator 1.54% : 0.000004s : 25: predicate.tuple_list_get_item_depend_reorder 3.60% : 0.000009s : 39: predicate.tuple_list_get_item_eliminator 1.38% : 0.000004s : 25: predicate.tuple_list_get_set_item_eliminator 2.07% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 1.84% : 0.000005s : 31: predicate.tuple_to_list_eliminator_ 2.32% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 2.97% : 0.000008s : 56: predicate.updatestate_useless_node_eliminater 0.38% : 0.000001s : 4: predicate.value_based_eliminate 0.61% : 0.000002s : 8: predicate.virtual_dataset_eliminate 0.53% : 0.000001s : 8: predicate.virtual_output_eliminate 0.21% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.31% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000988 15 57.09% : 0.000564s : 7: func_graph_cloner_run.FuncGraphClonerGraph 42.91% : 0.000424s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.087029 192 0.00% : 0.000004s : 1: ForceFp32Comm 32.43% : 0.028220s : 1: add_attr 32.40% : 0.028201s : 1: add_attr_with_inline 0.01% : 0.000004s : 1: add_comm_op_reuse_tag 0.07% : 0.000062s : 1: add_recomputation 0.01% : 0.000004s : 1: assign_add_opt 0.10% : 0.000089s : 1: auto_monad 0.03% : 0.000023s : 1: auto_monad_reorder 0.00% : 0.000003s : 1: begin_end_overlap_inline 0.01% : 0.000006s : 1: bias_add_comm_swap 0.64% : 0.000555s : 1: bootstrap 0.03% : 0.000028s : 1: cconv 0.01% : 0.000005s : 1: comm_op_add_attrs 0.02% : 0.000018s : 1: control_data_broadcast_order 0.01% : 0.000010s : 1: convert_after_rewriter 0.03% : 0.000027s : 1: cse_after_recomputation 0.01% : 0.000007s : 1: dataset_repeat_opt 0.01% : 0.000006s : 1: detach_backward 0.01% : 0.000010s : 1: environ_conv 0.07% : 0.000058s : 1: event_method 0.01% : 0.000005s : 1: full_micro_interleaved_order_control 0.01% : 0.000006s : 1: get_jit_bprop_graph 0.01% : 0.000011s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.01% : 0.000006s : 1: inline 0.01% : 0.000011s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.01% : 0.000006s : 1: label_fine_grained_interleaved_index 0.01% : 0.000010s : 1: label_micro_interleaved_index 0.58% : 0.000507s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.01% : 0.000006s : 1: micro_interleaved_order_control 1.00% : 0.000867s : 1: mutable_eliminate 0.01% : 0.000008s : 1: offloading_packed_experts 0.02% : 0.000017s : 1: opt.transform.loop_unroll_optimizer 0.02% : 0.000019s : 1: opt.transform.mutable_eliminate 1.90% : 0.001656s : 78: opt.transform.opt_a 0.03% : 0.000030s : 1: opt.transform.opt_after_cconv 0.03% : 0.000029s : 1: opt.transform.opt_after_jit_grad 0.17% : 0.000151s : 28: opt.transform.opt_b 0.06% : 0.000052s : 2: opt.transform.opt_trans_graph 0.05% : 0.000040s : 4: opt.transform.symbol_engine_opt 4.43% : 0.003853s : 1: opt_a 0.13% : 0.000117s : 1: opt_after_cconv 0.74% : 0.000640s : 1: opt_after_jit_grad 0.31% : 0.000274s : 1: opt_b 7.50% : 0.006530s : 1: optimize 0.03% : 0.000023s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000009s : 1: order_py_execute_after_rewriter 0.03% : 0.000026s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000007s : 1: overlap_grad_ring_attention 0.01% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.01% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.01% : 0.000005s : 1: overlap_recompute_comm 0.01% : 0.000008s : 1: parallel-infer-symbol 0.01% : 0.000007s : 1: parallel-infer-symbol-second 0.01% : 0.000005s : 1: partial_unused_args_eliminate 0.01% : 0.000005s : 1: pipeline_parallel_scheduler 0.01% : 0.000005s : 1: pipeline_split 3.65% : 0.003176s : 1: pre_auto_parallel 0.09% : 0.000075s : 1: py_interpret_to_execute 0.02% : 0.000019s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.02% : 0.000018s : 1: remove_dup_value 0.67% : 0.000587s : 1: renormalize.infer 0.67% : 0.000585s : 1: renormalize.specialize 0.01% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000009s : 1: rewriter_after_jit_bprop_graph 0.05% : 0.000045s : 1: rewriter_after_opt_a 0.15% : 0.000131s : 1: rewriter_before_opt_a 0.01% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.01% : 0.000006s : 1: slice_recompute_activation 0.01% : 0.000006s : 1: split_layernorm_comm 0.01% : 0.000005s : 1: split_matmul_comm_elemetwise 0.01% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.10% : 0.000089s : 1: symbol_engine_optimizer 0.11% : 0.000092s : 1: tuple_transform 11.29% : 0.009821s : 1: type_inference TotalTime = 0.140261, [21] [bootstrap]: 0.0366212 [type_inference]: 0.0546324 [event_method]: 6.763e-05 [auto_monad]: 0.00013578 [graph_reusing]: 9.05999e-06 [inline]: 3.30998e-06 [add_attr]: 0.00389307, [1] [add_attr_with_inline]: 0.00388018, [1] [Cycle 1]: 0.0001079, [2] [tag_attr]: 4.809e-05 [meta_addattr_fg_expand]: 1.049e-05 [parallel-infer-symbol]: 3.95e-06 [pre_auto_parallel]: 6.568e-05 [insert-virtual-dataset]: 2.37999e-06 [parallel-infer-symbol-second]: 1.00001e-06 [dataset_repeat_opt]: 2.19999e-06 [pipeline_split]: 1.68002e-06 [optimize]: 0.0439657, [53] [py_interpret_to_execute]: 4.835e-05 [rewriter_before_opt_a]: 0.00018417 [opt_a]: 0.041504, [3] [Cycle 1]: 0.0375537, [45] [expand_dump_flag]: 4.94e-06 [switch_simplify]: 8.308e-05 [loop_unroll]: 6.63e-05 [a_1]: 0.0299718 [with_stream_mark]: 3.143e-05 [recompute_prepare]: 2.806e-05 [updatestate_depend_eliminate]: 1.021e-05 [updatestate_assign_eliminate]: 7.7e-06 [updatestate_loads_eliminate]: 6.81001e-06 [parameter_eliminate]: 2.18002e-06 [a_2]: 0.00025674 [accelerated_algorithm]: 3.524e-05 [shard]: 1.60999e-06 [meta_shard_fg_expand]: 5.29998e-06 [shard_inline]: 1.661e-05 [merge_send_recv]: 1.701e-05 [auto_parallel]: 1.549e-05 [parallel]: 1.695e-05 [flash_sp]: 1.115e-05 [merge_comm]: 1.112e-05 [allreduce_fusion]: 9.94999e-06 [matmul_add_comm_reduction]: 3.199e-05 [allreduce_slice_to_reducescatter]: 9.80013e-07 [virtual_shard_identity]: 2.015e-05 [virtual_dataset]: 1.715e-05 [get_grad_eliminate_]: 1.655e-05 [virtual_output]: 1.577e-05 [merge_forward]: 1.009e-05 [cell_reuse_recompute_pass]: 1.62999e-06 [offload_activation]: 1.849e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.302e-05 [merge_recompute_call_nodes]: 9.20001e-07 [before_grad]: 2.944e-05 [set_forward_comm_id_for_comm_node_pass]: 1.126e-05 [meta_fg_expand]: 0.00197913 [flash_sp_send_recv_attached]: 7.21001e-06 [receive_attached]: 2.95002e-06 [after_resolve]: 7.179e-05 [a_after_grad]: 9.356e-05 [renormalize]: 0.00355403 [add_forward_monad_depend]: 1.445e-05 [auto_monad_grad]: 6.66e-06 [auto_monad_eliminator]: 6.316e-05 [cse]: 0.00022263 [a_3]: 0.00035738 [Cycle 2]: 0.00328207, [45] [expand_dump_flag]: 3.25e-06 [switch_simplify]: 4.992e-05 [loop_unroll]: 4.52e-05 [a_1]: 0.00167822 [with_stream_mark]: 1.881e-05 [recompute_prepare]: 1.131e-05 [updatestate_depend_eliminate]: 4.2e-06 [updatestate_assign_eliminate]: 3.75e-06 [updatestate_loads_eliminate]: 2.63e-06 [parameter_eliminate]: 1.92999e-06 [a_2]: 7.252e-05 [accelerated_algorithm]: 1.277e-05 [shard]: 1.97999e-06 [meta_shard_fg_expand]: 2.69001e-06 [shard_inline]: 6.21e-06 [merge_send_recv]: 9.32999e-06 [auto_parallel]: 1.089e-05 [parallel]: 9.08002e-06 [flash_sp]: 3.80003e-06 [merge_comm]: 3.68e-06 [allreduce_fusion]: 3.25e-06 [matmul_add_comm_reduction]: 9.74999e-06 [allreduce_slice_to_reducescatter]: 6.69999e-07 [virtual_shard_identity]: 8.32e-06 [virtual_dataset]: 6.20002e-06 [get_grad_eliminate_]: 6.39001e-06 [virtual_output]: 5.66e-06 [merge_forward]: 4.1e-06 [cell_reuse_recompute_pass]: 1.61998e-06 [offload_activation]: 1.023e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.191e-05 [merge_recompute_call_nodes]: 1.32e-06 [before_grad]: 1.066e-05 [set_forward_comm_id_for_comm_node_pass]: 3.57997e-06 [meta_fg_expand]: 0.000117 [flash_sp_send_recv_attached]: 2.10002e-06 [receive_attached]: 2.46e-06 [after_resolve]: 1.879e-05 [a_after_grad]: 1.111e-05 [renormalize]: 0.00072989 [add_forward_monad_depend]: 6.73998e-06 [auto_monad_grad]: 2.09999e-06 [auto_monad_eliminator]: 1.648e-05 [cse]: 2.449e-05 [a_3]: 4.608e-05 [Cycle 3]: 0.00064635, [45] [expand_dump_flag]: 2.37001e-06 [switch_simplify]: 7.89002e-06 [loop_unroll]: 6.00002e-06 [a_1]: 0.00012327 [with_stream_mark]: 1.26e-05 [recompute_prepare]: 6.25002e-06 [updatestate_depend_eliminate]: 3.28e-06 [updatestate_assign_eliminate]: 2.61e-06 [updatestate_loads_eliminate]: 2.51998e-06 [parameter_eliminate]: 1.29e-06 [a_2]: 6.69e-05 [accelerated_algorithm]: 1.465e-05 [shard]: 1.08001e-06 [meta_shard_fg_expand]: 1.64998e-06 [shard_inline]: 6.69999e-06 [merge_send_recv]: 5.12e-06 [auto_parallel]: 6.46e-06 [parallel]: 5.87999e-06 [flash_sp]: 1.19998e-06 [merge_comm]: 3.16001e-06 [allreduce_fusion]: 3.14001e-06 [matmul_add_comm_reduction]: 6.53e-06 [allreduce_slice_to_reducescatter]: 3.60014e-07 [virtual_shard_identity]: 8.42e-06 [virtual_dataset]: 5.76e-06 [get_grad_eliminate_]: 5.52001e-06 [virtual_output]: 5.44e-06 [merge_forward]: 4.1e-06 [cell_reuse_recompute_pass]: 2.29001e-06 [offload_activation]: 7.68999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.216e-05 [merge_recompute_call_nodes]: 6.50005e-07 [before_grad]: 9.45001e-06 [set_forward_comm_id_for_comm_node_pass]: 3.3e-06 [meta_fg_expand]: 1.91e-06 [flash_sp_send_recv_attached]: 1.22e-06 [receive_attached]: 1.79998e-06 [after_resolve]: 1.064e-05 [a_after_grad]: 8.48001e-06 [renormalize]: 1.00001e-07 [add_forward_monad_depend]: 1.87001e-06 [auto_monad_grad]: 1.72001e-06 [auto_monad_eliminator]: 9.91e-06 [cse]: 1.499e-05 [a_3]: 3.4e-05 [py_interpret_to_execute_after_opt_a]: 1.317e-05 [slice_cell_reuse_recomputed_activation]: 2.09e-06 [rewriter_after_opt_a]: 4.03e-05 [convert_after_rewriter]: 7.63001e-06 [order_py_execute_after_rewriter]: 5.57999e-06 [mutable_eliminate]: 0.00071693 [opt_b]: 0.00020378, [1] [Cycle 1]: 0.00019621, [7] [b_1]: 0.00011826 [b_2]: 9.15999e-06 [updatestate_depend_eliminate]: 7.15e-06 [updatestate_assign_eliminate]: 2.62001e-06 [updatestate_loads_eliminate]: 2.13002e-06 [renormalize]: 4.89992e-07 [cse]: 2.091e-05 [optimize_parallel_all_gather_comm]: 1.681e-05 [overlap_param_gather]: 1.68002e-06 [cconv]: 2.802e-05 [loop_unroll]: 0.00047716 [opt_after_cconv]: 0.00011026, [1] [Cycle 1]: 0.00010397, [7] [c_1]: 3.293e-05 [parameter_eliminate]: 3.29001e-06 [updatestate_depend_eliminate]: 7.43999e-06 [updatestate_assign_eliminate]: 2.79001e-06 [updatestate_loads_eliminate]: 2.34999e-06 [cse]: 1.979e-05 [renormalize]: 2.59985e-07 [remove_dup_value]: 1.385e-05 [tuple_transform]: 7.881e-05, [1] [Cycle 1]: 7.44e-05, [4] [d_1]: 4.548e-05 [none_parameter_eliminate]: 1.84e-06 [renormalize]: 1.39989e-07 [switch_simplify]: 6.93e-06 [partial_unused_args_eliminate]: 1.72001e-06 [add_recomputation]: 4.917e-05 [cse_after_recomputation]: 2.234e-05, [1] [Cycle 1]: 1.749e-05, [1] [cse]: 1.193e-05 [environ_conv]: 9.46e-06 [swap_dp_allreduce_reducescatter]: 5.14e-06 [bias_add_comm_swap]: 2.71e-06 [label_micro_interleaved_index]: 4.34002e-06 [label_fine_grained_interleaved_index]: 2.64001e-06 [merge_cast_opt]: 1.30001e-06 [slice_recompute_activation]: 2.22001e-06 [micro_interleaved_order_control]: 2.37001e-06 [assign_add_opt]: 1.17999e-06 [ForceFp32Comm]: 9.00007e-07 [remove_cast_before_assign_add]: 1.00001e-06 [full_micro_interleaved_order_control]: 2.66e-06 [reorder_send_recv_between_fp_bp]: 2.44999e-06 [comm_op_add_attrs]: 1.05999e-06 [add_comm_op_reuse_tag]: 9.89996e-07 [interleave_split_concat_branches]: 1.11997e-06 [interleave_parallel_branches]: 1.02998e-06 [overlap_opt_shard_in_pipeline]: 1.37999e-06 [overlap_opt_shard_grad_in_pipeline]: 1.96e-06 [control_data_broadcast_order]: 1.402e-05 [grouped_pairwise_exchange_alltoall]: 1.36002e-06 [offloading_packed_experts]: 4.60001e-06 [overlap_recompute_and_grad_model_parallel]: 4.68001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.25999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.32e-06 [overlap_recompute_comm]: 2.46e-06 [overlap_grad_ring_attention]: 4.67e-06 [overlap_grad_flash_sp]: 2.05e-05 [begin_end_overlap_inline]: 5.79981e-07 [split_matmul_comm_elemetwise]: 2.12001e-06 [split_layernorm_comm]: 1.69998e-06 [handle_group_info]: 1.05999e-06 [symbol_engine_optimizer]: 9.392e-05, [1] [Cycle 1]: 8.883e-05, [6] [build]: 1.126e-05 [elim_shapecalc]: 1.461e-05 [elim_not_effective]: 1.383e-05 [opt_reshape]: 7.08998e-06 [fold_const_symbol]: 9.20001e-06 [renormalize]: 2.09984e-07 [detach_backward]: 2.49001e-06 [pipeline_parallel_scheduler]: 1.51002e-06 [auto_monad_reorder]: 2.133e-05 [get_jit_bprop_graph]: 2.31e-06 [rewriter_after_jit_bprop_graph]: 4.73001e-06 [opt_after_jit_grad]: 0.00058089 [validate]: 4.606e-05 Sums bootstrap : 0.036621s : 27.13% type_inference : 0.054632s : 40.48% event_method : 0.000068s : 0.05% auto_monad : 0.000136s : 0.10% graph_reusing : 0.000009s : 0.01% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000048s : 0.04% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000010s : 0.01% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000066s : 0.05% insert-virtual-dataset : 0.000002s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000048s : 0.04% optimize.rewriter_before_opt_a : 0.000184s : 0.14% optimize.opt_a.expand_dump_flag : 0.000011s : 0.01% optimize.opt_a.switch_simplify : 0.000141s : 0.10% optimize.opt_a.loop_unroll : 0.000118s : 0.09% optimize.opt_a.a_1 : 0.031773s : 23.54% optimize.opt_a.with_stream_mark : 0.000063s : 0.05% optimize.opt_a.recompute_prepare : 0.000046s : 0.03% optimize.opt_a.updatestate_depend_eliminate : 0.000018s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000014s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000012s : 0.01% optimize.opt_a.parameter_eliminate : 0.000005s : 0.00% optimize.opt_a.a_2 : 0.000396s : 0.29% optimize.opt_a.accelerated_algorithm : 0.000063s : 0.05% optimize.opt_a.shard : 0.000005s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000010s : 0.01% optimize.opt_a.shard_inline : 0.000030s : 0.02% optimize.opt_a.merge_send_recv : 0.000031s : 0.02% optimize.opt_a.auto_parallel : 0.000033s : 0.02% optimize.opt_a.parallel : 0.000032s : 0.02% optimize.opt_a.flash_sp : 0.000016s : 0.01% optimize.opt_a.merge_comm : 0.000018s : 0.01% optimize.opt_a.allreduce_fusion : 0.000016s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000048s : 0.04% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000037s : 0.03% optimize.opt_a.virtual_dataset : 0.000029s : 0.02% optimize.opt_a.get_grad_eliminate_ : 0.000028s : 0.02% optimize.opt_a.virtual_output : 0.000027s : 0.02% optimize.opt_a.merge_forward : 0.000018s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000006s : 0.00% optimize.opt_a.offload_activation : 0.000036s : 0.03% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000057s : 0.04% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.00% optimize.opt_a.before_grad : 0.000050s : 0.04% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000018s : 0.01% optimize.opt_a.meta_fg_expand : 0.002098s : 1.55% optimize.opt_a.flash_sp_send_recv_attached : 0.000011s : 0.01% optimize.opt_a.receive_attached : 0.000007s : 0.01% optimize.opt_a.after_resolve : 0.000101s : 0.07% optimize.opt_a.a_after_grad : 0.000113s : 0.08% optimize.opt_a.renormalize : 0.004284s : 3.17% optimize.opt_a.add_forward_monad_depend : 0.000023s : 0.02% optimize.opt_a.auto_monad_grad : 0.000010s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000090s : 0.07% optimize.opt_a.cse : 0.000262s : 0.19% optimize.opt_a.a_3 : 0.000437s : 0.32% optimize.py_interpret_to_execute_after_opt_a : 0.000013s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000040s : 0.03% optimize.convert_after_rewriter : 0.000008s : 0.01% optimize.order_py_execute_after_rewriter : 0.000006s : 0.00% optimize.mutable_eliminate : 0.000717s : 0.53% optimize.opt_b.b_1 : 0.000118s : 0.09% optimize.opt_b.b_2 : 0.000009s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000007s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000021s : 0.02% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.01% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000028s : 0.02% optimize.loop_unroll : 0.000477s : 0.35% optimize.opt_after_cconv.c_1 : 0.000033s : 0.02% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000007s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000020s : 0.01% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000014s : 0.01% optimize.tuple_transform.d_1 : 0.000045s : 0.03% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000049s : 0.04% optimize.cse_after_recomputation.cse : 0.000012s : 0.01% optimize.environ_conv : 0.000009s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.00% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000004s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000003s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000002s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000014s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000001s : 0.00% optimize.offloading_packed_experts : 0.000005s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000005s : 0.00% optimize.overlap_grad_flash_sp : 0.000021s : 0.02% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000011s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000015s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000014s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.01% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000009s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000021s : 0.02% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000005s : 0.00% opt_after_jit_grad : 0.000581s : 0.43% validate : 0.000046s : 0.03% Time group info: ------[substitution.] 0.001038 190 8.58% : 0.000089s : 6: substitution.arithmetic_simplify 0.23% : 0.000002s : 2: substitution.elim_not_effective 0.60% : 0.000006s : 6: substitution.float_depend_g_call 0.48% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.12% : 0.000001s : 2: substitution.fold_const_symbol 0.63% : 0.000007s : 4: substitution.graph_param_transform 0.37% : 0.000004s : 2: substitution.incorporate_call 0.19% : 0.000002s : 2: substitution.incorporate_call_switch 58.85% : 0.000611s : 16: substitution.inline 2.24% : 0.000023s : 2: substitution.inline_without_move 0.99% : 0.000010s : 14: substitution.j_node_and_user_rematch 1.89% : 0.000020s : 3: substitution.less_batch_normalization 1.51% : 0.000016s : 11: substitution.minmaximum_grad 0.68% : 0.000007s : 6: substitution.partial_eliminate 1.13% : 0.000012s : 14: substitution.remove_not_recompute_node 3.33% : 0.000035s : 10: substitution.replace_applicator 1.12% : 0.000012s : 12: substitution.replace_old_param 0.34% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.13% : 0.000032s : 11: substitution.tuple_list_convert_item_index_to_positive 1.47% : 0.000015s : 11: substitution.tuple_list_get_item_const_eliminator 2.05% : 0.000021s : 11: substitution.tuple_list_get_item_depend_reorder 8.07% : 0.000084s : 30: substitution.tuple_list_get_item_eliminator 2.01% : 0.000021s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.054526 2 95.97% : 0.052330s : 1: type_inference.infer 4.03% : 0.002196s : 1: type_inference.specialize ------[replace.] 0.000300 34 4.14% : 0.000012s : 2: replace.arithmetic_simplify 55.83% : 0.000168s : 16: replace.inline 40.03% : 0.000120s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.000694 34 7.36% : 0.000051s : 2: match.arithmetic_simplify 86.29% : 0.000598s : 16: match.inline 6.35% : 0.000044s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.028930 4766 0.03% : 0.000008s : 58: predicate.accumulaten_eliminater 0.01% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.01% : 0.000003s : 23: predicate.addn_check_dump 0.03% : 0.000008s : 58: predicate.addn_zero_filter 0.03% : 0.000008s : 58: predicate.adjust_all_reduce_mul_add 0.05% : 0.000015s : 83: predicate.arithmetic_simplify 0.03% : 0.000008s : 60: predicate.cast_eliminate 0.03% : 0.000008s : 55: predicate.check_bprop_eliminate 0.01% : 0.000003s : 23: predicate.compare_switch_simplify 0.00% : 0.000000s : 4: predicate.const_output_eliminate 0.01% : 0.000003s : 23: predicate.depend_value_elim 0.03% : 0.000008s : 60: predicate.dict_get_item_const_eliminator 0.03% : 0.000009s : 60: predicate.dict_get_item_eliminator 0.03% : 0.000008s : 60: predicate.dict_set_item_eliminator 0.01% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.00% : 0.000001s : 4: predicate.elim_not_effective 0.00% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 0.03% : 0.000008s : 64: predicate.environ_add_const_eliminate 0.03% : 0.000008s : 64: predicate.environ_get_add_eliminate 0.03% : 0.000009s : 64: predicate.environ_get_depend_swap 0.04% : 0.000012s : 87: predicate.environ_get_eliminate 0.03% : 0.000008s : 64: predicate.environ_get_set_eliminate 0.04% : 0.000013s : 92: predicate.exchange_switch_depend_value 0.07% : 0.000019s : 92: predicate.float_depend_g_call 0.01% : 0.000003s : 23: predicate.float_environ_get_switch 0.01% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.00% : 0.000000s : 4: predicate.fold_const_symbol 0.02% : 0.000005s : 23: predicate.get_grad_eliminate 0.00% : 0.000000s : 4: predicate.graph_param_transform 0.01% : 0.000003s : 23: predicate.incorporate_call 0.01% : 0.000003s : 23: predicate.incorporate_call_switch 0.15% : 0.000043s : 204: predicate.inline 0.03% : 0.000010s : 46: predicate.inline_without_move 0.01% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.02% : 0.000005s : 23: predicate.less_batch_normalization 0.04% : 0.000012s : 84: predicate.list_to_tuple_eliminator_ 0.06% : 0.000018s : 142: predicate.load_eliminater 0.01% : 0.000002s : 4: predicate.loop_unroll_after_grad 0.06% : 0.000017s : 128: predicate.loop_unroll_before_grad 0.04% : 0.000011s : 68: predicate.make_slice_get_slice_eliminator 0.01% : 0.000003s : 23: predicate.merge_addn 0.02% : 0.000007s : 55: predicate.micro_step_allgather_replace 0.02% : 0.000007s : 55: predicate.mini_step_allgather_replace 0.03% : 0.000008s : 60: predicate.minmaximum_grad 0.01% : 0.000002s : 4: predicate.mutable_eliminate 0.00% : 0.000001s : 4: predicate.opt_reshape 0.00% : 0.000001s : 4: predicate.parallel_virtual_node 0.07% : 0.000020s : 92: predicate.partial_defer_inline 0.04% : 0.000012s : 80: predicate.partial_eliminate 0.02% : 0.000007s : 58: predicate.print_const_string_wrapper 0.01% : 0.000003s : 23: predicate.reduce_all_const_elim 0.04% : 0.000011s : 60: predicate.reduce_eliminate 0.06% : 0.000018s : 142: predicate.redundant_stop_gradient_eliminater 0.01% : 0.000002s : 23: predicate.remove_not_recompute_node 0.05% : 0.000013s : 131: predicate.replace_applicator 0.02% : 0.000005s : 46: predicate.replace_old_param 0.00% : 0.000001s : 4: predicate.reset_defer_inline 0.03% : 0.000008s : 60: predicate.reshape_eliminate 0.03% : 0.000008s : 55: predicate.row_tensor_add_zeros_like 0.00% : 0.000001s : 4: predicate.row_tensor_eliminate 0.03% : 0.000009s : 55: predicate.same_eliminate 0.01% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.02% : 0.000005s : 23: predicate.shard_identity_eliminate 0.00% : 0.000001s : 8: predicate.special_op_eliminate 0.01% : 0.000004s : 23: predicate.specialize_transform 0.03% : 0.000009s : 55: predicate.split_environ_get_set_with_tuple_value 0.03% : 0.000008s : 46: predicate.stack_unstack_eliminate 0.00% : 0.000001s : 4: predicate.switch_call_monad_eliminater 0.05% : 0.000013s : 92: predicate.switch_defer_inline 0.07% : 0.000020s : 147: predicate.switch_layer_defer_inline 0.13% : 0.000036s : 247: predicate.switch_simplify 0.03% : 0.000008s : 60: predicate.tile_eliminate 0.03% : 0.000008s : 60: predicate.transpose_eliminate 0.04% : 0.000010s : 68: predicate.tuple_list_convert_item_index_to_positive 0.04% : 0.000011s : 68: predicate.tuple_list_get_item_const_eliminator 0.03% : 0.000010s : 68: predicate.tuple_list_get_item_depend_reorder 97.61% : 0.028238s : 107: predicate.tuple_list_get_item_eliminator 0.04% : 0.000010s : 68: predicate.tuple_list_get_set_item_eliminator 0.05% : 0.000015s : 91: predicate.tuple_list_set_item_eliminator 0.04% : 0.000011s : 84: predicate.tuple_to_list_eliminator_ 0.06% : 0.000018s : 142: predicate.updatestate_pure_node_eliminater 0.08% : 0.000022s : 165: predicate.updatestate_useless_node_eliminater 0.00% : 0.000001s : 4: predicate.value_based_eliminate 0.01% : 0.000004s : 23: predicate.virtual_dataset_eliminate 0.01% : 0.000003s : 23: predicate.virtual_output_eliminate 0.00% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.00% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002542 37 59.94% : 0.001523s : 17: func_graph_cloner_run.FuncGraphClonerGraph 40.06% : 0.001018s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.225827 233 0.00% : 0.000004s : 1: ForceFp32Comm 1.73% : 0.003899s : 1: add_attr 1.72% : 0.003885s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.02% : 0.000054s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.06% : 0.000143s : 1: auto_monad 0.01% : 0.000026s : 1: auto_monad_reorder 0.00% : 0.000003s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 16.24% : 0.036664s : 1: bootstrap 0.01% : 0.000032s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.01% : 0.000018s : 1: control_data_broadcast_order 0.00% : 0.000011s : 1: convert_after_rewriter 0.01% : 0.000026s : 1: cse_after_recomputation 0.00% : 0.000006s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.01% : 0.000013s : 1: environ_conv 0.03% : 0.000076s : 1: event_method 0.00% : 0.000006s : 1: full_micro_interleaved_order_control 0.00% : 0.000007s : 1: get_jit_bprop_graph 0.01% : 0.000013s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000007s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000007s : 1: label_micro_interleaved_index 0.22% : 0.000486s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.32% : 0.000727s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.01% : 0.000017s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000018s : 1: opt.transform.mutable_eliminate 14.75% : 0.033302s : 117: opt.transform.opt_a 0.01% : 0.000031s : 1: opt.transform.opt_after_cconv 0.01% : 0.000030s : 1: opt.transform.opt_after_jit_grad 0.04% : 0.000099s : 28: opt.transform.opt_b 0.02% : 0.000050s : 2: opt.transform.opt_trans_graph 0.02% : 0.000041s : 4: opt.transform.symbol_engine_opt 18.38% : 0.041507s : 1: opt_a 0.05% : 0.000114s : 1: opt_after_cconv 0.26% : 0.000592s : 1: opt_after_jit_grad 0.09% : 0.000208s : 1: opt_b 19.47% : 0.043973s : 1: optimize 0.01% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000009s : 1: order_py_execute_after_rewriter 0.01% : 0.000025s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000008s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.03% : 0.000071s : 1: pre_auto_parallel 0.02% : 0.000053s : 1: py_interpret_to_execute 0.01% : 0.000017s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.01% : 0.000018s : 1: remove_dup_value 1.04% : 0.002360s : 2: renormalize.infer 0.84% : 0.001902s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000008s : 1: rewriter_after_jit_bprop_graph 0.02% : 0.000045s : 1: rewriter_after_opt_a 0.08% : 0.000190s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000004s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.04% : 0.000097s : 1: symbol_engine_optimizer 0.04% : 0.000082s : 1: tuple_transform 24.20% : 0.054660s : 1: type_inference TotalTime = 0.0565337, [21] [bootstrap]: 0.00046416 [type_inference]: 0.0375186 [event_method]: 7.664e-05 [auto_monad]: 0.00011748 [graph_reusing]: 8.50999e-06 [inline]: 2.73003e-06 [add_attr]: 0.00362431, [1] [add_attr_with_inline]: 0.00361339, [1] [Cycle 1]: 9.663e-05, [2] [tag_attr]: 4.206e-05 [meta_addattr_fg_expand]: 1.012e-05 [parallel-infer-symbol]: 3.33998e-06 [pre_auto_parallel]: 6.169e-05 [insert-virtual-dataset]: 2.85002e-06 [parallel-infer-symbol-second]: 8.30012e-07 [dataset_repeat_opt]: 2.14999e-06 [pipeline_split]: 1.55999e-06 [optimize]: 0.01395, [53] [py_interpret_to_execute]: 4.627e-05 [rewriter_before_opt_a]: 0.00016182 [opt_a]: 0.0117666, [3] [Cycle 1]: 0.00867527, [45] [expand_dump_flag]: 3.83999e-06 [switch_simplify]: 8.271e-05 [loop_unroll]: 6.754e-05 [a_1]: 0.00153895 [with_stream_mark]: 3.194e-05 [recompute_prepare]: 2.523e-05 [updatestate_depend_eliminate]: 9.40001e-06 [updatestate_assign_eliminate]: 6.88e-06 [updatestate_loads_eliminate]: 6.69999e-06 [parameter_eliminate]: 2.96001e-06 [a_2]: 0.00021272 [accelerated_algorithm]: 1.502e-05 [shard]: 2.06998e-06 [meta_shard_fg_expand]: 4.37e-06 [shard_inline]: 1.409e-05 [merge_send_recv]: 1.728e-05 [auto_parallel]: 1.303e-05 [parallel]: 1.896e-05 [flash_sp]: 1.241e-05 [merge_comm]: 8.89e-06 [allreduce_fusion]: 8.62e-06 [matmul_add_comm_reduction]: 3.154e-05 [allreduce_slice_to_reducescatter]: 6.40022e-07 [virtual_shard_identity]: 1.808e-05 [virtual_dataset]: 1.436e-05 [get_grad_eliminate_]: 1.361e-05 [virtual_output]: 1.398e-05 [merge_forward]: 9.67001e-06 [cell_reuse_recompute_pass]: 1.25999e-06 [offload_activation]: 1.745e-05 [cell_reuse_handle_not_recompute_node_pass]: 2.729e-05 [merge_recompute_call_nodes]: 1.54e-06 [before_grad]: 2.493e-05 [set_forward_comm_id_for_comm_node_pass]: 9.31998e-06 [meta_fg_expand]: 0.00170494 [flash_sp_send_recv_attached]: 4.44002e-06 [receive_attached]: 2.26e-06 [after_resolve]: 6.492e-05 [a_after_grad]: 8.755e-05 [renormalize]: 0.00359112 [add_forward_monad_depend]: 1.226e-05 [auto_monad_grad]: 7.2e-06 [auto_monad_eliminator]: 5.954e-05 [cse]: 0.00020947 [a_3]: 0.00032321 [Cycle 2]: 0.00260717, [45] [expand_dump_flag]: 2.91e-06 [switch_simplify]: 4.442e-05 [loop_unroll]: 4.08e-05 [a_1]: 0.00135918 [with_stream_mark]: 1.758e-05 [recompute_prepare]: 7.58001e-06 [updatestate_depend_eliminate]: 2.91e-06 [updatestate_assign_eliminate]: 2.46e-06 [updatestate_loads_eliminate]: 2.01998e-06 [parameter_eliminate]: 1.62999e-06 [a_2]: 4.681e-05 [accelerated_algorithm]: 4.91002e-06 [shard]: 1.52999e-06 [meta_shard_fg_expand]: 2.31e-06 [shard_inline]: 4.80001e-06 [merge_send_recv]: 6.73e-06 [auto_parallel]: 7.08998e-06 [parallel]: 8.59998e-06 [flash_sp]: 3.64002e-06 [merge_comm]: 2.79001e-06 [allreduce_fusion]: 2.50002e-06 [matmul_add_comm_reduction]: 6.90998e-06 [allreduce_slice_to_reducescatter]: 6.59988e-07 [virtual_shard_identity]: 5.52001e-06 [virtual_dataset]: 4.34002e-06 [get_grad_eliminate_]: 4.62e-06 [virtual_output]: 4.22e-06 [merge_forward]: 3.09001e-06 [cell_reuse_recompute_pass]: 1.47001e-06 [offload_activation]: 7.36999e-06 [cell_reuse_handle_not_recompute_node_pass]: 9.20999e-06 [merge_recompute_call_nodes]: 1.35999e-06 [before_grad]: 6.57002e-06 [set_forward_comm_id_for_comm_node_pass]: 2.79999e-06 [meta_fg_expand]: 0.00011193 [flash_sp_send_recv_attached]: 1.12e-06 [receive_attached]: 1.75001e-06 [after_resolve]: 1.007e-05 [a_after_grad]: 6.42001e-06 [renormalize]: 0.00054007 [add_forward_monad_depend]: 4.52e-06 [auto_monad_grad]: 1.46002e-06 [auto_monad_eliminator]: 9.97999e-06 [cse]: 1.556e-05 [a_3]: 3.079e-05 [Cycle 3]: 0.00046679, [45] [expand_dump_flag]: 1.50001e-06 [switch_simplify]: 5.62001e-06 [loop_unroll]: 4.35999e-06 [a_1]: 6.652e-05 [with_stream_mark]: 7.83001e-06 [recompute_prepare]: 4.3e-06 [updatestate_depend_eliminate]: 2.35002e-06 [updatestate_assign_eliminate]: 1.91e-06 [updatestate_loads_eliminate]: 1.52001e-06 [parameter_eliminate]: 8.39995e-07 [a_2]: 4.246e-05 [accelerated_algorithm]: 3.97e-06 [shard]: 1.24e-06 [meta_shard_fg_expand]: 1.23002e-06 [shard_inline]: 4.13999e-06 [merge_send_recv]: 4.08999e-06 [auto_parallel]: 4.97e-06 [parallel]: 4.40999e-06 [flash_sp]: 1.16002e-06 [merge_comm]: 2.16e-06 [allreduce_fusion]: 2.07999e-06 [matmul_add_comm_reduction]: 3.93999e-06 [allreduce_slice_to_reducescatter]: 4.39992e-07 [virtual_shard_identity]: 4.97999e-06 [virtual_dataset]: 3.98999e-06 [get_grad_eliminate_]: 3.8e-06 [virtual_output]: 4.01001e-06 [merge_forward]: 2.06e-06 [cell_reuse_recompute_pass]: 1.43002e-06 [offload_activation]: 4.57e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.009e-05 [merge_recompute_call_nodes]: 7.10017e-07 [before_grad]: 6.51e-06 [set_forward_comm_id_for_comm_node_pass]: 2.39999e-06 [meta_fg_expand]: 1.42999e-06 [flash_sp_send_recv_attached]: 9.89996e-07 [receive_attached]: 1.05999e-06 [after_resolve]: 7.24001e-06 [a_after_grad]: 5.77999e-06 [renormalize]: 1.10012e-07 [add_forward_monad_depend]: 1.10999e-06 [auto_monad_grad]: 8.39995e-07 [auto_monad_eliminator]: 4.4e-06 [cse]: 8.92999e-06 [a_3]: 2.215e-05 [py_interpret_to_execute_after_opt_a]: 1.119e-05 [slice_cell_reuse_recomputed_activation]: 2.22999e-06 [rewriter_after_opt_a]: 3.112e-05 [convert_after_rewriter]: 6.88e-06 [order_py_execute_after_rewriter]: 4.37e-06 [mutable_eliminate]: 0.0006518 [opt_b]: 0.00015211, [1] [Cycle 1]: 0.0001448, [7] [b_1]: 8.061e-05 [b_2]: 6.49001e-06 [updatestate_depend_eliminate]: 4.51002e-06 [updatestate_assign_eliminate]: 2.11e-06 [updatestate_loads_eliminate]: 1.56998e-06 [renormalize]: 2.3999e-07 [cse]: 1.487e-05 [optimize_parallel_all_gather_comm]: 1.596e-05 [overlap_param_gather]: 1.82001e-06 [cconv]: 2.279e-05 [loop_unroll]: 0.00046356 [opt_after_cconv]: 8.609e-05, [1] [Cycle 1]: 7.995e-05, [7] [c_1]: 2.081e-05 [parameter_eliminate]: 2.27999e-06 [updatestate_depend_eliminate]: 4.30999e-06 [updatestate_assign_eliminate]: 1.84998e-06 [updatestate_loads_eliminate]: 1.66e-06 [cse]: 1.517e-05 [renormalize]: 3.00002e-07 [remove_dup_value]: 1.246e-05 [tuple_transform]: 5.841e-05, [1] [Cycle 1]: 5.372e-05, [4] [d_1]: 2.868e-05 [none_parameter_eliminate]: 1.49e-06 [renormalize]: 1.8999e-07 [switch_simplify]: 4.95001e-06 [partial_unused_args_eliminate]: 2.02999e-06 [add_recomputation]: 3.403e-05 [cse_after_recomputation]: 3.007e-05, [1] [Cycle 1]: 2.483e-05, [1] [cse]: 7.35e-06 [environ_conv]: 5.06997e-06 [swap_dp_allreduce_reducescatter]: 4.58001e-06 [bias_add_comm_swap]: 2.74001e-06 [label_micro_interleaved_index]: 5.28002e-06 [label_fine_grained_interleaved_index]: 2.77002e-06 [merge_cast_opt]: 1.40001e-06 [slice_recompute_activation]: 2.04e-06 [micro_interleaved_order_control]: 2.38002e-06 [assign_add_opt]: 1.24e-06 [ForceFp32Comm]: 9.20001e-07 [remove_cast_before_assign_add]: 1.00001e-06 [full_micro_interleaved_order_control]: 2.26e-06 [reorder_send_recv_between_fp_bp]: 2.92002e-06 [comm_op_add_attrs]: 1.11002e-06 [add_comm_op_reuse_tag]: 1.34e-06 [interleave_split_concat_branches]: 1.24e-06 [interleave_parallel_branches]: 1.09e-06 [overlap_opt_shard_in_pipeline]: 1.35001e-06 [overlap_opt_shard_grad_in_pipeline]: 1.96e-06 [control_data_broadcast_order]: 9.94999e-06 [grouped_pairwise_exchange_alltoall]: 1.62999e-06 [offloading_packed_experts]: 2.89999e-06 [overlap_recompute_and_grad_model_parallel]: 3.67002e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.40999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.33002e-06 [overlap_recompute_comm]: 2.12999e-06 [overlap_grad_ring_attention]: 3.23e-06 [overlap_grad_flash_sp]: 1.545e-05 [begin_end_overlap_inline]: 5.69999e-07 [split_matmul_comm_elemetwise]: 2.22001e-06 [split_layernorm_comm]: 1.81e-06 [handle_group_info]: 1.62999e-06 [symbol_engine_optimizer]: 6.854e-05, [1] [Cycle 1]: 6.392e-05, [6] [build]: 2.48e-06 [elim_shapecalc]: 9.44998e-06 [elim_not_effective]: 9.85002e-06 [opt_reshape]: 5.37001e-06 [fold_const_symbol]: 6.98e-06 [renormalize]: 2.09984e-07 [detach_backward]: 2.44999e-06 [pipeline_parallel_scheduler]: 1.59e-06 [auto_monad_reorder]: 1.404e-05 [get_jit_bprop_graph]: 1.60999e-06 [rewriter_after_jit_bprop_graph]: 4.32e-06 [opt_after_jit_grad]: 0.00046541 [validate]: 3.122e-05 Sums bootstrap : 0.000464s : 0.90% type_inference : 0.037519s : 72.69% event_method : 0.000077s : 0.15% auto_monad : 0.000117s : 0.23% graph_reusing : 0.000009s : 0.02% inline : 0.000003s : 0.01% add_attr.add_attr_with_inline.tag_attr : 0.000042s : 0.08% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000010s : 0.02% parallel-infer-symbol : 0.000003s : 0.01% pre_auto_parallel : 0.000062s : 0.12% insert-virtual-dataset : 0.000003s : 0.01% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000046s : 0.09% optimize.rewriter_before_opt_a : 0.000162s : 0.31% optimize.opt_a.expand_dump_flag : 0.000008s : 0.02% optimize.opt_a.switch_simplify : 0.000133s : 0.26% optimize.opt_a.loop_unroll : 0.000113s : 0.22% optimize.opt_a.a_1 : 0.002965s : 5.74% optimize.opt_a.with_stream_mark : 0.000057s : 0.11% optimize.opt_a.recompute_prepare : 0.000037s : 0.07% optimize.opt_a.updatestate_depend_eliminate : 0.000015s : 0.03% optimize.opt_a.updatestate_assign_eliminate : 0.000011s : 0.02% optimize.opt_a.updatestate_loads_eliminate : 0.000010s : 0.02% optimize.opt_a.parameter_eliminate : 0.000005s : 0.01% optimize.opt_a.a_2 : 0.000302s : 0.59% optimize.opt_a.accelerated_algorithm : 0.000024s : 0.05% optimize.opt_a.shard : 0.000005s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000008s : 0.02% optimize.opt_a.shard_inline : 0.000023s : 0.04% optimize.opt_a.merge_send_recv : 0.000028s : 0.05% optimize.opt_a.auto_parallel : 0.000025s : 0.05% optimize.opt_a.parallel : 0.000032s : 0.06% optimize.opt_a.flash_sp : 0.000017s : 0.03% optimize.opt_a.merge_comm : 0.000014s : 0.03% optimize.opt_a.allreduce_fusion : 0.000013s : 0.03% optimize.opt_a.matmul_add_comm_reduction : 0.000042s : 0.08% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000029s : 0.06% optimize.opt_a.virtual_dataset : 0.000023s : 0.04% optimize.opt_a.get_grad_eliminate_ : 0.000022s : 0.04% optimize.opt_a.virtual_output : 0.000022s : 0.04% optimize.opt_a.merge_forward : 0.000015s : 0.03% optimize.opt_a.cell_reuse_recompute_pass : 0.000004s : 0.01% optimize.opt_a.offload_activation : 0.000029s : 0.06% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000047s : 0.09% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.01% optimize.opt_a.before_grad : 0.000038s : 0.07% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000015s : 0.03% optimize.opt_a.meta_fg_expand : 0.001818s : 3.52% optimize.opt_a.flash_sp_send_recv_attached : 0.000007s : 0.01% optimize.opt_a.receive_attached : 0.000005s : 0.01% optimize.opt_a.after_resolve : 0.000082s : 0.16% optimize.opt_a.a_after_grad : 0.000100s : 0.19% optimize.opt_a.renormalize : 0.004131s : 8.00% optimize.opt_a.add_forward_monad_depend : 0.000018s : 0.03% optimize.opt_a.auto_monad_grad : 0.000010s : 0.02% optimize.opt_a.auto_monad_eliminator : 0.000074s : 0.14% optimize.opt_a.cse : 0.000234s : 0.45% optimize.opt_a.a_3 : 0.000376s : 0.73% optimize.py_interpret_to_execute_after_opt_a : 0.000011s : 0.02% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000031s : 0.06% optimize.convert_after_rewriter : 0.000007s : 0.01% optimize.order_py_execute_after_rewriter : 0.000004s : 0.01% optimize.mutable_eliminate : 0.000652s : 1.26% optimize.opt_b.b_1 : 0.000081s : 0.16% optimize.opt_b.b_2 : 0.000006s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000005s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000015s : 0.03% optimize.optimize_parallel_all_gather_comm : 0.000016s : 0.03% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000023s : 0.04% optimize.loop_unroll : 0.000464s : 0.90% optimize.opt_after_cconv.c_1 : 0.000021s : 0.04% optimize.opt_after_cconv.parameter_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000004s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000015s : 0.03% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000012s : 0.02% optimize.tuple_transform.d_1 : 0.000029s : 0.06% optimize.tuple_transform.none_parameter_eliminate : 0.000001s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000005s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000034s : 0.07% optimize.cse_after_recomputation.cse : 0.000007s : 0.01% optimize.environ_conv : 0.000005s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.01% optimize.bias_add_comm_swap : 0.000003s : 0.01% optimize.label_micro_interleaved_index : 0.000005s : 0.01% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.01% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.01% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000010s : 0.02% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000003s : 0.01% optimize.overlap_recompute_and_grad_model_parallel : 0.000004s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000003s : 0.01% optimize.overlap_grad_flash_sp : 0.000015s : 0.03% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000002s : 0.00% optimize.symbol_engine_optimizer.build : 0.000002s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000009s : 0.02% optimize.symbol_engine_optimizer.elim_not_effective : 0.000010s : 0.02% optimize.symbol_engine_optimizer.opt_reshape : 0.000005s : 0.01% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000007s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000014s : 0.03% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.01% opt_after_jit_grad : 0.000465s : 0.90% validate : 0.000031s : 0.06% Time group info: ------[substitution.] 0.000838 143 7.76% : 0.000065s : 5: substitution.arithmetic_simplify 0.17% : 0.000001s : 1: substitution.elim_not_effective 0.60% : 0.000005s : 6: substitution.float_depend_g_call 0.46% : 0.000004s : 2: substitution.float_tuple_getitem_switch 0.11% : 0.000001s : 1: substitution.fold_const_symbol 0.53% : 0.000004s : 2: substitution.graph_param_transform 0.33% : 0.000003s : 2: substitution.incorporate_call 0.23% : 0.000002s : 2: substitution.incorporate_call_switch 63.32% : 0.000530s : 16: substitution.inline 2.39% : 0.000020s : 2: substitution.inline_without_move 0.91% : 0.000008s : 10: substitution.j_node_and_user_rematch 1.36% : 0.000011s : 7: substitution.minmaximum_grad 0.77% : 0.000006s : 6: substitution.partial_eliminate 1.10% : 0.000009s : 10: substitution.remove_not_recompute_node 3.41% : 0.000029s : 10: substitution.replace_applicator 1.12% : 0.000009s : 10: substitution.replace_old_param 0.35% : 0.000003s : 1: substitution.set_cell_output_no_recompute 2.65% : 0.000022s : 7: substitution.tuple_list_convert_item_index_to_positive 1.30% : 0.000011s : 7: substitution.tuple_list_get_item_const_eliminator 1.73% : 0.000014s : 7: substitution.tuple_list_get_item_depend_reorder 7.67% : 0.000064s : 22: substitution.tuple_list_get_item_eliminator 1.73% : 0.000015s : 7: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.037399 2 94.85% : 0.035472s : 1: type_inference.infer 5.15% : 0.001927s : 1: type_inference.specialize ------[replace.] 0.000259 30 2.41% : 0.000006s : 1: replace.arithmetic_simplify 56.61% : 0.000147s : 16: replace.inline 40.98% : 0.000106s : 13: replace.tuple_list_get_item_eliminator ------[match.] 0.000589 30 5.91% : 0.000035s : 1: match.arithmetic_simplify 88.18% : 0.000519s : 16: match.inline 5.91% : 0.000035s : 13: match.tuple_list_get_item_eliminator ------[predicate.] 0.000573 3865 1.04% : 0.000006s : 48: predicate.accumulaten_eliminater 0.22% : 0.000001s : 2: predicate.ad_related_special_op_eliminate 0.35% : 0.000002s : 17: predicate.addn_check_dump 1.08% : 0.000006s : 48: predicate.addn_zero_filter 1.06% : 0.000006s : 48: predicate.adjust_all_reduce_mul_add 1.97% : 0.000011s : 66: predicate.arithmetic_simplify 1.14% : 0.000007s : 49: predicate.cast_eliminate 1.15% : 0.000007s : 48: predicate.check_bprop_eliminate 0.36% : 0.000002s : 17: predicate.compare_switch_simplify 0.03% : 0.000000s : 2: predicate.const_output_eliminate 0.43% : 0.000002s : 17: predicate.depend_value_elim 1.18% : 0.000007s : 49: predicate.dict_get_item_const_eliminator 1.23% : 0.000007s : 49: predicate.dict_get_item_eliminator 1.15% : 0.000007s : 49: predicate.dict_set_item_eliminator 0.25% : 0.000001s : 4: predicate.dumpgradient_eliminate 0.05% : 0.000000s : 2: predicate.elim_not_effective 0.11% : 0.000001s : 2: predicate.elim_shapecalc_of_broadcastargs 1.13% : 0.000006s : 51: predicate.environ_add_const_eliminate 1.09% : 0.000006s : 51: predicate.environ_get_add_eliminate 3.88% : 0.000022s : 51: predicate.environ_get_depend_swap 1.56% : 0.000009s : 68: predicate.environ_get_eliminate 1.09% : 0.000006s : 51: predicate.environ_get_set_eliminate 1.80% : 0.000010s : 78: predicate.exchange_switch_depend_value 2.75% : 0.000016s : 78: predicate.float_depend_g_call 0.36% : 0.000002s : 17: predicate.float_environ_get_switch 0.45% : 0.000003s : 19: predicate.float_tuple_getitem_switch 0.04% : 0.000000s : 2: predicate.fold_const_symbol 0.53% : 0.000003s : 17: predicate.get_grad_eliminate 0.04% : 0.000000s : 2: predicate.graph_param_transform 0.40% : 0.000002s : 17: predicate.incorporate_call 0.35% : 0.000002s : 17: predicate.incorporate_call_switch 5.71% : 0.000033s : 164: predicate.inline 1.41% : 0.000008s : 40: predicate.inline_without_move 0.27% : 0.000002s : 17: predicate.j_node_and_user_rematch 0.55% : 0.000003s : 17: predicate.less_batch_normalization 1.61% : 0.000009s : 66: predicate.list_to_tuple_eliminator_ 2.53% : 0.000014s : 114: predicate.load_eliminater 0.29% : 0.000002s : 2: predicate.loop_unroll_after_grad 2.80% : 0.000016s : 121: predicate.loop_unroll_before_grad 1.24% : 0.000007s : 53: predicate.make_slice_get_slice_eliminator 0.40% : 0.000002s : 17: predicate.merge_addn 1.13% : 0.000006s : 48: predicate.micro_step_allgather_replace 1.12% : 0.000006s : 48: predicate.mini_step_allgather_replace 1.08% : 0.000006s : 49: predicate.minmaximum_grad 0.29% : 0.000002s : 2: predicate.mutable_eliminate 0.06% : 0.000000s : 2: predicate.opt_reshape 0.08% : 0.000000s : 2: predicate.parallel_virtual_node 2.40% : 0.000014s : 78: predicate.partial_defer_inline 1.68% : 0.000010s : 64: predicate.partial_eliminate 1.06% : 0.000006s : 48: predicate.print_const_string_wrapper 0.46% : 0.000003s : 17: predicate.reduce_all_const_elim 1.39% : 0.000008s : 49: predicate.reduce_eliminate 2.51% : 0.000014s : 114: predicate.redundant_stop_gradient_eliminater 0.25% : 0.000001s : 17: predicate.remove_not_recompute_node 1.96% : 0.000011s : 110: predicate.replace_applicator 0.66% : 0.000004s : 40: predicate.replace_old_param 0.06% : 0.000000s : 2: predicate.reset_defer_inline 1.15% : 0.000007s : 49: predicate.reshape_eliminate 1.14% : 0.000007s : 48: predicate.row_tensor_add_zeros_like 0.11% : 0.000001s : 2: predicate.row_tensor_eliminate 1.31% : 0.000008s : 48: predicate.same_eliminate 0.29% : 0.000002s : 17: predicate.set_cell_output_no_recompute 0.52% : 0.000003s : 17: predicate.shard_identity_eliminate 0.15% : 0.000001s : 4: predicate.special_op_eliminate 0.51% : 0.000003s : 17: predicate.specialize_transform 1.35% : 0.000008s : 48: predicate.split_environ_get_set_with_tuple_value 1.22% : 0.000007s : 40: predicate.stack_unstack_eliminate 0.07% : 0.000000s : 2: predicate.switch_call_monad_eliminater 2.04% : 0.000012s : 78: predicate.switch_defer_inline 3.01% : 0.000017s : 126: predicate.switch_layer_defer_inline 5.39% : 0.000031s : 218: predicate.switch_simplify 1.12% : 0.000006s : 49: predicate.tile_eliminate 1.04% : 0.000006s : 49: predicate.transpose_eliminate 1.41% : 0.000008s : 53: predicate.tuple_list_convert_item_index_to_positive 1.48% : 0.000009s : 53: predicate.tuple_list_get_item_const_eliminator 1.24% : 0.000007s : 53: predicate.tuple_list_get_item_depend_reorder 2.88% : 0.000016s : 83: predicate.tuple_list_get_item_eliminator 1.37% : 0.000008s : 53: predicate.tuple_list_get_set_item_eliminator 1.80% : 0.000010s : 70: predicate.tuple_list_set_item_eliminator 1.54% : 0.000009s : 66: predicate.tuple_to_list_eliminator_ 2.43% : 0.000014s : 114: predicate.updatestate_pure_node_eliminater 2.94% : 0.000017s : 131: predicate.updatestate_useless_node_eliminater 0.10% : 0.000001s : 2: predicate.value_based_eliminate 0.44% : 0.000003s : 17: predicate.virtual_dataset_eliminate 0.48% : 0.000003s : 17: predicate.virtual_output_eliminate 0.06% : 0.000000s : 2: predicate.virtual_view_grad_eliminate 0.12% : 0.000001s : 2: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002282 37 63.25% : 0.001443s : 17: func_graph_cloner_run.FuncGraphClonerGraph 36.75% : 0.000839s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.082481 233 0.00% : 0.000004s : 1: ForceFp32Comm 4.40% : 0.003630s : 1: add_attr 4.39% : 0.003617s : 1: add_attr_with_inline 0.01% : 0.000004s : 1: add_comm_op_reuse_tag 0.05% : 0.000038s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.15% : 0.000125s : 1: auto_monad 0.02% : 0.000018s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.01% : 0.000006s : 1: bias_add_comm_swap 0.60% : 0.000495s : 1: bootstrap 0.03% : 0.000026s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.02% : 0.000013s : 1: control_data_broadcast_order 0.01% : 0.000010s : 1: convert_after_rewriter 0.04% : 0.000033s : 1: cse_after_recomputation 0.01% : 0.000005s : 1: dataset_repeat_opt 0.01% : 0.000006s : 1: detach_backward 0.01% : 0.000008s : 1: environ_conv 0.10% : 0.000086s : 1: event_method 0.01% : 0.000005s : 1: full_micro_interleaved_order_control 0.01% : 0.000005s : 1: get_jit_bprop_graph 0.02% : 0.000013s : 1: graph_reusing 0.01% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.01% : 0.000005s : 1: handle_group_info 0.01% : 0.000006s : 1: inline 0.01% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.01% : 0.000007s : 1: label_fine_grained_interleaved_index 0.01% : 0.000008s : 1: label_micro_interleaved_index 0.57% : 0.000472s : 1: loop_unroll 0.01% : 0.000004s : 1: merge_cast_opt 0.01% : 0.000005s : 1: micro_interleaved_order_control 0.80% : 0.000660s : 1: mutable_eliminate 0.01% : 0.000006s : 1: offloading_packed_experts 0.01% : 0.000011s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000011s : 1: opt.transform.mutable_eliminate 5.10% : 0.004204s : 117: opt.transform.opt_a 0.02% : 0.000019s : 1: opt.transform.opt_after_cconv 0.02% : 0.000018s : 1: opt.transform.opt_after_jit_grad 0.07% : 0.000059s : 28: opt.transform.opt_b 0.04% : 0.000031s : 2: opt.transform.opt_trans_graph 0.03% : 0.000028s : 4: opt.transform.symbol_engine_opt 14.27% : 0.011770s : 1: opt_a 0.11% : 0.000090s : 1: opt_after_cconv 0.58% : 0.000475s : 1: opt_after_jit_grad 0.19% : 0.000156s : 1: opt_b 16.92% : 0.013955s : 1: optimize 0.02% : 0.000020s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000008s : 1: order_py_execute_after_rewriter 0.02% : 0.000019s : 1: overlap_grad_flash_sp 0.01% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000006s : 1: overlap_grad_ring_attention 0.01% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.01% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000007s : 1: overlap_recompute_and_grad_model_parallel 0.01% : 0.000005s : 1: overlap_recompute_comm 0.01% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.01% : 0.000005s : 1: partial_unused_args_eliminate 0.01% : 0.000005s : 1: pipeline_parallel_scheduler 0.01% : 0.000004s : 1: pipeline_split 0.08% : 0.000067s : 1: pre_auto_parallel 0.06% : 0.000051s : 1: py_interpret_to_execute 0.02% : 0.000015s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.02% : 0.000016s : 1: remove_dup_value 3.03% : 0.002501s : 2: renormalize.infer 1.95% : 0.001612s : 2: renormalize.specialize 0.01% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000008s : 1: rewriter_after_jit_bprop_graph 0.04% : 0.000035s : 1: rewriter_after_opt_a 0.20% : 0.000168s : 1: rewriter_before_opt_a 0.01% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.01% : 0.000005s : 1: slice_recompute_activation 0.01% : 0.000005s : 1: split_layernorm_comm 0.01% : 0.000005s : 1: split_matmul_comm_elemetwise 0.01% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.09% : 0.000071s : 1: symbol_engine_optimizer 0.07% : 0.000062s : 1: tuple_transform 45.52% : 0.037548s : 1: type_inference TotalTime = 0.0552441, [21] [bootstrap]: 0.00048369 [type_inference]: 0.0185437 [event_method]: 7.305e-05 [auto_monad]: 0.00011949 [graph_reusing]: 9.37999e-06 [inline]: 3.3e-06 [add_attr]: 0.00377554, [1] [add_attr_with_inline]: 0.00376434, [1] [Cycle 1]: 0.00010288, [2] [tag_attr]: 4.531e-05 [meta_addattr_fg_expand]: 1.057e-05 [parallel-infer-symbol]: 3.26001e-06 [pre_auto_parallel]: 6.229e-05 [insert-virtual-dataset]: 2.50997e-06 [parallel-infer-symbol-second]: 7.89994e-07 [dataset_repeat_opt]: 2.31998e-06 [pipeline_split]: 1.82001e-06 [optimize]: 0.0313625, [53] [py_interpret_to_execute]: 4.599e-05 [rewriter_before_opt_a]: 0.00017156 [opt_a]: 0.028943, [3] [Cycle 1]: 0.0251647, [45] [expand_dump_flag]: 4.22e-06 [switch_simplify]: 8.201e-05 [loop_unroll]: 6.414e-05 [a_1]: 0.00154918 [with_stream_mark]: 3.606e-05 [recompute_prepare]: 3.01e-05 [updatestate_depend_eliminate]: 9.09e-06 [updatestate_assign_eliminate]: 7.1e-06 [updatestate_loads_eliminate]: 6.65002e-06 [parameter_eliminate]: 2.96001e-06 [a_2]: 0.0002207 [accelerated_algorithm]: 1.765e-05 [shard]: 1.94e-06 [meta_shard_fg_expand]: 4.43999e-06 [shard_inline]: 1.476e-05 [merge_send_recv]: 1.759e-05 [auto_parallel]: 1.575e-05 [parallel]: 2.77e-05 [flash_sp]: 1.408e-05 [merge_comm]: 1.071e-05 [allreduce_fusion]: 7.96001e-06 [matmul_add_comm_reduction]: 3.326e-05 [allreduce_slice_to_reducescatter]: 8.59989e-07 [virtual_shard_identity]: 1.84e-05 [virtual_dataset]: 1.557e-05 [get_grad_eliminate_]: 1.502e-05 [virtual_output]: 1.386e-05 [merge_forward]: 1.035e-05 [cell_reuse_recompute_pass]: 2.07001e-06 [offload_activation]: 1.83e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.164e-05 [merge_recompute_call_nodes]: 1.69e-06 [before_grad]: 2.597e-05 [set_forward_comm_id_for_comm_node_pass]: 1.079e-05 [meta_fg_expand]: 0.0181434 [flash_sp_send_recv_attached]: 5.15999e-06 [receive_attached]: 2.94999e-06 [after_resolve]: 8.417e-05 [a_after_grad]: 9.169e-05 [renormalize]: 0.00342299 [add_forward_monad_depend]: 1.54e-05 [auto_monad_grad]: 7.40998e-06 [auto_monad_eliminator]: 6.624e-05 [cse]: 0.00023788 [a_3]: 0.00034898 [Cycle 2]: 0.00324022, [45] [expand_dump_flag]: 3.16001e-06 [switch_simplify]: 4.887e-05 [loop_unroll]: 4.579e-05 [a_1]: 0.00140336 [with_stream_mark]: 2.63e-05 [recompute_prepare]: 1.157e-05 [updatestate_depend_eliminate]: 4.20999e-06 [updatestate_assign_eliminate]: 2.48e-06 [updatestate_loads_eliminate]: 2.48e-06 [parameter_eliminate]: 2.07001e-06 [a_2]: 5.007e-05 [accelerated_algorithm]: 5.77001e-06 [shard]: 2.27001e-06 [meta_shard_fg_expand]: 2.22999e-06 [shard_inline]: 4.67998e-06 [merge_send_recv]: 7.83999e-06 [auto_parallel]: 9.82001e-06 [parallel]: 9.39998e-06 [flash_sp]: 3.91999e-06 [merge_comm]: 3.03e-06 [allreduce_fusion]: 2.68e-06 [matmul_add_comm_reduction]: 1.023e-05 [allreduce_slice_to_reducescatter]: 8.90024e-07 [virtual_shard_identity]: 7.11001e-06 [virtual_dataset]: 4.27998e-06 [get_grad_eliminate_]: 4.74998e-06 [virtual_output]: 4.72998e-06 [merge_forward]: 3.32002e-06 [cell_reuse_recompute_pass]: 1.99e-06 [offload_activation]: 9.31e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.162e-05 [merge_recompute_call_nodes]: 1.52001e-06 [before_grad]: 7.38999e-06 [set_forward_comm_id_for_comm_node_pass]: 2.98e-06 [meta_fg_expand]: 0.00014485 [flash_sp_send_recv_attached]: 2.01e-06 [receive_attached]: 2.59999e-06 [after_resolve]: 1.639e-05 [a_after_grad]: 7.18998e-06 [renormalize]: 0.00094118 [add_forward_monad_depend]: 7.06001e-06 [auto_monad_grad]: 2.26e-06 [auto_monad_eliminator]: 1.583e-05 [cse]: 2.869e-05 [a_3]: 3.823e-05 [Cycle 3]: 0.00051668, [45] [expand_dump_flag]: 2.81e-06 [switch_simplify]: 6.16e-06 [loop_unroll]: 4.85001e-06 [a_1]: 6.796e-05 [with_stream_mark]: 1.107e-05 [recompute_prepare]: 5.32001e-06 [updatestate_depend_eliminate]: 3.01001e-06 [updatestate_assign_eliminate]: 2.61e-06 [updatestate_loads_eliminate]: 1.95001e-06 [parameter_eliminate]: 1.45001e-06 [a_2]: 4.421e-05 [accelerated_algorithm]: 4.51002e-06 [shard]: 1.68002e-06 [meta_shard_fg_expand]: 1.42999e-06 [shard_inline]: 4.18001e-06 [merge_send_recv]: 4.98001e-06 [auto_parallel]: 6.21e-06 [parallel]: 6.51e-06 [flash_sp]: 1.20999e-06 [merge_comm]: 2.43e-06 [allreduce_fusion]: 2.46e-06 [matmul_add_comm_reduction]: 6.06e-06 [allreduce_slice_to_reducescatter]: 3.59985e-07 [virtual_shard_identity]: 6.12999e-06 [virtual_dataset]: 4.35999e-06 [get_grad_eliminate_]: 4.52e-06 [virtual_output]: 4.13999e-06 [merge_forward]: 2.77002e-06 [cell_reuse_recompute_pass]: 2.81999e-06 [offload_activation]: 5.92999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.136e-05 [merge_recompute_call_nodes]: 1.04998e-06 [before_grad]: 7.26999e-06 [set_forward_comm_id_for_comm_node_pass]: 4.20999e-06 [meta_fg_expand]: 1.76e-06 [flash_sp_send_recv_attached]: 1.47001e-06 [receive_attached]: 1.74e-06 [after_resolve]: 8.14002e-06 [a_after_grad]: 6.65002e-06 [renormalize]: 6.99947e-08 [add_forward_monad_depend]: 1.80001e-06 [auto_monad_grad]: 1.82999e-06 [auto_monad_eliminator]: 6.64999e-06 [cse]: 1.11e-05 [a_3]: 2.402e-05 [py_interpret_to_execute_after_opt_a]: 1.336e-05 [slice_cell_reuse_recomputed_activation]: 2.16e-06 [rewriter_after_opt_a]: 3.409e-05 [convert_after_rewriter]: 6.70002e-06 [order_py_execute_after_rewriter]: 4.22e-06 [mutable_eliminate]: 0.00075211 [opt_b]: 0.0001634, [1] [Cycle 1]: 0.00015541, [7] [b_1]: 8.329e-05 [b_2]: 5.94e-06 [updatestate_depend_eliminate]: 6.73998e-06 [updatestate_assign_eliminate]: 2.43998e-06 [updatestate_loads_eliminate]: 1.64e-06 [renormalize]: 6.19999e-07 [cse]: 1.762e-05 [optimize_parallel_all_gather_comm]: 1.807e-05 [overlap_param_gather]: 1.80001e-06 [cconv]: 2.903e-05 [loop_unroll]: 0.00050171 [opt_after_cconv]: 0.00010085, [1] [Cycle 1]: 9.406e-05, [7] [c_1]: 2.26e-05 [parameter_eliminate]: 3.73001e-06 [updatestate_depend_eliminate]: 6.11e-06 [updatestate_assign_eliminate]: 2.31e-06 [updatestate_loads_eliminate]: 1.62001e-06 [cse]: 2.005e-05 [renormalize]: 3.39991e-07 [remove_dup_value]: 1.261e-05 [tuple_transform]: 6.696e-05, [1] [Cycle 1]: 6.152e-05, [4] [d_1]: 3.245e-05 [none_parameter_eliminate]: 1.80001e-06 [renormalize]: 2.3999e-07 [switch_simplify]: 5.81e-06 [partial_unused_args_eliminate]: 1.71e-06 [add_recomputation]: 3.994e-05 [cse_after_recomputation]: 2.11e-05, [1] [Cycle 1]: 1.651e-05, [1] [cse]: 1.018e-05 [environ_conv]: 5.59e-06 [swap_dp_allreduce_reducescatter]: 4.80001e-06 [bias_add_comm_swap]: 2.94999e-06 [label_micro_interleaved_index]: 4.94003e-06 [label_fine_grained_interleaved_index]: 2.88e-06 [merge_cast_opt]: 1.37e-06 [slice_recompute_activation]: 2.09999e-06 [micro_interleaved_order_control]: 2.31e-06 [assign_add_opt]: 1.25999e-06 [ForceFp32Comm]: 8.79983e-07 [remove_cast_before_assign_add]: 1.07e-06 [full_micro_interleaved_order_control]: 2.19999e-06 [reorder_send_recv_between_fp_bp]: 2.91999e-06 [comm_op_add_attrs]: 1.44e-06 [add_comm_op_reuse_tag]: 1.27999e-06 [interleave_split_concat_branches]: 1.19e-06 [interleave_parallel_branches]: 1.37999e-06 [overlap_opt_shard_in_pipeline]: 1.22e-06 [overlap_opt_shard_grad_in_pipeline]: 1.63002e-06 [control_data_broadcast_order]: 1.122e-05 [grouped_pairwise_exchange_alltoall]: 1.84e-06 [offloading_packed_experts]: 3.4e-06 [overlap_recompute_and_grad_model_parallel]: 3.95e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.32999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.44998e-06 [overlap_recompute_comm]: 2.21e-06 [overlap_grad_ring_attention]: 3.09999e-06 [overlap_grad_flash_sp]: 1.86e-05 [begin_end_overlap_inline]: 5.00004e-07 [split_matmul_comm_elemetwise]: 2.45997e-06 [split_layernorm_comm]: 1.89e-06 [handle_group_info]: 1.05999e-06 [symbol_engine_optimizer]: 7.784e-05, [1] [Cycle 1]: 7.222e-05, [6] [build]: 3.65e-06 [elim_shapecalc]: 1.168e-05 [elim_not_effective]: 1.026e-05 [opt_reshape]: 5.54998e-06 [fold_const_symbol]: 7.33999e-06 [renormalize]: 1.59984e-07 [detach_backward]: 2.35002e-06 [pipeline_parallel_scheduler]: 1.67001e-06 [auto_monad_reorder]: 1.639e-05 [get_jit_bprop_graph]: 2.22999e-06 [rewriter_after_jit_bprop_graph]: 5.81e-06 [opt_after_jit_grad]: 0.00054508 [validate]: 3.781e-05 Sums bootstrap : 0.000484s : 0.97% type_inference : 0.018544s : 37.10% event_method : 0.000073s : 0.15% auto_monad : 0.000119s : 0.24% graph_reusing : 0.000009s : 0.02% inline : 0.000003s : 0.01% add_attr.add_attr_with_inline.tag_attr : 0.000045s : 0.09% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000011s : 0.02% parallel-infer-symbol : 0.000003s : 0.01% pre_auto_parallel : 0.000062s : 0.12% insert-virtual-dataset : 0.000003s : 0.01% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000046s : 0.09% optimize.rewriter_before_opt_a : 0.000172s : 0.34% optimize.opt_a.expand_dump_flag : 0.000010s : 0.02% optimize.opt_a.switch_simplify : 0.000137s : 0.27% optimize.opt_a.loop_unroll : 0.000115s : 0.23% optimize.opt_a.a_1 : 0.003020s : 6.04% optimize.opt_a.with_stream_mark : 0.000073s : 0.15% optimize.opt_a.recompute_prepare : 0.000047s : 0.09% optimize.opt_a.updatestate_depend_eliminate : 0.000016s : 0.03% optimize.opt_a.updatestate_assign_eliminate : 0.000012s : 0.02% optimize.opt_a.updatestate_loads_eliminate : 0.000011s : 0.02% optimize.opt_a.parameter_eliminate : 0.000006s : 0.01% optimize.opt_a.a_2 : 0.000315s : 0.63% optimize.opt_a.accelerated_algorithm : 0.000028s : 0.06% optimize.opt_a.shard : 0.000006s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000008s : 0.02% optimize.opt_a.shard_inline : 0.000024s : 0.05% optimize.opt_a.merge_send_recv : 0.000030s : 0.06% optimize.opt_a.auto_parallel : 0.000032s : 0.06% optimize.opt_a.parallel : 0.000044s : 0.09% optimize.opt_a.flash_sp : 0.000019s : 0.04% optimize.opt_a.merge_comm : 0.000016s : 0.03% optimize.opt_a.allreduce_fusion : 0.000013s : 0.03% optimize.opt_a.matmul_add_comm_reduction : 0.000050s : 0.10% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000032s : 0.06% optimize.opt_a.virtual_dataset : 0.000024s : 0.05% optimize.opt_a.get_grad_eliminate_ : 0.000024s : 0.05% optimize.opt_a.virtual_output : 0.000023s : 0.05% optimize.opt_a.merge_forward : 0.000016s : 0.03% optimize.opt_a.cell_reuse_recompute_pass : 0.000007s : 0.01% optimize.opt_a.offload_activation : 0.000034s : 0.07% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000055s : 0.11% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.01% optimize.opt_a.before_grad : 0.000041s : 0.08% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000018s : 0.04% optimize.opt_a.meta_fg_expand : 0.018290s : 36.60% optimize.opt_a.flash_sp_send_recv_attached : 0.000009s : 0.02% optimize.opt_a.receive_attached : 0.000007s : 0.01% optimize.opt_a.after_resolve : 0.000109s : 0.22% optimize.opt_a.a_after_grad : 0.000106s : 0.21% optimize.opt_a.renormalize : 0.004364s : 8.73% optimize.opt_a.add_forward_monad_depend : 0.000024s : 0.05% optimize.opt_a.auto_monad_grad : 0.000011s : 0.02% optimize.opt_a.auto_monad_eliminator : 0.000089s : 0.18% optimize.opt_a.cse : 0.000278s : 0.56% optimize.opt_a.a_3 : 0.000411s : 0.82% optimize.py_interpret_to_execute_after_opt_a : 0.000013s : 0.03% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000034s : 0.07% optimize.convert_after_rewriter : 0.000007s : 0.01% optimize.order_py_execute_after_rewriter : 0.000004s : 0.01% optimize.mutable_eliminate : 0.000752s : 1.50% optimize.opt_b.b_1 : 0.000083s : 0.17% optimize.opt_b.b_2 : 0.000006s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000007s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000018s : 0.04% optimize.optimize_parallel_all_gather_comm : 0.000018s : 0.04% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000029s : 0.06% optimize.loop_unroll : 0.000502s : 1.00% optimize.opt_after_cconv.c_1 : 0.000023s : 0.05% optimize.opt_after_cconv.parameter_eliminate : 0.000004s : 0.01% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000020s : 0.04% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000013s : 0.03% optimize.tuple_transform.d_1 : 0.000032s : 0.06% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000006s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000040s : 0.08% optimize.cse_after_recomputation.cse : 0.000010s : 0.02% optimize.environ_conv : 0.000006s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.01% optimize.bias_add_comm_swap : 0.000003s : 0.01% optimize.label_micro_interleaved_index : 0.000005s : 0.01% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.01% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.01% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000011s : 0.02% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000003s : 0.01% optimize.overlap_recompute_and_grad_model_parallel : 0.000004s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000003s : 0.01% optimize.overlap_grad_flash_sp : 0.000019s : 0.04% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000004s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000012s : 0.02% optimize.symbol_engine_optimizer.elim_not_effective : 0.000010s : 0.02% optimize.symbol_engine_optimizer.opt_reshape : 0.000006s : 0.01% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000007s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000016s : 0.03% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000006s : 0.01% opt_after_jit_grad : 0.000545s : 1.09% validate : 0.000038s : 0.08% Time group info: ------[substitution.] 0.000947 143 8.39% : 0.000079s : 5: substitution.arithmetic_simplify 0.15% : 0.000001s : 1: substitution.elim_not_effective 0.57% : 0.000005s : 6: substitution.float_depend_g_call 0.44% : 0.000004s : 2: substitution.float_tuple_getitem_switch 0.11% : 0.000001s : 1: substitution.fold_const_symbol 0.60% : 0.000006s : 2: substitution.graph_param_transform 0.40% : 0.000004s : 2: substitution.incorporate_call 0.23% : 0.000002s : 2: substitution.incorporate_call_switch 62.11% : 0.000588s : 16: substitution.inline 2.56% : 0.000024s : 2: substitution.inline_without_move 0.94% : 0.000009s : 10: substitution.j_node_and_user_rematch 1.26% : 0.000012s : 7: substitution.minmaximum_grad 0.83% : 0.000008s : 6: substitution.partial_eliminate 1.05% : 0.000010s : 10: substitution.remove_not_recompute_node 3.89% : 0.000037s : 10: substitution.replace_applicator 1.38% : 0.000013s : 10: substitution.replace_old_param 0.46% : 0.000004s : 1: substitution.set_cell_output_no_recompute 2.63% : 0.000025s : 7: substitution.tuple_list_convert_item_index_to_positive 1.23% : 0.000012s : 7: substitution.tuple_list_get_item_const_eliminator 1.67% : 0.000016s : 7: substitution.tuple_list_get_item_depend_reorder 7.50% : 0.000071s : 22: substitution.tuple_list_get_item_eliminator 1.62% : 0.000015s : 7: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.018426 2 88.80% : 0.016362s : 1: type_inference.infer 11.20% : 0.002063s : 1: type_inference.specialize ------[replace.] 0.000276 30 3.08% : 0.000009s : 1: replace.arithmetic_simplify 57.96% : 0.000160s : 16: replace.inline 38.96% : 0.000107s : 13: replace.tuple_list_get_item_eliminator ------[match.] 0.000656 30 6.31% : 0.000041s : 1: match.arithmetic_simplify 87.62% : 0.000575s : 16: match.inline 6.07% : 0.000040s : 13: match.tuple_list_get_item_eliminator ------[predicate.] 0.000581 3865 1.12% : 0.000006s : 48: predicate.accumulaten_eliminater 0.18% : 0.000001s : 2: predicate.ad_related_special_op_eliminate 0.38% : 0.000002s : 17: predicate.addn_check_dump 1.09% : 0.000006s : 48: predicate.addn_zero_filter 1.01% : 0.000006s : 48: predicate.adjust_all_reduce_mul_add 2.04% : 0.000012s : 66: predicate.arithmetic_simplify 1.15% : 0.000007s : 49: predicate.cast_eliminate 1.15% : 0.000007s : 48: predicate.check_bprop_eliminate 0.36% : 0.000002s : 17: predicate.compare_switch_simplify 0.03% : 0.000000s : 2: predicate.const_output_eliminate 0.43% : 0.000003s : 17: predicate.depend_value_elim 1.15% : 0.000007s : 49: predicate.dict_get_item_const_eliminator 1.28% : 0.000007s : 49: predicate.dict_get_item_eliminator 1.09% : 0.000006s : 49: predicate.dict_set_item_eliminator 0.30% : 0.000002s : 4: predicate.dumpgradient_eliminate 0.08% : 0.000000s : 2: predicate.elim_not_effective 0.09% : 0.000001s : 2: predicate.elim_shapecalc_of_broadcastargs 1.15% : 0.000007s : 51: predicate.environ_add_const_eliminate 1.08% : 0.000006s : 51: predicate.environ_get_add_eliminate 1.12% : 0.000006s : 51: predicate.environ_get_depend_swap 1.51% : 0.000009s : 68: predicate.environ_get_eliminate 1.07% : 0.000006s : 51: predicate.environ_get_set_eliminate 1.79% : 0.000010s : 78: predicate.exchange_switch_depend_value 2.83% : 0.000016s : 78: predicate.float_depend_g_call 0.36% : 0.000002s : 17: predicate.float_environ_get_switch 0.46% : 0.000003s : 19: predicate.float_tuple_getitem_switch 0.04% : 0.000000s : 2: predicate.fold_const_symbol 0.64% : 0.000004s : 17: predicate.get_grad_eliminate 0.04% : 0.000000s : 2: predicate.graph_param_transform 0.39% : 0.000002s : 17: predicate.incorporate_call 0.36% : 0.000002s : 17: predicate.incorporate_call_switch 6.12% : 0.000036s : 164: predicate.inline 1.63% : 0.000009s : 40: predicate.inline_without_move 0.22% : 0.000001s : 17: predicate.j_node_and_user_rematch 0.64% : 0.000004s : 17: predicate.less_batch_normalization 1.50% : 0.000009s : 66: predicate.list_to_tuple_eliminator_ 2.49% : 0.000014s : 114: predicate.load_eliminater 0.34% : 0.000002s : 2: predicate.loop_unroll_after_grad 2.95% : 0.000017s : 121: predicate.loop_unroll_before_grad 1.38% : 0.000008s : 53: predicate.make_slice_get_slice_eliminator 0.43% : 0.000002s : 17: predicate.merge_addn 1.13% : 0.000007s : 48: predicate.micro_step_allgather_replace 1.11% : 0.000006s : 48: predicate.mini_step_allgather_replace 1.06% : 0.000006s : 49: predicate.minmaximum_grad 0.38% : 0.000002s : 2: predicate.mutable_eliminate 0.07% : 0.000000s : 2: predicate.opt_reshape 0.08% : 0.000000s : 2: predicate.parallel_virtual_node 2.49% : 0.000014s : 78: predicate.partial_defer_inline 1.65% : 0.000010s : 64: predicate.partial_eliminate 1.04% : 0.000006s : 48: predicate.print_const_string_wrapper 0.43% : 0.000002s : 17: predicate.reduce_all_const_elim 1.40% : 0.000008s : 49: predicate.reduce_eliminate 2.42% : 0.000014s : 114: predicate.redundant_stop_gradient_eliminater 0.31% : 0.000002s : 17: predicate.remove_not_recompute_node 2.01% : 0.000012s : 110: predicate.replace_applicator 0.73% : 0.000004s : 40: predicate.replace_old_param 0.12% : 0.000001s : 2: predicate.reset_defer_inline 1.16% : 0.000007s : 49: predicate.reshape_eliminate 1.19% : 0.000007s : 48: predicate.row_tensor_add_zeros_like 0.07% : 0.000000s : 2: predicate.row_tensor_eliminate 1.50% : 0.000009s : 48: predicate.same_eliminate 0.31% : 0.000002s : 17: predicate.set_cell_output_no_recompute 0.61% : 0.000004s : 17: predicate.shard_identity_eliminate 0.15% : 0.000001s : 4: predicate.special_op_eliminate 0.50% : 0.000003s : 17: predicate.specialize_transform 1.44% : 0.000008s : 48: predicate.split_environ_get_set_with_tuple_value 1.32% : 0.000008s : 40: predicate.stack_unstack_eliminate 0.07% : 0.000000s : 2: predicate.switch_call_monad_eliminater 1.99% : 0.000012s : 78: predicate.switch_defer_inline 3.04% : 0.000018s : 126: predicate.switch_layer_defer_inline 5.70% : 0.000033s : 218: predicate.switch_simplify 1.10% : 0.000006s : 49: predicate.tile_eliminate 1.11% : 0.000006s : 49: predicate.transpose_eliminate 1.55% : 0.000009s : 53: predicate.tuple_list_convert_item_index_to_positive 1.50% : 0.000009s : 53: predicate.tuple_list_get_item_const_eliminator 1.33% : 0.000008s : 53: predicate.tuple_list_get_item_depend_reorder 3.10% : 0.000018s : 83: predicate.tuple_list_get_item_eliminator 1.44% : 0.000008s : 53: predicate.tuple_list_get_set_item_eliminator 1.85% : 0.000011s : 70: predicate.tuple_list_set_item_eliminator 1.56% : 0.000009s : 66: predicate.tuple_to_list_eliminator_ 2.41% : 0.000014s : 114: predicate.updatestate_pure_node_eliminater 2.83% : 0.000016s : 131: predicate.updatestate_useless_node_eliminater 0.06% : 0.000000s : 2: predicate.value_based_eliminate 0.47% : 0.000003s : 17: predicate.virtual_dataset_eliminate 0.53% : 0.000003s : 17: predicate.virtual_output_eliminate 0.05% : 0.000000s : 2: predicate.virtual_view_grad_eliminate 0.17% : 0.000001s : 2: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002475 37 61.01% : 0.001510s : 17: func_graph_cloner_run.FuncGraphClonerGraph 38.99% : 0.000965s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.099140 233 0.00% : 0.000004s : 1: ForceFp32Comm 3.82% : 0.003783s : 1: add_attr 3.80% : 0.003769s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.05% : 0.000045s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.13% : 0.000127s : 1: auto_monad 0.02% : 0.000020s : 1: auto_monad_reorder 0.00% : 0.000005s : 1: begin_end_overlap_inline 0.01% : 0.000006s : 1: bias_add_comm_swap 0.52% : 0.000517s : 1: bootstrap 0.03% : 0.000033s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.01% : 0.000015s : 1: control_data_broadcast_order 0.01% : 0.000010s : 1: convert_after_rewriter 0.02% : 0.000024s : 1: cse_after_recomputation 0.01% : 0.000005s : 1: dataset_repeat_opt 0.01% : 0.000006s : 1: detach_backward 0.01% : 0.000009s : 1: environ_conv 0.08% : 0.000082s : 1: event_method 0.01% : 0.000005s : 1: full_micro_interleaved_order_control 0.01% : 0.000006s : 1: get_jit_bprop_graph 0.01% : 0.000014s : 1: graph_reusing 0.01% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.01% : 0.000007s : 1: inline 0.01% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.01% : 0.000007s : 1: label_fine_grained_interleaved_index 0.01% : 0.000008s : 1: label_micro_interleaved_index 0.52% : 0.000511s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.01% : 0.000005s : 1: micro_interleaved_order_control 0.77% : 0.000763s : 1: mutable_eliminate 0.01% : 0.000007s : 1: offloading_packed_experts 0.02% : 0.000015s : 1: opt.transform.loop_unroll_optimizer 0.02% : 0.000016s : 1: opt.transform.mutable_eliminate 4.39% : 0.004355s : 117: opt.transform.opt_a 0.02% : 0.000021s : 1: opt.transform.opt_after_cconv 0.02% : 0.000020s : 1: opt.transform.opt_after_jit_grad 0.06% : 0.000061s : 28: opt.transform.opt_b 0.04% : 0.000036s : 2: opt.transform.opt_trans_graph 0.03% : 0.000031s : 4: opt.transform.symbol_engine_opt 29.20% : 0.028946s : 1: opt_a 0.11% : 0.000105s : 1: opt_after_cconv 0.56% : 0.000555s : 1: opt_after_jit_grad 0.17% : 0.000167s : 1: opt_b 31.64% : 0.031369s : 1: optimize 0.02% : 0.000022s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000008s : 1: order_py_execute_after_rewriter 0.02% : 0.000022s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000006s : 1: overlap_grad_ring_attention 0.00% : 0.000004s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.01% : 0.000006s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000007s : 1: overlap_recompute_and_grad_model_parallel 0.01% : 0.000005s : 1: overlap_recompute_comm 0.01% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.01% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.07% : 0.000067s : 1: pre_auto_parallel 0.05% : 0.000050s : 1: py_interpret_to_execute 0.02% : 0.000017s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.03% : 0.000028s : 1: remove_dup_value 2.42% : 0.002396s : 2: renormalize.infer 1.96% : 0.001944s : 2: renormalize.specialize 0.01% : 0.000007s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000010s : 1: rewriter_after_jit_bprop_graph 0.04% : 0.000038s : 1: rewriter_after_opt_a 0.18% : 0.000177s : 1: rewriter_before_opt_a 0.01% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.01% : 0.000005s : 1: split_matmul_comm_elemetwise 0.01% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.08% : 0.000081s : 1: symbol_engine_optimizer 0.07% : 0.000070s : 1: tuple_transform 18.74% : 0.018574s : 1: type_inference TotalTime = 0.101071, [21] [bootstrap]: 0.00044381 [type_inference]: 0.0906376 [event_method]: 4.679e-05 [auto_monad]: 6.818e-05 [graph_reusing]: 5.96e-06 [inline]: 2.59001e-06 [add_attr]: 0.00364497, [1] [add_attr_with_inline]: 0.00363422, [1] [Cycle 1]: 6.926e-05, [2] [tag_attr]: 2.903e-05 [meta_addattr_fg_expand]: 6.98e-06 [parallel-infer-symbol]: 3.67998e-06 [pre_auto_parallel]: 4.215e-05 [insert-virtual-dataset]: 2.69999e-06 [parallel-infer-symbol-second]: 8.30012e-07 [dataset_repeat_opt]: 2.24999e-06 [pipeline_split]: 1.64e-06 [optimize]: 0.00545954, [53] [py_interpret_to_execute]: 3.124e-05 [rewriter_before_opt_a]: 0.00010363 [opt_a]: 0.00332237, [2] [Cycle 1]: 0.0025189, [45] [expand_dump_flag]: 3.74002e-06 [switch_simplify]: 5.191e-05 [loop_unroll]: 3.947e-05 [a_1]: 0.00087245 [with_stream_mark]: 1.666e-05 [recompute_prepare]: 8.55999e-06 [updatestate_depend_eliminate]: 4.14997e-06 [updatestate_assign_eliminate]: 3.23e-06 [updatestate_loads_eliminate]: 2.73e-06 [parameter_eliminate]: 2.14e-06 [a_2]: 8.052e-05 [accelerated_algorithm]: 6.86001e-06 [shard]: 2.04e-06 [meta_shard_fg_expand]: 1.72001e-06 [shard_inline]: 6.05002e-06 [merge_send_recv]: 8.78001e-06 [auto_parallel]: 7.11001e-06 [parallel]: 1.842e-05 [flash_sp]: 8.38999e-06 [merge_comm]: 4.09002e-06 [allreduce_fusion]: 3.7e-06 [matmul_add_comm_reduction]: 9.73998e-06 [allreduce_slice_to_reducescatter]: 7.79983e-07 [virtual_shard_identity]: 7.51999e-06 [virtual_dataset]: 6.46e-06 [get_grad_eliminate_]: 6.21998e-06 [virtual_output]: 6.14001e-06 [merge_forward]: 3.83999e-06 [cell_reuse_recompute_pass]: 1.00999e-06 [offload_activation]: 1.074e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.188e-05 [merge_recompute_call_nodes]: 1.79e-06 [before_grad]: 1.032e-05 [set_forward_comm_id_for_comm_node_pass]: 3.53e-06 [meta_fg_expand]: 2.78e-06 [flash_sp_send_recv_attached]: 2.45002e-06 [receive_attached]: 2.16e-06 [after_resolve]: 1.194e-05 [a_after_grad]: 9.36e-06 [renormalize]: 0.00092758 [add_forward_monad_depend]: 5.56998e-06 [auto_monad_grad]: 2.43e-06 [auto_monad_eliminator]: 1.536e-05 [cse]: 2.198e-05 [a_3]: 4.655e-05 [Cycle 2]: 0.00079316, [45] [expand_dump_flag]: 1.72001e-06 [switch_simplify]: 7.68999e-06 [loop_unroll]: 5.84999e-06 [a_1]: 0.00014635 [with_stream_mark]: 1.322e-05 [recompute_prepare]: 6.66e-06 [updatestate_depend_eliminate]: 3.13e-06 [updatestate_assign_eliminate]: 2.69999e-06 [updatestate_loads_eliminate]: 2.93e-06 [parameter_eliminate]: 1.09e-06 [a_2]: 7.103e-05 [accelerated_algorithm]: 6.41e-06 [shard]: 1.32999e-06 [meta_shard_fg_expand]: 1.50001e-06 [shard_inline]: 5.94999e-06 [merge_send_recv]: 5.59e-06 [auto_parallel]: 5.49e-06 [parallel]: 6.31998e-06 [flash_sp]: 3.63e-06 [merge_comm]: 2.93e-06 [allreduce_fusion]: 2.86e-06 [matmul_add_comm_reduction]: 6.46e-06 [allreduce_slice_to_reducescatter]: 4.90021e-07 [virtual_shard_identity]: 6.44999e-06 [virtual_dataset]: 5.46e-06 [get_grad_eliminate_]: 5.49998e-06 [virtual_output]: 5.05001e-06 [merge_forward]: 3.32002e-06 [cell_reuse_recompute_pass]: 1.56998e-06 [offload_activation]: 6.81999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.109e-05 [merge_recompute_call_nodes]: 1.19e-06 [before_grad]: 8.33001e-06 [set_forward_comm_id_for_comm_node_pass]: 3.36999e-06 [meta_fg_expand]: 2.48e-06 [flash_sp_send_recv_attached]: 1.01002e-06 [receive_attached]: 1.42e-06 [after_resolve]: 1.017e-05 [a_after_grad]: 8.50999e-06 [renormalize]: 8.00064e-08 [add_forward_monad_depend]: 1.10001e-06 [auto_monad_grad]: 1.14e-06 [auto_monad_eliminator]: 6.96999e-06 [cse]: 6.406e-05 [a_3]: 4.024e-05 [py_interpret_to_execute_after_opt_a]: 1.369e-05 [slice_cell_reuse_recomputed_activation]: 2.27999e-06 [rewriter_after_opt_a]: 3.657e-05 [convert_after_rewriter]: 7.63999e-06 [order_py_execute_after_rewriter]: 4.81002e-06 [mutable_eliminate]: 0.00058859 [opt_b]: 0.00019387, [1] [Cycle 1]: 0.00018679, [7] [b_1]: 0.00011397 [b_2]: 7.65998e-06 [updatestate_depend_eliminate]: 6.31e-06 [updatestate_assign_eliminate]: 2.70997e-06 [updatestate_loads_eliminate]: 2.46e-06 [renormalize]: 5.70028e-07 [cse]: 1.798e-05 [optimize_parallel_all_gather_comm]: 1.684e-05 [overlap_param_gather]: 1.84998e-06 [cconv]: 2.877e-05 [loop_unroll]: 0.00043523 [opt_after_cconv]: 0.0001009, [1] [Cycle 1]: 9.496e-05, [7] [c_1]: 3.032e-05 [parameter_eliminate]: 2.81e-06 [updatestate_depend_eliminate]: 5.38002e-06 [updatestate_assign_eliminate]: 3.01999e-06 [updatestate_loads_eliminate]: 2.26e-06 [cse]: 1.725e-05 [renormalize]: 4.10015e-07 [remove_dup_value]: 1.326e-05 [tuple_transform]: 7.438e-05, [1] [Cycle 1]: 7.02e-05, [4] [d_1]: 4.305e-05 [none_parameter_eliminate]: 1.39e-06 [renormalize]: 2.30008e-07 [switch_simplify]: 6.74001e-06 [partial_unused_args_eliminate]: 1.71e-06 [add_recomputation]: 4.759e-05 [cse_after_recomputation]: 2.254e-05, [1] [Cycle 1]: 1.774e-05, [1] [cse]: 1.18e-05 [environ_conv]: 4.76997e-06 [swap_dp_allreduce_reducescatter]: 5.51e-06 [bias_add_comm_swap]: 2.61999e-06 [label_micro_interleaved_index]: 4.92e-06 [label_fine_grained_interleaved_index]: 2.70997e-06 [merge_cast_opt]: 1.42999e-06 [slice_recompute_activation]: 1.97001e-06 [micro_interleaved_order_control]: 2.32999e-06 [assign_add_opt]: 1.23002e-06 [ForceFp32Comm]: 7.80012e-07 [remove_cast_before_assign_add]: 1.18001e-06 [full_micro_interleaved_order_control]: 2.73e-06 [reorder_send_recv_between_fp_bp]: 2.83998e-06 [comm_op_add_attrs]: 1.10999e-06 [add_comm_op_reuse_tag]: 1.05999e-06 [interleave_split_concat_branches]: 1.27e-06 [interleave_parallel_branches]: 1.10001e-06 [overlap_opt_shard_in_pipeline]: 1.18001e-06 [overlap_opt_shard_grad_in_pipeline]: 1.66e-06 [control_data_broadcast_order]: 1.247e-05 [grouped_pairwise_exchange_alltoall]: 1.53002e-06 [offloading_packed_experts]: 4.07e-06 [overlap_recompute_and_grad_model_parallel]: 4.71002e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.30999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.37999e-06 [overlap_recompute_comm]: 2.36998e-06 [overlap_grad_ring_attention]: 4.16001e-06 [overlap_grad_flash_sp]: 1.872e-05 [begin_end_overlap_inline]: 5.29981e-07 [split_matmul_comm_elemetwise]: 1.94999e-06 [split_layernorm_comm]: 2.01998e-06 [handle_group_info]: 1.59e-06 [symbol_engine_optimizer]: 7.588e-05, [1] [Cycle 1]: 7.139e-05, [6] [build]: 3.21001e-06 [elim_shapecalc]: 1.124e-05 [elim_not_effective]: 1.292e-05 [opt_reshape]: 6.43998e-06 [fold_const_symbol]: 9.16002e-06 [renormalize]: 2.09984e-07 [detach_backward]: 2.29001e-06 [pipeline_parallel_scheduler]: 1.60999e-06 [auto_monad_reorder]: 1.638e-05 [get_jit_bprop_graph]: 1.99999e-06 [rewriter_after_jit_bprop_graph]: 4.72e-06 [opt_after_jit_grad]: 0.00047355 [validate]: 3.759e-05 Sums bootstrap : 0.000444s : 0.46% type_inference : 0.090638s : 94.05% event_method : 0.000047s : 0.05% auto_monad : 0.000068s : 0.07% graph_reusing : 0.000006s : 0.01% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000029s : 0.03% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.01% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000042s : 0.04% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000031s : 0.03% optimize.rewriter_before_opt_a : 0.000104s : 0.11% optimize.opt_a.expand_dump_flag : 0.000005s : 0.01% optimize.opt_a.switch_simplify : 0.000060s : 0.06% optimize.opt_a.loop_unroll : 0.000045s : 0.05% optimize.opt_a.a_1 : 0.001019s : 1.06% optimize.opt_a.with_stream_mark : 0.000030s : 0.03% optimize.opt_a.recompute_prepare : 0.000015s : 0.02% optimize.opt_a.updatestate_depend_eliminate : 0.000007s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.01% optimize.opt_a.parameter_eliminate : 0.000003s : 0.00% optimize.opt_a.a_2 : 0.000152s : 0.16% optimize.opt_a.accelerated_algorithm : 0.000013s : 0.01% optimize.opt_a.shard : 0.000003s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000003s : 0.00% optimize.opt_a.shard_inline : 0.000012s : 0.01% optimize.opt_a.merge_send_recv : 0.000014s : 0.01% optimize.opt_a.auto_parallel : 0.000013s : 0.01% optimize.opt_a.parallel : 0.000025s : 0.03% optimize.opt_a.flash_sp : 0.000012s : 0.01% optimize.opt_a.merge_comm : 0.000007s : 0.01% optimize.opt_a.allreduce_fusion : 0.000007s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000016s : 0.02% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000014s : 0.01% optimize.opt_a.virtual_dataset : 0.000012s : 0.01% optimize.opt_a.get_grad_eliminate_ : 0.000012s : 0.01% optimize.opt_a.virtual_output : 0.000011s : 0.01% optimize.opt_a.merge_forward : 0.000007s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000003s : 0.00% optimize.opt_a.offload_activation : 0.000018s : 0.02% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000023s : 0.02% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.00% optimize.opt_a.before_grad : 0.000019s : 0.02% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000007s : 0.01% optimize.opt_a.meta_fg_expand : 0.000005s : 0.01% optimize.opt_a.flash_sp_send_recv_attached : 0.000003s : 0.00% optimize.opt_a.receive_attached : 0.000004s : 0.00% optimize.opt_a.after_resolve : 0.000022s : 0.02% optimize.opt_a.a_after_grad : 0.000018s : 0.02% optimize.opt_a.renormalize : 0.000928s : 0.96% optimize.opt_a.add_forward_monad_depend : 0.000007s : 0.01% optimize.opt_a.auto_monad_grad : 0.000004s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000022s : 0.02% optimize.opt_a.cse : 0.000086s : 0.09% optimize.opt_a.a_3 : 0.000087s : 0.09% optimize.py_interpret_to_execute_after_opt_a : 0.000014s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000037s : 0.04% optimize.convert_after_rewriter : 0.000008s : 0.01% optimize.order_py_execute_after_rewriter : 0.000005s : 0.00% optimize.mutable_eliminate : 0.000589s : 0.61% optimize.opt_b.b_1 : 0.000114s : 0.12% optimize.opt_b.b_2 : 0.000008s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000006s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000018s : 0.02% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.02% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000029s : 0.03% optimize.loop_unroll : 0.000435s : 0.45% optimize.opt_after_cconv.c_1 : 0.000030s : 0.03% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000005s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000017s : 0.02% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000013s : 0.01% optimize.tuple_transform.d_1 : 0.000043s : 0.04% optimize.tuple_transform.none_parameter_eliminate : 0.000001s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000048s : 0.05% optimize.cse_after_recomputation.cse : 0.000012s : 0.01% optimize.environ_conv : 0.000005s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.01% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.01% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000003s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000012s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000019s : 0.02% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000002s : 0.00% optimize.symbol_engine_optimizer.build : 0.000003s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000011s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000013s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000006s : 0.01% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000009s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000016s : 0.02% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000005s : 0.00% opt_after_jit_grad : 0.000474s : 0.49% validate : 0.000038s : 0.04% Time group info: ------[substitution.] 0.000276 40 12.73% : 0.000035s : 8: substitution.arithmetic_simplify 0.71% : 0.000002s : 2: substitution.elim_not_effective 0.44% : 0.000001s : 2: substitution.fold_const_symbol 2.30% : 0.000006s : 4: substitution.graph_param_transform 72.29% : 0.000200s : 6: substitution.inline 1.14% : 0.000003s : 4: substitution.j_node_and_user_rematch 1.77% : 0.000005s : 4: substitution.remove_not_recompute_node 1.79% : 0.000005s : 4: substitution.replace_old_param 6.82% : 0.000019s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.090556 2 98.65% : 0.089333s : 1: type_inference.infer 1.35% : 0.001223s : 1: type_inference.specialize ------[replace.] 0.000088 12 59.75% : 0.000053s : 6: replace.inline 40.25% : 0.000035s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000212 12 92.37% : 0.000196s : 6: match.inline 7.63% : 0.000016s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000240 1534 0.92% : 0.000002s : 17: predicate.accumulaten_eliminater 0.58% : 0.000001s : 4: predicate.ad_related_special_op_eliminate 0.37% : 0.000001s : 8: predicate.addn_check_dump 0.98% : 0.000002s : 17: predicate.addn_zero_filter 0.89% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.22% : 0.000005s : 25: predicate.arithmetic_simplify 0.98% : 0.000002s : 17: predicate.cast_eliminate 0.49% : 0.000001s : 8: predicate.check_bprop_eliminate 0.39% : 0.000001s : 8: predicate.compare_switch_simplify 0.17% : 0.000000s : 4: predicate.const_output_eliminate 0.43% : 0.000001s : 8: predicate.depend_value_elim 0.97% : 0.000002s : 17: predicate.dict_get_item_const_eliminator 1.19% : 0.000003s : 17: predicate.dict_get_item_eliminator 0.94% : 0.000002s : 17: predicate.dict_set_item_eliminator 0.91% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.21% : 0.000001s : 4: predicate.elim_not_effective 0.33% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.14% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.17% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.10% : 0.000003s : 21: predicate.environ_get_depend_swap 1.58% : 0.000004s : 29: predicate.environ_get_eliminate 1.09% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.61% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.53% : 0.000006s : 29: predicate.float_depend_g_call 0.43% : 0.000001s : 8: predicate.float_environ_get_switch 0.62% : 0.000001s : 12: predicate.float_tuple_getitem_switch 0.15% : 0.000000s : 4: predicate.fold_const_symbol 0.61% : 0.000001s : 8: predicate.get_grad_eliminate 0.17% : 0.000000s : 4: predicate.graph_param_transform 0.44% : 0.000001s : 8: predicate.incorporate_call 0.38% : 0.000001s : 8: predicate.incorporate_call_switch 5.82% : 0.000014s : 70: predicate.inline 0.60% : 0.000001s : 8: predicate.inline_without_move 0.27% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.81% : 0.000002s : 8: predicate.less_batch_normalization 1.78% : 0.000004s : 31: predicate.list_to_tuple_eliminator_ 2.46% : 0.000006s : 48: predicate.load_eliminater 0.87% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.69% : 0.000006s : 47: predicate.loop_unroll_before_grad 1.51% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.43% : 0.000001s : 8: predicate.merge_addn 0.42% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.47% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.92% : 0.000002s : 17: predicate.minmaximum_grad 0.87% : 0.000002s : 4: predicate.mutable_eliminate 0.27% : 0.000001s : 4: predicate.opt_reshape 0.26% : 0.000001s : 4: predicate.parallel_virtual_node 2.15% : 0.000005s : 29: predicate.partial_defer_inline 1.65% : 0.000004s : 27: predicate.partial_eliminate 0.97% : 0.000002s : 17: predicate.print_const_string_wrapper 0.44% : 0.000001s : 8: predicate.reduce_all_const_elim 1.21% : 0.000003s : 17: predicate.reduce_eliminate 2.52% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.32% : 0.000001s : 8: predicate.remove_not_recompute_node 1.43% : 0.000003s : 31: predicate.replace_applicator 0.47% : 0.000001s : 8: predicate.replace_old_param 0.22% : 0.000001s : 4: predicate.reset_defer_inline 1.02% : 0.000002s : 17: predicate.reshape_eliminate 0.46% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.28% : 0.000001s : 4: predicate.row_tensor_eliminate 0.74% : 0.000002s : 8: predicate.same_eliminate 0.33% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.72% : 0.000002s : 8: predicate.shard_identity_eliminate 0.55% : 0.000001s : 8: predicate.special_op_eliminate 0.74% : 0.000002s : 8: predicate.specialize_transform 0.88% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.69% : 0.000002s : 8: predicate.stack_unstack_eliminate 0.28% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.74% : 0.000004s : 29: predicate.switch_defer_inline 2.10% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.35% : 0.000013s : 88: predicate.switch_simplify 0.93% : 0.000002s : 17: predicate.tile_eliminate 1.00% : 0.000002s : 17: predicate.transpose_eliminate 1.51% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.51% : 0.000004s : 25: predicate.tuple_list_get_item_const_eliminator 5.61% : 0.000013s : 25: predicate.tuple_list_get_item_depend_reorder 3.26% : 0.000008s : 39: predicate.tuple_list_get_item_eliminator 1.35% : 0.000003s : 25: predicate.tuple_list_get_set_item_eliminator 2.02% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 1.74% : 0.000004s : 31: predicate.tuple_to_list_eliminator_ 2.43% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 2.97% : 0.000007s : 56: predicate.updatestate_useless_node_eliminater 0.27% : 0.000001s : 4: predicate.value_based_eliminate 0.58% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.48% : 0.000001s : 8: predicate.virtual_output_eliminate 0.22% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.40% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000813 15 50.71% : 0.000412s : 7: func_graph_cloner_run.FuncGraphClonerGraph 49.29% : 0.000401s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.112675 192 0.00% : 0.000004s : 1: ForceFp32Comm 3.24% : 0.003651s : 1: add_attr 3.23% : 0.003639s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.05% : 0.000052s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.07% : 0.000074s : 1: auto_monad 0.02% : 0.000020s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.01% : 0.000006s : 1: bias_add_comm_swap 0.42% : 0.000475s : 1: bootstrap 0.03% : 0.000032s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.01% : 0.000016s : 1: control_data_broadcast_order 0.01% : 0.000011s : 1: convert_after_rewriter 0.02% : 0.000026s : 1: cse_after_recomputation 0.00% : 0.000006s : 1: dataset_repeat_opt 0.01% : 0.000006s : 1: detach_backward 0.01% : 0.000008s : 1: environ_conv 0.05% : 0.000055s : 1: event_method 0.00% : 0.000006s : 1: full_micro_interleaved_order_control 0.00% : 0.000006s : 1: get_jit_bprop_graph 0.01% : 0.000010s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.01% : 0.000006s : 1: inline 0.01% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.01% : 0.000008s : 1: label_micro_interleaved_index 0.39% : 0.000443s : 1: loop_unroll 0.00% : 0.000005s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.53% : 0.000597s : 1: mutable_eliminate 0.01% : 0.000007s : 1: offloading_packed_experts 0.01% : 0.000013s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000014s : 1: opt.transform.mutable_eliminate 1.28% : 0.001447s : 78: opt.transform.opt_a 0.03% : 0.000029s : 1: opt.transform.opt_after_cconv 0.02% : 0.000025s : 1: opt.transform.opt_after_jit_grad 0.08% : 0.000095s : 28: opt.transform.opt_b 0.04% : 0.000048s : 2: opt.transform.opt_trans_graph 0.03% : 0.000036s : 4: opt.transform.symbol_engine_opt 2.95% : 0.003325s : 1: opt_a 0.09% : 0.000104s : 1: opt_after_cconv 0.43% : 0.000482s : 1: opt_after_jit_grad 0.17% : 0.000197s : 1: opt_b 4.85% : 0.005464s : 1: optimize 0.02% : 0.000020s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000008s : 1: order_py_execute_after_rewriter 0.02% : 0.000022s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000004s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000005s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000006s : 1: overlap_recompute_comm 0.01% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.04% : 0.000047s : 1: pre_auto_parallel 0.03% : 0.000035s : 1: py_interpret_to_execute 0.02% : 0.000017s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.01% : 0.000017s : 1: remove_dup_value 0.41% : 0.000467s : 1: renormalize.infer 0.40% : 0.000451s : 1: renormalize.specialize 0.01% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000008s : 1: rewriter_after_jit_bprop_graph 0.04% : 0.000040s : 1: rewriter_after_opt_a 0.10% : 0.000108s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.01% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.07% : 0.000079s : 1: symbol_engine_optimizer 0.08% : 0.000089s : 1: tuple_transform 80.47% : 0.090665s : 1: type_inference TotalTime = 0.256578, [21] [bootstrap]: 0.0845151 [type_inference]: 0.0160141 [event_method]: 6.657e-05 [auto_monad]: 0.00013153 [graph_reusing]: 8.72e-06 [inline]: 2.41998e-06 [add_attr]: 0.00342352, [1] [add_attr_with_inline]: 0.00341322, [1] [Cycle 1]: 8.804e-05, [2] [tag_attr]: 4.272e-05 [meta_addattr_fg_expand]: 1.05e-05 [parallel-infer-symbol]: 3.8e-06 [pre_auto_parallel]: 0.0122225 [insert-virtual-dataset]: 3.35e-06 [parallel-infer-symbol-second]: 3.6e-06 [dataset_repeat_opt]: 3.49001e-06 [pipeline_split]: 1.76e-06 [optimize]: 0.139366, [53] [py_interpret_to_execute]: 8.621e-05 [rewriter_before_opt_a]: 0.00020605 [opt_a]: 0.136912, [3] [Cycle 1]: 0.00907849, [45] [expand_dump_flag]: 6.84001e-06 [switch_simplify]: 8.711e-05 [loop_unroll]: 6.669e-05 [a_1]: 0.00164573 [with_stream_mark]: 3.703e-05 [recompute_prepare]: 2.707e-05 [updatestate_depend_eliminate]: 1.105e-05 [updatestate_assign_eliminate]: 8.67e-06 [updatestate_loads_eliminate]: 8.12e-06 [parameter_eliminate]: 3.23e-06 [a_2]: 0.00025228 [accelerated_algorithm]: 3.815e-05 [shard]: 2.09e-06 [meta_shard_fg_expand]: 5.29e-06 [shard_inline]: 1.612e-05 [merge_send_recv]: 1.879e-05 [auto_parallel]: 1.534e-05 [parallel]: 2.058e-05 [flash_sp]: 1.336e-05 [merge_comm]: 9.76e-06 [allreduce_fusion]: 9.42001e-06 [matmul_add_comm_reduction]: 3.388e-05 [allreduce_slice_to_reducescatter]: 9.09989e-07 [virtual_shard_identity]: 1.966e-05 [virtual_dataset]: 1.631e-05 [get_grad_eliminate_]: 1.662e-05 [virtual_output]: 1.531e-05 [merge_forward]: 1.093e-05 [cell_reuse_recompute_pass]: 2.29001e-06 [offload_activation]: 2.023e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.119e-05 [merge_recompute_call_nodes]: 1.81e-06 [before_grad]: 2.925e-05 [set_forward_comm_id_for_comm_node_pass]: 1.058e-05 [meta_fg_expand]: 0.00201011 [flash_sp_send_recv_attached]: 7.41001e-06 [receive_attached]: 2.66e-06 [after_resolve]: 7.792e-05 [a_after_grad]: 9.317e-05 [renormalize]: 0.00334364 [add_forward_monad_depend]: 1.349e-05 [auto_monad_grad]: 7.2e-06 [auto_monad_eliminator]: 6.233e-05 [cse]: 0.0002388 [a_3]: 0.00035032 [Cycle 2]: 0.127126, [45] [expand_dump_flag]: 3.46999e-06 [switch_simplify]: 4.976e-05 [loop_unroll]: 4.355e-05 [a_1]: 0.00164738 [with_stream_mark]: 1.532e-05 [recompute_prepare]: 8.94998e-06 [updatestate_depend_eliminate]: 4.28001e-06 [updatestate_assign_eliminate]: 3.86999e-06 [updatestate_loads_eliminate]: 2.60002e-06 [parameter_eliminate]: 1.82001e-06 [a_2]: 7.159e-05 [accelerated_algorithm]: 1.213e-05 [shard]: 2.41998e-06 [meta_shard_fg_expand]: 2.59999e-06 [shard_inline]: 6.68e-06 [merge_send_recv]: 7.7e-06 [auto_parallel]: 9.20001e-06 [parallel]: 9.77001e-06 [flash_sp]: 3.7e-06 [merge_comm]: 3.73001e-06 [allreduce_fusion]: 3.2e-06 [matmul_add_comm_reduction]: 8.96002e-06 [allreduce_slice_to_reducescatter]: 7.89994e-07 [virtual_shard_identity]: 8.91002e-06 [virtual_dataset]: 7.38e-06 [get_grad_eliminate_]: 6.96001e-06 [virtual_output]: 6.21e-06 [merge_forward]: 4.13001e-06 [cell_reuse_recompute_pass]: 1.57001e-06 [offload_activation]: 1.011e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.135e-05 [merge_recompute_call_nodes]: 1.50001e-06 [before_grad]: 9.36e-06 [set_forward_comm_id_for_comm_node_pass]: 3.66001e-06 [meta_fg_expand]: 0.00011512 [flash_sp_send_recv_attached]: 1.62999e-06 [receive_attached]: 1.87999e-06 [after_resolve]: 1.391e-05 [a_after_grad]: 1.019e-05 [renormalize]: 0.124601 [add_forward_monad_depend]: 8.65001e-06 [auto_monad_grad]: 2.31998e-06 [auto_monad_eliminator]: 1.663e-05 [cse]: 2.146e-05 [a_3]: 5.977e-05 [Cycle 3]: 0.00068265, [45] [expand_dump_flag]: 2.63e-06 [switch_simplify]: 9.07999e-06 [loop_unroll]: 6.88998e-06 [a_1]: 0.00013453 [with_stream_mark]: 1.564e-05 [recompute_prepare]: 6.33e-06 [updatestate_depend_eliminate]: 3.50998e-06 [updatestate_assign_eliminate]: 3.28998e-06 [updatestate_loads_eliminate]: 2.74999e-06 [parameter_eliminate]: 1.92999e-06 [a_2]: 6.827e-05 [accelerated_algorithm]: 1.533e-05 [shard]: 2.04e-06 [meta_shard_fg_expand]: 2.32999e-06 [shard_inline]: 6.08002e-06 [merge_send_recv]: 8.35999e-06 [auto_parallel]: 1.044e-05 [parallel]: 1.018e-05 [flash_sp]: 2.01e-06 [merge_comm]: 3.06999e-06 [allreduce_fusion]: 3.56001e-06 [matmul_add_comm_reduction]: 7.67998e-06 [allreduce_slice_to_reducescatter]: 6.30011e-07 [virtual_shard_identity]: 7.26001e-06 [virtual_dataset]: 5.75001e-06 [get_grad_eliminate_]: 5.57001e-06 [virtual_output]: 5.66e-06 [merge_forward]: 4.48999e-06 [cell_reuse_recompute_pass]: 3.48e-06 [offload_activation]: 9.80002e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.191e-05 [merge_recompute_call_nodes]: 1.68002e-06 [before_grad]: 9.59e-06 [set_forward_comm_id_for_comm_node_pass]: 3.86999e-06 [meta_fg_expand]: 2.64001e-06 [flash_sp_send_recv_attached]: 1.73002e-06 [receive_attached]: 2.36e-06 [after_resolve]: 1.232e-05 [a_after_grad]: 9.10001e-06 [renormalize]: 1.00001e-07 [add_forward_monad_depend]: 1.29e-06 [auto_monad_grad]: 8.70001e-07 [auto_monad_eliminator]: 6.74001e-06 [cse]: 1.412e-05 [a_3]: 3.408e-05 [py_interpret_to_execute_after_opt_a]: 1.809e-05 [slice_cell_reuse_recomputed_activation]: 2.01998e-06 [rewriter_after_opt_a]: 3.776e-05 [convert_after_rewriter]: 7.42002e-06 [order_py_execute_after_rewriter]: 5.59e-06 [mutable_eliminate]: 0.00070513 [opt_b]: 0.00019768, [1] [Cycle 1]: 0.00019032, [7] [b_1]: 0.00011713 [b_2]: 7.51999e-06 [updatestate_depend_eliminate]: 5.91e-06 [updatestate_assign_eliminate]: 2.86999e-06 [updatestate_loads_eliminate]: 2.56e-06 [renormalize]: 5.69999e-07 [cse]: 1.721e-05 [optimize_parallel_all_gather_comm]: 2.804e-05 [overlap_param_gather]: 2.04e-06 [cconv]: 2.172e-05 [loop_unroll]: 0.00044627 [opt_after_cconv]: 0.00010711, [1] [Cycle 1]: 0.00010063, [7] [c_1]: 3.251e-05 [parameter_eliminate]: 3.21999e-06 [updatestate_depend_eliminate]: 5.87001e-06 [updatestate_assign_eliminate]: 2.98998e-06 [updatestate_loads_eliminate]: 2.78e-06 [cse]: 1.769e-05 [renormalize]: 4.99975e-07 [remove_dup_value]: 1.25e-05 [tuple_transform]: 7.541e-05, [1] [Cycle 1]: 7.063e-05, [4] [d_1]: 4.288e-05 [none_parameter_eliminate]: 1.57001e-06 [renormalize]: 1.69995e-07 [switch_simplify]: 7.46001e-06 [partial_unused_args_eliminate]: 1.61998e-06 [add_recomputation]: 4.266e-05 [cse_after_recomputation]: 2.358e-05, [1] [Cycle 1]: 1.844e-05, [1] [cse]: 1.219e-05 [environ_conv]: 8.19002e-06 [swap_dp_allreduce_reducescatter]: 6.10002e-06 [bias_add_comm_swap]: 2.44999e-06 [label_micro_interleaved_index]: 4.84e-06 [label_fine_grained_interleaved_index]: 2.71999e-06 [merge_cast_opt]: 1.24998e-06 [slice_recompute_activation]: 2.03002e-06 [micro_interleaved_order_control]: 2.49999e-06 [assign_add_opt]: 1.22e-06 [ForceFp32Comm]: 9.00007e-07 [remove_cast_before_assign_add]: 1.02998e-06 [full_micro_interleaved_order_control]: 2.53003e-06 [reorder_send_recv_between_fp_bp]: 2.81e-06 [comm_op_add_attrs]: 1.09003e-06 [add_comm_op_reuse_tag]: 9.79984e-07 [interleave_split_concat_branches]: 1.20999e-06 [interleave_parallel_branches]: 1.14998e-06 [overlap_opt_shard_in_pipeline]: 1.19998e-06 [overlap_opt_shard_grad_in_pipeline]: 1.60999e-06 [control_data_broadcast_order]: 1.226e-05 [grouped_pairwise_exchange_alltoall]: 1.96e-06 [offloading_packed_experts]: 3.74002e-06 [overlap_recompute_and_grad_model_parallel]: 4.64998e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.22e-06 [overlap_recompute_allgather_and_fa_grad]: 1.34e-06 [overlap_recompute_comm]: 2.14e-06 [overlap_grad_ring_attention]: 4.22e-06 [overlap_grad_flash_sp]: 2.052e-05 [begin_end_overlap_inline]: 5.19998e-07 [split_matmul_comm_elemetwise]: 2.51e-06 [split_layernorm_comm]: 2.26e-06 [handle_group_info]: 1.46002e-06 [symbol_engine_optimizer]: 8.207e-05, [1] [Cycle 1]: 7.663e-05, [6] [build]: 8.23001e-06 [elim_shapecalc]: 1.012e-05 [elim_not_effective]: 1.328e-05 [opt_reshape]: 6.82002e-06 [fold_const_symbol]: 9.34998e-06 [renormalize]: 2.10013e-07 [detach_backward]: 2.34999e-06 [pipeline_parallel_scheduler]: 1.69e-06 [auto_monad_reorder]: 1.781e-05 [get_jit_bprop_graph]: 2.88e-06 [rewriter_after_jit_bprop_graph]: 4.02e-06 [opt_after_jit_grad]: 0.0004983 [validate]: 3.987e-05 Sums bootstrap : 0.084515s : 33.58% type_inference : 0.016014s : 6.36% event_method : 0.000067s : 0.03% auto_monad : 0.000132s : 0.05% graph_reusing : 0.000009s : 0.00% inline : 0.000002s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000043s : 0.02% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000010s : 0.00% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.012222s : 4.86% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000004s : 0.00% dataset_repeat_opt : 0.000003s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000086s : 0.03% optimize.rewriter_before_opt_a : 0.000206s : 0.08% optimize.opt_a.expand_dump_flag : 0.000013s : 0.01% optimize.opt_a.switch_simplify : 0.000146s : 0.06% optimize.opt_a.loop_unroll : 0.000117s : 0.05% optimize.opt_a.a_1 : 0.003428s : 1.36% optimize.opt_a.with_stream_mark : 0.000068s : 0.03% optimize.opt_a.recompute_prepare : 0.000042s : 0.02% optimize.opt_a.updatestate_depend_eliminate : 0.000019s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000016s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000013s : 0.01% optimize.opt_a.parameter_eliminate : 0.000007s : 0.00% optimize.opt_a.a_2 : 0.000392s : 0.16% optimize.opt_a.accelerated_algorithm : 0.000066s : 0.03% optimize.opt_a.shard : 0.000007s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000010s : 0.00% optimize.opt_a.shard_inline : 0.000029s : 0.01% optimize.opt_a.merge_send_recv : 0.000035s : 0.01% optimize.opt_a.auto_parallel : 0.000035s : 0.01% optimize.opt_a.parallel : 0.000041s : 0.02% optimize.opt_a.flash_sp : 0.000019s : 0.01% optimize.opt_a.merge_comm : 0.000017s : 0.01% optimize.opt_a.allreduce_fusion : 0.000016s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000051s : 0.02% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000036s : 0.01% optimize.opt_a.virtual_dataset : 0.000029s : 0.01% optimize.opt_a.get_grad_eliminate_ : 0.000029s : 0.01% optimize.opt_a.virtual_output : 0.000027s : 0.01% optimize.opt_a.merge_forward : 0.000020s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000007s : 0.00% optimize.opt_a.offload_activation : 0.000040s : 0.02% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000054s : 0.02% optimize.opt_a.merge_recompute_call_nodes : 0.000005s : 0.00% optimize.opt_a.before_grad : 0.000048s : 0.02% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000018s : 0.01% optimize.opt_a.meta_fg_expand : 0.002128s : 0.85% optimize.opt_a.flash_sp_send_recv_attached : 0.000011s : 0.00% optimize.opt_a.receive_attached : 0.000007s : 0.00% optimize.opt_a.after_resolve : 0.000104s : 0.04% optimize.opt_a.a_after_grad : 0.000112s : 0.04% optimize.opt_a.renormalize : 0.127945s : 50.83% optimize.opt_a.add_forward_monad_depend : 0.000023s : 0.01% optimize.opt_a.auto_monad_grad : 0.000010s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000086s : 0.03% optimize.opt_a.cse : 0.000274s : 0.11% optimize.opt_a.a_3 : 0.000444s : 0.18% optimize.py_interpret_to_execute_after_opt_a : 0.000018s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000038s : 0.02% optimize.convert_after_rewriter : 0.000007s : 0.00% optimize.order_py_execute_after_rewriter : 0.000006s : 0.00% optimize.mutable_eliminate : 0.000705s : 0.28% optimize.opt_b.b_1 : 0.000117s : 0.05% optimize.opt_b.b_2 : 0.000008s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000006s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000017s : 0.01% optimize.optimize_parallel_all_gather_comm : 0.000028s : 0.01% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000022s : 0.01% optimize.loop_unroll : 0.000446s : 0.18% optimize.opt_after_cconv.c_1 : 0.000033s : 0.01% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.cse : 0.000018s : 0.01% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000013s : 0.00% optimize.tuple_transform.d_1 : 0.000043s : 0.02% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000043s : 0.02% optimize.cse_after_recomputation.cse : 0.000012s : 0.00% optimize.environ_conv : 0.000008s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.00% optimize.bias_add_comm_swap : 0.000002s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000003s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000012s : 0.00% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000021s : 0.01% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000003s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000008s : 0.00% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000010s : 0.00% optimize.symbol_engine_optimizer.elim_not_effective : 0.000013s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000009s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000018s : 0.01% get_jit_bprop_graph : 0.000003s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.00% opt_after_jit_grad : 0.000498s : 0.20% validate : 0.000040s : 0.02% Time group info: ------[substitution.] 0.000990 190 8.39% : 0.000083s : 6: substitution.arithmetic_simplify 0.21% : 0.000002s : 2: substitution.elim_not_effective 0.64% : 0.000006s : 6: substitution.float_depend_g_call 0.50% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.14% : 0.000001s : 2: substitution.fold_const_symbol 0.60% : 0.000006s : 4: substitution.graph_param_transform 0.34% : 0.000003s : 2: substitution.incorporate_call 0.24% : 0.000002s : 2: substitution.incorporate_call_switch 57.71% : 0.000571s : 16: substitution.inline 2.47% : 0.000024s : 2: substitution.inline_without_move 1.01% : 0.000010s : 14: substitution.j_node_and_user_rematch 2.24% : 0.000022s : 3: substitution.less_batch_normalization 1.56% : 0.000015s : 11: substitution.minmaximum_grad 0.76% : 0.000007s : 6: substitution.partial_eliminate 1.20% : 0.000012s : 14: substitution.remove_not_recompute_node 3.09% : 0.000031s : 10: substitution.replace_applicator 1.40% : 0.000014s : 12: substitution.replace_old_param 0.30% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.38% : 0.000033s : 11: substitution.tuple_list_convert_item_index_to_positive 1.56% : 0.000015s : 11: substitution.tuple_list_get_item_const_eliminator 2.00% : 0.000020s : 11: substitution.tuple_list_get_item_depend_reorder 8.21% : 0.000081s : 30: substitution.tuple_list_get_item_eliminator 2.07% : 0.000020s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.015912 2 88.21% : 0.014037s : 1: type_inference.infer 11.79% : 0.001875s : 1: type_inference.specialize ------[replace.] 0.000283 34 3.97% : 0.000011s : 2: replace.arithmetic_simplify 56.38% : 0.000160s : 16: replace.inline 39.66% : 0.000112s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.000649 34 7.17% : 0.000047s : 2: match.arithmetic_simplify 86.42% : 0.000560s : 16: match.inline 6.41% : 0.000042s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.000705 4766 1.12% : 0.000008s : 58: predicate.accumulaten_eliminater 0.22% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.44% : 0.000003s : 23: predicate.addn_check_dump 1.16% : 0.000008s : 58: predicate.addn_zero_filter 1.03% : 0.000007s : 58: predicate.adjust_all_reduce_mul_add 2.02% : 0.000014s : 83: predicate.arithmetic_simplify 1.22% : 0.000009s : 60: predicate.cast_eliminate 1.03% : 0.000007s : 55: predicate.check_bprop_eliminate 0.43% : 0.000003s : 23: predicate.compare_switch_simplify 0.05% : 0.000000s : 4: predicate.const_output_eliminate 0.48% : 0.000003s : 23: predicate.depend_value_elim 1.24% : 0.000009s : 60: predicate.dict_get_item_const_eliminator 1.29% : 0.000009s : 60: predicate.dict_get_item_eliminator 1.22% : 0.000009s : 60: predicate.dict_set_item_eliminator 0.26% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.07% : 0.000000s : 4: predicate.elim_not_effective 0.13% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.20% : 0.000008s : 64: predicate.environ_add_const_eliminate 1.12% : 0.000008s : 64: predicate.environ_get_add_eliminate 1.14% : 0.000008s : 64: predicate.environ_get_depend_swap 1.56% : 0.000011s : 87: predicate.environ_get_eliminate 1.15% : 0.000008s : 64: predicate.environ_get_set_eliminate 1.75% : 0.000012s : 92: predicate.exchange_switch_depend_value 2.64% : 0.000019s : 92: predicate.float_depend_g_call 0.44% : 0.000003s : 23: predicate.float_environ_get_switch 0.65% : 0.000005s : 27: predicate.float_tuple_getitem_switch 0.06% : 0.000000s : 4: predicate.fold_const_symbol 0.61% : 0.000004s : 23: predicate.get_grad_eliminate 0.06% : 0.000000s : 4: predicate.graph_param_transform 0.44% : 0.000003s : 23: predicate.incorporate_call 0.38% : 0.000003s : 23: predicate.incorporate_call_switch 5.67% : 0.000040s : 204: predicate.inline 1.26% : 0.000009s : 46: predicate.inline_without_move 0.25% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.66% : 0.000005s : 23: predicate.less_batch_normalization 1.66% : 0.000012s : 84: predicate.list_to_tuple_eliminator_ 2.55% : 0.000018s : 142: predicate.load_eliminater 0.27% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.48% : 0.000017s : 128: predicate.loop_unroll_before_grad 1.37% : 0.000010s : 68: predicate.make_slice_get_slice_eliminator 0.54% : 0.000004s : 23: predicate.merge_addn 1.02% : 0.000007s : 55: predicate.micro_step_allgather_replace 1.06% : 0.000007s : 55: predicate.mini_step_allgather_replace 1.13% : 0.000008s : 60: predicate.minmaximum_grad 0.29% : 0.000002s : 4: predicate.mutable_eliminate 0.09% : 0.000001s : 4: predicate.opt_reshape 0.10% : 0.000001s : 4: predicate.parallel_virtual_node 2.42% : 0.000017s : 92: predicate.partial_defer_inline 1.74% : 0.000012s : 80: predicate.partial_eliminate 1.10% : 0.000008s : 58: predicate.print_const_string_wrapper 0.43% : 0.000003s : 23: predicate.reduce_all_const_elim 1.68% : 0.000012s : 60: predicate.reduce_eliminate 2.63% : 0.000019s : 142: predicate.redundant_stop_gradient_eliminater 0.35% : 0.000002s : 23: predicate.remove_not_recompute_node 1.89% : 0.000013s : 131: predicate.replace_applicator 0.63% : 0.000004s : 46: predicate.replace_old_param 0.08% : 0.000001s : 4: predicate.reset_defer_inline 1.21% : 0.000009s : 60: predicate.reshape_eliminate 1.09% : 0.000008s : 55: predicate.row_tensor_add_zeros_like 0.11% : 0.000001s : 4: predicate.row_tensor_eliminate 1.27% : 0.000009s : 55: predicate.same_eliminate 0.31% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.60% : 0.000004s : 23: predicate.shard_identity_eliminate 0.24% : 0.000002s : 8: predicate.special_op_eliminate 0.50% : 0.000004s : 23: predicate.specialize_transform 1.33% : 0.000009s : 55: predicate.split_environ_get_set_with_tuple_value 1.25% : 0.000009s : 46: predicate.stack_unstack_eliminate 0.09% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.96% : 0.000014s : 92: predicate.switch_defer_inline 2.92% : 0.000021s : 147: predicate.switch_layer_defer_inline 5.29% : 0.000037s : 247: predicate.switch_simplify 1.24% : 0.000009s : 60: predicate.tile_eliminate 1.10% : 0.000008s : 60: predicate.transpose_eliminate 1.49% : 0.000011s : 68: predicate.tuple_list_convert_item_index_to_positive 1.58% : 0.000011s : 68: predicate.tuple_list_get_item_const_eliminator 1.41% : 0.000010s : 68: predicate.tuple_list_get_item_depend_reorder 3.10% : 0.000022s : 107: predicate.tuple_list_get_item_eliminator 1.39% : 0.000010s : 68: predicate.tuple_list_get_set_item_eliminator 1.98% : 0.000014s : 91: predicate.tuple_list_set_item_eliminator 1.65% : 0.000012s : 84: predicate.tuple_to_list_eliminator_ 2.58% : 0.000018s : 142: predicate.updatestate_pure_node_eliminater 2.99% : 0.000021s : 165: predicate.updatestate_useless_node_eliminater 0.12% : 0.000001s : 4: predicate.value_based_eliminate 0.55% : 0.000004s : 23: predicate.virtual_dataset_eliminate 0.49% : 0.000003s : 23: predicate.virtual_output_eliminate 0.07% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.17% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002211 37 58.23% : 0.001288s : 17: func_graph_cloner_run.FuncGraphClonerGraph 41.77% : 0.000924s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.532341 233 0.00% : 0.000004s : 1: ForceFp32Comm 0.64% : 0.003429s : 1: add_attr 0.64% : 0.003417s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.01% : 0.000047s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.03% : 0.000139s : 1: auto_monad 0.00% : 0.000022s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 15.88% : 0.084558s : 1: bootstrap 0.00% : 0.000026s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000016s : 1: control_data_broadcast_order 0.00% : 0.000011s : 1: convert_after_rewriter 0.01% : 0.000027s : 1: cse_after_recomputation 0.00% : 0.000007s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000011s : 1: environ_conv 0.01% : 0.000076s : 1: event_method 0.00% : 0.000007s : 1: full_micro_interleaved_order_control 0.00% : 0.000007s : 1: get_jit_bprop_graph 0.00% : 0.000013s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000012s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.09% : 0.000454s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.13% : 0.000714s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.00% : 0.000014s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000014s : 1: opt.transform.mutable_eliminate 0.93% : 0.004948s : 117: opt.transform.opt_a 0.01% : 0.000031s : 1: opt.transform.opt_after_cconv 0.00% : 0.000024s : 1: opt.transform.opt_after_jit_grad 0.02% : 0.000097s : 28: opt.transform.opt_b 0.01% : 0.000048s : 2: opt.transform.opt_trans_graph 0.01% : 0.000036s : 4: opt.transform.symbol_engine_opt 25.72% : 0.136915s : 1: opt_a 0.02% : 0.000111s : 1: opt_after_cconv 0.10% : 0.000508s : 1: opt_after_jit_grad 0.04% : 0.000201s : 1: opt_b 26.18% : 0.139371s : 1: optimize 0.01% : 0.000032s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000009s : 1: order_py_execute_after_rewriter 0.00% : 0.000025s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000004s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000007s : 1: parallel-infer-symbol-second 0.00% : 0.000006s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 2.30% : 0.012265s : 1: pre_auto_parallel 0.02% : 0.000093s : 1: py_interpret_to_execute 0.00% : 0.000022s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000016s : 1: remove_dup_value 23.65% : 0.125894s : 2: renormalize.infer 0.38% : 0.002015s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000007s : 1: rewriter_after_jit_bprop_graph 0.01% : 0.000041s : 1: rewriter_after_opt_a 0.04% : 0.000212s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.02% : 0.000085s : 1: symbol_engine_optimizer 0.01% : 0.000078s : 1: tuple_transform 3.01% : 0.016044s : 1: type_inference TotalTime = 0.206413, [21] [bootstrap]: 0.00050775 [type_inference]: 0.101874 [event_method]: 6.452e-05 [auto_monad]: 0.00014131 [graph_reusing]: 9.44e-06 [inline]: 2.59999e-06 [add_attr]: 0.0036964, [1] [add_attr_with_inline]: 0.00368441, [1] [Cycle 1]: 9.587e-05, [2] [tag_attr]: 4.47e-05 [meta_addattr_fg_expand]: 1.087e-05 [parallel-infer-symbol]: 4.00998e-06 [pre_auto_parallel]: 6.145e-05 [insert-virtual-dataset]: 2.51e-06 [parallel-infer-symbol-second]: 8.29983e-07 [dataset_repeat_opt]: 2.11e-06 [pipeline_split]: 1.54998e-06 [optimize]: 0.0993238, [53] [py_interpret_to_execute]: 4.794e-05 [rewriter_before_opt_a]: 0.00017894 [opt_a]: 0.0967908, [3] [Cycle 1]: 0.0930772, [45] [expand_dump_flag]: 4.35e-06 [switch_simplify]: 8.227e-05 [loop_unroll]: 6.726e-05 [a_1]: 0.00160493 [with_stream_mark]: 3.651e-05 [recompute_prepare]: 2.772e-05 [updatestate_depend_eliminate]: 1.147e-05 [updatestate_assign_eliminate]: 8.21002e-06 [updatestate_loads_eliminate]: 7.78999e-06 [parameter_eliminate]: 3.41999e-06 [a_2]: 0.00025195 [accelerated_algorithm]: 3.524e-05 [shard]: 2.43002e-06 [meta_shard_fg_expand]: 4.57e-06 [shard_inline]: 1.669e-05 [merge_send_recv]: 1.836e-05 [auto_parallel]: 1.468e-05 [parallel]: 2.024e-05 [flash_sp]: 1.206e-05 [merge_comm]: 1.017e-05 [allreduce_fusion]: 8.97e-06 [matmul_add_comm_reduction]: 3.406e-05 [allreduce_slice_to_reducescatter]: 7.7e-07 [virtual_shard_identity]: 1.907e-05 [virtual_dataset]: 1.642e-05 [get_grad_eliminate_]: 1.559e-05 [virtual_output]: 1.544e-05 [merge_forward]: 1.162e-05 [cell_reuse_recompute_pass]: 1.34e-06 [offload_activation]: 1.94e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.151e-05 [merge_recompute_call_nodes]: 1.80001e-06 [before_grad]: 2.919e-05 [set_forward_comm_id_for_comm_node_pass]: 1.055e-05 [meta_fg_expand]: 0.00184261 [flash_sp_send_recv_attached]: 6.33e-06 [receive_attached]: 2.67001e-06 [after_resolve]: 6.809e-05 [a_after_grad]: 8.845e-05 [renormalize]: 0.0876075 [add_forward_monad_depend]: 1.29e-05 [auto_monad_grad]: 6.58998e-06 [auto_monad_eliminator]: 6.153e-05 [cse]: 0.00022957 [a_3]: 0.00035602 [Cycle 2]: 0.00308903, [45] [expand_dump_flag]: 3.33e-06 [switch_simplify]: 4.756e-05 [loop_unroll]: 4.288e-05 [a_1]: 0.00162819 [with_stream_mark]: 1.684e-05 [recompute_prepare]: 8.87e-06 [updatestate_depend_eliminate]: 4.15e-06 [updatestate_assign_eliminate]: 3.58e-06 [updatestate_loads_eliminate]: 2.69001e-06 [parameter_eliminate]: 2.26998e-06 [a_2]: 6.916e-05 [accelerated_algorithm]: 1.161e-05 [shard]: 2.67001e-06 [meta_shard_fg_expand]: 2.89001e-06 [shard_inline]: 6.13998e-06 [merge_send_recv]: 9.02e-06 [auto_parallel]: 1.074e-05 [parallel]: 9.19e-06 [flash_sp]: 3.9e-06 [merge_comm]: 3.73001e-06 [allreduce_fusion]: 3.13998e-06 [matmul_add_comm_reduction]: 9.14998e-06 [allreduce_slice_to_reducescatter]: 6.69999e-07 [virtual_shard_identity]: 7.58999e-06 [virtual_dataset]: 5.96998e-06 [get_grad_eliminate_]: 5.57001e-06 [virtual_output]: 5.38002e-06 [merge_forward]: 4.35999e-06 [cell_reuse_recompute_pass]: 1.09e-06 [offload_activation]: 1.01e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.107e-05 [merge_recompute_call_nodes]: 1.54e-06 [before_grad]: 9.61998e-06 [set_forward_comm_id_for_comm_node_pass]: 3.56999e-06 [meta_fg_expand]: 0.00012193 [flash_sp_send_recv_attached]: 2.29999e-06 [receive_attached]: 2.74001e-06 [after_resolve]: 1.368e-05 [a_after_grad]: 9.92999e-06 [renormalize]: 0.00064129 [add_forward_monad_depend]: 4.46002e-06 [auto_monad_grad]: 1.87999e-06 [auto_monad_eliminator]: 1.127e-05 [cse]: 2.094e-05 [a_3]: 4.239e-05 [Cycle 3]: 0.00060494, [45] [expand_dump_flag]: 1.29e-06 [switch_simplify]: 7.26001e-06 [loop_unroll]: 5.69e-06 [a_1]: 0.00012247 [with_stream_mark]: 8.65999e-06 [recompute_prepare]: 5.94e-06 [updatestate_depend_eliminate]: 3.25002e-06 [updatestate_assign_eliminate]: 2.29999e-06 [updatestate_loads_eliminate]: 2.26998e-06 [parameter_eliminate]: 9.79984e-07 [a_2]: 6.582e-05 [accelerated_algorithm]: 1.105e-05 [shard]: 9.70002e-07 [meta_shard_fg_expand]: 1.25999e-06 [shard_inline]: 5.88002e-06 [merge_send_recv]: 4.65999e-06 [auto_parallel]: 6.11998e-06 [parallel]: 4.48001e-06 [flash_sp]: 9.70002e-07 [merge_comm]: 2.84999e-06 [allreduce_fusion]: 2.87002e-06 [matmul_add_comm_reduction]: 5.80002e-06 [allreduce_slice_to_reducescatter]: 3.69997e-07 [virtual_shard_identity]: 6.74001e-06 [virtual_dataset]: 5.74999e-06 [get_grad_eliminate_]: 5.45001e-06 [virtual_output]: 5.31998e-06 [merge_forward]: 2.86999e-06 [cell_reuse_recompute_pass]: 1.40001e-06 [offload_activation]: 6.54999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.096e-05 [merge_recompute_call_nodes]: 8.50006e-07 [before_grad]: 9.04e-06 [set_forward_comm_id_for_comm_node_pass]: 3.25002e-06 [meta_fg_expand]: 1.80001e-06 [flash_sp_send_recv_attached]: 9.20001e-07 [receive_attached]: 9.89996e-07 [after_resolve]: 9.68002e-06 [a_after_grad]: 8.67e-06 [renormalize]: 8.9989e-08 [add_forward_monad_depend]: 1.24998e-06 [auto_monad_grad]: 1.04e-06 [auto_monad_eliminator]: 5.82999e-06 [cse]: 1.403e-05 [a_3]: 3.288e-05 [py_interpret_to_execute_after_opt_a]: 1.193e-05 [slice_cell_reuse_recomputed_activation]: 1.93002e-06 [rewriter_after_opt_a]: 4.297e-05 [convert_after_rewriter]: 6.84999e-06 [order_py_execute_after_rewriter]: 5.35001e-06 [mutable_eliminate]: 0.00072367 [opt_b]: 0.00033582, [1] [Cycle 1]: 0.00032834, [7] [b_1]: 0.0002466 [b_2]: 8.82e-06 [updatestate_depend_eliminate]: 6.50002e-06 [updatestate_assign_eliminate]: 2.60002e-06 [updatestate_loads_eliminate]: 2.24001e-06 [renormalize]: 2.50002e-07 [cse]: 1.89e-05 [optimize_parallel_all_gather_comm]: 1.681e-05 [overlap_param_gather]: 2.17001e-06 [cconv]: 2.414e-05 [loop_unroll]: 0.0004463 [opt_after_cconv]: 9.77e-05, [1] [Cycle 1]: 9.17e-05, [7] [c_1]: 2.884e-05 [parameter_eliminate]: 2.34001e-06 [updatestate_depend_eliminate]: 5.04998e-06 [updatestate_assign_eliminate]: 2.63998e-06 [updatestate_loads_eliminate]: 2.68e-06 [cse]: 1.568e-05 [renormalize]: 3.59985e-07 [remove_dup_value]: 1.297e-05 [tuple_transform]: 8.373e-05, [1] [Cycle 1]: 7.914e-05, [4] [d_1]: 4.135e-05 [none_parameter_eliminate]: 1.52999e-06 [renormalize]: 2.40019e-07 [switch_simplify]: 1.74e-05 [partial_unused_args_eliminate]: 1.68002e-06 [add_recomputation]: 4.753e-05 [cse_after_recomputation]: 2.276e-05, [1] [Cycle 1]: 1.804e-05, [1] [cse]: 1.218e-05 [environ_conv]: 9.12001e-06 [swap_dp_allreduce_reducescatter]: 5.52001e-06 [bias_add_comm_swap]: 3.13e-06 [label_micro_interleaved_index]: 4.63001e-06 [label_fine_grained_interleaved_index]: 2.98998e-06 [merge_cast_opt]: 1.35001e-06 [slice_recompute_activation]: 2.44999e-06 [micro_interleaved_order_control]: 2.18002e-06 [assign_add_opt]: 1.27e-06 [ForceFp32Comm]: 8.79983e-07 [remove_cast_before_assign_add]: 9.90025e-07 [full_micro_interleaved_order_control]: 2.01998e-06 [reorder_send_recv_between_fp_bp]: 2.83e-06 [comm_op_add_attrs]: 1.04998e-06 [add_comm_op_reuse_tag]: 9.80013e-07 [interleave_split_concat_branches]: 1.24e-06 [interleave_parallel_branches]: 1.39e-06 [overlap_opt_shard_in_pipeline]: 1.52999e-06 [overlap_opt_shard_grad_in_pipeline]: 1.69998e-06 [control_data_broadcast_order]: 1.248e-05 [grouped_pairwise_exchange_alltoall]: 1.67001e-06 [offloading_packed_experts]: 4.15999e-06 [overlap_recompute_and_grad_model_parallel]: 5.04998e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.22e-06 [overlap_recompute_allgather_and_fa_grad]: 1.32999e-06 [overlap_recompute_comm]: 2.24999e-06 [overlap_grad_ring_attention]: 3.95e-06 [overlap_grad_flash_sp]: 2.016e-05 [begin_end_overlap_inline]: 7.2e-07 [split_matmul_comm_elemetwise]: 2.29999e-06 [split_layernorm_comm]: 1.82999e-06 [handle_group_info]: 1.12e-06 [symbol_engine_optimizer]: 8.46e-05, [1] [Cycle 1]: 7.971e-05, [6] [build]: 1.012e-05 [elim_shapecalc]: 1.05e-05 [elim_not_effective]: 1.36e-05 [opt_reshape]: 7.08e-06 [fold_const_symbol]: 9.50001e-06 [renormalize]: 2.3999e-07 [detach_backward]: 2.19999e-06 [pipeline_parallel_scheduler]: 1.62001e-06 [auto_monad_reorder]: 1.816e-05 [get_jit_bprop_graph]: 1.83002e-06 [rewriter_after_jit_bprop_graph]: 4.08001e-06 [opt_after_jit_grad]: 0.00046918 [validate]: 4.484e-05 Sums bootstrap : 0.000508s : 0.25% type_inference : 0.101874s : 50.58% event_method : 0.000065s : 0.03% auto_monad : 0.000141s : 0.07% graph_reusing : 0.000009s : 0.00% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000045s : 0.02% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000011s : 0.01% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000061s : 0.03% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000048s : 0.02% optimize.rewriter_before_opt_a : 0.000179s : 0.09% optimize.opt_a.expand_dump_flag : 0.000009s : 0.00% optimize.opt_a.switch_simplify : 0.000137s : 0.07% optimize.opt_a.loop_unroll : 0.000116s : 0.06% optimize.opt_a.a_1 : 0.003356s : 1.67% optimize.opt_a.with_stream_mark : 0.000062s : 0.03% optimize.opt_a.recompute_prepare : 0.000043s : 0.02% optimize.opt_a.updatestate_depend_eliminate : 0.000019s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000014s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000013s : 0.01% optimize.opt_a.parameter_eliminate : 0.000007s : 0.00% optimize.opt_a.a_2 : 0.000387s : 0.19% optimize.opt_a.accelerated_algorithm : 0.000058s : 0.03% optimize.opt_a.shard : 0.000006s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000009s : 0.00% optimize.opt_a.shard_inline : 0.000029s : 0.01% optimize.opt_a.merge_send_recv : 0.000032s : 0.02% optimize.opt_a.auto_parallel : 0.000032s : 0.02% optimize.opt_a.parallel : 0.000034s : 0.02% optimize.opt_a.flash_sp : 0.000017s : 0.01% optimize.opt_a.merge_comm : 0.000017s : 0.01% optimize.opt_a.allreduce_fusion : 0.000015s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000049s : 0.02% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000033s : 0.02% optimize.opt_a.virtual_dataset : 0.000028s : 0.01% optimize.opt_a.get_grad_eliminate_ : 0.000027s : 0.01% optimize.opt_a.virtual_output : 0.000026s : 0.01% optimize.opt_a.merge_forward : 0.000019s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000004s : 0.00% optimize.opt_a.offload_activation : 0.000036s : 0.02% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000054s : 0.03% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.00% optimize.opt_a.before_grad : 0.000048s : 0.02% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000017s : 0.01% optimize.opt_a.meta_fg_expand : 0.001966s : 0.98% optimize.opt_a.flash_sp_send_recv_attached : 0.000010s : 0.00% optimize.opt_a.receive_attached : 0.000006s : 0.00% optimize.opt_a.after_resolve : 0.000091s : 0.05% optimize.opt_a.a_after_grad : 0.000107s : 0.05% optimize.opt_a.renormalize : 0.088249s : 43.82% optimize.opt_a.add_forward_monad_depend : 0.000019s : 0.01% optimize.opt_a.auto_monad_grad : 0.000010s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000079s : 0.04% optimize.opt_a.cse : 0.000265s : 0.13% optimize.opt_a.a_3 : 0.000431s : 0.21% optimize.py_interpret_to_execute_after_opt_a : 0.000012s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000043s : 0.02% optimize.convert_after_rewriter : 0.000007s : 0.00% optimize.order_py_execute_after_rewriter : 0.000005s : 0.00% optimize.mutable_eliminate : 0.000724s : 0.36% optimize.opt_b.b_1 : 0.000247s : 0.12% optimize.opt_b.b_2 : 0.000009s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000007s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000019s : 0.01% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.01% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000024s : 0.01% optimize.loop_unroll : 0.000446s : 0.22% optimize.opt_after_cconv.c_1 : 0.000029s : 0.01% optimize.opt_after_cconv.parameter_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000005s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.cse : 0.000016s : 0.01% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000013s : 0.01% optimize.tuple_transform.d_1 : 0.000041s : 0.02% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000017s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000048s : 0.02% optimize.cse_after_recomputation.cse : 0.000012s : 0.01% optimize.environ_conv : 0.000009s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.00% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000002s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000012s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000020s : 0.01% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000010s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000011s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000014s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000018s : 0.01% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.00% opt_after_jit_grad : 0.000469s : 0.23% validate : 0.000045s : 0.02% Time group info: ------[substitution.] 0.000931 190 8.96% : 0.000083s : 6: substitution.arithmetic_simplify 0.22% : 0.000002s : 2: substitution.elim_not_effective 0.63% : 0.000006s : 6: substitution.float_depend_g_call 0.54% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.15% : 0.000001s : 2: substitution.fold_const_symbol 0.62% : 0.000006s : 4: substitution.graph_param_transform 0.30% : 0.000003s : 2: substitution.incorporate_call 0.25% : 0.000002s : 2: substitution.incorporate_call_switch 57.04% : 0.000531s : 16: substitution.inline 2.13% : 0.000020s : 2: substitution.inline_without_move 0.98% : 0.000009s : 14: substitution.j_node_and_user_rematch 2.06% : 0.000019s : 3: substitution.less_batch_normalization 1.64% : 0.000015s : 11: substitution.minmaximum_grad 0.78% : 0.000007s : 6: substitution.partial_eliminate 1.25% : 0.000012s : 14: substitution.remove_not_recompute_node 3.31% : 0.000031s : 10: substitution.replace_applicator 1.12% : 0.000010s : 12: substitution.replace_old_param 0.30% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.30% : 0.000031s : 11: substitution.tuple_list_convert_item_index_to_positive 1.62% : 0.000015s : 11: substitution.tuple_list_get_item_const_eliminator 2.17% : 0.000020s : 11: substitution.tuple_list_get_item_depend_reorder 8.47% : 0.000079s : 30: substitution.tuple_list_get_item_eliminator 2.15% : 0.000020s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.101759 2 97.97% : 0.099696s : 1: type_inference.infer 2.03% : 0.002064s : 1: type_inference.specialize ------[replace.] 0.000274 34 3.97% : 0.000011s : 2: replace.arithmetic_simplify 55.20% : 0.000151s : 16: replace.inline 40.82% : 0.000112s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.000609 34 7.96% : 0.000048s : 2: match.arithmetic_simplify 85.69% : 0.000522s : 16: match.inline 6.36% : 0.000039s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.000677 4766 1.06% : 0.000007s : 58: predicate.accumulaten_eliminater 0.27% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.41% : 0.000003s : 23: predicate.addn_check_dump 1.10% : 0.000007s : 58: predicate.addn_zero_filter 1.02% : 0.000007s : 58: predicate.adjust_all_reduce_mul_add 2.15% : 0.000015s : 83: predicate.arithmetic_simplify 1.19% : 0.000008s : 60: predicate.cast_eliminate 1.12% : 0.000008s : 55: predicate.check_bprop_eliminate 0.42% : 0.000003s : 23: predicate.compare_switch_simplify 0.06% : 0.000000s : 4: predicate.const_output_eliminate 0.44% : 0.000003s : 23: predicate.depend_value_elim 1.22% : 0.000008s : 60: predicate.dict_get_item_const_eliminator 1.28% : 0.000009s : 60: predicate.dict_get_item_eliminator 1.21% : 0.000008s : 60: predicate.dict_set_item_eliminator 0.26% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.07% : 0.000000s : 4: predicate.elim_not_effective 0.14% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.30% : 0.000009s : 64: predicate.environ_add_const_eliminate 1.16% : 0.000008s : 64: predicate.environ_get_add_eliminate 1.19% : 0.000008s : 64: predicate.environ_get_depend_swap 1.68% : 0.000011s : 87: predicate.environ_get_eliminate 1.18% : 0.000008s : 64: predicate.environ_get_set_eliminate 1.84% : 0.000012s : 92: predicate.exchange_switch_depend_value 2.62% : 0.000018s : 92: predicate.float_depend_g_call 0.42% : 0.000003s : 23: predicate.float_environ_get_switch 0.52% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.05% : 0.000000s : 4: predicate.fold_const_symbol 0.53% : 0.000004s : 23: predicate.get_grad_eliminate 0.06% : 0.000000s : 4: predicate.graph_param_transform 0.44% : 0.000003s : 23: predicate.incorporate_call 0.39% : 0.000003s : 23: predicate.incorporate_call_switch 5.73% : 0.000039s : 204: predicate.inline 1.29% : 0.000009s : 46: predicate.inline_without_move 0.26% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.64% : 0.000004s : 23: predicate.less_batch_normalization 1.65% : 0.000011s : 84: predicate.list_to_tuple_eliminator_ 2.57% : 0.000017s : 142: predicate.load_eliminater 0.31% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.52% : 0.000017s : 128: predicate.loop_unroll_before_grad 1.35% : 0.000009s : 68: predicate.make_slice_get_slice_eliminator 0.47% : 0.000003s : 23: predicate.merge_addn 1.09% : 0.000007s : 55: predicate.micro_step_allgather_replace 1.06% : 0.000007s : 55: predicate.mini_step_allgather_replace 1.16% : 0.000008s : 60: predicate.minmaximum_grad 0.34% : 0.000002s : 4: predicate.mutable_eliminate 0.10% : 0.000001s : 4: predicate.opt_reshape 0.12% : 0.000001s : 4: predicate.parallel_virtual_node 2.37% : 0.000016s : 92: predicate.partial_defer_inline 1.76% : 0.000012s : 80: predicate.partial_eliminate 1.08% : 0.000007s : 58: predicate.print_const_string_wrapper 0.49% : 0.000003s : 23: predicate.reduce_all_const_elim 1.46% : 0.000010s : 60: predicate.reduce_eliminate 2.61% : 0.000018s : 142: predicate.redundant_stop_gradient_eliminater 0.28% : 0.000002s : 23: predicate.remove_not_recompute_node 1.87% : 0.000013s : 131: predicate.replace_applicator 0.58% : 0.000004s : 46: predicate.replace_old_param 0.08% : 0.000001s : 4: predicate.reset_defer_inline 1.23% : 0.000008s : 60: predicate.reshape_eliminate 1.10% : 0.000007s : 55: predicate.row_tensor_add_zeros_like 0.11% : 0.000001s : 4: predicate.row_tensor_eliminate 1.37% : 0.000009s : 55: predicate.same_eliminate 0.31% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.56% : 0.000004s : 23: predicate.shard_identity_eliminate 0.22% : 0.000001s : 8: predicate.special_op_eliminate 0.60% : 0.000004s : 23: predicate.specialize_transform 1.39% : 0.000009s : 55: predicate.split_environ_get_set_with_tuple_value 1.16% : 0.000008s : 46: predicate.stack_unstack_eliminate 0.09% : 0.000001s : 4: predicate.switch_call_monad_eliminater 2.05% : 0.000014s : 92: predicate.switch_defer_inline 2.99% : 0.000020s : 147: predicate.switch_layer_defer_inline 5.12% : 0.000035s : 247: predicate.switch_simplify 1.18% : 0.000008s : 60: predicate.tile_eliminate 1.11% : 0.000008s : 60: predicate.transpose_eliminate 1.50% : 0.000010s : 68: predicate.tuple_list_convert_item_index_to_positive 1.50% : 0.000010s : 68: predicate.tuple_list_get_item_const_eliminator 1.34% : 0.000009s : 68: predicate.tuple_list_get_item_depend_reorder 2.96% : 0.000020s : 107: predicate.tuple_list_get_item_eliminator 1.48% : 0.000010s : 68: predicate.tuple_list_get_set_item_eliminator 1.99% : 0.000013s : 91: predicate.tuple_list_set_item_eliminator 1.60% : 0.000011s : 84: predicate.tuple_to_list_eliminator_ 2.56% : 0.000017s : 142: predicate.updatestate_pure_node_eliminater 3.04% : 0.000021s : 165: predicate.updatestate_useless_node_eliminater 0.14% : 0.000001s : 4: predicate.value_based_eliminate 0.57% : 0.000004s : 23: predicate.virtual_dataset_eliminate 0.47% : 0.000003s : 23: predicate.virtual_output_eliminate 0.09% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.13% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002201 37 59.31% : 0.001305s : 17: func_graph_cloner_run.FuncGraphClonerGraph 40.69% : 0.000895s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.402743 233 0.00% : 0.000004s : 1: ForceFp32Comm 0.92% : 0.003703s : 1: add_attr 0.92% : 0.003689s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.01% : 0.000052s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.04% : 0.000149s : 1: auto_monad 0.01% : 0.000022s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 0.13% : 0.000539s : 1: bootstrap 0.01% : 0.000027s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000016s : 1: control_data_broadcast_order 0.00% : 0.000010s : 1: convert_after_rewriter 0.01% : 0.000026s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000012s : 1: environ_conv 0.02% : 0.000073s : 1: event_method 0.00% : 0.000006s : 1: full_micro_interleaved_order_control 0.00% : 0.000006s : 1: get_jit_bprop_graph 0.00% : 0.000014s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.11% : 0.000454s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.18% : 0.000732s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.00% : 0.000013s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000015s : 1: opt.transform.mutable_eliminate 1.20% : 0.004828s : 117: opt.transform.opt_a 0.01% : 0.000028s : 1: opt.transform.opt_after_cconv 0.01% : 0.000023s : 1: opt.transform.opt_after_jit_grad 0.06% : 0.000224s : 28: opt.transform.opt_b 0.01% : 0.000056s : 2: opt.transform.opt_trans_graph 0.01% : 0.000037s : 4: opt.transform.symbol_engine_opt 24.03% : 0.096794s : 1: opt_a 0.03% : 0.000101s : 1: opt_after_cconv 0.12% : 0.000478s : 1: opt_after_jit_grad 0.08% : 0.000339s : 1: opt_b 24.66% : 0.099330s : 1: optimize 0.01% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000009s : 1: order_py_execute_after_rewriter 0.01% : 0.000024s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000004s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000006s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000004s : 1: pipeline_split 0.02% : 0.000066s : 1: pre_auto_parallel 0.01% : 0.000052s : 1: py_interpret_to_execute 0.00% : 0.000015s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000017s : 1: remove_dup_value 21.47% : 0.086485s : 2: renormalize.infer 0.43% : 0.001744s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000007s : 1: rewriter_after_jit_bprop_graph 0.01% : 0.000047s : 1: rewriter_after_opt_a 0.05% : 0.000184s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.02% : 0.000087s : 1: symbol_engine_optimizer 0.02% : 0.000087s : 1: tuple_transform 25.30% : 0.101901s : 1: type_inference TotalTime = 0.192967, [21] [bootstrap]: 0.00049459 [type_inference]: 0.112129 [event_method]: 6.319e-05 [auto_monad]: 0.00012948 [graph_reusing]: 8.52e-06 [inline]: 2.53998e-06 [add_attr]: 0.003616, [1] [add_attr_with_inline]: 0.00360503, [1] [Cycle 1]: 9.563e-05, [2] [tag_attr]: 4.551e-05 [meta_addattr_fg_expand]: 1.092e-05 [parallel-infer-symbol]: 3.88999e-06 [pre_auto_parallel]: 6.281e-05 [insert-virtual-dataset]: 2.39001e-06 [parallel-infer-symbol-second]: 6.40022e-07 [dataset_repeat_opt]: 1.93997e-06 [pipeline_split]: 1.65001e-06 [optimize]: 0.0757055, [53] [py_interpret_to_execute]: 4.473e-05 [rewriter_before_opt_a]: 0.00017312 [opt_a]: 0.0730687, [3] [Cycle 1]: 0.0690961, [45] [expand_dump_flag]: 4.28999e-06 [switch_simplify]: 8.355e-05 [loop_unroll]: 6.766e-05 [a_1]: 0.00157453 [with_stream_mark]: 3.172e-05 [recompute_prepare]: 2.349e-05 [updatestate_depend_eliminate]: 1.079e-05 [updatestate_assign_eliminate]: 8.38001e-06 [updatestate_loads_eliminate]: 7.85e-06 [parameter_eliminate]: 2.81999e-06 [a_2]: 0.00024837 [accelerated_algorithm]: 3.491e-05 [shard]: 2.24001e-06 [meta_shard_fg_expand]: 4.40999e-06 [shard_inline]: 1.727e-05 [merge_send_recv]: 1.974e-05 [auto_parallel]: 1.452e-05 [parallel]: 1.975e-05 [flash_sp]: 1.139e-05 [merge_comm]: 1.05e-05 [allreduce_fusion]: 9.47999e-06 [matmul_add_comm_reduction]: 3.377e-05 [allreduce_slice_to_reducescatter]: 8.50006e-07 [virtual_shard_identity]: 1.932e-05 [virtual_dataset]: 1.683e-05 [get_grad_eliminate_]: 1.609e-05 [virtual_output]: 1.54e-05 [merge_forward]: 1.042e-05 [cell_reuse_recompute_pass]: 1.35001e-06 [offload_activation]: 1.997e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.207e-05 [merge_recompute_call_nodes]: 1.44e-06 [before_grad]: 2.868e-05 [set_forward_comm_id_for_comm_node_pass]: 1.079e-05 [meta_fg_expand]: 0.00176776 [flash_sp_send_recv_attached]: 6.23e-06 [receive_attached]: 2.42001e-06 [after_resolve]: 6.585e-05 [a_after_grad]: 8.824e-05 [renormalize]: 0.0636948 [add_forward_monad_depend]: 1.617e-05 [auto_monad_grad]: 6.76e-06 [auto_monad_eliminator]: 7.056e-05 [cse]: 0.00024056 [a_3]: 0.00036456 [Cycle 2]: 0.0033333, [45] [expand_dump_flag]: 3.49001e-06 [switch_simplify]: 4.812e-05 [loop_unroll]: 4.353e-05 [a_1]: 0.00174922 [with_stream_mark]: 2.092e-05 [recompute_prepare]: 1.198e-05 [updatestate_depend_eliminate]: 4.4e-06 [updatestate_assign_eliminate]: 3.16001e-06 [updatestate_loads_eliminate]: 2.81e-06 [parameter_eliminate]: 1.65001e-06 [a_2]: 7.153e-05 [accelerated_algorithm]: 1.249e-05 [shard]: 2.31998e-06 [meta_shard_fg_expand]: 2.22001e-06 [shard_inline]: 6.58e-06 [merge_send_recv]: 8.28001e-06 [auto_parallel]: 1.064e-05 [parallel]: 9.91998e-06 [flash_sp]: 4.33999e-06 [merge_comm]: 3.30998e-06 [allreduce_fusion]: 3.50998e-06 [matmul_add_comm_reduction]: 9.79999e-06 [allreduce_slice_to_reducescatter]: 7.29982e-07 [virtual_shard_identity]: 8.49998e-06 [virtual_dataset]: 6.22001e-06 [get_grad_eliminate_]: 5.67001e-06 [virtual_output]: 5.51e-06 [merge_forward]: 3.99002e-06 [cell_reuse_recompute_pass]: 1.62001e-06 [offload_activation]: 1.026e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.128e-05 [merge_recompute_call_nodes]: 1.44998e-06 [before_grad]: 1.006e-05 [set_forward_comm_id_for_comm_node_pass]: 3.65998e-06 [meta_fg_expand]: 0.00013203 [flash_sp_send_recv_attached]: 2.04e-06 [receive_attached]: 2.63998e-06 [after_resolve]: 1.752e-05 [a_after_grad]: 9.86998e-06 [renormalize]: 0.0007147 [add_forward_monad_depend]: 6.18002e-06 [auto_monad_grad]: 2.21e-06 [auto_monad_eliminator]: 1.421e-05 [cse]: 2.463e-05 [a_3]: 4.488e-05 [Cycle 3]: 0.0006191, [45] [expand_dump_flag]: 1.89e-06 [switch_simplify]: 7.65998e-06 [loop_unroll]: 6.03998e-06 [a_1]: 0.00012202 [with_stream_mark]: 1.147e-05 [recompute_prepare]: 5.87001e-06 [updatestate_depend_eliminate]: 3.63999e-06 [updatestate_assign_eliminate]: 2.37001e-06 [updatestate_loads_eliminate]: 2.24001e-06 [parameter_eliminate]: 1.05001e-06 [a_2]: 6.498e-05 [accelerated_algorithm]: 1.21e-05 [shard]: 1.05999e-06 [meta_shard_fg_expand]: 1.68002e-06 [shard_inline]: 5.92001e-06 [merge_send_recv]: 5.49e-06 [auto_parallel]: 6.42001e-06 [parallel]: 5.69999e-06 [flash_sp]: 1.09e-06 [merge_comm]: 3.78999e-06 [allreduce_fusion]: 2.81e-06 [matmul_add_comm_reduction]: 5.54998e-06 [allreduce_slice_to_reducescatter]: 4.00003e-07 [virtual_shard_identity]: 7.76001e-06 [virtual_dataset]: 5.74e-06 [get_grad_eliminate_]: 5.46e-06 [virtual_output]: 5.42001e-06 [merge_forward]: 3.85e-06 [cell_reuse_recompute_pass]: 1.37e-06 [offload_activation]: 8.02998e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.1e-05 [merge_recompute_call_nodes]: 8.40024e-07 [before_grad]: 8.70999e-06 [set_forward_comm_id_for_comm_node_pass]: 3.48999e-06 [meta_fg_expand]: 2.42001e-06 [flash_sp_send_recv_attached]: 9.79984e-07 [receive_attached]: 1.15001e-06 [after_resolve]: 9.96e-06 [a_after_grad]: 8.60001e-06 [renormalize]: 7.99773e-08 [add_forward_monad_depend]: 1.09e-06 [auto_monad_grad]: 8.89995e-07 [auto_monad_eliminator]: 8.50001e-06 [cse]: 1.299e-05 [a_3]: 3.301e-05 [py_interpret_to_execute_after_opt_a]: 1.202e-05 [slice_cell_reuse_recomputed_activation]: 2.10002e-06 [rewriter_after_opt_a]: 3.961e-05 [convert_after_rewriter]: 7.46001e-06 [order_py_execute_after_rewriter]: 5.44e-06 [mutable_eliminate]: 0.0007519 [opt_b]: 0.00019753, [1] [Cycle 1]: 0.00018976, [7] [b_1]: 0.00011495 [b_2]: 8.62e-06 [updatestate_depend_eliminate]: 6.49999e-06 [updatestate_assign_eliminate]: 2.50002e-06 [updatestate_loads_eliminate]: 2.41e-06 [renormalize]: 7.09988e-07 [cse]: 1.948e-05 [optimize_parallel_all_gather_comm]: 0.00016494 [overlap_param_gather]: 3.13e-06 [cconv]: 2.749e-05 [loop_unroll]: 0.00046843 [opt_after_cconv]: 0.00010629, [1] [Cycle 1]: 9.96e-05, [7] [c_1]: 3.105e-05 [parameter_eliminate]: 3.48e-06 [updatestate_depend_eliminate]: 6.14999e-06 [updatestate_assign_eliminate]: 2.61999e-06 [updatestate_loads_eliminate]: 2.44001e-06 [cse]: 1.861e-05 [renormalize]: 5.00004e-07 [remove_dup_value]: 1.396e-05 [tuple_transform]: 8.988e-05, [1] [Cycle 1]: 8.455e-05, [4] [d_1]: 4.363e-05 [none_parameter_eliminate]: 2.42001e-06 [renormalize]: 2.59985e-07 [switch_simplify]: 9.11002e-06 [partial_unused_args_eliminate]: 2.28002e-06 [add_recomputation]: 5.756e-05 [cse_after_recomputation]: 2.299e-05, [1] [Cycle 1]: 1.784e-05, [1] [cse]: 1.21e-05 [environ_conv]: 1.013e-05 [swap_dp_allreduce_reducescatter]: 5.32001e-06 [bias_add_comm_swap]: 2.53998e-06 [label_micro_interleaved_index]: 4.32998e-06 [label_fine_grained_interleaved_index]: 2.51e-06 [merge_cast_opt]: 1.29998e-06 [slice_recompute_activation]: 2.07001e-06 [micro_interleaved_order_control]: 2.34001e-06 [assign_add_opt]: 1.25001e-06 [ForceFp32Comm]: 1.35999e-06 [remove_cast_before_assign_add]: 1.15001e-06 [full_micro_interleaved_order_control]: 2.07001e-06 [reorder_send_recv_between_fp_bp]: 2.58e-06 [comm_op_add_attrs]: 1.26002e-06 [add_comm_op_reuse_tag]: 1.04e-06 [interleave_split_concat_branches]: 1.22e-06 [interleave_parallel_branches]: 1.05999e-06 [overlap_opt_shard_in_pipeline]: 1.71002e-06 [overlap_opt_shard_grad_in_pipeline]: 1.63002e-06 [control_data_broadcast_order]: 1.319e-05 [grouped_pairwise_exchange_alltoall]: 1.56002e-06 [offloading_packed_experts]: 4.37e-06 [overlap_recompute_and_grad_model_parallel]: 5.07e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.27999e-06 [overlap_recompute_allgather_and_fa_grad]: 1.39e-06 [overlap_recompute_comm]: 2.32999e-06 [overlap_grad_ring_attention]: 4.23999e-06 [overlap_grad_flash_sp]: 2.195e-05 [begin_end_overlap_inline]: 8.2e-07 [split_matmul_comm_elemetwise]: 2.38998e-06 [split_layernorm_comm]: 2.16e-06 [handle_group_info]: 1.05999e-06 [symbol_engine_optimizer]: 9.446e-05, [1] [Cycle 1]: 8.88e-05, [6] [build]: 1.171e-05 [elim_shapecalc]: 1.27e-05 [elim_not_effective]: 1.547e-05 [opt_reshape]: 7.63001e-06 [fold_const_symbol]: 1.037e-05 [renormalize]: 2.19996e-07 [detach_backward]: 2.68e-06 [pipeline_parallel_scheduler]: 1.60999e-06 [auto_monad_reorder]: 1.951e-05 [get_jit_bprop_graph]: 1.81e-06 [rewriter_after_jit_bprop_graph]: 5.09998e-06 [opt_after_jit_grad]: 0.00048574 [validate]: 4.711e-05 Sums bootstrap : 0.000495s : 0.26% type_inference : 0.112129s : 59.65% event_method : 0.000063s : 0.03% auto_monad : 0.000129s : 0.07% graph_reusing : 0.000009s : 0.00% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000046s : 0.02% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000011s : 0.01% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000063s : 0.03% insert-virtual-dataset : 0.000002s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000045s : 0.02% optimize.rewriter_before_opt_a : 0.000173s : 0.09% optimize.opt_a.expand_dump_flag : 0.000010s : 0.01% optimize.opt_a.switch_simplify : 0.000139s : 0.07% optimize.opt_a.loop_unroll : 0.000117s : 0.06% optimize.opt_a.a_1 : 0.003446s : 1.83% optimize.opt_a.with_stream_mark : 0.000064s : 0.03% optimize.opt_a.recompute_prepare : 0.000041s : 0.02% optimize.opt_a.updatestate_depend_eliminate : 0.000019s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000014s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000013s : 0.01% optimize.opt_a.parameter_eliminate : 0.000006s : 0.00% optimize.opt_a.a_2 : 0.000385s : 0.20% optimize.opt_a.accelerated_algorithm : 0.000060s : 0.03% optimize.opt_a.shard : 0.000006s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000008s : 0.00% optimize.opt_a.shard_inline : 0.000030s : 0.02% optimize.opt_a.merge_send_recv : 0.000034s : 0.02% optimize.opt_a.auto_parallel : 0.000032s : 0.02% optimize.opt_a.parallel : 0.000035s : 0.02% optimize.opt_a.flash_sp : 0.000017s : 0.01% optimize.opt_a.merge_comm : 0.000018s : 0.01% optimize.opt_a.allreduce_fusion : 0.000016s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000049s : 0.03% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000036s : 0.02% optimize.opt_a.virtual_dataset : 0.000029s : 0.02% optimize.opt_a.get_grad_eliminate_ : 0.000027s : 0.01% optimize.opt_a.virtual_output : 0.000026s : 0.01% optimize.opt_a.merge_forward : 0.000018s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000004s : 0.00% optimize.opt_a.offload_activation : 0.000038s : 0.02% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000054s : 0.03% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.00% optimize.opt_a.before_grad : 0.000047s : 0.03% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000018s : 0.01% optimize.opt_a.meta_fg_expand : 0.001902s : 1.01% optimize.opt_a.flash_sp_send_recv_attached : 0.000009s : 0.00% optimize.opt_a.receive_attached : 0.000006s : 0.00% optimize.opt_a.after_resolve : 0.000093s : 0.05% optimize.opt_a.a_after_grad : 0.000107s : 0.06% optimize.opt_a.renormalize : 0.064410s : 34.26% optimize.opt_a.add_forward_monad_depend : 0.000023s : 0.01% optimize.opt_a.auto_monad_grad : 0.000010s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000093s : 0.05% optimize.opt_a.cse : 0.000278s : 0.15% optimize.opt_a.a_3 : 0.000442s : 0.24% optimize.py_interpret_to_execute_after_opt_a : 0.000012s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000040s : 0.02% optimize.convert_after_rewriter : 0.000007s : 0.00% optimize.order_py_execute_after_rewriter : 0.000005s : 0.00% optimize.mutable_eliminate : 0.000752s : 0.40% optimize.opt_b.b_1 : 0.000115s : 0.06% optimize.opt_b.b_2 : 0.000009s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000006s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000019s : 0.01% optimize.optimize_parallel_all_gather_comm : 0.000165s : 0.09% optimize.overlap_param_gather : 0.000003s : 0.00% optimize.cconv : 0.000027s : 0.01% optimize.loop_unroll : 0.000468s : 0.25% optimize.opt_after_cconv.c_1 : 0.000031s : 0.02% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000019s : 0.01% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000014s : 0.01% optimize.tuple_transform.d_1 : 0.000044s : 0.02% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000009s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000058s : 0.03% optimize.cse_after_recomputation.cse : 0.000012s : 0.01% optimize.environ_conv : 0.000010s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.00% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000004s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000002s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000013s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000022s : 0.01% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000012s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000013s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000015s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000008s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000020s : 0.01% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000005s : 0.00% opt_after_jit_grad : 0.000486s : 0.26% validate : 0.000047s : 0.03% Time group info: ------[substitution.] 0.000974 190 9.14% : 0.000089s : 6: substitution.arithmetic_simplify 0.23% : 0.000002s : 2: substitution.elim_not_effective 0.65% : 0.000006s : 6: substitution.float_depend_g_call 0.45% : 0.000004s : 3: substitution.float_tuple_getitem_switch 0.14% : 0.000001s : 2: substitution.fold_const_symbol 0.60% : 0.000006s : 4: substitution.graph_param_transform 0.29% : 0.000003s : 2: substitution.incorporate_call 0.24% : 0.000002s : 2: substitution.incorporate_call_switch 56.87% : 0.000554s : 16: substitution.inline 2.06% : 0.000020s : 2: substitution.inline_without_move 0.95% : 0.000009s : 14: substitution.j_node_and_user_rematch 1.97% : 0.000019s : 3: substitution.less_batch_normalization 1.71% : 0.000017s : 11: substitution.minmaximum_grad 0.83% : 0.000008s : 6: substitution.partial_eliminate 1.14% : 0.000011s : 14: substitution.remove_not_recompute_node 3.56% : 0.000035s : 10: substitution.replace_applicator 1.26% : 0.000012s : 12: substitution.replace_old_param 0.28% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.37% : 0.000033s : 11: substitution.tuple_list_convert_item_index_to_positive 1.66% : 0.000016s : 11: substitution.tuple_list_get_item_const_eliminator 2.17% : 0.000021s : 11: substitution.tuple_list_get_item_depend_reorder 8.23% : 0.000080s : 30: substitution.tuple_list_get_item_eliminator 2.21% : 0.000021s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.112040 2 98.40% : 0.110245s : 1: type_inference.infer 1.60% : 0.001795s : 1: type_inference.specialize ------[replace.] 0.000285 34 4.74% : 0.000014s : 2: replace.arithmetic_simplify 54.80% : 0.000156s : 16: replace.inline 40.45% : 0.000115s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.000639 34 8.50% : 0.000054s : 2: match.arithmetic_simplify 84.88% : 0.000543s : 16: match.inline 6.62% : 0.000042s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.000695 4766 1.16% : 0.000008s : 58: predicate.accumulaten_eliminater 0.23% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.42% : 0.000003s : 23: predicate.addn_check_dump 1.14% : 0.000008s : 58: predicate.addn_zero_filter 1.02% : 0.000007s : 58: predicate.adjust_all_reduce_mul_add 2.05% : 0.000014s : 83: predicate.arithmetic_simplify 1.18% : 0.000008s : 60: predicate.cast_eliminate 1.11% : 0.000008s : 55: predicate.check_bprop_eliminate 0.40% : 0.000003s : 23: predicate.compare_switch_simplify 0.05% : 0.000000s : 4: predicate.const_output_eliminate 0.42% : 0.000003s : 23: predicate.depend_value_elim 1.20% : 0.000008s : 60: predicate.dict_get_item_const_eliminator 1.30% : 0.000009s : 60: predicate.dict_get_item_eliminator 1.14% : 0.000008s : 60: predicate.dict_set_item_eliminator 0.32% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.09% : 0.000001s : 4: predicate.elim_not_effective 0.14% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.20% : 0.000008s : 64: predicate.environ_add_const_eliminate 1.17% : 0.000008s : 64: predicate.environ_get_add_eliminate 1.16% : 0.000008s : 64: predicate.environ_get_depend_swap 1.63% : 0.000011s : 87: predicate.environ_get_eliminate 1.18% : 0.000008s : 64: predicate.environ_get_set_eliminate 1.82% : 0.000013s : 92: predicate.exchange_switch_depend_value 2.64% : 0.000018s : 92: predicate.float_depend_g_call 0.39% : 0.000003s : 23: predicate.float_environ_get_switch 0.49% : 0.000003s : 27: predicate.float_tuple_getitem_switch 0.07% : 0.000000s : 4: predicate.fold_const_symbol 0.49% : 0.000003s : 23: predicate.get_grad_eliminate 0.06% : 0.000000s : 4: predicate.graph_param_transform 0.43% : 0.000003s : 23: predicate.incorporate_call 0.39% : 0.000003s : 23: predicate.incorporate_call_switch 5.88% : 0.000041s : 204: predicate.inline 1.32% : 0.000009s : 46: predicate.inline_without_move 0.25% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.66% : 0.000005s : 23: predicate.less_batch_normalization 1.66% : 0.000012s : 84: predicate.list_to_tuple_eliminator_ 2.62% : 0.000018s : 142: predicate.load_eliminater 0.30% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.50% : 0.000017s : 128: predicate.loop_unroll_before_grad 1.34% : 0.000009s : 68: predicate.make_slice_get_slice_eliminator 0.48% : 0.000003s : 23: predicate.merge_addn 1.04% : 0.000007s : 55: predicate.micro_step_allgather_replace 1.09% : 0.000008s : 55: predicate.mini_step_allgather_replace 1.14% : 0.000008s : 60: predicate.minmaximum_grad 0.32% : 0.000002s : 4: predicate.mutable_eliminate 0.13% : 0.000001s : 4: predicate.opt_reshape 0.14% : 0.000001s : 4: predicate.parallel_virtual_node 2.41% : 0.000017s : 92: predicate.partial_defer_inline 1.83% : 0.000013s : 80: predicate.partial_eliminate 1.14% : 0.000008s : 58: predicate.print_const_string_wrapper 0.44% : 0.000003s : 23: predicate.reduce_all_const_elim 1.42% : 0.000010s : 60: predicate.reduce_eliminate 2.62% : 0.000018s : 142: predicate.redundant_stop_gradient_eliminater 0.34% : 0.000002s : 23: predicate.remove_not_recompute_node 1.89% : 0.000013s : 131: predicate.replace_applicator 0.59% : 0.000004s : 46: predicate.replace_old_param 0.07% : 0.000000s : 4: predicate.reset_defer_inline 1.13% : 0.000008s : 60: predicate.reshape_eliminate 1.14% : 0.000008s : 55: predicate.row_tensor_add_zeros_like 0.13% : 0.000001s : 4: predicate.row_tensor_eliminate 1.29% : 0.000009s : 55: predicate.same_eliminate 0.34% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.54% : 0.000004s : 23: predicate.shard_identity_eliminate 0.21% : 0.000001s : 8: predicate.special_op_eliminate 0.52% : 0.000004s : 23: predicate.specialize_transform 1.42% : 0.000010s : 55: predicate.split_environ_get_set_with_tuple_value 1.14% : 0.000008s : 46: predicate.stack_unstack_eliminate 0.09% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.98% : 0.000014s : 92: predicate.switch_defer_inline 2.94% : 0.000020s : 147: predicate.switch_layer_defer_inline 5.06% : 0.000035s : 247: predicate.switch_simplify 1.24% : 0.000009s : 60: predicate.tile_eliminate 1.19% : 0.000008s : 60: predicate.transpose_eliminate 1.50% : 0.000010s : 68: predicate.tuple_list_convert_item_index_to_positive 1.60% : 0.000011s : 68: predicate.tuple_list_get_item_const_eliminator 1.31% : 0.000009s : 68: predicate.tuple_list_get_item_depend_reorder 2.98% : 0.000021s : 107: predicate.tuple_list_get_item_eliminator 1.46% : 0.000010s : 68: predicate.tuple_list_get_set_item_eliminator 1.95% : 0.000014s : 91: predicate.tuple_list_set_item_eliminator 1.70% : 0.000012s : 84: predicate.tuple_to_list_eliminator_ 2.71% : 0.000019s : 142: predicate.updatestate_pure_node_eliminater 3.07% : 0.000021s : 165: predicate.updatestate_useless_node_eliminater 0.11% : 0.000001s : 4: predicate.value_based_eliminate 0.48% : 0.000003s : 23: predicate.virtual_dataset_eliminate 0.48% : 0.000003s : 23: predicate.virtual_output_eliminate 0.07% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.13% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002283 37 57.65% : 0.001316s : 17: func_graph_cloner_run.FuncGraphClonerGraph 42.35% : 0.000967s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.341736 233 0.00% : 0.000004s : 1: ForceFp32Comm 1.06% : 0.003623s : 1: add_attr 1.06% : 0.003610s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.02% : 0.000062s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.04% : 0.000138s : 1: auto_monad 0.01% : 0.000024s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000005s : 1: bias_add_comm_swap 0.15% : 0.000526s : 1: bootstrap 0.01% : 0.000031s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000017s : 1: control_data_broadcast_order 0.00% : 0.000011s : 1: convert_after_rewriter 0.01% : 0.000026s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000014s : 1: environ_conv 0.02% : 0.000072s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000006s : 1: get_jit_bprop_graph 0.00% : 0.000013s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000007s : 1: label_micro_interleaved_index 0.14% : 0.000476s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.22% : 0.000761s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.00% : 0.000015s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000016s : 1: opt.transform.mutable_eliminate 1.44% : 0.004934s : 117: opt.transform.opt_a 0.01% : 0.000030s : 1: opt.transform.opt_after_cconv 0.01% : 0.000025s : 1: opt.transform.opt_after_jit_grad 0.03% : 0.000096s : 28: opt.transform.opt_b 0.01% : 0.000050s : 2: opt.transform.opt_trans_graph 0.01% : 0.000041s : 4: opt.transform.symbol_engine_opt 21.38% : 0.073072s : 1: opt_a 0.03% : 0.000110s : 1: opt_after_cconv 0.14% : 0.000495s : 1: opt_after_jit_grad 0.06% : 0.000202s : 1: opt_b 22.16% : 0.075712s : 1: optimize 0.05% : 0.000173s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000009s : 1: order_py_execute_after_rewriter 0.01% : 0.000026s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000004s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000007s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000006s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.02% : 0.000068s : 1: pre_auto_parallel 0.01% : 0.000049s : 1: py_interpret_to_execute 0.00% : 0.000016s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.01% : 0.000018s : 1: remove_dup_value 18.22% : 0.062264s : 2: renormalize.infer 0.62% : 0.002123s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000009s : 1: rewriter_after_jit_bprop_graph 0.01% : 0.000044s : 1: rewriter_after_opt_a 0.05% : 0.000178s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000006s : 1: split_matmul_comm_elemetwise 0.00% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.03% : 0.000097s : 1: symbol_engine_optimizer 0.03% : 0.000093s : 1: tuple_transform 32.82% : 0.112153s : 1: type_inference TotalTime = 0.178811, [21] [bootstrap]: 0.00039076 [type_inference]: 0.0832918 [event_method]: 6.131e-05 [auto_monad]: 0.00012685 [graph_reusing]: 9.21998e-06 [inline]: 2.58e-06 [add_attr]: 0.00323279, [1] [add_attr_with_inline]: 0.00322326, [1] [Cycle 1]: 8.837e-05, [2] [tag_attr]: 4.126e-05 [meta_addattr_fg_expand]: 1.068e-05 [parallel-infer-symbol]: 3.37002e-06 [pre_auto_parallel]: 5.611e-05 [insert-virtual-dataset]: 2.53e-06 [parallel-infer-symbol-second]: 9.09989e-07 [dataset_repeat_opt]: 1.92001e-06 [pipeline_split]: 1.58002e-06 [optimize]: 0.0908959, [53] [py_interpret_to_execute]: 4.036e-05 [rewriter_before_opt_a]: 0.00015991 [opt_a]: 0.0884273, [3] [Cycle 1]: 0.0846349, [45] [expand_dump_flag]: 3.71001e-06 [switch_simplify]: 8.225e-05 [loop_unroll]: 6.736e-05 [a_1]: 0.00154446 [with_stream_mark]: 2.95e-05 [recompute_prepare]: 2.47e-05 [updatestate_depend_eliminate]: 9.87999e-06 [updatestate_assign_eliminate]: 8.33001e-06 [updatestate_loads_eliminate]: 7.8e-06 [parameter_eliminate]: 2.78998e-06 [a_2]: 0.00025141 [accelerated_algorithm]: 3.349e-05 [shard]: 1.97001e-06 [meta_shard_fg_expand]: 4.53999e-06 [shard_inline]: 1.712e-05 [merge_send_recv]: 1.926e-05 [auto_parallel]: 1.317e-05 [parallel]: 1.83e-05 [flash_sp]: 1.055e-05 [merge_comm]: 1.007e-05 [allreduce_fusion]: 9.25999e-06 [matmul_add_comm_reduction]: 2.929e-05 [allreduce_slice_to_reducescatter]: 6.49976e-07 [virtual_shard_identity]: 1.909e-05 [virtual_dataset]: 1.611e-05 [get_grad_eliminate_]: 1.544e-05 [virtual_output]: 1.52e-05 [merge_forward]: 1.022e-05 [cell_reuse_recompute_pass]: 1.45999e-06 [offload_activation]: 2.003e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.122e-05 [merge_recompute_call_nodes]: 1.37e-06 [before_grad]: 2.873e-05 [set_forward_comm_id_for_comm_node_pass]: 1.005e-05 [meta_fg_expand]: 0.00156689 [flash_sp_send_recv_attached]: 6.68003e-06 [receive_attached]: 2.61e-06 [after_resolve]: 6.574e-05 [a_after_grad]: 8.67e-05 [renormalize]: 0.0795027 [add_forward_monad_depend]: 1.497e-05 [auto_monad_grad]: 7.34002e-06 [auto_monad_eliminator]: 6.867e-05 [cse]: 0.0002298 [a_3]: 0.00035529 [Cycle 2]: 0.00314467, [45] [expand_dump_flag]: 3.38e-06 [switch_simplify]: 4.832e-05 [loop_unroll]: 4.406e-05 [a_1]: 0.00165098 [with_stream_mark]: 1.763e-05 [recompute_prepare]: 9.22001e-06 [updatestate_depend_eliminate]: 4.70001e-06 [updatestate_assign_eliminate]: 3.53999e-06 [updatestate_loads_eliminate]: 2.64001e-06 [parameter_eliminate]: 1.81e-06 [a_2]: 6.891e-05 [accelerated_algorithm]: 1.212e-05 [shard]: 2.39999e-06 [meta_shard_fg_expand]: 2.09e-06 [shard_inline]: 6.66999e-06 [merge_send_recv]: 8.65001e-06 [auto_parallel]: 1.024e-05 [parallel]: 9.11998e-06 [flash_sp]: 3.95e-06 [merge_comm]: 3.53e-06 [allreduce_fusion]: 3.65e-06 [matmul_add_comm_reduction]: 9.21998e-06 [allreduce_slice_to_reducescatter]: 6.19999e-07 [virtual_shard_identity]: 8.06001e-06 [virtual_dataset]: 5.78997e-06 [get_grad_eliminate_]: 5.84e-06 [virtual_output]: 5.46e-06 [merge_forward]: 4.31002e-06 [cell_reuse_recompute_pass]: 1.25001e-06 [offload_activation]: 1.04e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.169e-05 [merge_recompute_call_nodes]: 1.42e-06 [before_grad]: 9.17999e-06 [set_forward_comm_id_for_comm_node_pass]: 3.55998e-06 [meta_fg_expand]: 0.00012274 [flash_sp_send_recv_attached]: 1.78002e-06 [receive_attached]: 2.26998e-06 [after_resolve]: 1.507e-05 [a_after_grad]: 9.58002e-06 [renormalize]: 0.00065554 [add_forward_monad_depend]: 4.94e-06 [auto_monad_grad]: 2.63e-06 [auto_monad_eliminator]: 1.208e-05 [cse]: 2.261e-05 [a_3]: 4.392e-05 [Cycle 3]: 0.00062924, [45] [expand_dump_flag]: 2.04999e-06 [switch_simplify]: 8.18001e-06 [loop_unroll]: 5.79e-06 [a_1]: 0.00012162 [with_stream_mark]: 1.093e-05 [recompute_prepare]: 5.79e-06 [updatestate_depend_eliminate]: 3.45003e-06 [updatestate_assign_eliminate]: 2.60002e-06 [updatestate_loads_eliminate]: 2.59999e-06 [parameter_eliminate]: 1.04998e-06 [a_2]: 6.605e-05 [accelerated_algorithm]: 1.185e-05 [shard]: 1.17e-06 [meta_shard_fg_expand]: 1.47001e-06 [shard_inline]: 5.80002e-06 [merge_send_recv]: 5.82999e-06 [auto_parallel]: 8.27e-06 [parallel]: 6.64999e-06 [flash_sp]: 1.35999e-06 [merge_comm]: 3.09999e-06 [allreduce_fusion]: 3.08e-06 [matmul_add_comm_reduction]: 6.43998e-06 [allreduce_slice_to_reducescatter]: 4.69998e-07 [virtual_shard_identity]: 7.05e-06 [virtual_dataset]: 5.54e-06 [get_grad_eliminate_]: 5.79999e-06 [virtual_output]: 6.06998e-06 [merge_forward]: 2.98998e-06 [cell_reuse_recompute_pass]: 1.64998e-06 [offload_activation]: 8.52e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.13e-05 [merge_recompute_call_nodes]: 1.12e-06 [before_grad]: 9.12001e-06 [set_forward_comm_id_for_comm_node_pass]: 3.71001e-06 [meta_fg_expand]: 2.30002e-06 [flash_sp_send_recv_attached]: 1.01997e-06 [receive_attached]: 1.34003e-06 [after_resolve]: 1.062e-05 [a_after_grad]: 8.52e-06 [renormalize]: 8.00064e-08 [add_forward_monad_depend]: 1.42e-06 [auto_monad_grad]: 1.34e-06 [auto_monad_eliminator]: 8.01001e-06 [cse]: 1.365e-05 [a_3]: 3.452e-05 [py_interpret_to_execute_after_opt_a]: 1.299e-05 [slice_cell_reuse_recomputed_activation]: 2.29999e-06 [rewriter_after_opt_a]: 3.998e-05 [convert_after_rewriter]: 7.43e-06 [order_py_execute_after_rewriter]: 5.44e-06 [mutable_eliminate]: 0.00076048 [opt_b]: 0.00020396, [1] [Cycle 1]: 0.00019607, [7] [b_1]: 0.00011487 [b_2]: 7.87e-06 [updatestate_depend_eliminate]: 7.56999e-06 [updatestate_assign_eliminate]: 2.99001e-06 [updatestate_loads_eliminate]: 2.63e-06 [renormalize]: 7.2e-07 [cse]: 2.268e-05 [optimize_parallel_all_gather_comm]: 1.768e-05 [overlap_param_gather]: 2.00002e-06 [cconv]: 2.785e-05 [loop_unroll]: 0.00048669 [opt_after_cconv]: 0.00010469, [1] [Cycle 1]: 9.846e-05, [7] [c_1]: 2.979e-05 [parameter_eliminate]: 3.21001e-06 [updatestate_depend_eliminate]: 5.91e-06 [updatestate_assign_eliminate]: 2.60002e-06 [updatestate_loads_eliminate]: 2.31e-06 [cse]: 1.888e-05 [renormalize]: 2.80008e-07 [remove_dup_value]: 1.382e-05 [tuple_transform]: 8.27e-05, [1] [Cycle 1]: 7.789e-05, [4] [d_1]: 4.241e-05 [none_parameter_eliminate]: 1.53002e-06 [renormalize]: 1.59984e-07 [switch_simplify]: 1.601e-05 [partial_unused_args_eliminate]: 1.92001e-06 [add_recomputation]: 5.213e-05 [cse_after_recomputation]: 2.36e-05, [1] [Cycle 1]: 1.881e-05, [1] [cse]: 1.317e-05 [environ_conv]: 9.10999e-06 [swap_dp_allreduce_reducescatter]: 5.47001e-06 [bias_add_comm_swap]: 2.26e-06 [label_micro_interleaved_index]: 4.96002e-06 [label_fine_grained_interleaved_index]: 2.98e-06 [merge_cast_opt]: 1.27e-06 [slice_recompute_activation]: 2.11e-06 [micro_interleaved_order_control]: 2.42001e-06 [assign_add_opt]: 1.26002e-06 [ForceFp32Comm]: 8.79983e-07 [remove_cast_before_assign_add]: 1.03001e-06 [full_micro_interleaved_order_control]: 2.66e-06 [reorder_send_recv_between_fp_bp]: 2.63e-06 [comm_op_add_attrs]: 1.23002e-06 [add_comm_op_reuse_tag]: 1.13001e-06 [interleave_split_concat_branches]: 1.24e-06 [interleave_parallel_branches]: 1.07e-06 [overlap_opt_shard_in_pipeline]: 1.19998e-06 [overlap_opt_shard_grad_in_pipeline]: 1.70001e-06 [control_data_broadcast_order]: 1.4e-05 [grouped_pairwise_exchange_alltoall]: 1.74998e-06 [offloading_packed_experts]: 3.98999e-06 [overlap_recompute_and_grad_model_parallel]: 4.41002e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.31002e-06 [overlap_recompute_allgather_and_fa_grad]: 1.32e-06 [overlap_recompute_comm]: 2.07001e-06 [overlap_grad_ring_attention]: 4.23999e-06 [overlap_grad_flash_sp]: 2.037e-05 [begin_end_overlap_inline]: 5.19998e-07 [split_matmul_comm_elemetwise]: 2.11e-06 [split_layernorm_comm]: 1.64e-06 [handle_group_info]: 1.07e-06 [symbol_engine_optimizer]: 8.689e-05, [1] [Cycle 1]: 8.184e-05, [6] [build]: 1.149e-05 [elim_shapecalc]: 1.106e-05 [elim_not_effective]: 1.34e-05 [opt_reshape]: 7.25e-06 [fold_const_symbol]: 9.86e-06 [renormalize]: 2.10013e-07 [detach_backward]: 2.52001e-06 [pipeline_parallel_scheduler]: 1.46998e-06 [auto_monad_reorder]: 1.874e-05 [get_jit_bprop_graph]: 2.19999e-06 [rewriter_after_jit_bprop_graph]: 4.48999e-06 [opt_after_jit_grad]: 0.00049713 [validate]: 4.694e-05 Sums bootstrap : 0.000391s : 0.22% type_inference : 0.083292s : 47.80% event_method : 0.000061s : 0.04% auto_monad : 0.000127s : 0.07% graph_reusing : 0.000009s : 0.01% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000041s : 0.02% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000011s : 0.01% parallel-infer-symbol : 0.000003s : 0.00% pre_auto_parallel : 0.000056s : 0.03% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000040s : 0.02% optimize.rewriter_before_opt_a : 0.000160s : 0.09% optimize.opt_a.expand_dump_flag : 0.000009s : 0.01% optimize.opt_a.switch_simplify : 0.000139s : 0.08% optimize.opt_a.loop_unroll : 0.000117s : 0.07% optimize.opt_a.a_1 : 0.003317s : 1.90% optimize.opt_a.with_stream_mark : 0.000058s : 0.03% optimize.opt_a.recompute_prepare : 0.000040s : 0.02% optimize.opt_a.updatestate_depend_eliminate : 0.000018s : 0.01% optimize.opt_a.updatestate_assign_eliminate : 0.000014s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000013s : 0.01% optimize.opt_a.parameter_eliminate : 0.000006s : 0.00% optimize.opt_a.a_2 : 0.000386s : 0.22% optimize.opt_a.accelerated_algorithm : 0.000057s : 0.03% optimize.opt_a.shard : 0.000006s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000008s : 0.00% optimize.opt_a.shard_inline : 0.000030s : 0.02% optimize.opt_a.merge_send_recv : 0.000034s : 0.02% optimize.opt_a.auto_parallel : 0.000032s : 0.02% optimize.opt_a.parallel : 0.000034s : 0.02% optimize.opt_a.flash_sp : 0.000016s : 0.01% optimize.opt_a.merge_comm : 0.000017s : 0.01% optimize.opt_a.allreduce_fusion : 0.000016s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000045s : 0.03% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000034s : 0.02% optimize.opt_a.virtual_dataset : 0.000027s : 0.02% optimize.opt_a.get_grad_eliminate_ : 0.000027s : 0.02% optimize.opt_a.virtual_output : 0.000027s : 0.02% optimize.opt_a.merge_forward : 0.000018s : 0.01% optimize.opt_a.cell_reuse_recompute_pass : 0.000004s : 0.00% optimize.opt_a.offload_activation : 0.000039s : 0.02% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000054s : 0.03% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.00% optimize.opt_a.before_grad : 0.000047s : 0.03% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000017s : 0.01% optimize.opt_a.meta_fg_expand : 0.001692s : 0.97% optimize.opt_a.flash_sp_send_recv_attached : 0.000009s : 0.01% optimize.opt_a.receive_attached : 0.000006s : 0.00% optimize.opt_a.after_resolve : 0.000091s : 0.05% optimize.opt_a.a_after_grad : 0.000105s : 0.06% optimize.opt_a.renormalize : 0.080158s : 46.00% optimize.opt_a.add_forward_monad_depend : 0.000021s : 0.01% optimize.opt_a.auto_monad_grad : 0.000011s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000089s : 0.05% optimize.opt_a.cse : 0.000266s : 0.15% optimize.opt_a.a_3 : 0.000434s : 0.25% optimize.py_interpret_to_execute_after_opt_a : 0.000013s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000040s : 0.02% optimize.convert_after_rewriter : 0.000007s : 0.00% optimize.order_py_execute_after_rewriter : 0.000005s : 0.00% optimize.mutable_eliminate : 0.000760s : 0.44% optimize.opt_b.b_1 : 0.000115s : 0.07% optimize.opt_b.b_2 : 0.000008s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000008s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000023s : 0.01% optimize.optimize_parallel_all_gather_comm : 0.000018s : 0.01% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000028s : 0.02% optimize.loop_unroll : 0.000487s : 0.28% optimize.opt_after_cconv.c_1 : 0.000030s : 0.02% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000019s : 0.01% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000014s : 0.01% optimize.tuple_transform.d_1 : 0.000042s : 0.02% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000016s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000052s : 0.03% optimize.cse_after_recomputation.cse : 0.000013s : 0.01% optimize.environ_conv : 0.000009s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.00% optimize.bias_add_comm_swap : 0.000002s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000003s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000014s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000004s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000020s : 0.01% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000011s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000011s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000013s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000001s : 0.00% auto_monad_reorder : 0.000019s : 0.01% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.00% opt_after_jit_grad : 0.000497s : 0.29% validate : 0.000047s : 0.03% Time group info: ------[substitution.] 0.000891 190 9.24% : 0.000082s : 6: substitution.arithmetic_simplify 0.23% : 0.000002s : 2: substitution.elim_not_effective 0.56% : 0.000005s : 6: substitution.float_depend_g_call 0.48% : 0.000004s : 3: substitution.float_tuple_getitem_switch 0.14% : 0.000001s : 2: substitution.fold_const_symbol 0.64% : 0.000006s : 4: substitution.graph_param_transform 0.36% : 0.000003s : 2: substitution.incorporate_call 0.22% : 0.000002s : 2: substitution.incorporate_call_switch 55.16% : 0.000492s : 16: substitution.inline 2.10% : 0.000019s : 2: substitution.inline_without_move 0.99% : 0.000009s : 14: substitution.j_node_and_user_rematch 2.04% : 0.000018s : 3: substitution.less_batch_normalization 1.72% : 0.000015s : 11: substitution.minmaximum_grad 0.85% : 0.000008s : 6: substitution.partial_eliminate 1.30% : 0.000012s : 14: substitution.remove_not_recompute_node 3.54% : 0.000032s : 10: substitution.replace_applicator 1.25% : 0.000011s : 12: substitution.replace_old_param 0.36% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.53% : 0.000031s : 11: substitution.tuple_list_convert_item_index_to_positive 1.72% : 0.000015s : 11: substitution.tuple_list_get_item_const_eliminator 2.30% : 0.000020s : 11: substitution.tuple_list_get_item_depend_reorder 9.01% : 0.000080s : 30: substitution.tuple_list_get_item_eliminator 2.28% : 0.000020s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.083204 2 97.90% : 0.081454s : 1: type_inference.infer 2.10% : 0.001750s : 1: type_inference.specialize ------[replace.] 0.000275 34 4.34% : 0.000012s : 2: replace.arithmetic_simplify 55.33% : 0.000152s : 16: replace.inline 40.33% : 0.000111s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.000572 34 8.51% : 0.000049s : 2: match.arithmetic_simplify 84.37% : 0.000482s : 16: match.inline 7.13% : 0.000041s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.000683 4766 1.07% : 0.000007s : 58: predicate.accumulaten_eliminater 0.22% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.42% : 0.000003s : 23: predicate.addn_check_dump 1.10% : 0.000007s : 58: predicate.addn_zero_filter 1.03% : 0.000007s : 58: predicate.adjust_all_reduce_mul_add 2.05% : 0.000014s : 83: predicate.arithmetic_simplify 1.19% : 0.000008s : 60: predicate.cast_eliminate 1.07% : 0.000007s : 55: predicate.check_bprop_eliminate 0.42% : 0.000003s : 23: predicate.compare_switch_simplify 0.05% : 0.000000s : 4: predicate.const_output_eliminate 0.45% : 0.000003s : 23: predicate.depend_value_elim 1.20% : 0.000008s : 60: predicate.dict_get_item_const_eliminator 1.27% : 0.000009s : 60: predicate.dict_get_item_eliminator 1.15% : 0.000008s : 60: predicate.dict_set_item_eliminator 0.36% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.06% : 0.000000s : 4: predicate.elim_not_effective 0.16% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.17% : 0.000008s : 64: predicate.environ_add_const_eliminate 1.15% : 0.000008s : 64: predicate.environ_get_add_eliminate 1.17% : 0.000008s : 64: predicate.environ_get_depend_swap 1.62% : 0.000011s : 87: predicate.environ_get_eliminate 1.14% : 0.000008s : 64: predicate.environ_get_set_eliminate 1.77% : 0.000012s : 92: predicate.exchange_switch_depend_value 2.62% : 0.000018s : 92: predicate.float_depend_g_call 0.47% : 0.000003s : 23: predicate.float_environ_get_switch 0.58% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.05% : 0.000000s : 4: predicate.fold_const_symbol 0.53% : 0.000004s : 23: predicate.get_grad_eliminate 0.06% : 0.000000s : 4: predicate.graph_param_transform 0.45% : 0.000003s : 23: predicate.incorporate_call 0.40% : 0.000003s : 23: predicate.incorporate_call_switch 5.72% : 0.000039s : 204: predicate.inline 1.30% : 0.000009s : 46: predicate.inline_without_move 0.26% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.75% : 0.000005s : 23: predicate.less_batch_normalization 1.68% : 0.000011s : 84: predicate.list_to_tuple_eliminator_ 2.58% : 0.000018s : 142: predicate.load_eliminater 0.37% : 0.000003s : 4: predicate.loop_unroll_after_grad 2.49% : 0.000017s : 128: predicate.loop_unroll_before_grad 1.40% : 0.000010s : 68: predicate.make_slice_get_slice_eliminator 0.44% : 0.000003s : 23: predicate.merge_addn 1.04% : 0.000007s : 55: predicate.micro_step_allgather_replace 1.14% : 0.000008s : 55: predicate.mini_step_allgather_replace 1.13% : 0.000008s : 60: predicate.minmaximum_grad 0.45% : 0.000003s : 4: predicate.mutable_eliminate 0.12% : 0.000001s : 4: predicate.opt_reshape 0.09% : 0.000001s : 4: predicate.parallel_virtual_node 2.31% : 0.000016s : 92: predicate.partial_defer_inline 1.78% : 0.000012s : 80: predicate.partial_eliminate 1.08% : 0.000007s : 58: predicate.print_const_string_wrapper 0.45% : 0.000003s : 23: predicate.reduce_all_const_elim 1.43% : 0.000010s : 60: predicate.reduce_eliminate 2.62% : 0.000018s : 142: predicate.redundant_stop_gradient_eliminater 0.34% : 0.000002s : 23: predicate.remove_not_recompute_node 1.92% : 0.000013s : 131: predicate.replace_applicator 0.64% : 0.000004s : 46: predicate.replace_old_param 0.07% : 0.000001s : 4: predicate.reset_defer_inline 1.22% : 0.000008s : 60: predicate.reshape_eliminate 1.08% : 0.000007s : 55: predicate.row_tensor_add_zeros_like 0.09% : 0.000001s : 4: predicate.row_tensor_eliminate 1.40% : 0.000010s : 55: predicate.same_eliminate 0.33% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.60% : 0.000004s : 23: predicate.shard_identity_eliminate 0.21% : 0.000001s : 8: predicate.special_op_eliminate 0.53% : 0.000004s : 23: predicate.specialize_transform 1.38% : 0.000009s : 55: predicate.split_environ_get_set_with_tuple_value 1.16% : 0.000008s : 46: predicate.stack_unstack_eliminate 0.10% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.97% : 0.000013s : 92: predicate.switch_defer_inline 3.04% : 0.000021s : 147: predicate.switch_layer_defer_inline 5.11% : 0.000035s : 247: predicate.switch_simplify 1.09% : 0.000007s : 60: predicate.tile_eliminate 1.13% : 0.000008s : 60: predicate.transpose_eliminate 1.47% : 0.000010s : 68: predicate.tuple_list_convert_item_index_to_positive 1.63% : 0.000011s : 68: predicate.tuple_list_get_item_const_eliminator 1.35% : 0.000009s : 68: predicate.tuple_list_get_item_depend_reorder 3.18% : 0.000022s : 107: predicate.tuple_list_get_item_eliminator 1.44% : 0.000010s : 68: predicate.tuple_list_get_set_item_eliminator 1.93% : 0.000013s : 91: predicate.tuple_list_set_item_eliminator 1.58% : 0.000011s : 84: predicate.tuple_to_list_eliminator_ 2.56% : 0.000017s : 142: predicate.updatestate_pure_node_eliminater 3.10% : 0.000021s : 165: predicate.updatestate_useless_node_eliminater 0.11% : 0.000001s : 4: predicate.value_based_eliminate 0.51% : 0.000003s : 23: predicate.virtual_dataset_eliminate 0.45% : 0.000003s : 23: predicate.virtual_output_eliminate 0.08% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.13% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002069 37 56.82% : 0.001175s : 17: func_graph_cloner_run.FuncGraphClonerGraph 43.18% : 0.000893s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.358008 233 0.00% : 0.000004s : 1: ForceFp32Comm 0.90% : 0.003239s : 1: add_attr 0.90% : 0.003227s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.02% : 0.000057s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.04% : 0.000135s : 1: auto_monad 0.01% : 0.000022s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000005s : 1: bias_add_comm_swap 0.12% : 0.000413s : 1: bootstrap 0.01% : 0.000031s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.00% : 0.000018s : 1: control_data_broadcast_order 0.00% : 0.000011s : 1: convert_after_rewriter 0.01% : 0.000027s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000012s : 1: environ_conv 0.02% : 0.000070s : 1: event_method 0.00% : 0.000006s : 1: full_micro_interleaved_order_control 0.00% : 0.000006s : 1: get_jit_bprop_graph 0.00% : 0.000013s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.14% : 0.000495s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.22% : 0.000770s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.00% : 0.000016s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000018s : 1: opt.transform.mutable_eliminate 1.34% : 0.004789s : 117: opt.transform.opt_a 0.01% : 0.000029s : 1: opt.transform.opt_after_cconv 0.01% : 0.000026s : 1: opt.transform.opt_after_jit_grad 0.03% : 0.000095s : 28: opt.transform.opt_b 0.02% : 0.000056s : 2: opt.transform.opt_trans_graph 0.01% : 0.000037s : 4: opt.transform.symbol_engine_opt 24.70% : 0.088431s : 1: opt_a 0.03% : 0.000108s : 1: opt_after_cconv 0.14% : 0.000507s : 1: opt_after_jit_grad 0.06% : 0.000208s : 1: opt_b 25.39% : 0.090902s : 1: optimize 0.01% : 0.000022s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000009s : 1: order_py_execute_after_rewriter 0.01% : 0.000024s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000004s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000007s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000007s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.02% : 0.000061s : 1: pre_auto_parallel 0.01% : 0.000044s : 1: py_interpret_to_execute 0.00% : 0.000017s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000018s : 1: remove_dup_value 21.83% : 0.078170s : 2: renormalize.infer 0.55% : 0.001968s : 2: renormalize.specialize 0.00% : 0.000005s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000008s : 1: rewriter_after_jit_bprop_graph 0.01% : 0.000044s : 1: rewriter_after_opt_a 0.05% : 0.000165s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000004s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.03% : 0.000090s : 1: symbol_engine_optimizer 0.02% : 0.000086s : 1: tuple_transform 23.27% : 0.083312s : 1: type_inference TotalTime = 0.0561502, [21] [bootstrap]: 0.00040464 [type_inference]: 0.0373835 [event_method]: 6.596e-05 [auto_monad]: 0.00012613 [graph_reusing]: 8.77e-06 [inline]: 2.57001e-06 [add_attr]: 0.00342221, [1] [add_attr_with_inline]: 0.00341216, [1] [Cycle 1]: 8.477e-05, [2] [tag_attr]: 4.088e-05 [meta_addattr_fg_expand]: 1.107e-05 [parallel-infer-symbol]: 4.55999e-06 [pre_auto_parallel]: 5.746e-05 [insert-virtual-dataset]: 3.33e-06 [parallel-infer-symbol-second]: 8.39995e-07 [dataset_repeat_opt]: 1.97999e-06 [pipeline_split]: 1.86998e-06 [optimize]: 0.0139024, [53] [py_interpret_to_execute]: 4.423e-05 [rewriter_before_opt_a]: 0.00017706 [opt_a]: 0.0115089, [3] [Cycle 1]: 0.00798697, [45] [expand_dump_flag]: 4.68999e-06 [switch_simplify]: 8.56e-05 [loop_unroll]: 7.169e-05 [a_1]: 0.00153413 [with_stream_mark]: 2.586e-05 [recompute_prepare]: 2.297e-05 [updatestate_depend_eliminate]: 9.41003e-06 [updatestate_assign_eliminate]: 8.35001e-06 [updatestate_loads_eliminate]: 7.90998e-06 [parameter_eliminate]: 2.78e-06 [a_2]: 0.00024481 [accelerated_algorithm]: 3.197e-05 [shard]: 1.72001e-06 [meta_shard_fg_expand]: 3.94002e-06 [shard_inline]: 1.713e-05 [merge_send_recv]: 1.778e-05 [auto_parallel]: 1.193e-05 [parallel]: 1.874e-05 [flash_sp]: 9.64e-06 [merge_comm]: 9.84999e-06 [allreduce_fusion]: 9.39e-06 [matmul_add_comm_reduction]: 3.053e-05 [allreduce_slice_to_reducescatter]: 6.30011e-07 [virtual_shard_identity]: 1.787e-05 [virtual_dataset]: 1.629e-05 [get_grad_eliminate_]: 1.526e-05 [virtual_output]: 1.554e-05 [merge_forward]: 1.026e-05 [cell_reuse_recompute_pass]: 1.23002e-06 [offload_activation]: 1.903e-05 [cell_reuse_handle_not_recompute_node_pass]: 2.935e-05 [merge_recompute_call_nodes]: 1.74998e-06 [before_grad]: 2.877e-05 [set_forward_comm_id_for_comm_node_pass]: 1.039e-05 [meta_fg_expand]: 0.00154581 [flash_sp_send_recv_attached]: 6.16998e-06 [receive_attached]: 2.84999e-06 [after_resolve]: 6.192e-05 [a_after_grad]: 8.398e-05 [renormalize]: 0.0030211 [add_forward_monad_depend]: 9.51e-06 [auto_monad_grad]: 6.78e-06 [auto_monad_eliminator]: 5.695e-05 [cse]: 0.00020027 [a_3]: 0.00035961 [Cycle 2]: 0.00288626, [45] [expand_dump_flag]: 1.96998e-06 [switch_simplify]: 4.692e-05 [loop_unroll]: 4.276e-05 [a_1]: 0.00156714 [with_stream_mark]: 1.399e-05 [recompute_prepare]: 8.94e-06 [updatestate_depend_eliminate]: 3.6e-06 [updatestate_assign_eliminate]: 2.78e-06 [updatestate_loads_eliminate]: 2.23002e-06 [parameter_eliminate]: 1.66998e-06 [a_2]: 6.814e-05 [accelerated_algorithm]: 1.077e-05 [shard]: 1.21997e-06 [meta_shard_fg_expand]: 1.98002e-06 [shard_inline]: 6.39001e-06 [merge_send_recv]: 5.71998e-06 [auto_parallel]: 6.44001e-06 [parallel]: 6.07999e-06 [flash_sp]: 3.35e-06 [merge_comm]: 3.33e-06 [allreduce_fusion]: 3.18998e-06 [matmul_add_comm_reduction]: 6.36998e-06 [allreduce_slice_to_reducescatter]: 9.10019e-07 [virtual_shard_identity]: 7.28e-06 [virtual_dataset]: 6.11998e-06 [get_grad_eliminate_]: 5.66e-06 [virtual_output]: 5.55001e-06 [merge_forward]: 3.18998e-06 [cell_reuse_recompute_pass]: 1.18001e-06 [offload_activation]: 7.98999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.065e-05 [merge_recompute_call_nodes]: 1.18001e-06 [before_grad]: 9.47999e-06 [set_forward_comm_id_for_comm_node_pass]: 3.20002e-06 [meta_fg_expand]: 9.54e-05 [flash_sp_send_recv_attached]: 1.17999e-06 [receive_attached]: 1.40001e-06 [after_resolve]: 1.301e-05 [a_after_grad]: 9.76e-06 [renormalize]: 0.0005643 [add_forward_monad_depend]: 4.84e-06 [auto_monad_grad]: 2.29999e-06 [auto_monad_eliminator]: 1.146e-05 [cse]: 1.735e-05 [a_3]: 4.324e-05 [Cycle 3]: 0.00061876, [45] [expand_dump_flag]: 1.19e-06 [switch_simplify]: 8.55001e-06 [loop_unroll]: 5.77001e-06 [a_1]: 0.00013023 [with_stream_mark]: 9.21998e-06 [recompute_prepare]: 6.76e-06 [updatestate_depend_eliminate]: 3.24001e-06 [updatestate_assign_eliminate]: 2.37001e-06 [updatestate_loads_eliminate]: 2.09999e-06 [parameter_eliminate]: 8.39995e-07 [a_2]: 6.512e-05 [accelerated_algorithm]: 1.106e-05 [shard]: 9.29984e-07 [meta_shard_fg_expand]: 1.25001e-06 [shard_inline]: 5.69999e-06 [merge_send_recv]: 4.68001e-06 [auto_parallel]: 6.14999e-06 [parallel]: 5.23002e-06 [flash_sp]: 1.39e-06 [merge_comm]: 3.31001e-06 [allreduce_fusion]: 2.70002e-06 [matmul_add_comm_reduction]: 5.45001e-06 [allreduce_slice_to_reducescatter]: 3.50003e-07 [virtual_shard_identity]: 6.97002e-06 [virtual_dataset]: 5.62001e-06 [get_grad_eliminate_]: 5.43002e-06 [virtual_output]: 5.61e-06 [merge_forward]: 2.80002e-06 [cell_reuse_recompute_pass]: 1.62001e-06 [offload_activation]: 6.94001e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.086e-05 [merge_recompute_call_nodes]: 1.00001e-06 [before_grad]: 8.62e-06 [set_forward_comm_id_for_comm_node_pass]: 3.46001e-06 [meta_fg_expand]: 1.89999e-06 [flash_sp_send_recv_attached]: 8.29983e-07 [receive_attached]: 1.05999e-06 [after_resolve]: 9.87999e-06 [a_after_grad]: 8.61002e-06 [renormalize]: 8.9989e-08 [add_forward_monad_depend]: 1.15999e-06 [auto_monad_grad]: 9.00007e-07 [auto_monad_eliminator]: 7.07997e-06 [cse]: 1.187e-05 [a_3]: 3.593e-05 [py_interpret_to_execute_after_opt_a]: 1.251e-05 [slice_cell_reuse_recomputed_activation]: 2.53e-06 [rewriter_after_opt_a]: 3.712e-05 [convert_after_rewriter]: 7.93999e-06 [order_py_execute_after_rewriter]: 5.60001e-06 [mutable_eliminate]: 0.00070657 [opt_b]: 0.00020326, [1] [Cycle 1]: 0.00019519, [7] [b_1]: 0.00012292 [b_2]: 8.27e-06 [updatestate_depend_eliminate]: 5.52999e-06 [updatestate_assign_eliminate]: 2.46998e-06 [updatestate_loads_eliminate]: 2.14999e-06 [renormalize]: 2.3999e-07 [cse]: 1.791e-05 [optimize_parallel_all_gather_comm]: 1.75e-05 [overlap_param_gather]: 1.79998e-06 [cconv]: 2.223e-05 [loop_unroll]: 0.00045934 [opt_after_cconv]: 0.00010119, [1] [Cycle 1]: 9.52e-05, [7] [c_1]: 2.999e-05 [parameter_eliminate]: 2.64001e-06 [updatestate_depend_eliminate]: 5.70001e-06 [updatestate_assign_eliminate]: 2.41998e-06 [updatestate_loads_eliminate]: 2.53e-06 [cse]: 1.857e-05 [renormalize]: 3.50003e-07 [remove_dup_value]: 1.314e-05 [tuple_transform]: 9.126e-05, [1] [Cycle 1]: 8.635e-05, [4] [d_1]: 5.742e-05 [none_parameter_eliminate]: 1.72999e-06 [renormalize]: 2.30008e-07 [switch_simplify]: 7.7e-06 [partial_unused_args_eliminate]: 1.59e-06 [add_recomputation]: 4.6e-05 [cse_after_recomputation]: 2.326e-05, [1] [Cycle 1]: 1.834e-05, [1] [cse]: 1.257e-05 [environ_conv]: 8.3e-06 [swap_dp_allreduce_reducescatter]: 5.54e-06 [bias_add_comm_swap]: 2.99001e-06 [label_micro_interleaved_index]: 4.60001e-06 [label_fine_grained_interleaved_index]: 3.09999e-06 [merge_cast_opt]: 1.34e-06 [slice_recompute_activation]: 2.24999e-06 [micro_interleaved_order_control]: 2.44001e-06 [assign_add_opt]: 1.29e-06 [ForceFp32Comm]: 1.17e-06 [remove_cast_before_assign_add]: 1.06002e-06 [full_micro_interleaved_order_control]: 2.43e-06 [reorder_send_recv_between_fp_bp]: 2.93e-06 [comm_op_add_attrs]: 1.12e-06 [add_comm_op_reuse_tag]: 1.25001e-06 [interleave_split_concat_branches]: 1.20999e-06 [interleave_parallel_branches]: 1.10001e-06 [overlap_opt_shard_in_pipeline]: 1.29e-06 [overlap_opt_shard_grad_in_pipeline]: 1.92001e-06 [control_data_broadcast_order]: 1.215e-05 [grouped_pairwise_exchange_alltoall]: 1.59998e-06 [offloading_packed_experts]: 3.94002e-06 [overlap_recompute_and_grad_model_parallel]: 4.72e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.57001e-06 [overlap_recompute_allgather_and_fa_grad]: 1.35001e-06 [overlap_recompute_comm]: 2.26e-06 [overlap_grad_ring_attention]: 4.25999e-06 [overlap_grad_flash_sp]: 1.814e-05 [begin_end_overlap_inline]: 7.50006e-07 [split_matmul_comm_elemetwise]: 2.21e-06 [split_layernorm_comm]: 1.70001e-06 [handle_group_info]: 1.05999e-06 [symbol_engine_optimizer]: 8.228e-05, [1] [Cycle 1]: 7.734e-05, [6] [build]: 6.74001e-06 [elim_shapecalc]: 1.042e-05 [elim_not_effective]: 1.323e-05 [opt_reshape]: 6.91001e-06 [fold_const_symbol]: 9.68997e-06 [renormalize]: 2.00002e-07 [detach_backward]: 2.48e-06 [pipeline_parallel_scheduler]: 1.69e-06 [auto_monad_reorder]: 1.734e-05 [get_jit_bprop_graph]: 1.57999e-06 [rewriter_after_jit_bprop_graph]: 4.85999e-06 [opt_after_jit_grad]: 0.00050677 [validate]: 4.297e-05 Sums bootstrap : 0.000405s : 0.79% type_inference : 0.037384s : 72.64% event_method : 0.000066s : 0.13% auto_monad : 0.000126s : 0.25% graph_reusing : 0.000009s : 0.02% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000041s : 0.08% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000011s : 0.02% parallel-infer-symbol : 0.000005s : 0.01% pre_auto_parallel : 0.000057s : 0.11% insert-virtual-dataset : 0.000003s : 0.01% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000044s : 0.09% optimize.rewriter_before_opt_a : 0.000177s : 0.34% optimize.opt_a.expand_dump_flag : 0.000008s : 0.02% optimize.opt_a.switch_simplify : 0.000141s : 0.27% optimize.opt_a.loop_unroll : 0.000120s : 0.23% optimize.opt_a.a_1 : 0.003232s : 6.28% optimize.opt_a.with_stream_mark : 0.000049s : 0.10% optimize.opt_a.recompute_prepare : 0.000039s : 0.08% optimize.opt_a.updatestate_depend_eliminate : 0.000016s : 0.03% optimize.opt_a.updatestate_assign_eliminate : 0.000014s : 0.03% optimize.opt_a.updatestate_loads_eliminate : 0.000012s : 0.02% optimize.opt_a.parameter_eliminate : 0.000005s : 0.01% optimize.opt_a.a_2 : 0.000378s : 0.73% optimize.opt_a.accelerated_algorithm : 0.000054s : 0.10% optimize.opt_a.shard : 0.000004s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000007s : 0.01% optimize.opt_a.shard_inline : 0.000029s : 0.06% optimize.opt_a.merge_send_recv : 0.000028s : 0.05% optimize.opt_a.auto_parallel : 0.000025s : 0.05% optimize.opt_a.parallel : 0.000030s : 0.06% optimize.opt_a.flash_sp : 0.000014s : 0.03% optimize.opt_a.merge_comm : 0.000016s : 0.03% optimize.opt_a.allreduce_fusion : 0.000015s : 0.03% optimize.opt_a.matmul_add_comm_reduction : 0.000042s : 0.08% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000032s : 0.06% optimize.opt_a.virtual_dataset : 0.000028s : 0.05% optimize.opt_a.get_grad_eliminate_ : 0.000026s : 0.05% optimize.opt_a.virtual_output : 0.000027s : 0.05% optimize.opt_a.merge_forward : 0.000016s : 0.03% optimize.opt_a.cell_reuse_recompute_pass : 0.000004s : 0.01% optimize.opt_a.offload_activation : 0.000034s : 0.07% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000051s : 0.10% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.01% optimize.opt_a.before_grad : 0.000047s : 0.09% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000017s : 0.03% optimize.opt_a.meta_fg_expand : 0.001643s : 3.19% optimize.opt_a.flash_sp_send_recv_attached : 0.000008s : 0.02% optimize.opt_a.receive_attached : 0.000005s : 0.01% optimize.opt_a.after_resolve : 0.000085s : 0.16% optimize.opt_a.a_after_grad : 0.000102s : 0.20% optimize.opt_a.renormalize : 0.003585s : 6.97% optimize.opt_a.add_forward_monad_depend : 0.000016s : 0.03% optimize.opt_a.auto_monad_grad : 0.000010s : 0.02% optimize.opt_a.auto_monad_eliminator : 0.000075s : 0.15% optimize.opt_a.cse : 0.000229s : 0.45% optimize.opt_a.a_3 : 0.000439s : 0.85% optimize.py_interpret_to_execute_after_opt_a : 0.000013s : 0.02% optimize.slice_cell_reuse_recomputed_activation : 0.000003s : 0.00% optimize.rewriter_after_opt_a : 0.000037s : 0.07% optimize.convert_after_rewriter : 0.000008s : 0.02% optimize.order_py_execute_after_rewriter : 0.000006s : 0.01% optimize.mutable_eliminate : 0.000707s : 1.37% optimize.opt_b.b_1 : 0.000123s : 0.24% optimize.opt_b.b_2 : 0.000008s : 0.02% optimize.opt_b.updatestate_depend_eliminate : 0.000006s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000018s : 0.03% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.03% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000022s : 0.04% optimize.loop_unroll : 0.000459s : 0.89% optimize.opt_after_cconv.c_1 : 0.000030s : 0.06% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.cse : 0.000019s : 0.04% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000013s : 0.03% optimize.tuple_transform.d_1 : 0.000057s : 0.11% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000008s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000046s : 0.09% optimize.cse_after_recomputation.cse : 0.000013s : 0.02% optimize.environ_conv : 0.000008s : 0.02% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.01% optimize.bias_add_comm_swap : 0.000003s : 0.01% optimize.label_micro_interleaved_index : 0.000005s : 0.01% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.01% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.01% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000012s : 0.02% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.01% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000002s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.01% optimize.overlap_grad_flash_sp : 0.000018s : 0.04% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000007s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000010s : 0.02% optimize.symbol_engine_optimizer.elim_not_effective : 0.000013s : 0.03% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.01% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.02% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000017s : 0.03% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000005s : 0.01% opt_after_jit_grad : 0.000507s : 0.98% validate : 0.000043s : 0.08% Time group info: ------[substitution.] 0.000839 190 8.01% : 0.000067s : 6: substitution.arithmetic_simplify 0.24% : 0.000002s : 2: substitution.elim_not_effective 0.59% : 0.000005s : 6: substitution.float_depend_g_call 0.53% : 0.000004s : 3: substitution.float_tuple_getitem_switch 0.17% : 0.000001s : 2: substitution.fold_const_symbol 0.76% : 0.000006s : 4: substitution.graph_param_transform 0.32% : 0.000003s : 2: substitution.incorporate_call 0.31% : 0.000003s : 2: substitution.incorporate_call_switch 57.08% : 0.000479s : 16: substitution.inline 2.09% : 0.000018s : 2: substitution.inline_without_move 1.11% : 0.000009s : 14: substitution.j_node_and_user_rematch 1.97% : 0.000017s : 3: substitution.less_batch_normalization 1.70% : 0.000014s : 11: substitution.minmaximum_grad 0.73% : 0.000006s : 6: substitution.partial_eliminate 1.27% : 0.000011s : 14: substitution.remove_not_recompute_node 3.28% : 0.000028s : 10: substitution.replace_applicator 1.17% : 0.000010s : 12: substitution.replace_old_param 0.32% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.57% : 0.000030s : 11: substitution.tuple_list_convert_item_index_to_positive 1.75% : 0.000015s : 11: substitution.tuple_list_get_item_const_eliminator 2.29% : 0.000019s : 11: substitution.tuple_list_get_item_depend_reorder 8.50% : 0.000071s : 30: substitution.tuple_list_get_item_eliminator 2.24% : 0.000019s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.037292 2 94.98% : 0.035421s : 1: type_inference.infer 5.02% : 0.001871s : 1: type_inference.specialize ------[replace.] 0.000262 34 4.20% : 0.000011s : 2: replace.arithmetic_simplify 54.37% : 0.000143s : 16: replace.inline 41.44% : 0.000109s : 16: replace.tuple_list_get_item_eliminator ------[match.] 0.000547 34 7.34% : 0.000040s : 2: match.arithmetic_simplify 85.84% : 0.000469s : 16: match.inline 6.82% : 0.000037s : 16: match.tuple_list_get_item_eliminator ------[predicate.] 0.000679 4766 1.12% : 0.000008s : 58: predicate.accumulaten_eliminater 0.29% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.42% : 0.000003s : 23: predicate.addn_check_dump 1.11% : 0.000008s : 58: predicate.addn_zero_filter 1.03% : 0.000007s : 58: predicate.adjust_all_reduce_mul_add 2.05% : 0.000014s : 83: predicate.arithmetic_simplify 1.14% : 0.000008s : 60: predicate.cast_eliminate 1.09% : 0.000007s : 55: predicate.check_bprop_eliminate 0.43% : 0.000003s : 23: predicate.compare_switch_simplify 0.06% : 0.000000s : 4: predicate.const_output_eliminate 0.45% : 0.000003s : 23: predicate.depend_value_elim 1.22% : 0.000008s : 60: predicate.dict_get_item_const_eliminator 1.32% : 0.000009s : 60: predicate.dict_get_item_eliminator 1.13% : 0.000008s : 60: predicate.dict_set_item_eliminator 0.25% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.07% : 0.000000s : 4: predicate.elim_not_effective 0.11% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.21% : 0.000008s : 64: predicate.environ_add_const_eliminate 1.19% : 0.000008s : 64: predicate.environ_get_add_eliminate 1.20% : 0.000008s : 64: predicate.environ_get_depend_swap 1.72% : 0.000012s : 87: predicate.environ_get_eliminate 1.26% : 0.000009s : 64: predicate.environ_get_set_eliminate 1.83% : 0.000012s : 92: predicate.exchange_switch_depend_value 2.52% : 0.000017s : 92: predicate.float_depend_g_call 0.42% : 0.000003s : 23: predicate.float_environ_get_switch 0.51% : 0.000003s : 27: predicate.float_tuple_getitem_switch 0.05% : 0.000000s : 4: predicate.fold_const_symbol 0.50% : 0.000003s : 23: predicate.get_grad_eliminate 0.06% : 0.000000s : 4: predicate.graph_param_transform 0.44% : 0.000003s : 23: predicate.incorporate_call 0.40% : 0.000003s : 23: predicate.incorporate_call_switch 5.57% : 0.000038s : 204: predicate.inline 1.34% : 0.000009s : 46: predicate.inline_without_move 0.25% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.63% : 0.000004s : 23: predicate.less_batch_normalization 1.64% : 0.000011s : 84: predicate.list_to_tuple_eliminator_ 2.59% : 0.000018s : 142: predicate.load_eliminater 0.28% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.64% : 0.000018s : 128: predicate.loop_unroll_before_grad 1.33% : 0.000009s : 68: predicate.make_slice_get_slice_eliminator 0.45% : 0.000003s : 23: predicate.merge_addn 1.18% : 0.000008s : 55: predicate.micro_step_allgather_replace 1.08% : 0.000007s : 55: predicate.mini_step_allgather_replace 1.14% : 0.000008s : 60: predicate.minmaximum_grad 0.34% : 0.000002s : 4: predicate.mutable_eliminate 0.10% : 0.000001s : 4: predicate.opt_reshape 0.10% : 0.000001s : 4: predicate.parallel_virtual_node 2.37% : 0.000016s : 92: predicate.partial_defer_inline 1.79% : 0.000012s : 80: predicate.partial_eliminate 1.09% : 0.000007s : 58: predicate.print_const_string_wrapper 0.44% : 0.000003s : 23: predicate.reduce_all_const_elim 1.46% : 0.000010s : 60: predicate.reduce_eliminate 2.63% : 0.000018s : 142: predicate.redundant_stop_gradient_eliminater 0.28% : 0.000002s : 23: predicate.remove_not_recompute_node 1.92% : 0.000013s : 131: predicate.replace_applicator 0.59% : 0.000004s : 46: predicate.replace_old_param 0.08% : 0.000001s : 4: predicate.reset_defer_inline 1.16% : 0.000008s : 60: predicate.reshape_eliminate 1.14% : 0.000008s : 55: predicate.row_tensor_add_zeros_like 0.10% : 0.000001s : 4: predicate.row_tensor_eliminate 1.28% : 0.000009s : 55: predicate.same_eliminate 0.34% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.53% : 0.000004s : 23: predicate.shard_identity_eliminate 0.27% : 0.000002s : 8: predicate.special_op_eliminate 0.54% : 0.000004s : 23: predicate.specialize_transform 1.32% : 0.000009s : 55: predicate.split_environ_get_set_with_tuple_value 1.16% : 0.000008s : 46: predicate.stack_unstack_eliminate 0.10% : 0.000001s : 4: predicate.switch_call_monad_eliminater 2.03% : 0.000014s : 92: predicate.switch_defer_inline 2.99% : 0.000020s : 147: predicate.switch_layer_defer_inline 5.31% : 0.000036s : 247: predicate.switch_simplify 1.12% : 0.000008s : 60: predicate.tile_eliminate 1.15% : 0.000008s : 60: predicate.transpose_eliminate 1.45% : 0.000010s : 68: predicate.tuple_list_convert_item_index_to_positive 1.69% : 0.000012s : 68: predicate.tuple_list_get_item_const_eliminator 1.39% : 0.000009s : 68: predicate.tuple_list_get_item_depend_reorder 2.93% : 0.000020s : 107: predicate.tuple_list_get_item_eliminator 1.50% : 0.000010s : 68: predicate.tuple_list_get_set_item_eliminator 1.92% : 0.000013s : 91: predicate.tuple_list_set_item_eliminator 1.66% : 0.000011s : 84: predicate.tuple_to_list_eliminator_ 2.55% : 0.000017s : 142: predicate.updatestate_pure_node_eliminater 3.14% : 0.000021s : 165: predicate.updatestate_useless_node_eliminater 0.11% : 0.000001s : 4: predicate.value_based_eliminate 0.53% : 0.000004s : 23: predicate.virtual_dataset_eliminate 0.48% : 0.000003s : 23: predicate.virtual_output_eliminate 0.08% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.13% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002103 37 62.28% : 0.001310s : 17: func_graph_cloner_run.FuncGraphClonerGraph 37.72% : 0.000793s : 20: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.081880 233 0.00% : 0.000004s : 1: ForceFp32Comm 4.19% : 0.003428s : 1: add_attr 4.17% : 0.003416s : 1: add_attr_with_inline 0.01% : 0.000004s : 1: add_comm_op_reuse_tag 0.06% : 0.000051s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.16% : 0.000134s : 1: auto_monad 0.03% : 0.000021s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.01% : 0.000006s : 1: bias_add_comm_swap 0.53% : 0.000435s : 1: bootstrap 0.03% : 0.000026s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.02% : 0.000016s : 1: control_data_broadcast_order 0.01% : 0.000011s : 1: convert_after_rewriter 0.03% : 0.000026s : 1: cse_after_recomputation 0.01% : 0.000005s : 1: dataset_repeat_opt 0.01% : 0.000006s : 1: detach_backward 0.01% : 0.000012s : 1: environ_conv 0.09% : 0.000075s : 1: event_method 0.01% : 0.000006s : 1: full_micro_interleaved_order_control 0.01% : 0.000005s : 1: get_jit_bprop_graph 0.02% : 0.000013s : 1: graph_reusing 0.01% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.01% : 0.000006s : 1: inline 0.01% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.01% : 0.000007s : 1: label_fine_grained_interleaved_index 0.01% : 0.000008s : 1: label_micro_interleaved_index 0.57% : 0.000468s : 1: loop_unroll 0.01% : 0.000004s : 1: merge_cast_opt 0.01% : 0.000005s : 1: micro_interleaved_order_control 0.88% : 0.000717s : 1: mutable_eliminate 0.01% : 0.000007s : 1: offloading_packed_experts 0.02% : 0.000013s : 1: opt.transform.loop_unroll_optimizer 0.02% : 0.000016s : 1: opt.transform.mutable_eliminate 5.74% : 0.004698s : 117: opt.transform.opt_a 0.04% : 0.000029s : 1: opt.transform.opt_after_cconv 0.03% : 0.000025s : 1: opt.transform.opt_after_jit_grad 0.12% : 0.000101s : 28: opt.transform.opt_b 0.08% : 0.000063s : 2: opt.transform.opt_trans_graph 0.04% : 0.000036s : 4: opt.transform.symbol_engine_opt 14.06% : 0.011512s : 1: opt_a 0.13% : 0.000105s : 1: opt_after_cconv 0.63% : 0.000516s : 1: opt_after_jit_grad 0.25% : 0.000207s : 1: opt_b 16.99% : 0.013908s : 1: optimize 0.03% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000009s : 1: order_py_execute_after_rewriter 0.03% : 0.000022s : 1: overlap_grad_flash_sp 0.01% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000007s : 1: overlap_grad_ring_attention 0.01% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.01% : 0.000005s : 1: overlap_param_gather 0.01% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.01% : 0.000005s : 1: overlap_recompute_comm 0.01% : 0.000009s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.01% : 0.000005s : 1: partial_unused_args_eliminate 0.01% : 0.000005s : 1: pipeline_parallel_scheduler 0.01% : 0.000007s : 1: pipeline_split 0.08% : 0.000062s : 1: pre_auto_parallel 0.06% : 0.000049s : 1: py_interpret_to_execute 0.02% : 0.000016s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.02% : 0.000017s : 1: remove_dup_value 2.39% : 0.001955s : 2: renormalize.infer 1.97% : 0.001615s : 2: renormalize.specialize 0.01% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000008s : 1: rewriter_after_jit_bprop_graph 0.05% : 0.000041s : 1: rewriter_after_opt_a 0.22% : 0.000182s : 1: rewriter_before_opt_a 0.01% : 0.000006s : 1: slice_cell_reuse_recomputed_activation 0.01% : 0.000005s : 1: slice_recompute_activation 0.01% : 0.000004s : 1: split_layernorm_comm 0.01% : 0.000005s : 1: split_matmul_comm_elemetwise 0.01% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.10% : 0.000085s : 1: symbol_engine_optimizer 0.12% : 0.000094s : 1: tuple_transform 45.69% : 0.037407s : 1: type_inference [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:45:31.177.735 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:45:31.178.092 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:45:31.351.437 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:45:31.351.723 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. TotalTime = 0.0189308, [21] [bootstrap]: 0.00043151 [type_inference]: 0.00869276 [event_method]: 4.766e-05 [auto_monad]: 7.238e-05 [graph_reusing]: 8.03999e-06 [inline]: 2.30002e-06 [add_attr]: 0.00344525, [1] [add_attr_with_inline]: 0.0034347, [1] [Cycle 1]: 7.727e-05, [2] [tag_attr]: 3.086e-05 [meta_addattr_fg_expand]: 6.83e-06 [parallel-infer-symbol]: 3.11001e-06 [pre_auto_parallel]: 4.4e-05 [insert-virtual-dataset]: 2.68998e-06 [parallel-infer-symbol-second]: 1.02e-06 [dataset_repeat_opt]: 2.17001e-06 [pipeline_split]: 1.67001e-06 [optimize]: 0.00531472, [53] [py_interpret_to_execute]: 3.189e-05 [rewriter_before_opt_a]: 0.00010452 [opt_a]: 0.00315175, [2] [Cycle 1]: 0.00250807, [45] [expand_dump_flag]: 3.56999e-06 [switch_simplify]: 5.5e-05 [loop_unroll]: 3.971e-05 [a_1]: 0.00091188 [with_stream_mark]: 2.009e-05 [recompute_prepare]: 9.76998e-06 [updatestate_depend_eliminate]: 4.65001e-06 [updatestate_assign_eliminate]: 3.32002e-06 [updatestate_loads_eliminate]: 2.88e-06 [parameter_eliminate]: 2.06e-06 [a_2]: 8.383e-05 [accelerated_algorithm]: 6.99001e-06 [shard]: 1.80001e-06 [meta_shard_fg_expand]: 2.04999e-06 [shard_inline]: 6.14999e-06 [merge_send_recv]: 8.47e-06 [auto_parallel]: 6.81001e-06 [parallel]: 2.044e-05 [flash_sp]: 9.27001e-06 [merge_comm]: 4.12e-06 [allreduce_fusion]: 3.47002e-06 [matmul_add_comm_reduction]: 9.44e-06 [allreduce_slice_to_reducescatter]: 6.39993e-07 [virtual_shard_identity]: 7.78999e-06 [virtual_dataset]: 6.49001e-06 [get_grad_eliminate_]: 6.12001e-06 [virtual_output]: 5.99e-06 [merge_forward]: 4.3e-06 [cell_reuse_recompute_pass]: 1.44e-06 [offload_activation]: 1.052e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.203e-05 [merge_recompute_call_nodes]: 1.81998e-06 [before_grad]: 1.057e-05 [set_forward_comm_id_for_comm_node_pass]: 4.28001e-06 [meta_fg_expand]: 2.74001e-06 [flash_sp_send_recv_attached]: 2.56e-06 [receive_attached]: 2.73003e-06 [after_resolve]: 1.2e-05 [a_after_grad]: 9.41e-06 [renormalize]: 0.00083478 [add_forward_monad_depend]: 5.77001e-06 [auto_monad_grad]: 2.44001e-06 [auto_monad_eliminator]: 1.559e-05 [cse]: 2.847e-05 [a_3]: 4.565e-05 [Cycle 2]: 0.00063292, [45] [expand_dump_flag]: 1.19998e-06 [switch_simplify]: 7.75e-06 [loop_unroll]: 5.88998e-06 [a_1]: 0.00013359 [with_stream_mark]: 1.248e-05 [recompute_prepare]: 6.63e-06 [updatestate_depend_eliminate]: 2.94999e-06 [updatestate_assign_eliminate]: 2.47001e-06 [updatestate_loads_eliminate]: 2.67001e-06 [parameter_eliminate]: 1.12e-06 [a_2]: 7.205e-05 [accelerated_algorithm]: 6.44001e-06 [shard]: 1.25999e-06 [meta_shard_fg_expand]: 1.34e-06 [shard_inline]: 5.92001e-06 [merge_send_recv]: 4.53001e-06 [auto_parallel]: 5.64998e-06 [parallel]: 4.62e-06 [flash_sp]: 3.42002e-06 [merge_comm]: 3.26001e-06 [allreduce_fusion]: 3.18998e-06 [matmul_add_comm_reduction]: 5.86e-06 [allreduce_slice_to_reducescatter]: 4.30009e-07 [virtual_shard_identity]: 6.38e-06 [virtual_dataset]: 5.70001e-06 [get_grad_eliminate_]: 5.40001e-06 [virtual_output]: 5.21998e-06 [merge_forward]: 3.11999e-06 [cell_reuse_recompute_pass]: 1.71e-06 [offload_activation]: 6.68e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.02e-05 [merge_recompute_call_nodes]: 7.7e-07 [before_grad]: 9.54e-06 [set_forward_comm_id_for_comm_node_pass]: 3.75e-06 [meta_fg_expand]: 1.92999e-06 [flash_sp_send_recv_attached]: 8.09989e-07 [receive_attached]: 1.21002e-06 [after_resolve]: 1.081e-05 [a_after_grad]: 9.03002e-06 [renormalize]: 7.99773e-08 [add_forward_monad_depend]: 1.45001e-06 [auto_monad_grad]: 8.40024e-07 [auto_monad_eliminator]: 7.03998e-06 [cse]: 1.299e-05 [a_3]: 3.401e-05 [py_interpret_to_execute_after_opt_a]: 1.074e-05 [slice_cell_reuse_recomputed_activation]: 2.24001e-06 [rewriter_after_opt_a]: 3.319e-05 [convert_after_rewriter]: 7.21001e-06 [order_py_execute_after_rewriter]: 5.20999e-06 [mutable_eliminate]: 0.00056585 [opt_b]: 0.00019258, [1] [Cycle 1]: 0.00018613, [7] [b_1]: 0.00011604 [b_2]: 7.58001e-06 [updatestate_depend_eliminate]: 5.74e-06 [updatestate_assign_eliminate]: 2.50002e-06 [updatestate_loads_eliminate]: 2.50997e-06 [renormalize]: 5.00004e-07 [cse]: 1.598e-05 [optimize_parallel_all_gather_comm]: 1.752e-05 [overlap_param_gather]: 1.69e-06 [cconv]: 2.573e-05 [loop_unroll]: 0.00043548 [opt_after_cconv]: 9.942e-05, [1] [Cycle 1]: 9.333e-05, [7] [c_1]: 2.935e-05 [parameter_eliminate]: 2.68e-06 [updatestate_depend_eliminate]: 5.55001e-06 [updatestate_assign_eliminate]: 2.59999e-06 [updatestate_loads_eliminate]: 2.27001e-06 [cse]: 1.665e-05 [renormalize]: 4.39992e-07 [remove_dup_value]: 1.434e-05 [tuple_transform]: 7.283e-05, [1] [Cycle 1]: 6.866e-05, [4] [d_1]: 4.173e-05 [none_parameter_eliminate]: 1.66002e-06 [renormalize]: 2.00002e-07 [switch_simplify]: 6.64001e-06 [partial_unused_args_eliminate]: 2.01e-06 [add_recomputation]: 7.698e-05 [cse_after_recomputation]: 2.294e-05, [1] [Cycle 1]: 1.82e-05, [1] [cse]: 1.251e-05 [environ_conv]: 5.05001e-06 [swap_dp_allreduce_reducescatter]: 5.71e-06 [bias_add_comm_swap]: 3.11999e-06 [label_micro_interleaved_index]: 4.57e-06 [label_fine_grained_interleaved_index]: 2.73003e-06 [merge_cast_opt]: 1.38002e-06 [slice_recompute_activation]: 2.19001e-06 [micro_interleaved_order_control]: 2.56e-06 [assign_add_opt]: 1.29e-06 [ForceFp32Comm]: 7.79983e-07 [remove_cast_before_assign_add]: 1.03001e-06 [full_micro_interleaved_order_control]: 2.25002e-06 [reorder_send_recv_between_fp_bp]: 2.77002e-06 [comm_op_add_attrs]: 1.27e-06 [add_comm_op_reuse_tag]: 1.17e-06 [interleave_split_concat_branches]: 1.25999e-06 [interleave_parallel_branches]: 1.04e-06 [overlap_opt_shard_in_pipeline]: 1.22999e-06 [overlap_opt_shard_grad_in_pipeline]: 1.80001e-06 [control_data_broadcast_order]: 1.321e-05 [grouped_pairwise_exchange_alltoall]: 1.901e-05 [offloading_packed_experts]: 4.52003e-06 [overlap_recompute_and_grad_model_parallel]: 5.25001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.30001e-06 [overlap_recompute_allgather_and_fa_grad]: 1.41002e-06 [overlap_recompute_comm]: 2.13002e-06 [overlap_grad_ring_attention]: 3.88999e-06 [overlap_grad_flash_sp]: 1.941e-05 [begin_end_overlap_inline]: 6.59988e-07 [split_matmul_comm_elemetwise]: 2.29001e-06 [split_layernorm_comm]: 1.61998e-06 [handle_group_info]: 1.22e-06 [symbol_engine_optimizer]: 7.978e-05, [1] [Cycle 1]: 7.505e-05, [6] [build]: 3.70998e-06 [elim_shapecalc]: 1.093e-05 [elim_not_effective]: 1.466e-05 [opt_reshape]: 7.45e-06 [fold_const_symbol]: 9.65002e-06 [renormalize]: 2.20025e-07 [detach_backward]: 2.12999e-06 [pipeline_parallel_scheduler]: 1.40001e-06 [auto_monad_reorder]: 1.694e-05 [get_jit_bprop_graph]: 1.76e-06 [rewriter_after_jit_bprop_graph]: 0.00015155 [opt_after_jit_grad]: 0.00048552 [validate]: 3.787e-05 Sums bootstrap : 0.000432s : 2.98% type_inference : 0.008693s : 59.95% event_method : 0.000048s : 0.33% auto_monad : 0.000072s : 0.50% graph_reusing : 0.000008s : 0.06% inline : 0.000002s : 0.02% add_attr.add_attr_with_inline.tag_attr : 0.000031s : 0.21% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.05% parallel-infer-symbol : 0.000003s : 0.02% pre_auto_parallel : 0.000044s : 0.30% insert-virtual-dataset : 0.000003s : 0.02% parallel-infer-symbol-second : 0.000001s : 0.01% dataset_repeat_opt : 0.000002s : 0.01% pipeline_split : 0.000002s : 0.01% optimize.py_interpret_to_execute : 0.000032s : 0.22% optimize.rewriter_before_opt_a : 0.000105s : 0.72% optimize.opt_a.expand_dump_flag : 0.000005s : 0.03% optimize.opt_a.switch_simplify : 0.000063s : 0.43% optimize.opt_a.loop_unroll : 0.000046s : 0.31% optimize.opt_a.a_1 : 0.001045s : 7.21% optimize.opt_a.with_stream_mark : 0.000033s : 0.22% optimize.opt_a.recompute_prepare : 0.000016s : 0.11% optimize.opt_a.updatestate_depend_eliminate : 0.000008s : 0.05% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.04% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.04% optimize.opt_a.parameter_eliminate : 0.000003s : 0.02% optimize.opt_a.a_2 : 0.000156s : 1.07% optimize.opt_a.accelerated_algorithm : 0.000013s : 0.09% optimize.opt_a.shard : 0.000003s : 0.02% optimize.opt_a.meta_shard_fg_expand : 0.000003s : 0.02% optimize.opt_a.shard_inline : 0.000012s : 0.08% optimize.opt_a.merge_send_recv : 0.000013s : 0.09% optimize.opt_a.auto_parallel : 0.000012s : 0.09% optimize.opt_a.parallel : 0.000025s : 0.17% optimize.opt_a.flash_sp : 0.000013s : 0.09% optimize.opt_a.merge_comm : 0.000007s : 0.05% optimize.opt_a.allreduce_fusion : 0.000007s : 0.05% optimize.opt_a.matmul_add_comm_reduction : 0.000015s : 0.11% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.01% optimize.opt_a.virtual_shard_identity : 0.000014s : 0.10% optimize.opt_a.virtual_dataset : 0.000012s : 0.08% optimize.opt_a.get_grad_eliminate_ : 0.000012s : 0.08% optimize.opt_a.virtual_output : 0.000011s : 0.08% optimize.opt_a.merge_forward : 0.000007s : 0.05% optimize.opt_a.cell_reuse_recompute_pass : 0.000003s : 0.02% optimize.opt_a.offload_activation : 0.000017s : 0.12% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000022s : 0.15% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.02% optimize.opt_a.before_grad : 0.000020s : 0.14% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000008s : 0.06% optimize.opt_a.meta_fg_expand : 0.000005s : 0.03% optimize.opt_a.flash_sp_send_recv_attached : 0.000003s : 0.02% optimize.opt_a.receive_attached : 0.000004s : 0.03% optimize.opt_a.after_resolve : 0.000023s : 0.16% optimize.opt_a.a_after_grad : 0.000018s : 0.13% optimize.opt_a.renormalize : 0.000835s : 5.76% optimize.opt_a.add_forward_monad_depend : 0.000007s : 0.05% optimize.opt_a.auto_monad_grad : 0.000003s : 0.02% optimize.opt_a.auto_monad_eliminator : 0.000023s : 0.16% optimize.opt_a.cse : 0.000041s : 0.29% optimize.opt_a.a_3 : 0.000080s : 0.55% optimize.py_interpret_to_execute_after_opt_a : 0.000011s : 0.07% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.02% optimize.rewriter_after_opt_a : 0.000033s : 0.23% optimize.convert_after_rewriter : 0.000007s : 0.05% optimize.order_py_execute_after_rewriter : 0.000005s : 0.04% optimize.mutable_eliminate : 0.000566s : 3.90% optimize.opt_b.b_1 : 0.000116s : 0.80% optimize.opt_b.b_2 : 0.000008s : 0.05% optimize.opt_b.updatestate_depend_eliminate : 0.000006s : 0.04% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.02% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.02% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000016s : 0.11% optimize.optimize_parallel_all_gather_comm : 0.000018s : 0.12% optimize.overlap_param_gather : 0.000002s : 0.01% optimize.cconv : 0.000026s : 0.18% optimize.loop_unroll : 0.000435s : 3.00% optimize.opt_after_cconv.c_1 : 0.000029s : 0.20% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.02% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.04% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.02% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.02% optimize.opt_after_cconv.cse : 0.000017s : 0.11% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000014s : 0.10% optimize.tuple_transform.d_1 : 0.000042s : 0.29% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.01% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.05% optimize.partial_unused_args_eliminate : 0.000002s : 0.01% optimize.add_recomputation : 0.000077s : 0.53% optimize.cse_after_recomputation.cse : 0.000013s : 0.09% optimize.environ_conv : 0.000005s : 0.03% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.04% optimize.bias_add_comm_swap : 0.000003s : 0.02% optimize.label_micro_interleaved_index : 0.000005s : 0.03% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.02% optimize.merge_cast_opt : 0.000001s : 0.01% optimize.slice_recompute_activation : 0.000002s : 0.02% optimize.micro_interleaved_order_control : 0.000003s : 0.02% optimize.assign_add_opt : 0.000001s : 0.01% optimize.ForceFp32Comm : 0.000001s : 0.01% optimize.remove_cast_before_assign_add : 0.000001s : 0.01% optimize.full_micro_interleaved_order_control : 0.000002s : 0.02% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.02% optimize.comm_op_add_attrs : 0.000001s : 0.01% optimize.add_comm_op_reuse_tag : 0.000001s : 0.01% optimize.interleave_split_concat_branches : 0.000001s : 0.01% optimize.interleave_parallel_branches : 0.000001s : 0.01% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.01% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.01% optimize.control_data_broadcast_order : 0.000013s : 0.09% optimize.grouped_pairwise_exchange_alltoall : 0.000019s : 0.13% optimize.offloading_packed_experts : 0.000005s : 0.03% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.04% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.01% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.01% optimize.overlap_recompute_comm : 0.000002s : 0.01% optimize.overlap_grad_ring_attention : 0.000004s : 0.03% optimize.overlap_grad_flash_sp : 0.000019s : 0.13% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.02% optimize.split_layernorm_comm : 0.000002s : 0.01% optimize.handle_group_info : 0.000001s : 0.01% optimize.symbol_engine_optimizer.build : 0.000004s : 0.03% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000011s : 0.08% optimize.symbol_engine_optimizer.elim_not_effective : 0.000015s : 0.10% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.05% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.07% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.01% pipeline_parallel_scheduler : 0.000001s : 0.01% auto_monad_reorder : 0.000017s : 0.12% get_jit_bprop_graph : 0.000002s : 0.01% rewriter_after_jit_bprop_graph : 0.000152s : 1.05% opt_after_jit_grad : 0.000486s : 3.35% validate : 0.000038s : 0.26% Time group info: ------[substitution.] 0.000287 40 12.14% : 0.000035s : 8: substitution.arithmetic_simplify 0.64% : 0.000002s : 2: substitution.elim_not_effective 0.48% : 0.000001s : 2: substitution.fold_const_symbol 2.09% : 0.000006s : 4: substitution.graph_param_transform 72.46% : 0.000208s : 6: substitution.inline 1.29% : 0.000004s : 4: substitution.j_node_and_user_rematch 1.63% : 0.000005s : 4: substitution.remove_not_recompute_node 1.81% : 0.000005s : 4: substitution.replace_old_param 7.45% : 0.000021s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.008627 2 88.05% : 0.007595s : 1: type_inference.infer 11.95% : 0.001031s : 1: type_inference.specialize ------[replace.] 0.000092 12 60.74% : 0.000056s : 6: replace.inline 39.26% : 0.000036s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000223 12 91.64% : 0.000204s : 6: match.inline 8.36% : 0.000019s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000230 1534 1.07% : 0.000002s : 17: predicate.accumulaten_eliminater 0.63% : 0.000001s : 4: predicate.ad_related_special_op_eliminate 0.47% : 0.000001s : 8: predicate.addn_check_dump 1.03% : 0.000002s : 17: predicate.addn_zero_filter 0.93% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.33% : 0.000005s : 25: predicate.arithmetic_simplify 1.07% : 0.000002s : 17: predicate.cast_eliminate 0.56% : 0.000001s : 8: predicate.check_bprop_eliminate 0.46% : 0.000001s : 8: predicate.compare_switch_simplify 0.16% : 0.000000s : 4: predicate.const_output_eliminate 0.49% : 0.000001s : 8: predicate.depend_value_elim 1.02% : 0.000002s : 17: predicate.dict_get_item_const_eliminator 1.13% : 0.000003s : 17: predicate.dict_get_item_eliminator 0.99% : 0.000002s : 17: predicate.dict_set_item_eliminator 0.91% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.22% : 0.000001s : 4: predicate.elim_not_effective 0.35% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.20% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.18% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.18% : 0.000003s : 21: predicate.environ_get_depend_swap 1.63% : 0.000004s : 29: predicate.environ_get_eliminate 1.12% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.75% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.56% : 0.000006s : 29: predicate.float_depend_g_call 0.50% : 0.000001s : 8: predicate.float_environ_get_switch 0.60% : 0.000001s : 12: predicate.float_tuple_getitem_switch 0.17% : 0.000000s : 4: predicate.fold_const_symbol 0.54% : 0.000001s : 8: predicate.get_grad_eliminate 0.18% : 0.000000s : 4: predicate.graph_param_transform 0.47% : 0.000001s : 8: predicate.incorporate_call 0.41% : 0.000001s : 8: predicate.incorporate_call_switch 6.35% : 0.000015s : 70: predicate.inline 0.73% : 0.000002s : 8: predicate.inline_without_move 0.28% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.84% : 0.000002s : 8: predicate.less_batch_normalization 1.89% : 0.000004s : 31: predicate.list_to_tuple_eliminator_ 2.70% : 0.000006s : 48: predicate.load_eliminater 0.80% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.71% : 0.000006s : 47: predicate.loop_unroll_before_grad 1.62% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.51% : 0.000001s : 8: predicate.merge_addn 0.47% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.49% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.89% : 0.000002s : 17: predicate.minmaximum_grad 0.95% : 0.000002s : 4: predicate.mutable_eliminate 0.28% : 0.000001s : 4: predicate.opt_reshape 0.33% : 0.000001s : 4: predicate.parallel_virtual_node 2.09% : 0.000005s : 29: predicate.partial_defer_inline 1.74% : 0.000004s : 27: predicate.partial_eliminate 0.98% : 0.000002s : 17: predicate.print_const_string_wrapper 0.61% : 0.000001s : 8: predicate.reduce_all_const_elim 1.27% : 0.000003s : 17: predicate.reduce_eliminate 2.60% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.35% : 0.000001s : 8: predicate.remove_not_recompute_node 1.47% : 0.000003s : 31: predicate.replace_applicator 0.49% : 0.000001s : 8: predicate.replace_old_param 0.26% : 0.000001s : 4: predicate.reset_defer_inline 1.00% : 0.000002s : 17: predicate.reshape_eliminate 0.54% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.45% : 0.000001s : 4: predicate.row_tensor_eliminate 0.66% : 0.000002s : 8: predicate.same_eliminate 0.37% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.57% : 0.000001s : 8: predicate.shard_identity_eliminate 0.54% : 0.000001s : 8: predicate.special_op_eliminate 0.57% : 0.000001s : 8: predicate.specialize_transform 0.74% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.63% : 0.000001s : 8: predicate.stack_unstack_eliminate 0.30% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.81% : 0.000004s : 29: predicate.switch_defer_inline 2.23% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.53% : 0.000013s : 88: predicate.switch_simplify 0.93% : 0.000002s : 17: predicate.tile_eliminate 1.05% : 0.000002s : 17: predicate.transpose_eliminate 1.61% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.49% : 0.000003s : 25: predicate.tuple_list_get_item_const_eliminator 1.44% : 0.000003s : 25: predicate.tuple_list_get_item_depend_reorder 3.43% : 0.000008s : 39: predicate.tuple_list_get_item_eliminator 1.53% : 0.000004s : 25: predicate.tuple_list_get_set_item_eliminator 2.04% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 1.72% : 0.000004s : 31: predicate.tuple_to_list_eliminator_ 2.58% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.13% : 0.000007s : 56: predicate.updatestate_useless_node_eliminater 0.44% : 0.000001s : 4: predicate.value_based_eliminate 0.58% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.50% : 0.000001s : 8: predicate.virtual_output_eliminate 0.23% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.31% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000765 15 54.24% : 0.000415s : 7: func_graph_cloner_run.FuncGraphClonerGraph 45.76% : 0.000350s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.030132 192 0.01% : 0.000004s : 1: ForceFp32Comm 11.45% : 0.003451s : 1: add_attr 11.41% : 0.003439s : 1: add_attr_with_inline 0.01% : 0.000004s : 1: add_comm_op_reuse_tag 0.27% : 0.000082s : 1: add_recomputation 0.01% : 0.000004s : 1: assign_add_opt 0.26% : 0.000079s : 1: auto_monad 0.07% : 0.000020s : 1: auto_monad_reorder 0.01% : 0.000004s : 1: begin_end_overlap_inline 0.02% : 0.000006s : 1: bias_add_comm_swap 1.53% : 0.000461s : 1: bootstrap 0.10% : 0.000029s : 1: cconv 0.02% : 0.000005s : 1: comm_op_add_attrs 0.05% : 0.000016s : 1: control_data_broadcast_order 0.03% : 0.000011s : 1: convert_after_rewriter 0.09% : 0.000026s : 1: cse_after_recomputation 0.02% : 0.000005s : 1: dataset_repeat_opt 0.02% : 0.000006s : 1: detach_backward 0.03% : 0.000008s : 1: environ_conv 0.18% : 0.000055s : 1: event_method 0.02% : 0.000005s : 1: full_micro_interleaved_order_control 0.02% : 0.000005s : 1: get_jit_bprop_graph 0.04% : 0.000012s : 1: graph_reusing 0.07% : 0.000023s : 1: grouped_pairwise_exchange_alltoall 0.01% : 0.000004s : 1: handle_group_info 0.02% : 0.000006s : 1: inline 0.02% : 0.000006s : 1: insert-virtual-dataset 0.01% : 0.000004s : 1: interleave_parallel_branches 0.01% : 0.000004s : 1: interleave_split_concat_branches 0.02% : 0.000006s : 1: label_fine_grained_interleaved_index 0.02% : 0.000007s : 1: label_micro_interleaved_index 1.47% : 0.000443s : 1: loop_unroll 0.06% : 0.000018s : 1: merge_cast_opt 0.02% : 0.000005s : 1: micro_interleaved_order_control 1.91% : 0.000574s : 1: mutable_eliminate 0.03% : 0.000008s : 1: offloading_packed_experts 0.04% : 0.000013s : 1: opt.transform.loop_unroll_optimizer 0.05% : 0.000016s : 1: opt.transform.mutable_eliminate 4.92% : 0.001481s : 78: opt.transform.opt_a 0.09% : 0.000028s : 1: opt.transform.opt_after_cconv 0.08% : 0.000024s : 1: opt.transform.opt_after_jit_grad 0.32% : 0.000096s : 28: opt.transform.opt_b 0.15% : 0.000046s : 2: opt.transform.opt_trans_graph 0.13% : 0.000039s : 4: opt.transform.symbol_engine_opt 10.47% : 0.003155s : 1: opt_a 0.34% : 0.000103s : 1: opt_after_cconv 1.65% : 0.000496s : 1: opt_after_jit_grad 0.65% : 0.000196s : 1: opt_b 17.65% : 0.005320s : 1: optimize 0.07% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.03% : 0.000008s : 1: order_py_execute_after_rewriter 0.08% : 0.000023s : 1: overlap_grad_flash_sp 0.01% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.02% : 0.000007s : 1: overlap_grad_ring_attention 0.02% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.01% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.02% : 0.000005s : 1: overlap_param_gather 0.01% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.03% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.02% : 0.000005s : 1: overlap_recompute_comm 0.02% : 0.000007s : 1: parallel-infer-symbol 0.01% : 0.000004s : 1: parallel-infer-symbol-second 0.02% : 0.000005s : 1: partial_unused_args_eliminate 0.02% : 0.000005s : 1: pipeline_parallel_scheduler 0.02% : 0.000005s : 1: pipeline_split 0.16% : 0.000049s : 1: pre_auto_parallel 0.12% : 0.000036s : 1: py_interpret_to_execute 0.05% : 0.000014s : 1: py_interpret_to_execute_after_opt_a 0.01% : 0.000004s : 1: remove_cast_before_assign_add 0.06% : 0.000018s : 1: remove_dup_value 1.42% : 0.000429s : 1: renormalize.infer 1.32% : 0.000397s : 1: renormalize.specialize 0.02% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.52% : 0.000157s : 1: rewriter_after_jit_bprop_graph 0.12% : 0.000037s : 1: rewriter_after_opt_a 0.36% : 0.000110s : 1: rewriter_before_opt_a 0.02% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.02% : 0.000005s : 1: slice_recompute_activation 0.01% : 0.000004s : 1: split_layernorm_comm 0.02% : 0.000005s : 1: split_matmul_comm_elemetwise 0.03% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.27% : 0.000083s : 1: symbol_engine_optimizer 0.25% : 0.000076s : 1: tuple_transform 28.91% : 0.008710s : 1: type_inference TotalTime = 0.107071, [21] [bootstrap]: 0.00038119 [type_inference]: 0.0655117 [event_method]: 8.438e-05 [auto_monad]: 0.0001254 [graph_reusing]: 9.52999e-06 [inline]: 2.39999e-06 [add_attr]: 0.00326253, [1] [add_attr_with_inline]: 0.00325353, [1] [Cycle 1]: 9.215e-05, [2] [tag_attr]: 4.456e-05 [meta_addattr_fg_expand]: 1.293e-05 [parallel-infer-symbol]: 3.28e-06 [pre_auto_parallel]: 6.293e-05 [insert-virtual-dataset]: 2.44999e-06 [parallel-infer-symbol-second]: 9.20001e-07 [dataset_repeat_opt]: 2.23998e-06 [pipeline_split]: 2.07999e-06 [optimize]: 0.0369158, [53] [py_interpret_to_execute]: 4.376e-05 [rewriter_before_opt_a]: 0.0001797 [opt_a]: 0.0344399, [3] [Cycle 1]: 0.0305924, [45] [expand_dump_flag]: 4.28001e-06 [switch_simplify]: 9.197e-05 [loop_unroll]: 7.641e-05 [a_1]: 0.00173992 [with_stream_mark]: 3.163e-05 [recompute_prepare]: 2.582e-05 [updatestate_depend_eliminate]: 1.038e-05 [updatestate_assign_eliminate]: 8.19002e-06 [updatestate_loads_eliminate]: 7.89002e-06 [parameter_eliminate]: 2.84999e-06 [a_2]: 0.00024964 [accelerated_algorithm]: 3.42e-05 [shard]: 2.14999e-06 [meta_shard_fg_expand]: 3.98001e-06 [shard_inline]: 1.669e-05 [merge_send_recv]: 1.839e-05 [auto_parallel]: 1.363e-05 [parallel]: 1.889e-05 [flash_sp]: 1.13e-05 [merge_comm]: 1.073e-05 [allreduce_fusion]: 9.66003e-06 [matmul_add_comm_reduction]: 3.224e-05 [allreduce_slice_to_reducescatter]: 6.40022e-07 [virtual_shard_identity]: 1.915e-05 [virtual_dataset]: 1.553e-05 [get_grad_eliminate_]: 1.526e-05 [virtual_output]: 1.512e-05 [merge_forward]: 1.022e-05 [cell_reuse_recompute_pass]: 1.23002e-06 [offload_activation]: 1.933e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.098e-05 [merge_recompute_call_nodes]: 1.76e-06 [before_grad]: 2.911e-05 [set_forward_comm_id_for_comm_node_pass]: 1.05e-05 [meta_fg_expand]: 0.0016618 [flash_sp_send_recv_attached]: 6.58e-06 [receive_attached]: 2.44999e-06 [after_resolve]: 6.873e-05 [a_after_grad]: 8.868e-05 [renormalize]: 0.0251545 [add_forward_monad_depend]: 1.687e-05 [auto_monad_grad]: 7.18998e-06 [auto_monad_eliminator]: 6.715e-05 [cse]: 0.00022867 [a_3]: 0.00035506 [Cycle 2]: 0.0032181, [45] [expand_dump_flag]: 3.31001e-06 [switch_simplify]: 4.767e-05 [loop_unroll]: 4.378e-05 [a_1]: 0.00165883 [with_stream_mark]: 1.967e-05 [recompute_prepare]: 1.029e-05 [updatestate_depend_eliminate]: 4.46002e-06 [updatestate_assign_eliminate]: 3.41001e-06 [updatestate_loads_eliminate]: 2.74001e-06 [parameter_eliminate]: 2.27001e-06 [a_2]: 7.062e-05 [accelerated_algorithm]: 1.177e-05 [shard]: 2.06e-06 [meta_shard_fg_expand]: 2.69999e-06 [shard_inline]: 7.66999e-06 [merge_send_recv]: 9.44e-06 [auto_parallel]: 1.074e-05 [parallel]: 1.013e-05 [flash_sp]: 4.05998e-06 [merge_comm]: 3.31999e-06 [allreduce_fusion]: 3.67002e-06 [matmul_add_comm_reduction]: 9.05999e-06 [allreduce_slice_to_reducescatter]: 6.09987e-07 [virtual_shard_identity]: 7.82e-06 [virtual_dataset]: 6.03998e-06 [get_grad_eliminate_]: 5.62999e-06 [virtual_output]: 6.02001e-06 [merge_forward]: 4.17e-06 [cell_reuse_recompute_pass]: 1.76998e-06 [offload_activation]: 1.034e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.192e-05 [merge_recompute_call_nodes]: 1.37e-06 [before_grad]: 9.91998e-06 [set_forward_comm_id_for_comm_node_pass]: 3.34001e-06 [meta_fg_expand]: 0.00015987 [flash_sp_send_recv_attached]: 1.91e-06 [receive_attached]: 2.49001e-06 [after_resolve]: 1.461e-05 [a_after_grad]: 1.025e-05 [renormalize]: 0.0006764 [add_forward_monad_depend]: 4.95001e-06 [auto_monad_grad]: 2.12999e-06 [auto_monad_eliminator]: 1.185e-05 [cse]: 2.133e-05 [a_3]: 4.342e-05 [Cycle 3]: 0.0006113, [45] [expand_dump_flag]: 2.01e-06 [switch_simplify]: 7.36999e-06 [loop_unroll]: 6.19001e-06 [a_1]: 0.00011992 [with_stream_mark]: 9.47999e-06 [recompute_prepare]: 5.97999e-06 [updatestate_depend_eliminate]: 3.33e-06 [updatestate_assign_eliminate]: 2.73e-06 [updatestate_loads_eliminate]: 2.31e-06 [parameter_eliminate]: 1.23002e-06 [a_2]: 6.582e-05 [accelerated_algorithm]: 1.154e-05 [shard]: 1.11002e-06 [meta_shard_fg_expand]: 1.54e-06 [shard_inline]: 5.96998e-06 [merge_send_recv]: 4.89e-06 [auto_parallel]: 5.79e-06 [parallel]: 5.30999e-06 [flash_sp]: 1.00001e-06 [merge_comm]: 3.21001e-06 [allreduce_fusion]: 3.04999e-06 [matmul_add_comm_reduction]: 6.09001e-06 [allreduce_slice_to_reducescatter]: 3.69997e-07 [virtual_shard_identity]: 6.64001e-06 [virtual_dataset]: 5.42999e-06 [get_grad_eliminate_]: 5.18002e-06 [virtual_output]: 5.34e-06 [merge_forward]: 3.21001e-06 [cell_reuse_recompute_pass]: 1.42999e-06 [offload_activation]: 7.17002e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.074e-05 [merge_recompute_call_nodes]: 6.60017e-07 [before_grad]: 8.42e-06 [set_forward_comm_id_for_comm_node_pass]: 3.5e-06 [meta_fg_expand]: 2.12999e-06 [flash_sp_send_recv_attached]: 1.15999e-06 [receive_attached]: 1.37999e-06 [after_resolve]: 1.038e-05 [a_after_grad]: 9.56e-06 [renormalize]: 9.00181e-08 [add_forward_monad_depend]: 1.10999e-06 [auto_monad_grad]: 1.34e-06 [auto_monad_eliminator]: 7.15e-06 [cse]: 1.246e-05 [a_3]: 3.388e-05 [py_interpret_to_execute_after_opt_a]: 1.169e-05 [slice_cell_reuse_recomputed_activation]: 1.87001e-06 [rewriter_after_opt_a]: 3.871e-05 [convert_after_rewriter]: 6.89001e-06 [order_py_execute_after_rewriter]: 4.68001e-06 [mutable_eliminate]: 0.00073722 [opt_b]: 0.00019584, [1] [Cycle 1]: 0.00018841, [7] [b_1]: 0.00011664 [b_2]: 7.98001e-06 [updatestate_depend_eliminate]: 6.28e-06 [updatestate_assign_eliminate]: 2.69999e-06 [updatestate_loads_eliminate]: 2.19001e-06 [renormalize]: 3.00002e-07 [cse]: 1.718e-05 [optimize_parallel_all_gather_comm]: 1.741e-05 [overlap_param_gather]: 1.81e-06 [cconv]: 2.643e-05 [loop_unroll]: 0.0005231 [opt_after_cconv]: 0.0001028, [1] [Cycle 1]: 9.676e-05, [7] [c_1]: 3.046e-05 [parameter_eliminate]: 3.27997e-06 [updatestate_depend_eliminate]: 5.85002e-06 [updatestate_assign_eliminate]: 2.49001e-06 [updatestate_loads_eliminate]: 2.34999e-06 [cse]: 1.661e-05 [renormalize]: 3.19997e-07 [remove_dup_value]: 1.299e-05 [tuple_transform]: 8.224e-05, [1] [Cycle 1]: 7.767e-05, [4] [d_1]: 4.182e-05 [none_parameter_eliminate]: 1.41998e-06 [renormalize]: 2.00002e-07 [switch_simplify]: 7.45e-06 [partial_unused_args_eliminate]: 2.04e-06 [add_recomputation]: 4.847e-05 [cse_after_recomputation]: 2.255e-05, [1] [Cycle 1]: 1.772e-05, [1] [cse]: 1.189e-05 [environ_conv]: 7.58999e-06 [swap_dp_allreduce_reducescatter]: 5.71e-06 [bias_add_comm_swap]: 2.34999e-06 [label_micro_interleaved_index]: 4.23999e-06 [label_fine_grained_interleaved_index]: 2.69999e-06 [merge_cast_opt]: 1.32e-06 [slice_recompute_activation]: 1.87999e-06 [micro_interleaved_order_control]: 2.48e-06 [assign_add_opt]: 1.39e-06 [ForceFp32Comm]: 9.00007e-07 [remove_cast_before_assign_add]: 1.00001e-06 [full_micro_interleaved_order_control]: 2.06e-06 [reorder_send_recv_between_fp_bp]: 2.71999e-06 [comm_op_add_attrs]: 1.39e-06 [add_comm_op_reuse_tag]: 1.08001e-06 [interleave_split_concat_branches]: 1.51002e-06 [interleave_parallel_branches]: 1.06002e-06 [overlap_opt_shard_in_pipeline]: 1.35999e-06 [overlap_opt_shard_grad_in_pipeline]: 1.75001e-06 [control_data_broadcast_order]: 1.229e-05 [grouped_pairwise_exchange_alltoall]: 1.71e-06 [offloading_packed_experts]: 3.97e-06 [overlap_recompute_and_grad_model_parallel]: 4.84e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.28002e-06 [overlap_recompute_allgather_and_fa_grad]: 1.37999e-06 [overlap_recompute_comm]: 2.37001e-06 [overlap_grad_ring_attention]: 4.27e-06 [overlap_grad_flash_sp]: 1.955e-05 [begin_end_overlap_inline]: 5.3001e-07 [split_matmul_comm_elemetwise]: 2.42001e-06 [split_layernorm_comm]: 1.75001e-06 [handle_group_info]: 1.07e-06 [symbol_engine_optimizer]: 8.421e-05, [1] [Cycle 1]: 7.928e-05, [6] [build]: 1.077e-05 [elim_shapecalc]: 1.04e-05 [elim_not_effective]: 1.304e-05 [opt_reshape]: 7.15e-06 [fold_const_symbol]: 9.31e-06 [renormalize]: 1.80007e-07 [detach_backward]: 2.27999e-06 [pipeline_parallel_scheduler]: 1.71e-06 [auto_monad_reorder]: 1.789e-05 [get_jit_bprop_graph]: 2.41e-06 [rewriter_after_jit_bprop_graph]: 4.01001e-06 [opt_after_jit_grad]: 0.00046536 [validate]: 4.256e-05 Sums bootstrap : 0.000381s : 0.37% type_inference : 0.065512s : 63.92% event_method : 0.000084s : 0.08% auto_monad : 0.000125s : 0.12% graph_reusing : 0.000010s : 0.01% inline : 0.000002s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000045s : 0.04% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000013s : 0.01% parallel-infer-symbol : 0.000003s : 0.00% pre_auto_parallel : 0.000063s : 0.06% insert-virtual-dataset : 0.000002s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000044s : 0.04% optimize.rewriter_before_opt_a : 0.000180s : 0.18% optimize.opt_a.expand_dump_flag : 0.000010s : 0.01% optimize.opt_a.switch_simplify : 0.000147s : 0.14% optimize.opt_a.loop_unroll : 0.000126s : 0.12% optimize.opt_a.a_1 : 0.003519s : 3.43% optimize.opt_a.with_stream_mark : 0.000061s : 0.06% optimize.opt_a.recompute_prepare : 0.000042s : 0.04% optimize.opt_a.updatestate_depend_eliminate : 0.000018s : 0.02% optimize.opt_a.updatestate_assign_eliminate : 0.000014s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000013s : 0.01% optimize.opt_a.parameter_eliminate : 0.000006s : 0.01% optimize.opt_a.a_2 : 0.000386s : 0.38% optimize.opt_a.accelerated_algorithm : 0.000058s : 0.06% optimize.opt_a.shard : 0.000005s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000008s : 0.01% optimize.opt_a.shard_inline : 0.000030s : 0.03% optimize.opt_a.merge_send_recv : 0.000033s : 0.03% optimize.opt_a.auto_parallel : 0.000030s : 0.03% optimize.opt_a.parallel : 0.000034s : 0.03% optimize.opt_a.flash_sp : 0.000016s : 0.02% optimize.opt_a.merge_comm : 0.000017s : 0.02% optimize.opt_a.allreduce_fusion : 0.000016s : 0.02% optimize.opt_a.matmul_add_comm_reduction : 0.000047s : 0.05% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000034s : 0.03% optimize.opt_a.virtual_dataset : 0.000027s : 0.03% optimize.opt_a.get_grad_eliminate_ : 0.000026s : 0.03% optimize.opt_a.virtual_output : 0.000026s : 0.03% optimize.opt_a.merge_forward : 0.000018s : 0.02% optimize.opt_a.cell_reuse_recompute_pass : 0.000004s : 0.00% optimize.opt_a.offload_activation : 0.000037s : 0.04% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000054s : 0.05% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.00% optimize.opt_a.before_grad : 0.000047s : 0.05% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000017s : 0.02% optimize.opt_a.meta_fg_expand : 0.001824s : 1.78% optimize.opt_a.flash_sp_send_recv_attached : 0.000010s : 0.01% optimize.opt_a.receive_attached : 0.000006s : 0.01% optimize.opt_a.after_resolve : 0.000094s : 0.09% optimize.opt_a.a_after_grad : 0.000108s : 0.11% optimize.opt_a.renormalize : 0.025831s : 25.20% optimize.opt_a.add_forward_monad_depend : 0.000023s : 0.02% optimize.opt_a.auto_monad_grad : 0.000011s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000086s : 0.08% optimize.opt_a.cse : 0.000262s : 0.26% optimize.opt_a.a_3 : 0.000432s : 0.42% optimize.py_interpret_to_execute_after_opt_a : 0.000012s : 0.01% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000039s : 0.04% optimize.convert_after_rewriter : 0.000007s : 0.01% optimize.order_py_execute_after_rewriter : 0.000005s : 0.00% optimize.mutable_eliminate : 0.000737s : 0.72% optimize.opt_b.b_1 : 0.000117s : 0.11% optimize.opt_b.b_2 : 0.000008s : 0.01% optimize.opt_b.updatestate_depend_eliminate : 0.000006s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000017s : 0.02% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.02% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000026s : 0.03% optimize.loop_unroll : 0.000523s : 0.51% optimize.opt_after_cconv.c_1 : 0.000030s : 0.03% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.cse : 0.000017s : 0.02% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000013s : 0.01% optimize.tuple_transform.d_1 : 0.000042s : 0.04% optimize.tuple_transform.none_parameter_eliminate : 0.000001s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000048s : 0.05% optimize.cse_after_recomputation.cse : 0.000012s : 0.01% optimize.environ_conv : 0.000008s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.01% optimize.bias_add_comm_swap : 0.000002s : 0.00% optimize.label_micro_interleaved_index : 0.000004s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000002s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000012s : 0.01% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000020s : 0.02% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000011s : 0.01% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000010s : 0.01% optimize.symbol_engine_optimizer.elim_not_effective : 0.000013s : 0.01% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.01% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000009s : 0.01% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000018s : 0.02% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.00% opt_after_jit_grad : 0.000465s : 0.45% validate : 0.000043s : 0.04% Time group info: ------[substitution.] 0.000961 195 8.88% : 0.000085s : 7: substitution.arithmetic_simplify 0.21% : 0.000002s : 2: substitution.elim_not_effective 0.55% : 0.000005s : 6: substitution.float_depend_g_call 0.49% : 0.000005s : 3: substitution.float_tuple_getitem_switch 0.14% : 0.000001s : 2: substitution.fold_const_symbol 0.57% : 0.000005s : 4: substitution.graph_param_transform 0.36% : 0.000003s : 2: substitution.incorporate_call 0.23% : 0.000002s : 2: substitution.incorporate_call_switch 57.33% : 0.000551s : 18: substitution.inline 2.08% : 0.000020s : 2: substitution.inline_without_move 0.98% : 0.000009s : 14: substitution.j_node_and_user_rematch 1.94% : 0.000019s : 3: substitution.less_batch_normalization 1.63% : 0.000016s : 11: substitution.minmaximum_grad 0.73% : 0.000007s : 6: substitution.partial_eliminate 1.18% : 0.000011s : 14: substitution.remove_not_recompute_node 3.30% : 0.000032s : 10: substitution.replace_applicator 1.10% : 0.000011s : 12: substitution.replace_old_param 0.31% : 0.000003s : 1: substitution.set_cell_output_no_recompute 3.33% : 0.000032s : 11: substitution.tuple_list_convert_item_index_to_positive 1.56% : 0.000015s : 11: substitution.tuple_list_get_item_const_eliminator 2.07% : 0.000020s : 11: substitution.tuple_list_get_item_depend_reorder 8.80% : 0.000085s : 32: substitution.tuple_list_get_item_eliminator 2.23% : 0.000021s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.065410 2 96.75% : 0.063284s : 1: type_inference.infer 3.25% : 0.002127s : 1: type_inference.specialize ------[replace.] 0.000294 38 3.96% : 0.000012s : 2: replace.arithmetic_simplify 55.20% : 0.000162s : 18: replace.inline 40.83% : 0.000120s : 18: replace.tuple_list_get_item_eliminator ------[match.] 0.000633 38 7.67% : 0.000049s : 2: match.arithmetic_simplify 85.37% : 0.000540s : 18: match.inline 6.96% : 0.000044s : 18: match.tuple_list_get_item_eliminator ------[predicate.] 0.000715 4976 1.07% : 0.000008s : 61: predicate.accumulaten_eliminater 0.22% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.44% : 0.000003s : 23: predicate.addn_check_dump 1.13% : 0.000008s : 61: predicate.addn_zero_filter 1.03% : 0.000007s : 61: predicate.adjust_all_reduce_mul_add 2.21% : 0.000016s : 86: predicate.arithmetic_simplify 1.14% : 0.000008s : 63: predicate.cast_eliminate 1.04% : 0.000007s : 55: predicate.check_bprop_eliminate 0.43% : 0.000003s : 23: predicate.compare_switch_simplify 0.05% : 0.000000s : 4: predicate.const_output_eliminate 0.41% : 0.000003s : 23: predicate.depend_value_elim 1.23% : 0.000009s : 63: predicate.dict_get_item_const_eliminator 1.32% : 0.000009s : 63: predicate.dict_get_item_eliminator 1.12% : 0.000008s : 63: predicate.dict_set_item_eliminator 0.27% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.07% : 0.000000s : 4: predicate.elim_not_effective 0.11% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.21% : 0.000009s : 67: predicate.environ_add_const_eliminate 1.14% : 0.000008s : 67: predicate.environ_get_add_eliminate 1.16% : 0.000008s : 67: predicate.environ_get_depend_swap 1.61% : 0.000012s : 90: predicate.environ_get_eliminate 1.20% : 0.000009s : 67: predicate.environ_get_set_eliminate 1.84% : 0.000013s : 99: predicate.exchange_switch_depend_value 2.69% : 0.000019s : 99: predicate.float_depend_g_call 0.40% : 0.000003s : 23: predicate.float_environ_get_switch 0.50% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.05% : 0.000000s : 4: predicate.fold_const_symbol 0.45% : 0.000003s : 23: predicate.get_grad_eliminate 0.06% : 0.000000s : 4: predicate.graph_param_transform 0.43% : 0.000003s : 23: predicate.incorporate_call 0.39% : 0.000003s : 23: predicate.incorporate_call_switch 5.71% : 0.000041s : 214: predicate.inline 1.30% : 0.000009s : 46: predicate.inline_without_move 0.23% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.60% : 0.000004s : 23: predicate.less_batch_normalization 1.65% : 0.000012s : 89: predicate.list_to_tuple_eliminator_ 2.64% : 0.000019s : 150: predicate.load_eliminater 0.27% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.60% : 0.000019s : 141: predicate.loop_unroll_before_grad 1.36% : 0.000010s : 71: predicate.make_slice_get_slice_eliminator 0.43% : 0.000003s : 23: predicate.merge_addn 0.99% : 0.000007s : 55: predicate.micro_step_allgather_replace 1.01% : 0.000007s : 55: predicate.mini_step_allgather_replace 1.13% : 0.000008s : 63: predicate.minmaximum_grad 0.36% : 0.000003s : 4: predicate.mutable_eliminate 0.11% : 0.000001s : 4: predicate.opt_reshape 0.15% : 0.000001s : 4: predicate.parallel_virtual_node 2.47% : 0.000018s : 99: predicate.partial_defer_inline 1.80% : 0.000013s : 85: predicate.partial_eliminate 1.08% : 0.000008s : 61: predicate.print_const_string_wrapper 0.44% : 0.000003s : 23: predicate.reduce_all_const_elim 1.46% : 0.000010s : 63: predicate.reduce_eliminate 2.76% : 0.000020s : 150: predicate.redundant_stop_gradient_eliminater 0.29% : 0.000002s : 23: predicate.remove_not_recompute_node 1.95% : 0.000014s : 136: predicate.replace_applicator 0.65% : 0.000005s : 46: predicate.replace_old_param 0.10% : 0.000001s : 4: predicate.reset_defer_inline 1.14% : 0.000008s : 63: predicate.reshape_eliminate 1.03% : 0.000007s : 55: predicate.row_tensor_add_zeros_like 0.11% : 0.000001s : 4: predicate.row_tensor_eliminate 1.22% : 0.000009s : 55: predicate.same_eliminate 0.32% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.53% : 0.000004s : 23: predicate.shard_identity_eliminate 0.17% : 0.000001s : 8: predicate.special_op_eliminate 0.53% : 0.000004s : 23: predicate.specialize_transform 1.31% : 0.000009s : 55: predicate.split_environ_get_set_with_tuple_value 1.08% : 0.000008s : 46: predicate.stack_unstack_eliminate 0.09% : 0.000001s : 4: predicate.switch_call_monad_eliminater 2.08% : 0.000015s : 99: predicate.switch_defer_inline 3.10% : 0.000022s : 154: predicate.switch_layer_defer_inline 5.33% : 0.000038s : 267: predicate.switch_simplify 1.16% : 0.000008s : 63: predicate.tile_eliminate 1.09% : 0.000008s : 63: predicate.transpose_eliminate 1.52% : 0.000011s : 71: predicate.tuple_list_convert_item_index_to_positive 1.65% : 0.000012s : 71: predicate.tuple_list_get_item_const_eliminator 1.43% : 0.000010s : 71: predicate.tuple_list_get_item_depend_reorder 3.07% : 0.000022s : 112: predicate.tuple_list_get_item_eliminator 1.44% : 0.000010s : 71: predicate.tuple_list_get_set_item_eliminator 2.03% : 0.000015s : 94: predicate.tuple_list_set_item_eliminator 1.71% : 0.000012s : 89: predicate.tuple_to_list_eliminator_ 2.56% : 0.000018s : 150: predicate.updatestate_pure_node_eliminater 3.10% : 0.000022s : 173: predicate.updatestate_useless_node_eliminater 0.11% : 0.000001s : 4: predicate.value_based_eliminate 0.49% : 0.000003s : 23: predicate.virtual_dataset_eliminate 0.47% : 0.000003s : 23: predicate.virtual_output_eliminate 0.07% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.11% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002256 41 55.24% : 0.001246s : 19: func_graph_cloner_run.FuncGraphClonerGraph 44.76% : 0.001010s : 22: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.178204 233 0.00% : 0.000004s : 1: ForceFp32Comm 1.83% : 0.003268s : 1: add_attr 1.83% : 0.003258s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.03% : 0.000053s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.07% : 0.000133s : 1: auto_monad 0.01% : 0.000022s : 1: auto_monad_reorder 0.00% : 0.000003s : 1: begin_end_overlap_inline 0.00% : 0.000005s : 1: bias_add_comm_swap 0.23% : 0.000404s : 1: bootstrap 0.02% : 0.000030s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.01% : 0.000016s : 1: control_data_broadcast_order 0.01% : 0.000010s : 1: convert_after_rewriter 0.01% : 0.000026s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.01% : 0.000011s : 1: environ_conv 0.05% : 0.000095s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000006s : 1: get_jit_bprop_graph 0.01% : 0.000014s : 1: graph_reusing 0.00% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000007s : 1: label_micro_interleaved_index 0.30% : 0.000531s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000006s : 1: micro_interleaved_order_control 0.42% : 0.000747s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.01% : 0.000014s : 1: opt.transform.loop_unroll_optimizer 0.01% : 0.000015s : 1: opt.transform.mutable_eliminate 2.82% : 0.005018s : 117: opt.transform.opt_a 0.02% : 0.000029s : 1: opt.transform.opt_after_cconv 0.01% : 0.000025s : 1: opt.transform.opt_after_jit_grad 0.05% : 0.000097s : 28: opt.transform.opt_b 0.03% : 0.000047s : 2: opt.transform.opt_trans_graph 0.02% : 0.000036s : 4: opt.transform.symbol_engine_opt 19.33% : 0.034443s : 1: opt_a 0.06% : 0.000106s : 1: opt_after_cconv 0.27% : 0.000475s : 1: opt_after_jit_grad 0.11% : 0.000200s : 1: opt_b 20.72% : 0.036921s : 1: optimize 0.01% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000008s : 1: order_py_execute_after_rewriter 0.01% : 0.000023s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000007s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.04% : 0.000068s : 1: pre_auto_parallel 0.03% : 0.000047s : 1: py_interpret_to_execute 0.01% : 0.000015s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.01% : 0.000017s : 1: remove_dup_value 13.31% : 0.023723s : 2: renormalize.infer 1.17% : 0.002082s : 2: renormalize.specialize 0.00% : 0.000005s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000007s : 1: rewriter_after_jit_bprop_graph 0.02% : 0.000043s : 1: rewriter_after_opt_a 0.10% : 0.000184s : 1: rewriter_before_opt_a 0.00% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.05% : 0.000087s : 1: symbol_engine_optimizer 0.05% : 0.000085s : 1: tuple_transform 36.78% : 0.065537s : 1: type_inference TotalTime = 0.0188994, [21] [bootstrap]: 0.00043879 [type_inference]: 0.0085871 [event_method]: 4.539e-05 [auto_monad]: 7.271e-05 [graph_reusing]: 7.2e-06 [inline]: 2.43e-06 [add_attr]: 0.00347074, [1] [add_attr_with_inline]: 0.00346161, [1] [Cycle 1]: 7.579e-05, [2] [tag_attr]: 3.031e-05 [meta_addattr_fg_expand]: 7.23e-06 [parallel-infer-symbol]: 3.56999e-06 [pre_auto_parallel]: 4.283e-05 [insert-virtual-dataset]: 2.48e-06 [parallel-infer-symbol-second]: 8.00006e-07 [dataset_repeat_opt]: 2.01998e-06 [pipeline_split]: 1.57999e-06 [optimize]: 0.00548036, [53] [py_interpret_to_execute]: 3.448e-05 [rewriter_before_opt_a]: 0.00010332 [opt_a]: 0.00314394, [2] [Cycle 1]: 0.00251696, [45] [expand_dump_flag]: 3.43999e-06 [switch_simplify]: 5.478e-05 [loop_unroll]: 3.877e-05 [a_1]: 0.00091286 [with_stream_mark]: 1.902e-05 [recompute_prepare]: 1.009e-05 [updatestate_depend_eliminate]: 4.15e-06 [updatestate_assign_eliminate]: 3.51999e-06 [updatestate_loads_eliminate]: 3.26999e-06 [parameter_eliminate]: 2.30002e-06 [a_2]: 8.034e-05 [accelerated_algorithm]: 7.06001e-06 [shard]: 2.48002e-06 [meta_shard_fg_expand]: 2.01e-06 [shard_inline]: 6.23e-06 [merge_send_recv]: 8.49998e-06 [auto_parallel]: 7.46999e-06 [parallel]: 1.871e-05 [flash_sp]: 8.42e-06 [merge_comm]: 3.7e-06 [allreduce_fusion]: 3.30998e-06 [matmul_add_comm_reduction]: 9.87999e-06 [allreduce_slice_to_reducescatter]: 6.89994e-07 [virtual_shard_identity]: 8.92999e-06 [virtual_dataset]: 7.19001e-06 [get_grad_eliminate_]: 5.83002e-06 [virtual_output]: 6.28998e-06 [merge_forward]: 4.35e-06 [cell_reuse_recompute_pass]: 1.64e-06 [offload_activation]: 1.06e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.156e-05 [merge_recompute_call_nodes]: 1.35999e-06 [before_grad]: 1.085e-05 [set_forward_comm_id_for_comm_node_pass]: 3.64002e-06 [meta_fg_expand]: 2.66e-06 [flash_sp_send_recv_attached]: 2.52001e-06 [receive_attached]: 2.36e-06 [after_resolve]: 1.3e-05 [a_after_grad]: 1.065e-05 [renormalize]: 0.00085377 [add_forward_monad_depend]: 5.34998e-06 [auto_monad_grad]: 2.16998e-06 [auto_monad_eliminator]: 1.579e-05 [cse]: 2.835e-05 [a_3]: 4.407e-05 [Cycle 2]: 0.00061677, [45] [expand_dump_flag]: 1.25001e-06 [switch_simplify]: 7.86001e-06 [loop_unroll]: 6.06e-06 [a_1]: 0.00013145 [with_stream_mark]: 1.174e-05 [recompute_prepare]: 5.97999e-06 [updatestate_depend_eliminate]: 3.09999e-06 [updatestate_assign_eliminate]: 2.58e-06 [updatestate_loads_eliminate]: 2.84999e-06 [parameter_eliminate]: 9.09989e-07 [a_2]: 6.952e-05 [accelerated_algorithm]: 5.83002e-06 [shard]: 1.32e-06 [meta_shard_fg_expand]: 1.54998e-06 [shard_inline]: 5.72001e-06 [merge_send_recv]: 4.51002e-06 [auto_parallel]: 5.30999e-06 [parallel]: 4.91002e-06 [flash_sp]: 3.16999e-06 [merge_comm]: 3.11001e-06 [allreduce_fusion]: 3.04999e-06 [matmul_add_comm_reduction]: 5.86003e-06 [allreduce_slice_to_reducescatter]: 5.50004e-07 [virtual_shard_identity]: 6.29001e-06 [virtual_dataset]: 5.39e-06 [get_grad_eliminate_]: 5.12e-06 [virtual_output]: 5.00999e-06 [merge_forward]: 2.88e-06 [cell_reuse_recompute_pass]: 1.29e-06 [offload_activation]: 6.94001e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.008e-05 [merge_recompute_call_nodes]: 7.2e-07 [before_grad]: 8.63001e-06 [set_forward_comm_id_for_comm_node_pass]: 3.67998e-06 [meta_fg_expand]: 1.91e-06 [flash_sp_send_recv_attached]: 9.5999e-07 [receive_attached]: 1.00999e-06 [after_resolve]: 9.61e-06 [a_after_grad]: 8.92999e-06 [renormalize]: 8.00064e-08 [add_forward_monad_depend]: 1.37e-06 [auto_monad_grad]: 9.70002e-07 [auto_monad_eliminator]: 6.62002e-06 [cse]: 1.294e-05 [a_3]: 3.285e-05 [py_interpret_to_execute_after_opt_a]: 1.021e-05 [slice_cell_reuse_recomputed_activation]: 2.31e-06 [rewriter_after_opt_a]: 3.439e-05 [convert_after_rewriter]: 7.8e-06 [order_py_execute_after_rewriter]: 4.95999e-06 [mutable_eliminate]: 0.00051809 [opt_b]: 0.00019433, [1] [Cycle 1]: 0.00018769, [7] [b_1]: 0.00011529 [b_2]: 7.85e-06 [updatestate_depend_eliminate]: 5.97001e-06 [updatestate_assign_eliminate]: 2.47001e-06 [updatestate_loads_eliminate]: 2.79001e-06 [renormalize]: 5.89993e-07 [cse]: 1.822e-05 [optimize_parallel_all_gather_comm]: 1.705e-05 [overlap_param_gather]: 1.89e-06 [cconv]: 2.609e-05 [loop_unroll]: 0.00045856 [opt_after_cconv]: 0.00010183, [1] [Cycle 1]: 9.554e-05, [7] [c_1]: 2.921e-05 [parameter_eliminate]: 3.24001e-06 [updatestate_depend_eliminate]: 5.32999e-06 [updatestate_assign_eliminate]: 2.51998e-06 [updatestate_loads_eliminate]: 2.19001e-06 [cse]: 1.743e-05 [renormalize]: 2.30008e-07 [remove_dup_value]: 1.416e-05 [tuple_transform]: 7.533e-05, [1] [Cycle 1]: 7.096e-05, [4] [d_1]: 4.377e-05 [none_parameter_eliminate]: 1.79998e-06 [renormalize]: 1.8999e-07 [switch_simplify]: 6.62002e-06 [partial_unused_args_eliminate]: 1.60999e-06 [add_recomputation]: 6.509e-05 [cse_after_recomputation]: 2.206e-05, [1] [Cycle 1]: 1.736e-05, [1] [cse]: 1.165e-05 [environ_conv]: 5.25001e-06 [swap_dp_allreduce_reducescatter]: 0.00019408 [bias_add_comm_swap]: 3.43999e-06 [label_micro_interleaved_index]: 4.89e-06 [label_fine_grained_interleaved_index]: 2.70002e-06 [merge_cast_opt]: 1.32e-06 [slice_recompute_activation]: 2.05002e-06 [micro_interleaved_order_control]: 2.19001e-06 [assign_add_opt]: 1.47999e-06 [ForceFp32Comm]: 9.49978e-07 [remove_cast_before_assign_add]: 8.10018e-07 [full_micro_interleaved_order_control]: 2.32999e-06 [reorder_send_recv_between_fp_bp]: 2.86e-06 [comm_op_add_attrs]: 1.09e-06 [add_comm_op_reuse_tag]: 1.02998e-06 [interleave_split_concat_branches]: 1.42999e-06 [interleave_parallel_branches]: 1.04e-06 [overlap_opt_shard_in_pipeline]: 1.17e-06 [overlap_opt_shard_grad_in_pipeline]: 1.83002e-06 [control_data_broadcast_order]: 1.77e-05 [grouped_pairwise_exchange_alltoall]: 1.54e-06 [offloading_packed_experts]: 4.22e-06 [overlap_recompute_and_grad_model_parallel]: 4.80001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.29e-06 [overlap_recompute_allgather_and_fa_grad]: 1.37999e-06 [overlap_recompute_comm]: 2.31e-06 [overlap_grad_ring_attention]: 3.88001e-06 [overlap_grad_flash_sp]: 2.1e-05 [begin_end_overlap_inline]: 5.09986e-07 [split_matmul_comm_elemetwise]: 2.60002e-06 [split_layernorm_comm]: 1.82001e-06 [handle_group_info]: 1.12999e-06 [symbol_engine_optimizer]: 8.959e-05, [1] [Cycle 1]: 8.357e-05, [6] [build]: 4.37e-06 [elim_shapecalc]: 1.385e-05 [elim_not_effective]: 1.46e-05 [opt_reshape]: 7.13e-06 [fold_const_symbol]: 9.94001e-06 [renormalize]: 2.50002e-07 [detach_backward]: 1.77999e-06 [pipeline_parallel_scheduler]: 1.84998e-06 [auto_monad_reorder]: 1.887e-05 [get_jit_bprop_graph]: 1.76003e-06 [rewriter_after_jit_bprop_graph]: 5.12e-06 [opt_after_jit_grad]: 0.0005074 [validate]: 4.002e-05 Sums bootstrap : 0.000439s : 3.04% type_inference : 0.008587s : 59.51% event_method : 0.000045s : 0.31% auto_monad : 0.000073s : 0.50% graph_reusing : 0.000007s : 0.05% inline : 0.000002s : 0.02% add_attr.add_attr_with_inline.tag_attr : 0.000030s : 0.21% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.05% parallel-infer-symbol : 0.000004s : 0.02% pre_auto_parallel : 0.000043s : 0.30% insert-virtual-dataset : 0.000002s : 0.02% parallel-infer-symbol-second : 0.000001s : 0.01% dataset_repeat_opt : 0.000002s : 0.01% pipeline_split : 0.000002s : 0.01% optimize.py_interpret_to_execute : 0.000034s : 0.24% optimize.rewriter_before_opt_a : 0.000103s : 0.72% optimize.opt_a.expand_dump_flag : 0.000005s : 0.03% optimize.opt_a.switch_simplify : 0.000063s : 0.43% optimize.opt_a.loop_unroll : 0.000045s : 0.31% optimize.opt_a.a_1 : 0.001044s : 7.24% optimize.opt_a.with_stream_mark : 0.000031s : 0.21% optimize.opt_a.recompute_prepare : 0.000016s : 0.11% optimize.opt_a.updatestate_depend_eliminate : 0.000007s : 0.05% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.04% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.04% optimize.opt_a.parameter_eliminate : 0.000003s : 0.02% optimize.opt_a.a_2 : 0.000150s : 1.04% optimize.opt_a.accelerated_algorithm : 0.000013s : 0.09% optimize.opt_a.shard : 0.000004s : 0.03% optimize.opt_a.meta_shard_fg_expand : 0.000004s : 0.02% optimize.opt_a.shard_inline : 0.000012s : 0.08% optimize.opt_a.merge_send_recv : 0.000013s : 0.09% optimize.opt_a.auto_parallel : 0.000013s : 0.09% optimize.opt_a.parallel : 0.000024s : 0.16% optimize.opt_a.flash_sp : 0.000012s : 0.08% optimize.opt_a.merge_comm : 0.000007s : 0.05% optimize.opt_a.allreduce_fusion : 0.000006s : 0.04% optimize.opt_a.matmul_add_comm_reduction : 0.000016s : 0.11% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.01% optimize.opt_a.virtual_shard_identity : 0.000015s : 0.11% optimize.opt_a.virtual_dataset : 0.000013s : 0.09% optimize.opt_a.get_grad_eliminate_ : 0.000011s : 0.08% optimize.opt_a.virtual_output : 0.000011s : 0.08% optimize.opt_a.merge_forward : 0.000007s : 0.05% optimize.opt_a.cell_reuse_recompute_pass : 0.000003s : 0.02% optimize.opt_a.offload_activation : 0.000018s : 0.12% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000022s : 0.15% optimize.opt_a.merge_recompute_call_nodes : 0.000002s : 0.01% optimize.opt_a.before_grad : 0.000019s : 0.13% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000007s : 0.05% optimize.opt_a.meta_fg_expand : 0.000005s : 0.03% optimize.opt_a.flash_sp_send_recv_attached : 0.000003s : 0.02% optimize.opt_a.receive_attached : 0.000003s : 0.02% optimize.opt_a.after_resolve : 0.000023s : 0.16% optimize.opt_a.a_after_grad : 0.000020s : 0.14% optimize.opt_a.renormalize : 0.000854s : 5.92% optimize.opt_a.add_forward_monad_depend : 0.000007s : 0.05% optimize.opt_a.auto_monad_grad : 0.000003s : 0.02% optimize.opt_a.auto_monad_eliminator : 0.000022s : 0.16% optimize.opt_a.cse : 0.000041s : 0.29% optimize.opt_a.a_3 : 0.000077s : 0.53% optimize.py_interpret_to_execute_after_opt_a : 0.000010s : 0.07% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.02% optimize.rewriter_after_opt_a : 0.000034s : 0.24% optimize.convert_after_rewriter : 0.000008s : 0.05% optimize.order_py_execute_after_rewriter : 0.000005s : 0.03% optimize.mutable_eliminate : 0.000518s : 3.59% optimize.opt_b.b_1 : 0.000115s : 0.80% optimize.opt_b.b_2 : 0.000008s : 0.05% optimize.opt_b.updatestate_depend_eliminate : 0.000006s : 0.04% optimize.opt_b.updatestate_assign_eliminate : 0.000002s : 0.02% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.02% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000018s : 0.13% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.12% optimize.overlap_param_gather : 0.000002s : 0.01% optimize.cconv : 0.000026s : 0.18% optimize.loop_unroll : 0.000459s : 3.18% optimize.opt_after_cconv.c_1 : 0.000029s : 0.20% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.02% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000005s : 0.04% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.02% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.02% optimize.opt_after_cconv.cse : 0.000017s : 0.12% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000014s : 0.10% optimize.tuple_transform.d_1 : 0.000044s : 0.30% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.01% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.05% optimize.partial_unused_args_eliminate : 0.000002s : 0.01% optimize.add_recomputation : 0.000065s : 0.45% optimize.cse_after_recomputation.cse : 0.000012s : 0.08% optimize.environ_conv : 0.000005s : 0.04% optimize.swap_dp_allreduce_reducescatter : 0.000194s : 1.34% optimize.bias_add_comm_swap : 0.000003s : 0.02% optimize.label_micro_interleaved_index : 0.000005s : 0.03% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.02% optimize.merge_cast_opt : 0.000001s : 0.01% optimize.slice_recompute_activation : 0.000002s : 0.01% optimize.micro_interleaved_order_control : 0.000002s : 0.02% optimize.assign_add_opt : 0.000001s : 0.01% optimize.ForceFp32Comm : 0.000001s : 0.01% optimize.remove_cast_before_assign_add : 0.000001s : 0.01% optimize.full_micro_interleaved_order_control : 0.000002s : 0.02% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.02% optimize.comm_op_add_attrs : 0.000001s : 0.01% optimize.add_comm_op_reuse_tag : 0.000001s : 0.01% optimize.interleave_split_concat_branches : 0.000001s : 0.01% optimize.interleave_parallel_branches : 0.000001s : 0.01% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.01% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.01% optimize.control_data_broadcast_order : 0.000018s : 0.12% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.01% optimize.offloading_packed_experts : 0.000004s : 0.03% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.03% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.01% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.01% optimize.overlap_recompute_comm : 0.000002s : 0.02% optimize.overlap_grad_ring_attention : 0.000004s : 0.03% optimize.overlap_grad_flash_sp : 0.000021s : 0.15% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000003s : 0.02% optimize.split_layernorm_comm : 0.000002s : 0.01% optimize.handle_group_info : 0.000001s : 0.01% optimize.symbol_engine_optimizer.build : 0.000004s : 0.03% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000014s : 0.10% optimize.symbol_engine_optimizer.elim_not_effective : 0.000015s : 0.10% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.05% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000010s : 0.07% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.01% pipeline_parallel_scheduler : 0.000002s : 0.01% auto_monad_reorder : 0.000019s : 0.13% get_jit_bprop_graph : 0.000002s : 0.01% rewriter_after_jit_bprop_graph : 0.000005s : 0.04% opt_after_jit_grad : 0.000507s : 3.52% validate : 0.000040s : 0.28% Time group info: ------[substitution.] 0.000280 40 11.96% : 0.000033s : 8: substitution.arithmetic_simplify 0.82% : 0.000002s : 2: substitution.elim_not_effective 0.56% : 0.000002s : 2: substitution.fold_const_symbol 2.04% : 0.000006s : 4: substitution.graph_param_transform 72.63% : 0.000203s : 6: substitution.inline 1.33% : 0.000004s : 4: substitution.j_node_and_user_rematch 1.56% : 0.000004s : 4: substitution.remove_not_recompute_node 1.69% : 0.000005s : 4: substitution.replace_old_param 7.41% : 0.000021s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.008519 2 88.40% : 0.007531s : 1: type_inference.infer 11.60% : 0.000988s : 1: type_inference.specialize ------[replace.] 0.000095 12 58.01% : 0.000055s : 6: replace.inline 41.99% : 0.000040s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000217 12 91.70% : 0.000199s : 6: match.inline 8.30% : 0.000018s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000228 1534 0.98% : 0.000002s : 17: predicate.accumulaten_eliminater 0.74% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.45% : 0.000001s : 8: predicate.addn_check_dump 0.97% : 0.000002s : 17: predicate.addn_zero_filter 0.98% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.34% : 0.000005s : 25: predicate.arithmetic_simplify 0.99% : 0.000002s : 17: predicate.cast_eliminate 0.49% : 0.000001s : 8: predicate.check_bprop_eliminate 0.43% : 0.000001s : 8: predicate.compare_switch_simplify 0.17% : 0.000000s : 4: predicate.const_output_eliminate 0.47% : 0.000001s : 8: predicate.depend_value_elim 1.03% : 0.000002s : 17: predicate.dict_get_item_const_eliminator 1.11% : 0.000003s : 17: predicate.dict_get_item_eliminator 0.98% : 0.000002s : 17: predicate.dict_set_item_eliminator 0.82% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.20% : 0.000000s : 4: predicate.elim_not_effective 0.36% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.33% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.14% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.16% : 0.000003s : 21: predicate.environ_get_depend_swap 1.72% : 0.000004s : 29: predicate.environ_get_eliminate 1.16% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.69% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.52% : 0.000006s : 29: predicate.float_depend_g_call 0.41% : 0.000001s : 8: predicate.float_environ_get_switch 0.64% : 0.000001s : 12: predicate.float_tuple_getitem_switch 0.16% : 0.000000s : 4: predicate.fold_const_symbol 0.57% : 0.000001s : 8: predicate.get_grad_eliminate 0.18% : 0.000000s : 4: predicate.graph_param_transform 0.46% : 0.000001s : 8: predicate.incorporate_call 0.39% : 0.000001s : 8: predicate.incorporate_call_switch 6.21% : 0.000014s : 70: predicate.inline 0.80% : 0.000002s : 8: predicate.inline_without_move 0.28% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.67% : 0.000002s : 8: predicate.less_batch_normalization 2.00% : 0.000005s : 31: predicate.list_to_tuple_eliminator_ 2.66% : 0.000006s : 48: predicate.load_eliminater 0.74% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.67% : 0.000006s : 47: predicate.loop_unroll_before_grad 1.59% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.56% : 0.000001s : 8: predicate.merge_addn 0.49% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.50% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.91% : 0.000002s : 17: predicate.minmaximum_grad 0.96% : 0.000002s : 4: predicate.mutable_eliminate 0.26% : 0.000001s : 4: predicate.opt_reshape 0.36% : 0.000001s : 4: predicate.parallel_virtual_node 2.14% : 0.000005s : 29: predicate.partial_defer_inline 1.76% : 0.000004s : 27: predicate.partial_eliminate 1.04% : 0.000002s : 17: predicate.print_const_string_wrapper 0.44% : 0.000001s : 8: predicate.reduce_all_const_elim 1.37% : 0.000003s : 17: predicate.reduce_eliminate 2.77% : 0.000006s : 48: predicate.redundant_stop_gradient_eliminater 0.36% : 0.000001s : 8: predicate.remove_not_recompute_node 1.43% : 0.000003s : 31: predicate.replace_applicator 0.50% : 0.000001s : 8: predicate.replace_old_param 0.22% : 0.000001s : 4: predicate.reset_defer_inline 1.02% : 0.000002s : 17: predicate.reshape_eliminate 0.49% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.37% : 0.000001s : 4: predicate.row_tensor_eliminate 0.57% : 0.000001s : 8: predicate.same_eliminate 0.37% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.68% : 0.000002s : 8: predicate.shard_identity_eliminate 0.53% : 0.000001s : 8: predicate.special_op_eliminate 0.59% : 0.000001s : 8: predicate.specialize_transform 0.69% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.61% : 0.000001s : 8: predicate.stack_unstack_eliminate 0.28% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.82% : 0.000004s : 29: predicate.switch_defer_inline 2.23% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.75% : 0.000013s : 88: predicate.switch_simplify 0.97% : 0.000002s : 17: predicate.tile_eliminate 1.07% : 0.000002s : 17: predicate.transpose_eliminate 1.66% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.59% : 0.000004s : 25: predicate.tuple_list_get_item_const_eliminator 1.44% : 0.000003s : 25: predicate.tuple_list_get_item_depend_reorder 3.20% : 0.000007s : 39: predicate.tuple_list_get_item_eliminator 1.58% : 0.000004s : 25: predicate.tuple_list_get_set_item_eliminator 2.15% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 1.80% : 0.000004s : 31: predicate.tuple_to_list_eliminator_ 2.58% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.26% : 0.000007s : 56: predicate.updatestate_useless_node_eliminater 0.29% : 0.000001s : 4: predicate.value_based_eliminate 0.54% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.51% : 0.000001s : 8: predicate.virtual_output_eliminate 0.23% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.40% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000766 15 56.40% : 0.000432s : 7: func_graph_cloner_run.FuncGraphClonerGraph 43.60% : 0.000334s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.030280 192 0.01% : 0.000004s : 1: ForceFp32Comm 11.48% : 0.003477s : 1: add_attr 11.44% : 0.003465s : 1: add_attr_with_inline 0.01% : 0.000004s : 1: add_comm_op_reuse_tag 0.23% : 0.000070s : 1: add_recomputation 0.02% : 0.000005s : 1: assign_add_opt 0.26% : 0.000078s : 1: auto_monad 0.08% : 0.000023s : 1: auto_monad_reorder 0.01% : 0.000003s : 1: begin_end_overlap_inline 0.02% : 0.000007s : 1: bias_add_comm_swap 1.55% : 0.000468s : 1: bootstrap 0.10% : 0.000030s : 1: cconv 0.01% : 0.000004s : 1: comm_op_add_attrs 0.07% : 0.000021s : 1: control_data_broadcast_order 0.04% : 0.000011s : 1: convert_after_rewriter 0.08% : 0.000025s : 1: cse_after_recomputation 0.02% : 0.000005s : 1: dataset_repeat_opt 0.02% : 0.000005s : 1: detach_backward 0.03% : 0.000008s : 1: environ_conv 0.18% : 0.000053s : 1: event_method 0.02% : 0.000005s : 1: full_micro_interleaved_order_control 0.02% : 0.000005s : 1: get_jit_bprop_graph 0.04% : 0.000011s : 1: graph_reusing 0.01% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.01% : 0.000004s : 1: handle_group_info 0.02% : 0.000006s : 1: inline 0.02% : 0.000006s : 1: insert-virtual-dataset 0.01% : 0.000004s : 1: interleave_parallel_branches 0.01% : 0.000004s : 1: interleave_split_concat_branches 0.02% : 0.000007s : 1: label_fine_grained_interleaved_index 0.03% : 0.000008s : 1: label_micro_interleaved_index 1.54% : 0.000466s : 1: loop_unroll 0.02% : 0.000005s : 1: merge_cast_opt 0.02% : 0.000005s : 1: micro_interleaved_order_control 1.74% : 0.000528s : 1: mutable_eliminate 0.03% : 0.000008s : 1: offloading_packed_experts 0.04% : 0.000013s : 1: opt.transform.loop_unroll_optimizer 0.05% : 0.000015s : 1: opt.transform.mutable_eliminate 4.86% : 0.001471s : 78: opt.transform.opt_a 0.09% : 0.000028s : 1: opt.transform.opt_after_cconv 0.08% : 0.000025s : 1: opt.transform.opt_after_jit_grad 0.31% : 0.000095s : 28: opt.transform.opt_b 0.16% : 0.000048s : 2: opt.transform.opt_trans_graph 0.13% : 0.000041s : 4: opt.transform.symbol_engine_opt 10.39% : 0.003147s : 1: opt_a 0.35% : 0.000106s : 1: opt_after_cconv 1.70% : 0.000516s : 1: opt_after_jit_grad 0.65% : 0.000198s : 1: opt_b 18.12% : 0.005486s : 1: optimize 0.07% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.03% : 0.000008s : 1: order_py_execute_after_rewriter 0.08% : 0.000024s : 1: overlap_grad_flash_sp 0.01% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.02% : 0.000007s : 1: overlap_grad_ring_attention 0.02% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.01% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.02% : 0.000005s : 1: overlap_param_gather 0.01% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.03% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.02% : 0.000005s : 1: overlap_recompute_comm 0.03% : 0.000008s : 1: parallel-infer-symbol 0.01% : 0.000004s : 1: parallel-infer-symbol-second 0.02% : 0.000005s : 1: partial_unused_args_eliminate 0.02% : 0.000005s : 1: pipeline_parallel_scheduler 0.01% : 0.000005s : 1: pipeline_split 0.16% : 0.000048s : 1: pre_auto_parallel 0.13% : 0.000039s : 1: py_interpret_to_execute 0.05% : 0.000014s : 1: py_interpret_to_execute_after_opt_a 0.01% : 0.000004s : 1: remove_cast_before_assign_add 0.06% : 0.000017s : 1: remove_dup_value 1.49% : 0.000452s : 1: renormalize.infer 1.30% : 0.000393s : 1: renormalize.specialize 0.02% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.03% : 0.000008s : 1: rewriter_after_jit_bprop_graph 0.13% : 0.000038s : 1: rewriter_after_opt_a 0.36% : 0.000108s : 1: rewriter_before_opt_a 0.02% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.02% : 0.000005s : 1: slice_recompute_activation 0.02% : 0.000005s : 1: split_layernorm_comm 0.02% : 0.000006s : 1: split_matmul_comm_elemetwise 0.68% : 0.000205s : 1: swap_dp_allreduce_reducescatter 0.31% : 0.000092s : 1: symbol_engine_optimizer 0.26% : 0.000078s : 1: tuple_transform 28.43% : 0.008608s : 1: type_inference TotalTime = 0.0368004, [21] [bootstrap]: 0.00041565 [type_inference]: 0.016354 [event_method]: 8.175e-05 [auto_monad]: 0.00012418 [graph_reusing]: 9.46998e-06 [inline]: 2.74001e-06 [add_attr]: 0.00328615, [1] [add_attr_with_inline]: 0.00327688, [1] [Cycle 1]: 8.716e-05, [2] [tag_attr]: 4.272e-05 [meta_addattr_fg_expand]: 1.204e-05 [parallel-infer-symbol]: 3.13e-06 [pre_auto_parallel]: 5.997e-05 [insert-virtual-dataset]: 2.34999e-06 [parallel-infer-symbol-second]: 7.39994e-07 [dataset_repeat_opt]: 2.01e-06 [pipeline_split]: 1.87999e-06 [optimize]: 0.0157675, [53] [py_interpret_to_execute]: 4.581e-05 [rewriter_before_opt_a]: 0.00017866 [opt_a]: 0.0134313, [3] [Cycle 1]: 0.00964307, [45] [expand_dump_flag]: 4.56002e-06 [switch_simplify]: 9.004e-05 [loop_unroll]: 7.61e-05 [a_1]: 0.00174257 [with_stream_mark]: 2.941e-05 [recompute_prepare]: 2.476e-05 [updatestate_depend_eliminate]: 9.62001e-06 [updatestate_assign_eliminate]: 8.52e-06 [updatestate_loads_eliminate]: 7.65e-06 [parameter_eliminate]: 2.78e-06 [a_2]: 0.00024918 [accelerated_algorithm]: 3.159e-05 [shard]: 1.92001e-06 [meta_shard_fg_expand]: 4.05998e-06 [shard_inline]: 1.704e-05 [merge_send_recv]: 1.789e-05 [auto_parallel]: 1.334e-05 [parallel]: 2.111e-05 [flash_sp]: 1.057e-05 [merge_comm]: 1.029e-05 [allreduce_fusion]: 9.58002e-06 [matmul_add_comm_reduction]: 3.162e-05 [allreduce_slice_to_reducescatter]: 1.14e-06 [virtual_shard_identity]: 1.891e-05 [virtual_dataset]: 1.585e-05 [get_grad_eliminate_]: 1.556e-05 [virtual_output]: 1.545e-05 [merge_forward]: 1.028e-05 [cell_reuse_recompute_pass]: 1.17999e-06 [offload_activation]: 1.959e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.191e-05 [merge_recompute_call_nodes]: 1.63002e-06 [before_grad]: 2.867e-05 [set_forward_comm_id_for_comm_node_pass]: 1.053e-05 [meta_fg_expand]: 0.00165809 [flash_sp_send_recv_attached]: 6.68998e-06 [receive_attached]: 2.34999e-06 [after_resolve]: 6.593e-05 [a_after_grad]: 8.698e-05 [renormalize]: 0.00429322 [add_forward_monad_depend]: 1.273e-05 [auto_monad_grad]: 6.21e-06 [auto_monad_eliminator]: 6.196e-05 [cse]: 0.00021411 [a_3]: 0.00034631 [Cycle 2]: 0.00315999, [45] [expand_dump_flag]: 2.83998e-06 [switch_simplify]: 4.771e-05 [loop_unroll]: 4.424e-05 [a_1]: 0.00164982 [with_stream_mark]: 1.909e-05 [recompute_prepare]: 1.04e-05 [updatestate_depend_eliminate]: 4.85999e-06 [updatestate_assign_eliminate]: 3.43e-06 [updatestate_loads_eliminate]: 2.51e-06 [parameter_eliminate]: 1.79e-06 [a_2]: 6.991e-05 [accelerated_algorithm]: 1.303e-05 [shard]: 2.48e-06 [meta_shard_fg_expand]: 2.68998e-06 [shard_inline]: 6.49999e-06 [merge_send_recv]: 7.3e-06 [auto_parallel]: 9.29e-06 [parallel]: 8.87e-06 [flash_sp]: 3.97002e-06 [merge_comm]: 3.52997e-06 [allreduce_fusion]: 3.43e-06 [matmul_add_comm_reduction]: 9.20999e-06 [allreduce_slice_to_reducescatter]: 6.29982e-07 [virtual_shard_identity]: 8.54e-06 [virtual_dataset]: 5.89e-06 [get_grad_eliminate_]: 6.24001e-06 [virtual_output]: 5.44998e-06 [merge_forward]: 4.23001e-06 [cell_reuse_recompute_pass]: 1.15999e-06 [offload_activation]: 9.62999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.206e-05 [merge_recompute_call_nodes]: 1.60999e-06 [before_grad]: 9.94001e-06 [set_forward_comm_id_for_comm_node_pass]: 3.71999e-06 [meta_fg_expand]: 0.00014953 [flash_sp_send_recv_attached]: 1.77999e-06 [receive_attached]: 2.51e-06 [after_resolve]: 1.514e-05 [a_after_grad]: 9.87999e-06 [renormalize]: 0.00064025 [add_forward_monad_depend]: 5.19e-06 [auto_monad_grad]: 2.63e-06 [auto_monad_eliminator]: 1.364e-05 [cse]: 2.236e-05 [a_3]: 4.396e-05 [Cycle 3]: 0.00061003, [45] [expand_dump_flag]: 1.50001e-06 [switch_simplify]: 8.03001e-06 [loop_unroll]: 6.02999e-06 [a_1]: 0.00012332 [with_stream_mark]: 9.07001e-06 [recompute_prepare]: 6.23002e-06 [updatestate_depend_eliminate]: 3.2e-06 [updatestate_assign_eliminate]: 2.65002e-06 [updatestate_loads_eliminate]: 2.16998e-06 [parameter_eliminate]: 1.02998e-06 [a_2]: 6.526e-05 [accelerated_algorithm]: 1.139e-05 [shard]: 9.89996e-07 [meta_shard_fg_expand]: 1.51998e-06 [shard_inline]: 6.21e-06 [merge_send_recv]: 5.22e-06 [auto_parallel]: 6.02999e-06 [parallel]: 5.27001e-06 [flash_sp]: 1.04e-06 [merge_comm]: 3.09999e-06 [allreduce_fusion]: 2.69001e-06 [matmul_add_comm_reduction]: 6.10002e-06 [allreduce_slice_to_reducescatter]: 3.69997e-07 [virtual_shard_identity]: 7.31999e-06 [virtual_dataset]: 5.69e-06 [get_grad_eliminate_]: 5.46e-06 [virtual_output]: 5.69999e-06 [merge_forward]: 3.31999e-06 [cell_reuse_recompute_pass]: 1.47001e-06 [offload_activation]: 7.55e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.11e-05 [merge_recompute_call_nodes]: 7.2e-07 [before_grad]: 8.58001e-06 [set_forward_comm_id_for_comm_node_pass]: 3.35e-06 [meta_fg_expand]: 1.79e-06 [flash_sp_send_recv_attached]: 1.15999e-06 [receive_attached]: 1.60999e-06 [after_resolve]: 9.69e-06 [a_after_grad]: 8.62998e-06 [renormalize]: 7.99773e-08 [add_forward_monad_depend]: 1.07998e-06 [auto_monad_grad]: 1.49e-06 [auto_monad_eliminator]: 6.36998e-06 [cse]: 1.215e-05 [a_3]: 3.37e-05 [py_interpret_to_execute_after_opt_a]: 1.131e-05 [slice_cell_reuse_recomputed_activation]: 2.21998e-06 [rewriter_after_opt_a]: 3.66e-05 [convert_after_rewriter]: 6.56999e-06 [order_py_execute_after_rewriter]: 5.35999e-06 [mutable_eliminate]: 0.00068667 [opt_b]: 0.00020109, [1] [Cycle 1]: 0.00019285, [7] [b_1]: 0.00011772 [b_2]: 8.37998e-06 [updatestate_depend_eliminate]: 6.61e-06 [updatestate_assign_eliminate]: 2.89001e-06 [updatestate_loads_eliminate]: 2.11e-06 [renormalize]: 4.99975e-07 [cse]: 1.935e-05 [optimize_parallel_all_gather_comm]: 1.74e-05 [overlap_param_gather]: 2.38998e-06 [cconv]: 2.374e-05 [loop_unroll]: 0.00042647 [opt_after_cconv]: 0.00010136, [1] [Cycle 1]: 9.518e-05, [7] [c_1]: 3.068e-05 [parameter_eliminate]: 2.98e-06 [updatestate_depend_eliminate]: 5.24e-06 [updatestate_assign_eliminate]: 2.64999e-06 [updatestate_loads_eliminate]: 2.34999e-06 [cse]: 1.733e-05 [renormalize]: 4.10015e-07 [remove_dup_value]: 1.408e-05 [tuple_transform]: 7.641e-05, [1] [Cycle 1]: 7.162e-05, [4] [d_1]: 4.401e-05 [none_parameter_eliminate]: 1.83002e-06 [renormalize]: 1.69995e-07 [switch_simplify]: 6.98e-06 [partial_unused_args_eliminate]: 1.97001e-06 [add_recomputation]: 4.645e-05 [cse_after_recomputation]: 3.16e-05, [1] [Cycle 1]: 1.832e-05, [1] [cse]: 1.237e-05 [environ_conv]: 8.15999e-06 [swap_dp_allreduce_reducescatter]: 5.67001e-06 [bias_add_comm_swap]: 2.53998e-06 [label_micro_interleaved_index]: 5.02e-06 [label_fine_grained_interleaved_index]: 3.18e-06 [merge_cast_opt]: 1.28002e-06 [slice_recompute_activation]: 1.93002e-06 [micro_interleaved_order_control]: 2.76e-06 [assign_add_opt]: 1.22999e-06 [ForceFp32Comm]: 9.80013e-07 [remove_cast_before_assign_add]: 1.01002e-06 [full_micro_interleaved_order_control]: 2.34999e-06 [reorder_send_recv_between_fp_bp]: 2.89999e-06 [comm_op_add_attrs]: 1.21997e-06 [add_comm_op_reuse_tag]: 1.04e-06 [interleave_split_concat_branches]: 1.25001e-06 [interleave_parallel_branches]: 1.10001e-06 [overlap_opt_shard_in_pipeline]: 1.22e-06 [overlap_opt_shard_grad_in_pipeline]: 1.90001e-06 [control_data_broadcast_order]: 1.343e-05 [grouped_pairwise_exchange_alltoall]: 1.67001e-06 [offloading_packed_experts]: 3.91999e-06 [overlap_recompute_and_grad_model_parallel]: 4.57998e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.27e-06 [overlap_recompute_allgather_and_fa_grad]: 1.42e-06 [overlap_recompute_comm]: 2.04e-06 [overlap_grad_ring_attention]: 4.47e-06 [overlap_grad_flash_sp]: 2.083e-05 [begin_end_overlap_inline]: 5.19998e-07 [split_matmul_comm_elemetwise]: 2.09999e-06 [split_layernorm_comm]: 1.83002e-06 [handle_group_info]: 1.17e-06 [symbol_engine_optimizer]: 8.283e-05, [1] [Cycle 1]: 7.838e-05, [6] [build]: 9.94999e-06 [elim_shapecalc]: 1.052e-05 [elim_not_effective]: 1.344e-05 [opt_reshape]: 6.58e-06 [fold_const_symbol]: 9.41998e-06 [renormalize]: 2.89991e-07 [detach_backward]: 1.79e-06 [pipeline_parallel_scheduler]: 1.49998e-06 [auto_monad_reorder]: 1.744e-05 [get_jit_bprop_graph]: 2.16998e-06 [rewriter_after_jit_bprop_graph]: 4.1e-06 [opt_after_jit_grad]: 0.00046769 [validate]: 4.206e-05 Sums bootstrap : 0.000416s : 1.29% type_inference : 0.016354s : 50.70% event_method : 0.000082s : 0.25% auto_monad : 0.000124s : 0.38% graph_reusing : 0.000009s : 0.03% inline : 0.000003s : 0.01% add_attr.add_attr_with_inline.tag_attr : 0.000043s : 0.13% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000012s : 0.04% parallel-infer-symbol : 0.000003s : 0.01% pre_auto_parallel : 0.000060s : 0.19% insert-virtual-dataset : 0.000002s : 0.01% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.01% pipeline_split : 0.000002s : 0.01% optimize.py_interpret_to_execute : 0.000046s : 0.14% optimize.rewriter_before_opt_a : 0.000179s : 0.55% optimize.opt_a.expand_dump_flag : 0.000009s : 0.03% optimize.opt_a.switch_simplify : 0.000146s : 0.45% optimize.opt_a.loop_unroll : 0.000126s : 0.39% optimize.opt_a.a_1 : 0.003516s : 10.90% optimize.opt_a.with_stream_mark : 0.000058s : 0.18% optimize.opt_a.recompute_prepare : 0.000041s : 0.13% optimize.opt_a.updatestate_depend_eliminate : 0.000018s : 0.05% optimize.opt_a.updatestate_assign_eliminate : 0.000015s : 0.05% optimize.opt_a.updatestate_loads_eliminate : 0.000012s : 0.04% optimize.opt_a.parameter_eliminate : 0.000006s : 0.02% optimize.opt_a.a_2 : 0.000384s : 1.19% optimize.opt_a.accelerated_algorithm : 0.000056s : 0.17% optimize.opt_a.shard : 0.000005s : 0.02% optimize.opt_a.meta_shard_fg_expand : 0.000008s : 0.03% optimize.opt_a.shard_inline : 0.000030s : 0.09% optimize.opt_a.merge_send_recv : 0.000030s : 0.09% optimize.opt_a.auto_parallel : 0.000029s : 0.09% optimize.opt_a.parallel : 0.000035s : 0.11% optimize.opt_a.flash_sp : 0.000016s : 0.05% optimize.opt_a.merge_comm : 0.000017s : 0.05% optimize.opt_a.allreduce_fusion : 0.000016s : 0.05% optimize.opt_a.matmul_add_comm_reduction : 0.000047s : 0.15% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.01% optimize.opt_a.virtual_shard_identity : 0.000035s : 0.11% optimize.opt_a.virtual_dataset : 0.000027s : 0.09% optimize.opt_a.get_grad_eliminate_ : 0.000027s : 0.08% optimize.opt_a.virtual_output : 0.000027s : 0.08% optimize.opt_a.merge_forward : 0.000018s : 0.06% optimize.opt_a.cell_reuse_recompute_pass : 0.000004s : 0.01% optimize.opt_a.offload_activation : 0.000037s : 0.11% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000055s : 0.17% optimize.opt_a.merge_recompute_call_nodes : 0.000004s : 0.01% optimize.opt_a.before_grad : 0.000047s : 0.15% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000018s : 0.05% optimize.opt_a.meta_fg_expand : 0.001809s : 5.61% optimize.opt_a.flash_sp_send_recv_attached : 0.000010s : 0.03% optimize.opt_a.receive_attached : 0.000006s : 0.02% optimize.opt_a.after_resolve : 0.000091s : 0.28% optimize.opt_a.a_after_grad : 0.000105s : 0.33% optimize.opt_a.renormalize : 0.004934s : 15.30% optimize.opt_a.add_forward_monad_depend : 0.000019s : 0.06% optimize.opt_a.auto_monad_grad : 0.000010s : 0.03% optimize.opt_a.auto_monad_eliminator : 0.000082s : 0.25% optimize.opt_a.cse : 0.000249s : 0.77% optimize.opt_a.a_3 : 0.000424s : 1.31% optimize.py_interpret_to_execute_after_opt_a : 0.000011s : 0.04% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.01% optimize.rewriter_after_opt_a : 0.000037s : 0.11% optimize.convert_after_rewriter : 0.000007s : 0.02% optimize.order_py_execute_after_rewriter : 0.000005s : 0.02% optimize.mutable_eliminate : 0.000687s : 2.13% optimize.opt_b.b_1 : 0.000118s : 0.36% optimize.opt_b.b_2 : 0.000008s : 0.03% optimize.opt_b.updatestate_depend_eliminate : 0.000007s : 0.02% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.01% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000019s : 0.06% optimize.optimize_parallel_all_gather_comm : 0.000017s : 0.05% optimize.overlap_param_gather : 0.000002s : 0.01% optimize.cconv : 0.000024s : 0.07% optimize.loop_unroll : 0.000426s : 1.32% optimize.opt_after_cconv.c_1 : 0.000031s : 0.10% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000005s : 0.02% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.01% optimize.opt_after_cconv.cse : 0.000017s : 0.05% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000014s : 0.04% optimize.tuple_transform.d_1 : 0.000044s : 0.14% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.01% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.02% optimize.partial_unused_args_eliminate : 0.000002s : 0.01% optimize.add_recomputation : 0.000046s : 0.14% optimize.cse_after_recomputation.cse : 0.000012s : 0.04% optimize.environ_conv : 0.000008s : 0.03% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.02% optimize.bias_add_comm_swap : 0.000003s : 0.01% optimize.label_micro_interleaved_index : 0.000005s : 0.02% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.01% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.01% optimize.micro_interleaved_order_control : 0.000003s : 0.01% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.01% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.01% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.01% optimize.control_data_broadcast_order : 0.000013s : 0.04% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.01% optimize.offloading_packed_experts : 0.000004s : 0.01% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.01% optimize.overlap_grad_ring_attention : 0.000004s : 0.01% optimize.overlap_grad_flash_sp : 0.000021s : 0.06% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.01% optimize.split_layernorm_comm : 0.000002s : 0.01% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000010s : 0.03% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000011s : 0.03% optimize.symbol_engine_optimizer.elim_not_effective : 0.000013s : 0.04% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.02% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000009s : 0.03% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.01% pipeline_parallel_scheduler : 0.000001s : 0.00% auto_monad_reorder : 0.000017s : 0.05% get_jit_bprop_graph : 0.000002s : 0.01% rewriter_after_jit_bprop_graph : 0.000004s : 0.01% opt_after_jit_grad : 0.000468s : 1.45% validate : 0.000042s : 0.13% Time group info: ------[substitution.] 0.000926 195 6.98% : 0.000065s : 7: substitution.arithmetic_simplify 0.22% : 0.000002s : 2: substitution.elim_not_effective 0.61% : 0.000006s : 6: substitution.float_depend_g_call 0.48% : 0.000004s : 3: substitution.float_tuple_getitem_switch 0.17% : 0.000002s : 2: substitution.fold_const_symbol 0.68% : 0.000006s : 4: substitution.graph_param_transform 0.29% : 0.000003s : 2: substitution.incorporate_call 0.23% : 0.000002s : 2: substitution.incorporate_call_switch 56.58% : 0.000524s : 18: substitution.inline 2.13% : 0.000020s : 2: substitution.inline_without_move 1.02% : 0.000009s : 14: substitution.j_node_and_user_rematch 1.86% : 0.000017s : 3: substitution.less_batch_normalization 1.63% : 0.000015s : 11: substitution.minmaximum_grad 0.69% : 0.000006s : 6: substitution.partial_eliminate 1.28% : 0.000012s : 14: substitution.remove_not_recompute_node 3.16% : 0.000029s : 10: substitution.replace_applicator 1.13% : 0.000010s : 12: substitution.replace_old_param 0.38% : 0.000003s : 1: substitution.set_cell_output_no_recompute 5.65% : 0.000052s : 11: substitution.tuple_list_convert_item_index_to_positive 1.67% : 0.000015s : 11: substitution.tuple_list_get_item_const_eliminator 2.17% : 0.000020s : 11: substitution.tuple_list_get_item_depend_reorder 8.79% : 0.000081s : 32: substitution.tuple_list_get_item_eliminator 2.19% : 0.000020s : 11: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.016261 2 86.35% : 0.014041s : 1: type_inference.infer 13.65% : 0.002219s : 1: type_inference.specialize ------[replace.] 0.000292 38 3.74% : 0.000011s : 2: replace.arithmetic_simplify 55.11% : 0.000161s : 18: replace.inline 41.14% : 0.000120s : 18: replace.tuple_list_get_item_eliminator ------[match.] 0.000586 38 5.22% : 0.000031s : 2: match.arithmetic_simplify 87.55% : 0.000513s : 18: match.inline 7.23% : 0.000042s : 18: match.tuple_list_get_item_eliminator ------[predicate.] 0.000706 4976 1.09% : 0.000008s : 61: predicate.accumulaten_eliminater 0.22% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.41% : 0.000003s : 23: predicate.addn_check_dump 1.17% : 0.000008s : 61: predicate.addn_zero_filter 1.02% : 0.000007s : 61: predicate.adjust_all_reduce_mul_add 1.98% : 0.000014s : 86: predicate.arithmetic_simplify 1.17% : 0.000008s : 63: predicate.cast_eliminate 1.03% : 0.000007s : 55: predicate.check_bprop_eliminate 0.41% : 0.000003s : 23: predicate.compare_switch_simplify 0.05% : 0.000000s : 4: predicate.const_output_eliminate 0.42% : 0.000003s : 23: predicate.depend_value_elim 1.28% : 0.000009s : 63: predicate.dict_get_item_const_eliminator 1.27% : 0.000009s : 63: predicate.dict_get_item_eliminator 1.15% : 0.000008s : 63: predicate.dict_set_item_eliminator 0.27% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.09% : 0.000001s : 4: predicate.elim_not_effective 0.10% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.22% : 0.000009s : 67: predicate.environ_add_const_eliminate 1.21% : 0.000009s : 67: predicate.environ_get_add_eliminate 1.16% : 0.000008s : 67: predicate.environ_get_depend_swap 1.61% : 0.000011s : 90: predicate.environ_get_eliminate 1.17% : 0.000008s : 67: predicate.environ_get_set_eliminate 1.96% : 0.000014s : 99: predicate.exchange_switch_depend_value 2.73% : 0.000019s : 99: predicate.float_depend_g_call 0.47% : 0.000003s : 23: predicate.float_environ_get_switch 0.51% : 0.000004s : 27: predicate.float_tuple_getitem_switch 0.05% : 0.000000s : 4: predicate.fold_const_symbol 0.47% : 0.000003s : 23: predicate.get_grad_eliminate 0.06% : 0.000000s : 4: predicate.graph_param_transform 0.43% : 0.000003s : 23: predicate.incorporate_call 0.39% : 0.000003s : 23: predicate.incorporate_call_switch 5.77% : 0.000041s : 214: predicate.inline 1.30% : 0.000009s : 46: predicate.inline_without_move 0.24% : 0.000002s : 23: predicate.j_node_and_user_rematch 0.69% : 0.000005s : 23: predicate.less_batch_normalization 1.68% : 0.000012s : 89: predicate.list_to_tuple_eliminator_ 2.60% : 0.000018s : 150: predicate.load_eliminater 0.28% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.68% : 0.000019s : 141: predicate.loop_unroll_before_grad 1.36% : 0.000010s : 71: predicate.make_slice_get_slice_eliminator 0.43% : 0.000003s : 23: predicate.merge_addn 1.02% : 0.000007s : 55: predicate.micro_step_allgather_replace 1.03% : 0.000007s : 55: predicate.mini_step_allgather_replace 1.16% : 0.000008s : 63: predicate.minmaximum_grad 0.33% : 0.000002s : 4: predicate.mutable_eliminate 0.10% : 0.000001s : 4: predicate.opt_reshape 0.13% : 0.000001s : 4: predicate.parallel_virtual_node 2.39% : 0.000017s : 99: predicate.partial_defer_inline 1.83% : 0.000013s : 85: predicate.partial_eliminate 1.12% : 0.000008s : 61: predicate.print_const_string_wrapper 0.42% : 0.000003s : 23: predicate.reduce_all_const_elim 1.47% : 0.000010s : 63: predicate.reduce_eliminate 2.63% : 0.000019s : 150: predicate.redundant_stop_gradient_eliminater 0.33% : 0.000002s : 23: predicate.remove_not_recompute_node 1.93% : 0.000014s : 136: predicate.replace_applicator 0.58% : 0.000004s : 46: predicate.replace_old_param 0.12% : 0.000001s : 4: predicate.reset_defer_inline 1.14% : 0.000008s : 63: predicate.reshape_eliminate 1.05% : 0.000007s : 55: predicate.row_tensor_add_zeros_like 0.14% : 0.000001s : 4: predicate.row_tensor_eliminate 1.17% : 0.000008s : 55: predicate.same_eliminate 0.32% : 0.000002s : 23: predicate.set_cell_output_no_recompute 0.60% : 0.000004s : 23: predicate.shard_identity_eliminate 0.23% : 0.000002s : 8: predicate.special_op_eliminate 0.57% : 0.000004s : 23: predicate.specialize_transform 1.44% : 0.000010s : 55: predicate.split_environ_get_set_with_tuple_value 1.11% : 0.000008s : 46: predicate.stack_unstack_eliminate 0.09% : 0.000001s : 4: predicate.switch_call_monad_eliminater 2.05% : 0.000014s : 99: predicate.switch_defer_inline 2.98% : 0.000021s : 154: predicate.switch_layer_defer_inline 5.29% : 0.000037s : 267: predicate.switch_simplify 1.18% : 0.000008s : 63: predicate.tile_eliminate 1.16% : 0.000008s : 63: predicate.transpose_eliminate 1.47% : 0.000010s : 71: predicate.tuple_list_convert_item_index_to_positive 1.59% : 0.000011s : 71: predicate.tuple_list_get_item_const_eliminator 1.35% : 0.000010s : 71: predicate.tuple_list_get_item_depend_reorder 2.99% : 0.000021s : 112: predicate.tuple_list_get_item_eliminator 1.39% : 0.000010s : 71: predicate.tuple_list_get_set_item_eliminator 1.93% : 0.000014s : 94: predicate.tuple_list_set_item_eliminator 1.60% : 0.000011s : 89: predicate.tuple_to_list_eliminator_ 2.63% : 0.000019s : 150: predicate.updatestate_pure_node_eliminater 3.12% : 0.000022s : 173: predicate.updatestate_useless_node_eliminater 0.14% : 0.000001s : 4: predicate.value_based_eliminate 0.47% : 0.000003s : 23: predicate.virtual_dataset_eliminate 0.46% : 0.000003s : 23: predicate.virtual_output_eliminate 0.07% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.11% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002136 41 59.02% : 0.001261s : 19: func_graph_cloner_run.FuncGraphClonerGraph 40.98% : 0.000875s : 22: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.065910 233 0.01% : 0.000004s : 1: ForceFp32Comm 4.99% : 0.003291s : 1: add_attr 4.98% : 0.003281s : 1: add_attr_with_inline 0.01% : 0.000004s : 1: add_comm_op_reuse_tag 0.08% : 0.000051s : 1: add_recomputation 0.01% : 0.000004s : 1: assign_add_opt 0.20% : 0.000132s : 1: auto_monad 0.03% : 0.000021s : 1: auto_monad_reorder 0.01% : 0.000003s : 1: begin_end_overlap_inline 0.01% : 0.000005s : 1: bias_add_comm_swap 0.66% : 0.000437s : 1: bootstrap 0.04% : 0.000027s : 1: cconv 0.01% : 0.000004s : 1: comm_op_add_attrs 0.03% : 0.000017s : 1: control_data_broadcast_order 0.01% : 0.000010s : 1: convert_after_rewriter 0.05% : 0.000035s : 1: cse_after_recomputation 0.01% : 0.000005s : 1: dataset_repeat_opt 0.01% : 0.000005s : 1: detach_backward 0.02% : 0.000012s : 1: environ_conv 0.14% : 0.000090s : 1: event_method 0.01% : 0.000006s : 1: full_micro_interleaved_order_control 0.01% : 0.000006s : 1: get_jit_bprop_graph 0.02% : 0.000014s : 1: graph_reusing 0.01% : 0.000004s : 1: grouped_pairwise_exchange_alltoall 0.01% : 0.000004s : 1: handle_group_info 0.01% : 0.000006s : 1: inline 0.01% : 0.000006s : 1: insert-virtual-dataset 0.01% : 0.000004s : 1: interleave_parallel_branches 0.01% : 0.000004s : 1: interleave_split_concat_branches 0.01% : 0.000006s : 1: label_fine_grained_interleaved_index 0.01% : 0.000008s : 1: label_micro_interleaved_index 0.66% : 0.000435s : 1: loop_unroll 0.01% : 0.000004s : 1: merge_cast_opt 0.01% : 0.000006s : 1: micro_interleaved_order_control 1.05% : 0.000695s : 1: mutable_eliminate 0.01% : 0.000007s : 1: offloading_packed_experts 0.02% : 0.000013s : 1: opt.transform.loop_unroll_optimizer 0.02% : 0.000016s : 1: opt.transform.mutable_eliminate 7.59% : 0.005004s : 117: opt.transform.opt_a 0.04% : 0.000029s : 1: opt.transform.opt_after_cconv 0.04% : 0.000024s : 1: opt.transform.opt_after_jit_grad 0.15% : 0.000098s : 28: opt.transform.opt_b 0.07% : 0.000049s : 2: opt.transform.opt_trans_graph 0.05% : 0.000036s : 4: opt.transform.symbol_engine_opt 20.38% : 0.013435s : 1: opt_a 0.16% : 0.000105s : 1: opt_after_cconv 0.72% : 0.000477s : 1: opt_after_jit_grad 0.31% : 0.000205s : 1: opt_b 23.93% : 0.015773s : 1: optimize 0.03% : 0.000021s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000009s : 1: order_py_execute_after_rewriter 0.04% : 0.000025s : 1: overlap_grad_flash_sp 0.01% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000007s : 1: overlap_grad_ring_attention 0.01% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.01% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.01% : 0.000005s : 1: overlap_param_gather 0.01% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000007s : 1: overlap_recompute_and_grad_model_parallel 0.01% : 0.000005s : 1: overlap_recompute_comm 0.01% : 0.000007s : 1: parallel-infer-symbol 0.01% : 0.000004s : 1: parallel-infer-symbol-second 0.01% : 0.000005s : 1: partial_unused_args_eliminate 0.01% : 0.000005s : 1: pipeline_parallel_scheduler 0.01% : 0.000005s : 1: pipeline_split 0.10% : 0.000065s : 1: pre_auto_parallel 0.08% : 0.000050s : 1: py_interpret_to_execute 0.02% : 0.000015s : 1: py_interpret_to_execute_after_opt_a 0.01% : 0.000004s : 1: remove_cast_before_assign_add 0.03% : 0.000018s : 1: remove_dup_value 4.85% : 0.003200s : 2: renormalize.infer 2.60% : 0.001716s : 2: renormalize.specialize 0.01% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000007s : 1: rewriter_after_jit_bprop_graph 0.06% : 0.000040s : 1: rewriter_after_opt_a 0.28% : 0.000184s : 1: rewriter_before_opt_a 0.01% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.01% : 0.000005s : 1: slice_recompute_activation 0.01% : 0.000005s : 1: split_layernorm_comm 0.01% : 0.000005s : 1: split_matmul_comm_elemetwise 0.01% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.13% : 0.000086s : 1: symbol_engine_optimizer 0.12% : 0.000080s : 1: tuple_transform 24.84% : 0.016372s : 1: type_inference [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:45:33.830.425 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:45:33.830.759 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. TotalTime = 0.423107, [21] [bootstrap]: 0.00039308 [type_inference]: 0.377798 [event_method]: 0.00010902 [auto_monad]: 0.00027727 [graph_reusing]: 2.461e-05 [inline]: 2.59999e-06 [add_attr]: 0.0280983, [1] [add_attr_with_inline]: 0.0280875, [1] [Cycle 1]: 0.00017276, [2] [tag_attr]: 0.00011404 [meta_addattr_fg_expand]: 2.73e-05 [parallel-infer-symbol]: 3.82002e-06 [pre_auto_parallel]: 0.00013954 [insert-virtual-dataset]: 2.44999e-06 [parallel-infer-symbol-second]: 3.46001e-06 [dataset_repeat_opt]: 2.02001e-06 [pipeline_split]: 1.63002e-06 [optimize]: 0.0155164, [53] [py_interpret_to_execute]: 0.00012001 [rewriter_before_opt_a]: 0.00041163 [opt_a]: 0.0127125, [2] [Cycle 1]: 0.0120021, [45] [expand_dump_flag]: 8.45001e-06 [switch_simplify]: 0.00035029 [loop_unroll]: 0.00015029 [a_1]: 0.00480979 [with_stream_mark]: 2.038e-05 [recompute_prepare]: 1.743e-05 [updatestate_depend_eliminate]: 6.97002e-06 [updatestate_assign_eliminate]: 4.96002e-06 [updatestate_loads_eliminate]: 5.07e-06 [parameter_eliminate]: 2.30002e-06 [a_2]: 0.00026664 [accelerated_algorithm]: 8.32e-06 [shard]: 2.74999e-06 [meta_shard_fg_expand]: 6.33e-06 [shard_inline]: 8.10999e-06 [merge_send_recv]: 8.64e-06 [auto_parallel]: 8.2e-06 [parallel]: 1.873e-05 [flash_sp]: 9.91998e-06 [merge_comm]: 3.84002e-06 [allreduce_fusion]: 3.28998e-06 [matmul_add_comm_reduction]: 9.31e-06 [allreduce_slice_to_reducescatter]: 7.39994e-07 [virtual_shard_identity]: 9.36002e-06 [virtual_dataset]: 6.46e-06 [get_grad_eliminate_]: 6.72002e-06 [virtual_output]: 6.31e-06 [merge_forward]: 4.42e-06 [cell_reuse_recompute_pass]: 1.20999e-06 [offload_activation]: 1.017e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.184e-05 [merge_recompute_call_nodes]: 1.43002e-06 [before_grad]: 1.067e-05 [set_forward_comm_id_for_comm_node_pass]: 3.66999e-06 [meta_fg_expand]: 5.74999e-06 [flash_sp_send_recv_attached]: 2.56998e-06 [receive_attached]: 2.19001e-06 [after_resolve]: 1.255e-05 [a_after_grad]: 9.87001e-06 [renormalize]: 0.00575424 [add_forward_monad_depend]: 7.23e-06 [auto_monad_grad]: 2.87002e-06 [auto_monad_eliminator]: 2.161e-05 [cse]: 3.263e-05 [a_3]: 5.449e-05 [Cycle 2]: 0.00069691, [45] [expand_dump_flag]: 2.15002e-06 [switch_simplify]: 8.71002e-06 [loop_unroll]: 6.45002e-06 [a_1]: 0.00015609 [with_stream_mark]: 1.474e-05 [recompute_prepare]: 7.46999e-06 [updatestate_depend_eliminate]: 3.73999e-06 [updatestate_assign_eliminate]: 3.13e-06 [updatestate_loads_eliminate]: 3.28e-06 [parameter_eliminate]: 1.25999e-06 [a_2]: 7.622e-05 [accelerated_algorithm]: 6.54999e-06 [shard]: 2.07999e-06 [meta_shard_fg_expand]: 1.92001e-06 [shard_inline]: 6.12999e-06 [merge_send_recv]: 7.73999e-06 [auto_parallel]: 6.80002e-06 [parallel]: 7.23e-06 [flash_sp]: 3.89002e-06 [merge_comm]: 3.48999e-06 [allreduce_fusion]: 3.18e-06 [matmul_add_comm_reduction]: 7.71001e-06 [allreduce_slice_to_reducescatter]: 7.40023e-07 [virtual_shard_identity]: 6.96001e-06 [virtual_dataset]: 6.14999e-06 [get_grad_eliminate_]: 5.49998e-06 [virtual_output]: 5.67999e-06 [merge_forward]: 3.26001e-06 [cell_reuse_recompute_pass]: 2.49001e-06 [offload_activation]: 9.15001e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.159e-05 [merge_recompute_call_nodes]: 1.19e-06 [before_grad]: 1.003e-05 [set_forward_comm_id_for_comm_node_pass]: 4.3e-06 [meta_fg_expand]: 2.69999e-06 [flash_sp_send_recv_attached]: 1.25999e-06 [receive_attached]: 1.96e-06 [after_resolve]: 1.345e-05 [a_after_grad]: 1.104e-05 [renormalize]: 7.99773e-08 [add_forward_monad_depend]: 1.12e-06 [auto_monad_grad]: 8.50006e-07 [auto_monad_eliminator]: 6.89999e-06 [cse]: 1.244e-05 [a_3]: 3.521e-05 [py_interpret_to_execute_after_opt_a]: 1.47e-05 [slice_cell_reuse_recomputed_activation]: 2.51e-06 [rewriter_after_opt_a]: 3.717e-05 [convert_after_rewriter]: 6.35002e-06 [order_py_execute_after_rewriter]: 5.84999e-06 [mutable_eliminate]: 0.00070635 [opt_b]: 0.00020037, [1] [Cycle 1]: 0.00019313, [7] [b_1]: 0.00012158 [b_2]: 8.20999e-06 [updatestate_depend_eliminate]: 5.47001e-06 [updatestate_assign_eliminate]: 2.51e-06 [updatestate_loads_eliminate]: 2.22001e-06 [renormalize]: 7.39994e-07 [cse]: 1.665e-05 [optimize_parallel_all_gather_comm]: 1.635e-05 [overlap_param_gather]: 1.92999e-06 [cconv]: 2.585e-05 [loop_unroll]: 0.00047989 [opt_after_cconv]: 0.00010275, [1] [Cycle 1]: 9.703e-05, [7] [c_1]: 3.326e-05 [parameter_eliminate]: 2.66e-06 [updatestate_depend_eliminate]: 5.36002e-06 [updatestate_assign_eliminate]: 2.54999e-06 [updatestate_loads_eliminate]: 2.76999e-06 [cse]: 1.629e-05 [renormalize]: 5.50004e-07 [remove_dup_value]: 1.361e-05 [tuple_transform]: 7.906e-05, [1] [Cycle 1]: 7.441e-05, [4] [d_1]: 4.617e-05 [none_parameter_eliminate]: 1.67999e-06 [renormalize]: 2.09984e-07 [switch_simplify]: 7.46001e-06 [partial_unused_args_eliminate]: 1.71e-06 [add_recomputation]: 4.77e-05 [cse_after_recomputation]: 2.124e-05, [1] [Cycle 1]: 1.682e-05, [1] [cse]: 1.138e-05 [environ_conv]: 4.62998e-06 [swap_dp_allreduce_reducescatter]: 6.17999e-06 [bias_add_comm_swap]: 2.55002e-06 [label_micro_interleaved_index]: 4.4e-06 [label_fine_grained_interleaved_index]: 3.03e-06 [merge_cast_opt]: 1.40999e-06 [slice_recompute_activation]: 2.27999e-06 [micro_interleaved_order_control]: 2.45002e-06 [assign_add_opt]: 1.17e-06 [ForceFp32Comm]: 8.70001e-07 [remove_cast_before_assign_add]: 1.36002e-06 [full_micro_interleaved_order_control]: 2.17999e-06 [reorder_send_recv_between_fp_bp]: 2.54001e-06 [comm_op_add_attrs]: 1.20999e-06 [add_comm_op_reuse_tag]: 1.02998e-06 [interleave_split_concat_branches]: 1.31002e-06 [interleave_parallel_branches]: 1.07e-06 [overlap_opt_shard_in_pipeline]: 1.40001e-06 [overlap_opt_shard_grad_in_pipeline]: 2.02999e-06 [control_data_broadcast_order]: 1.247e-05 [grouped_pairwise_exchange_alltoall]: 2.20002e-06 [offloading_packed_experts]: 3.78001e-06 [overlap_recompute_and_grad_model_parallel]: 4.46002e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.29e-06 [overlap_recompute_allgather_and_fa_grad]: 1.38002e-06 [overlap_recompute_comm]: 2.21998e-06 [overlap_grad_ring_attention]: 3.66999e-06 [overlap_grad_flash_sp]: 2.061e-05 [begin_end_overlap_inline]: 5.19998e-07 [split_matmul_comm_elemetwise]: 2.15002e-06 [split_layernorm_comm]: 1.63002e-06 [handle_group_info]: 1.00001e-06 [symbol_engine_optimizer]: 0.00017469, [1] [Cycle 1]: 0.00017018, [6] [build]: 8.619e-05 [elim_shapecalc]: 1.243e-05 [elim_not_effective]: 1.559e-05 [opt_reshape]: 7.47002e-06 [fold_const_symbol]: 1.542e-05 [renormalize]: 2.19996e-07 [detach_backward]: 2.24001e-06 [pipeline_parallel_scheduler]: 1.67999e-06 [auto_monad_reorder]: 1.693e-05 [get_jit_bprop_graph]: 1.62001e-06 [rewriter_after_jit_bprop_graph]: 3.69002e-06 [opt_after_jit_grad]: 0.00048945 [validate]: 3.796e-05 Sums bootstrap : 0.000393s : 0.10% type_inference : 0.377798s : 95.87% event_method : 0.000109s : 0.03% auto_monad : 0.000277s : 0.07% graph_reusing : 0.000025s : 0.01% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000114s : 0.03% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000027s : 0.01% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000140s : 0.04% insert-virtual-dataset : 0.000002s : 0.00% parallel-infer-symbol-second : 0.000003s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000120s : 0.03% optimize.rewriter_before_opt_a : 0.000412s : 0.10% optimize.opt_a.expand_dump_flag : 0.000011s : 0.00% optimize.opt_a.switch_simplify : 0.000359s : 0.09% optimize.opt_a.loop_unroll : 0.000157s : 0.04% optimize.opt_a.a_1 : 0.004966s : 1.26% optimize.opt_a.with_stream_mark : 0.000035s : 0.01% optimize.opt_a.recompute_prepare : 0.000025s : 0.01% optimize.opt_a.updatestate_depend_eliminate : 0.000011s : 0.00% optimize.opt_a.updatestate_assign_eliminate : 0.000008s : 0.00% optimize.opt_a.updatestate_loads_eliminate : 0.000008s : 0.00% optimize.opt_a.parameter_eliminate : 0.000004s : 0.00% optimize.opt_a.a_2 : 0.000343s : 0.09% optimize.opt_a.accelerated_algorithm : 0.000015s : 0.00% optimize.opt_a.shard : 0.000005s : 0.00% optimize.opt_a.meta_shard_fg_expand : 0.000008s : 0.00% optimize.opt_a.shard_inline : 0.000014s : 0.00% optimize.opt_a.merge_send_recv : 0.000016s : 0.00% optimize.opt_a.auto_parallel : 0.000015s : 0.00% optimize.opt_a.parallel : 0.000026s : 0.01% optimize.opt_a.flash_sp : 0.000014s : 0.00% optimize.opt_a.merge_comm : 0.000007s : 0.00% optimize.opt_a.allreduce_fusion : 0.000006s : 0.00% optimize.opt_a.matmul_add_comm_reduction : 0.000017s : 0.00% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000016s : 0.00% optimize.opt_a.virtual_dataset : 0.000013s : 0.00% optimize.opt_a.get_grad_eliminate_ : 0.000012s : 0.00% optimize.opt_a.virtual_output : 0.000012s : 0.00% optimize.opt_a.merge_forward : 0.000008s : 0.00% optimize.opt_a.cell_reuse_recompute_pass : 0.000004s : 0.00% optimize.opt_a.offload_activation : 0.000019s : 0.00% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000023s : 0.01% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.00% optimize.opt_a.before_grad : 0.000021s : 0.01% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000008s : 0.00% optimize.opt_a.meta_fg_expand : 0.000008s : 0.00% optimize.opt_a.flash_sp_send_recv_attached : 0.000004s : 0.00% optimize.opt_a.receive_attached : 0.000004s : 0.00% optimize.opt_a.after_resolve : 0.000026s : 0.01% optimize.opt_a.a_after_grad : 0.000021s : 0.01% optimize.opt_a.renormalize : 0.005754s : 1.46% optimize.opt_a.add_forward_monad_depend : 0.000008s : 0.00% optimize.opt_a.auto_monad_grad : 0.000004s : 0.00% optimize.opt_a.auto_monad_eliminator : 0.000029s : 0.01% optimize.opt_a.cse : 0.000045s : 0.01% optimize.opt_a.a_3 : 0.000090s : 0.02% optimize.py_interpret_to_execute_after_opt_a : 0.000015s : 0.00% optimize.slice_cell_reuse_recomputed_activation : 0.000003s : 0.00% optimize.rewriter_after_opt_a : 0.000037s : 0.01% optimize.convert_after_rewriter : 0.000006s : 0.00% optimize.order_py_execute_after_rewriter : 0.000006s : 0.00% optimize.mutable_eliminate : 0.000706s : 0.18% optimize.opt_b.b_1 : 0.000122s : 0.03% optimize.opt_b.b_2 : 0.000008s : 0.00% optimize.opt_b.updatestate_depend_eliminate : 0.000005s : 0.00% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.00% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000017s : 0.00% optimize.optimize_parallel_all_gather_comm : 0.000016s : 0.00% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000026s : 0.01% optimize.loop_unroll : 0.000480s : 0.12% optimize.opt_after_cconv.c_1 : 0.000033s : 0.01% optimize.opt_after_cconv.parameter_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000005s : 0.00% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.00% optimize.opt_after_cconv.cse : 0.000016s : 0.00% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000014s : 0.00% optimize.tuple_transform.d_1 : 0.000046s : 0.01% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.00% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000048s : 0.01% optimize.cse_after_recomputation.cse : 0.000011s : 0.00% optimize.environ_conv : 0.000005s : 0.00% optimize.swap_dp_allreduce_reducescatter : 0.000006s : 0.00% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000004s : 0.00% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000012s : 0.00% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.00% optimize.overlap_recompute_and_grad_model_parallel : 0.000004s : 0.00% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.00% optimize.overlap_grad_flash_sp : 0.000021s : 0.01% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000086s : 0.02% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000012s : 0.00% optimize.symbol_engine_optimizer.elim_not_effective : 0.000016s : 0.00% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.00% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000015s : 0.00% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000017s : 0.00% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.00% opt_after_jit_grad : 0.000489s : 0.12% validate : 0.000038s : 0.01% Time group info: ------[substitution.] 0.001397 325 2.98% : 0.000042s : 7: substitution.arithmetic_simplify 0.44% : 0.000006s : 4: substitution.depend_value_elim 0.30% : 0.000004s : 2: substitution.elim_not_effective 0.42% : 0.000006s : 2: substitution.fold_const_symbol 0.45% : 0.000006s : 4: substitution.graph_param_transform 60.56% : 0.000846s : 37: substitution.inline 0.26% : 0.000004s : 4: substitution.j_node_and_user_rematch 0.68% : 0.000009s : 8: substitution.minmaximum_grad 0.34% : 0.000005s : 4: substitution.remove_not_recompute_node 0.41% : 0.000006s : 4: substitution.replace_old_param 1.81% : 0.000025s : 10: substitution.switch_simplify 6.74% : 0.000094s : 42: substitution.tuple_list_convert_item_index_to_positive 3.78% : 0.000053s : 46: substitution.tuple_list_get_item_const_eliminator 7.43% : 0.000104s : 46: substitution.tuple_list_get_item_depend_reorder 8.52% : 0.000119s : 59: substitution.tuple_list_get_item_eliminator 4.89% : 0.000068s : 46: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.377587 2 97.46% : 0.368013s : 1: type_inference.infer 2.54% : 0.009574s : 1: type_inference.specialize ------[replace.] 0.000591 68 4.09% : 0.000024s : 4: replace.depend_value_elim 49.41% : 0.000292s : 37: replace.inline 19.89% : 0.000118s : 10: replace.switch_simplify 11.17% : 0.000066s : 4: replace.tuple_list_get_item_depend_reorder 15.44% : 0.000091s : 13: replace.tuple_list_get_item_eliminator ------[match.] 0.000919 68 0.45% : 0.000004s : 4: match.depend_value_elim 89.92% : 0.000826s : 37: match.inline 1.96% : 0.000018s : 10: match.switch_simplify 4.33% : 0.000040s : 4: match.tuple_list_get_item_depend_reorder 3.34% : 0.000031s : 13: match.tuple_list_get_item_eliminator ------[predicate.] 0.000777 4995 1.36% : 0.000011s : 74: predicate.accumulaten_eliminater 0.24% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.25% : 0.000002s : 16: predicate.addn_check_dump 1.48% : 0.000012s : 74: predicate.addn_zero_filter 1.28% : 0.000010s : 74: predicate.adjust_all_reduce_mul_add 2.29% : 0.000018s : 90: predicate.arithmetic_simplify 1.35% : 0.000011s : 74: predicate.cast_eliminate 0.15% : 0.000001s : 8: predicate.check_bprop_eliminate 0.26% : 0.000002s : 16: predicate.compare_switch_simplify 0.05% : 0.000000s : 4: predicate.const_output_eliminate 0.34% : 0.000003s : 18: predicate.depend_value_elim 1.43% : 0.000011s : 74: predicate.dict_get_item_const_eliminator 1.47% : 0.000011s : 74: predicate.dict_get_item_eliminator 1.32% : 0.000010s : 74: predicate.dict_set_item_eliminator 0.26% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.06% : 0.000000s : 4: predicate.elim_not_effective 0.11% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.44% : 0.000011s : 78: predicate.environ_add_const_eliminate 1.37% : 0.000011s : 78: predicate.environ_get_add_eliminate 1.35% : 0.000011s : 78: predicate.environ_get_depend_swap 1.65% : 0.000013s : 94: predicate.environ_get_eliminate 1.39% : 0.000011s : 78: predicate.environ_get_set_eliminate 2.39% : 0.000019s : 128: predicate.exchange_switch_depend_value 3.16% : 0.000025s : 128: predicate.float_depend_g_call 0.28% : 0.000002s : 16: predicate.float_environ_get_switch 0.34% : 0.000003s : 20: predicate.float_tuple_getitem_switch 0.05% : 0.000000s : 4: predicate.fold_const_symbol 0.17% : 0.000001s : 8: predicate.get_grad_eliminate 0.06% : 0.000000s : 4: predicate.graph_param_transform 0.30% : 0.000002s : 16: predicate.incorporate_call 0.26% : 0.000002s : 16: predicate.incorporate_call_switch 6.26% : 0.000049s : 234: predicate.inline 0.23% : 0.000002s : 8: predicate.inline_without_move 0.08% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.27% : 0.000002s : 8: predicate.less_batch_normalization 1.82% : 0.000014s : 99: predicate.list_to_tuple_eliminator_ 3.08% : 0.000024s : 173: predicate.load_eliminater 0.25% : 0.000002s : 4: predicate.loop_unroll_after_grad 3.33% : 0.000026s : 181: predicate.loop_unroll_before_grad 1.67% : 0.000013s : 86: predicate.make_slice_get_slice_eliminator 0.27% : 0.000002s : 16: predicate.merge_addn 0.13% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.15% : 0.000001s : 8: predicate.mini_step_allgather_replace 1.37% : 0.000011s : 74: predicate.minmaximum_grad 0.25% : 0.000002s : 4: predicate.mutable_eliminate 0.08% : 0.000001s : 4: predicate.opt_reshape 0.09% : 0.000001s : 4: predicate.parallel_virtual_node 3.04% : 0.000024s : 128: predicate.partial_defer_inline 1.92% : 0.000015s : 95: predicate.partial_eliminate 1.30% : 0.000010s : 74: predicate.print_const_string_wrapper 0.21% : 0.000002s : 12: predicate.reduce_all_const_elim 1.79% : 0.000014s : 74: predicate.reduce_eliminate 3.03% : 0.000024s : 173: predicate.redundant_stop_gradient_eliminater 0.14% : 0.000001s : 8: predicate.remove_not_recompute_node 1.36% : 0.000011s : 99: predicate.replace_applicator 0.15% : 0.000001s : 8: predicate.replace_old_param 0.07% : 0.000001s : 4: predicate.reset_defer_inline 1.38% : 0.000011s : 74: predicate.reshape_eliminate 0.16% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.13% : 0.000001s : 4: predicate.row_tensor_eliminate 0.20% : 0.000002s : 8: predicate.same_eliminate 0.19% : 0.000001s : 12: predicate.set_cell_output_no_recompute 0.20% : 0.000002s : 8: predicate.shard_identity_eliminate 0.25% : 0.000002s : 8: predicate.special_op_eliminate 0.36% : 0.000003s : 16: predicate.specialize_transform 0.24% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.23% : 0.000002s : 8: predicate.stack_unstack_eliminate 0.08% : 0.000001s : 4: predicate.switch_call_monad_eliminater 2.61% : 0.000020s : 128: predicate.switch_defer_inline 2.58% : 0.000020s : 136: predicate.switch_layer_defer_inline 6.85% : 0.000053s : 349: predicate.switch_simplify 1.29% : 0.000010s : 74: predicate.tile_eliminate 1.33% : 0.000010s : 74: predicate.transpose_eliminate 1.95% : 0.000015s : 82: predicate.tuple_list_convert_item_index_to_positive 1.97% : 0.000015s : 86: predicate.tuple_list_get_item_const_eliminator 1.79% : 0.000014s : 86: predicate.tuple_list_get_item_depend_reorder 3.13% : 0.000024s : 115: predicate.tuple_list_get_item_eliminator 1.80% : 0.000014s : 86: predicate.tuple_list_get_set_item_eliminator 2.25% : 0.000017s : 102: predicate.tuple_list_set_item_eliminator 1.74% : 0.000014s : 99: predicate.tuple_to_list_eliminator_ 3.06% : 0.000024s : 173: predicate.updatestate_pure_node_eliminater 3.35% : 0.000026s : 189: predicate.updatestate_useless_node_eliminater 0.09% : 0.000001s : 4: predicate.value_based_eliminate 0.16% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.17% : 0.000001s : 8: predicate.virtual_output_eliminate 0.08% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.10% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.009392 87 72.44% : 0.006803s : 46: func_graph_cloner_run.FuncGraphClonerGraph 27.56% : 0.002589s : 41: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.478618 209 0.00% : 0.000004s : 1: ForceFp32Comm 5.87% : 0.028104s : 1: add_attr 5.87% : 0.028092s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.01% : 0.000052s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.06% : 0.000287s : 1: auto_monad 0.00% : 0.000021s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000005s : 1: bias_add_comm_swap 0.09% : 0.000417s : 1: bootstrap 0.01% : 0.000030s : 1: cconv 0.00% : 0.000005s : 1: comm_op_add_attrs 0.00% : 0.000016s : 1: control_data_broadcast_order 0.00% : 0.000010s : 1: convert_after_rewriter 0.01% : 0.000024s : 1: cse_after_recomputation 0.00% : 0.000006s : 1: dataset_repeat_opt 0.00% : 0.000006s : 1: detach_backward 0.00% : 0.000008s : 1: environ_conv 0.02% : 0.000119s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000005s : 1: get_jit_bprop_graph 0.01% : 0.000031s : 1: graph_reusing 0.00% : 0.000006s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000006s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000007s : 1: label_micro_interleaved_index 0.10% : 0.000488s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.15% : 0.000715s : 1: mutable_eliminate 0.00% : 0.000007s : 1: offloading_packed_experts 0.00% : 0.000015s : 1: opt.transform.loop_unroll_optimizer 0.00% : 0.000015s : 1: opt.transform.mutable_eliminate 1.25% : 0.006000s : 95: opt.transform.opt_a 0.01% : 0.000032s : 1: opt.transform.opt_after_cconv 0.01% : 0.000026s : 1: opt.transform.opt_after_jit_grad 0.02% : 0.000101s : 28: opt.transform.opt_b 0.01% : 0.000051s : 2: opt.transform.opt_trans_graph 0.01% : 0.000047s : 4: opt.transform.symbol_engine_opt 2.66% : 0.012716s : 1: opt_a 0.02% : 0.000107s : 1: opt_after_cconv 0.10% : 0.000499s : 1: opt_after_jit_grad 0.04% : 0.000204s : 1: opt_b 3.24% : 0.015522s : 1: optimize 0.00% : 0.000020s : 1: optimize_parallel_all_gather_comm 0.00% : 0.000009s : 1: order_py_execute_after_rewriter 0.01% : 0.000024s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.00% : 0.000007s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000005s : 1: overlap_recompute_allgather_and_fa_grad 0.00% : 0.000007s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000005s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000007s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.03% : 0.000147s : 1: pre_auto_parallel 0.03% : 0.000126s : 1: py_interpret_to_execute 0.00% : 0.000018s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.00% : 0.000017s : 1: remove_dup_value 0.68% : 0.003238s : 1: renormalize.infer 0.52% : 0.002503s : 1: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000007s : 1: rewriter_after_jit_bprop_graph 0.01% : 0.000041s : 1: rewriter_after_opt_a 0.09% : 0.000418s : 1: rewriter_before_opt_a 0.00% : 0.000006s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000004s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.00% : 0.000009s : 1: swap_dp_allreduce_reducescatter 0.04% : 0.000178s : 1: symbol_engine_optimizer 0.02% : 0.000082s : 1: tuple_transform 78.94% : 0.377826s : 1: type_inference TotalTime = 0.0323496, [21] [bootstrap]: 0.00061986 [type_inference]: 0.02236 [event_method]: 5.194e-05 [auto_monad]: 7.367e-05 [graph_reusing]: 6.76999e-06 [inline]: 2.64999e-06 [add_attr]: 0.0033993, [1] [add_attr_with_inline]: 0.00339038, [1] [Cycle 1]: 7.585e-05, [2] [tag_attr]: 3.694e-05 [meta_addattr_fg_expand]: 6.97002e-06 [parallel-infer-symbol]: 3.18e-06 [pre_auto_parallel]: 4.149e-05 [insert-virtual-dataset]: 2.34999e-06 [parallel-infer-symbol-second]: 8.40024e-07 [dataset_repeat_opt]: 2.02999e-06 [pipeline_split]: 1.59e-06 [optimize]: 0.00507722, [53] [py_interpret_to_execute]: 2.982e-05 [rewriter_before_opt_a]: 0.00010235 [opt_a]: 0.00298627, [2] [Cycle 1]: 0.00235434, [45] [expand_dump_flag]: 3.31999e-06 [switch_simplify]: 5.18e-05 [loop_unroll]: 3.918e-05 [a_1]: 0.00087623 [with_stream_mark]: 1.408e-05 [recompute_prepare]: 8.31002e-06 [updatestate_depend_eliminate]: 3.85e-06 [updatestate_assign_eliminate]: 3.25e-06 [updatestate_loads_eliminate]: 3.45e-06 [parameter_eliminate]: 1.71e-06 [a_2]: 8.096e-05 [accelerated_algorithm]: 6.79999e-06 [shard]: 1.81998e-06 [meta_shard_fg_expand]: 1.94e-06 [shard_inline]: 6.16998e-06 [merge_send_recv]: 8.07e-06 [auto_parallel]: 6.02001e-06 [parallel]: 2.223e-05 [flash_sp]: 7.61999e-06 [merge_comm]: 3.93001e-06 [allreduce_fusion]: 3.31999e-06 [matmul_add_comm_reduction]: 8.87999e-06 [allreduce_slice_to_reducescatter]: 6.40022e-07 [virtual_shard_identity]: 7.78999e-06 [virtual_dataset]: 6.58e-06 [get_grad_eliminate_]: 5.76e-06 [virtual_output]: 5.87001e-06 [merge_forward]: 3.76999e-06 [cell_reuse_recompute_pass]: 1.18001e-06 [offload_activation]: 9.86998e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.151e-05 [merge_recompute_call_nodes]: 1.43002e-06 [before_grad]: 9.71e-06 [set_forward_comm_id_for_comm_node_pass]: 3.58999e-06 [meta_fg_expand]: 2.88e-06 [flash_sp_send_recv_attached]: 2.65002e-06 [receive_attached]: 2.34999e-06 [after_resolve]: 1.099e-05 [a_after_grad]: 9.39998e-06 [renormalize]: 0.00076232 [add_forward_monad_depend]: 5.29e-06 [auto_monad_grad]: 2.07001e-06 [auto_monad_eliminator]: 1.501e-05 [cse]: 2.857e-05 [a_3]: 4.32e-05 [Cycle 2]: 0.0006222, [45] [expand_dump_flag]: 1.25001e-06 [switch_simplify]: 7.23999e-06 [loop_unroll]: 5.97999e-06 [a_1]: 0.00013374 [with_stream_mark]: 1.098e-05 [recompute_prepare]: 6.29999e-06 [updatestate_depend_eliminate]: 3.08e-06 [updatestate_assign_eliminate]: 2.51998e-06 [updatestate_loads_eliminate]: 2.96001e-06 [parameter_eliminate]: 9.10019e-07 [a_2]: 7.128e-05 [accelerated_algorithm]: 6.09001e-06 [shard]: 1.19e-06 [meta_shard_fg_expand]: 1.30999e-06 [shard_inline]: 5.85002e-06 [merge_send_recv]: 4.29002e-06 [auto_parallel]: 5.68002e-06 [parallel]: 4.64002e-06 [flash_sp]: 3.48999e-06 [merge_comm]: 3.06999e-06 [allreduce_fusion]: 2.71999e-06 [matmul_add_comm_reduction]: 5.33002e-06 [allreduce_slice_to_reducescatter]: 3.89991e-07 [virtual_shard_identity]: 6.49001e-06 [virtual_dataset]: 5.54998e-06 [get_grad_eliminate_]: 5.24e-06 [virtual_output]: 5.16002e-06 [merge_forward]: 2.63e-06 [cell_reuse_recompute_pass]: 1.57999e-06 [offload_activation]: 6.26e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.018e-05 [merge_recompute_call_nodes]: 6.90023e-07 [before_grad]: 8.48999e-06 [set_forward_comm_id_for_comm_node_pass]: 3.73001e-06 [meta_fg_expand]: 1.99e-06 [flash_sp_send_recv_attached]: 9.90025e-07 [receive_attached]: 1.01002e-06 [after_resolve]: 1.021e-05 [a_after_grad]: 8.58001e-06 [renormalize]: 1.00001e-07 [add_forward_monad_depend]: 1.18001e-06 [auto_monad_grad]: 1.02998e-06 [auto_monad_eliminator]: 6.39999e-06 [cse]: 1.223e-05 [a_3]: 3.389e-05 [py_interpret_to_execute_after_opt_a]: 9.37001e-06 [slice_cell_reuse_recomputed_activation]: 2.24999e-06 [rewriter_after_opt_a]: 3.274e-05 [convert_after_rewriter]: 6.54001e-06 [order_py_execute_after_rewriter]: 5.75001e-06 [mutable_eliminate]: 0.00046586 [opt_b]: 0.00019297, [1] [Cycle 1]: 0.00018698, [7] [b_1]: 0.00011695 [b_2]: 8.3e-06 [updatestate_depend_eliminate]: 5.07e-06 [updatestate_assign_eliminate]: 2.53e-06 [updatestate_loads_eliminate]: 2.29001e-06 [renormalize]: 5.19998e-07 [cse]: 1.762e-05 [optimize_parallel_all_gather_comm]: 1.637e-05 [overlap_param_gather]: 1.68002e-06 [cconv]: 2.397e-05 [loop_unroll]: 0.00042431 [opt_after_cconv]: 9.827e-05, [1] [Cycle 1]: 9.274e-05, [7] [c_1]: 3.008e-05 [parameter_eliminate]: 2.37001e-06 [updatestate_depend_eliminate]: 5.20001e-06 [updatestate_assign_eliminate]: 3.23e-06 [updatestate_loads_eliminate]: 2.38998e-06 [cse]: 1.62e-05 [renormalize]: 5.10016e-07 [remove_dup_value]: 1.277e-05 [tuple_transform]: 7.248e-05, [1] [Cycle 1]: 6.779e-05, [4] [d_1]: 4.151e-05 [none_parameter_eliminate]: 1.50001e-06 [renormalize]: 1.69995e-07 [switch_simplify]: 6.62002e-06 [partial_unused_args_eliminate]: 1.86e-06 [add_recomputation]: 4.538e-05 [cse_after_recomputation]: 2.002e-05, [1] [Cycle 1]: 1.508e-05, [1] [cse]: 1.017e-05 [environ_conv]: 4.90001e-06 [swap_dp_allreduce_reducescatter]: 5.03002e-06 [bias_add_comm_swap]: 2.46998e-06 [label_micro_interleaved_index]: 4.32e-06 [label_fine_grained_interleaved_index]: 2.71999e-06 [merge_cast_opt]: 1.25001e-06 [slice_recompute_activation]: 2.34001e-06 [micro_interleaved_order_control]: 2.41e-06 [assign_add_opt]: 1.61998e-06 [ForceFp32Comm]: 1.07998e-06 [remove_cast_before_assign_add]: 1.44e-06 [full_micro_interleaved_order_control]: 2.04e-06 [reorder_send_recv_between_fp_bp]: 2.56998e-06 [comm_op_add_attrs]: 9.80013e-07 [add_comm_op_reuse_tag]: 9.39996e-07 [interleave_split_concat_branches]: 1.15001e-06 [interleave_parallel_branches]: 1.00001e-06 [overlap_opt_shard_in_pipeline]: 1.59998e-06 [overlap_opt_shard_grad_in_pipeline]: 1.92999e-06 [control_data_broadcast_order]: 1.278e-05 [grouped_pairwise_exchange_alltoall]: 2.01998e-06 [offloading_packed_experts]: 4.38001e-06 [overlap_recompute_and_grad_model_parallel]: 4.40999e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.61998e-06 [overlap_recompute_allgather_and_fa_grad]: 1.40999e-06 [overlap_recompute_comm]: 2.37999e-06 [overlap_grad_ring_attention]: 4.23001e-06 [overlap_grad_flash_sp]: 1.816e-05 [begin_end_overlap_inline]: 5.40022e-07 [split_matmul_comm_elemetwise]: 2.01e-06 [split_layernorm_comm]: 1.64e-06 [handle_group_info]: 1.09003e-06 [symbol_engine_optimizer]: 0.00018603, [1] [Cycle 1]: 0.00018148, [6] [build]: 0.00010135 [elim_shapecalc]: 1.253e-05 [elim_not_effective]: 1.463e-05 [opt_reshape]: 6.90998e-06 [fold_const_symbol]: 1.388e-05 [renormalize]: 2.49973e-07 [detach_backward]: 1.89999e-06 [pipeline_parallel_scheduler]: 1.70001e-06 [auto_monad_reorder]: 1.647e-05 [get_jit_bprop_graph]: 9.80013e-07 [rewriter_after_jit_bprop_graph]: 3.63999e-06 [opt_after_jit_grad]: 0.00046404 [validate]: 5.316e-05 Sums bootstrap : 0.000620s : 2.21% type_inference : 0.022360s : 79.88% event_method : 0.000052s : 0.19% auto_monad : 0.000074s : 0.26% graph_reusing : 0.000007s : 0.02% inline : 0.000003s : 0.01% add_attr.add_attr_with_inline.tag_attr : 0.000037s : 0.13% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.02% parallel-infer-symbol : 0.000003s : 0.01% pre_auto_parallel : 0.000041s : 0.15% insert-virtual-dataset : 0.000002s : 0.01% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.01% pipeline_split : 0.000002s : 0.01% optimize.py_interpret_to_execute : 0.000030s : 0.11% optimize.rewriter_before_opt_a : 0.000102s : 0.37% optimize.opt_a.expand_dump_flag : 0.000005s : 0.02% optimize.opt_a.switch_simplify : 0.000059s : 0.21% optimize.opt_a.loop_unroll : 0.000045s : 0.16% optimize.opt_a.a_1 : 0.001010s : 3.61% optimize.opt_a.with_stream_mark : 0.000025s : 0.09% optimize.opt_a.recompute_prepare : 0.000015s : 0.05% optimize.opt_a.updatestate_depend_eliminate : 0.000007s : 0.02% optimize.opt_a.updatestate_assign_eliminate : 0.000006s : 0.02% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.02% optimize.opt_a.parameter_eliminate : 0.000003s : 0.01% optimize.opt_a.a_2 : 0.000152s : 0.54% optimize.opt_a.accelerated_algorithm : 0.000013s : 0.05% optimize.opt_a.shard : 0.000003s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000003s : 0.01% optimize.opt_a.shard_inline : 0.000012s : 0.04% optimize.opt_a.merge_send_recv : 0.000012s : 0.04% optimize.opt_a.auto_parallel : 0.000012s : 0.04% optimize.opt_a.parallel : 0.000027s : 0.10% optimize.opt_a.flash_sp : 0.000011s : 0.04% optimize.opt_a.merge_comm : 0.000007s : 0.03% optimize.opt_a.allreduce_fusion : 0.000006s : 0.02% optimize.opt_a.matmul_add_comm_reduction : 0.000014s : 0.05% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000014s : 0.05% optimize.opt_a.virtual_dataset : 0.000012s : 0.04% optimize.opt_a.get_grad_eliminate_ : 0.000011s : 0.04% optimize.opt_a.virtual_output : 0.000011s : 0.04% optimize.opt_a.merge_forward : 0.000006s : 0.02% optimize.opt_a.cell_reuse_recompute_pass : 0.000003s : 0.01% optimize.opt_a.offload_activation : 0.000016s : 0.06% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000022s : 0.08% optimize.opt_a.merge_recompute_call_nodes : 0.000002s : 0.01% optimize.opt_a.before_grad : 0.000018s : 0.07% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000007s : 0.03% optimize.opt_a.meta_fg_expand : 0.000005s : 0.02% optimize.opt_a.flash_sp_send_recv_attached : 0.000004s : 0.01% optimize.opt_a.receive_attached : 0.000003s : 0.01% optimize.opt_a.after_resolve : 0.000021s : 0.08% optimize.opt_a.a_after_grad : 0.000018s : 0.06% optimize.opt_a.renormalize : 0.000762s : 2.72% optimize.opt_a.add_forward_monad_depend : 0.000006s : 0.02% optimize.opt_a.auto_monad_grad : 0.000003s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000021s : 0.08% optimize.opt_a.cse : 0.000041s : 0.15% optimize.opt_a.a_3 : 0.000077s : 0.28% optimize.py_interpret_to_execute_after_opt_a : 0.000009s : 0.03% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.01% optimize.rewriter_after_opt_a : 0.000033s : 0.12% optimize.convert_after_rewriter : 0.000007s : 0.02% optimize.order_py_execute_after_rewriter : 0.000006s : 0.02% optimize.mutable_eliminate : 0.000466s : 1.66% optimize.opt_b.b_1 : 0.000117s : 0.42% optimize.opt_b.b_2 : 0.000008s : 0.03% optimize.opt_b.updatestate_depend_eliminate : 0.000005s : 0.02% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_b.updatestate_loads_eliminate : 0.000002s : 0.01% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000018s : 0.06% optimize.optimize_parallel_all_gather_comm : 0.000016s : 0.06% optimize.overlap_param_gather : 0.000002s : 0.01% optimize.cconv : 0.000024s : 0.09% optimize.loop_unroll : 0.000424s : 1.52% optimize.opt_after_cconv.c_1 : 0.000030s : 0.11% optimize.opt_after_cconv.parameter_eliminate : 0.000002s : 0.01% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000005s : 0.02% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000002s : 0.01% optimize.opt_after_cconv.cse : 0.000016s : 0.06% optimize.opt_after_cconv.renormalize : 0.000001s : 0.00% optimize.remove_dup_value : 0.000013s : 0.05% optimize.tuple_transform.d_1 : 0.000042s : 0.15% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.01% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.02% optimize.partial_unused_args_eliminate : 0.000002s : 0.01% optimize.add_recomputation : 0.000045s : 0.16% optimize.cse_after_recomputation.cse : 0.000010s : 0.04% optimize.environ_conv : 0.000005s : 0.02% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.02% optimize.bias_add_comm_swap : 0.000002s : 0.01% optimize.label_micro_interleaved_index : 0.000004s : 0.02% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.01% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.01% optimize.micro_interleaved_order_control : 0.000002s : 0.01% optimize.assign_add_opt : 0.000002s : 0.01% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.01% optimize.full_micro_interleaved_order_control : 0.000002s : 0.01% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.01% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000002s : 0.01% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.01% optimize.control_data_broadcast_order : 0.000013s : 0.05% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.01% optimize.offloading_packed_experts : 0.000004s : 0.02% optimize.overlap_recompute_and_grad_model_parallel : 0.000004s : 0.02% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000002s : 0.01% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.01% optimize.overlap_recompute_comm : 0.000002s : 0.01% optimize.overlap_grad_ring_attention : 0.000004s : 0.02% optimize.overlap_grad_flash_sp : 0.000018s : 0.06% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.01% optimize.split_layernorm_comm : 0.000002s : 0.01% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000101s : 0.36% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000013s : 0.04% optimize.symbol_engine_optimizer.elim_not_effective : 0.000015s : 0.05% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.02% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000014s : 0.05% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.01% pipeline_parallel_scheduler : 0.000002s : 0.01% auto_monad_reorder : 0.000016s : 0.06% get_jit_bprop_graph : 0.000001s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.01% opt_after_jit_grad : 0.000464s : 1.66% validate : 0.000053s : 0.19% Time group info: ------[substitution.] 0.000276 40 11.84% : 0.000033s : 8: substitution.arithmetic_simplify 1.09% : 0.000003s : 2: substitution.elim_not_effective 1.96% : 0.000005s : 2: substitution.fold_const_symbol 2.05% : 0.000006s : 4: substitution.graph_param_transform 71.72% : 0.000198s : 6: substitution.inline 1.15% : 0.000003s : 4: substitution.j_node_and_user_rematch 1.68% : 0.000005s : 4: substitution.remove_not_recompute_node 1.52% : 0.000004s : 4: substitution.replace_old_param 6.99% : 0.000019s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.022293 2 95.37% : 0.021262s : 1: type_inference.infer 4.63% : 0.001032s : 1: type_inference.specialize ------[replace.] 0.000090 12 60.54% : 0.000054s : 6: replace.inline 39.46% : 0.000036s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000211 12 92.18% : 0.000194s : 6: match.inline 7.82% : 0.000017s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000232 1534 1.02% : 0.000002s : 17: predicate.accumulaten_eliminater 0.79% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.46% : 0.000001s : 8: predicate.addn_check_dump 1.10% : 0.000003s : 17: predicate.addn_zero_filter 0.92% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.22% : 0.000005s : 25: predicate.arithmetic_simplify 1.07% : 0.000002s : 17: predicate.cast_eliminate 0.55% : 0.000001s : 8: predicate.check_bprop_eliminate 0.42% : 0.000001s : 8: predicate.compare_switch_simplify 0.16% : 0.000000s : 4: predicate.const_output_eliminate 0.47% : 0.000001s : 8: predicate.depend_value_elim 0.97% : 0.000002s : 17: predicate.dict_get_item_const_eliminator 1.04% : 0.000002s : 17: predicate.dict_get_item_eliminator 0.93% : 0.000002s : 17: predicate.dict_set_item_eliminator 0.76% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.20% : 0.000000s : 4: predicate.elim_not_effective 0.42% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.16% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.18% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.14% : 0.000003s : 21: predicate.environ_get_depend_swap 1.73% : 0.000004s : 29: predicate.environ_get_eliminate 1.15% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.73% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.66% : 0.000006s : 29: predicate.float_depend_g_call 0.44% : 0.000001s : 8: predicate.float_environ_get_switch 0.69% : 0.000002s : 12: predicate.float_tuple_getitem_switch 0.16% : 0.000000s : 4: predicate.fold_const_symbol 0.50% : 0.000001s : 8: predicate.get_grad_eliminate 0.19% : 0.000000s : 4: predicate.graph_param_transform 0.48% : 0.000001s : 8: predicate.incorporate_call 0.38% : 0.000001s : 8: predicate.incorporate_call_switch 6.20% : 0.000014s : 70: predicate.inline 0.63% : 0.000001s : 8: predicate.inline_without_move 0.25% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.72% : 0.000002s : 8: predicate.less_batch_normalization 1.89% : 0.000004s : 31: predicate.list_to_tuple_eliminator_ 2.68% : 0.000006s : 48: predicate.load_eliminater 0.95% : 0.000002s : 4: predicate.loop_unroll_after_grad 2.71% : 0.000006s : 47: predicate.loop_unroll_before_grad 1.66% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.52% : 0.000001s : 8: predicate.merge_addn 0.47% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.45% : 0.000001s : 8: predicate.mini_step_allgather_replace 0.90% : 0.000002s : 17: predicate.minmaximum_grad 0.81% : 0.000002s : 4: predicate.mutable_eliminate 0.27% : 0.000001s : 4: predicate.opt_reshape 0.41% : 0.000001s : 4: predicate.parallel_virtual_node 2.20% : 0.000005s : 29: predicate.partial_defer_inline 1.69% : 0.000004s : 27: predicate.partial_eliminate 1.00% : 0.000002s : 17: predicate.print_const_string_wrapper 0.48% : 0.000001s : 8: predicate.reduce_all_const_elim 1.29% : 0.000003s : 17: predicate.reduce_eliminate 2.83% : 0.000007s : 48: predicate.redundant_stop_gradient_eliminater 0.34% : 0.000001s : 8: predicate.remove_not_recompute_node 1.47% : 0.000003s : 31: predicate.replace_applicator 0.40% : 0.000001s : 8: predicate.replace_old_param 0.30% : 0.000001s : 4: predicate.reset_defer_inline 0.95% : 0.000002s : 17: predicate.reshape_eliminate 0.53% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.31% : 0.000001s : 4: predicate.row_tensor_eliminate 0.57% : 0.000001s : 8: predicate.same_eliminate 0.35% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.65% : 0.000002s : 8: predicate.shard_identity_eliminate 0.55% : 0.000001s : 8: predicate.special_op_eliminate 0.62% : 0.000001s : 8: predicate.specialize_transform 0.70% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.57% : 0.000001s : 8: predicate.stack_unstack_eliminate 0.26% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.77% : 0.000004s : 29: predicate.switch_defer_inline 2.29% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.86% : 0.000014s : 88: predicate.switch_simplify 0.95% : 0.000002s : 17: predicate.tile_eliminate 1.06% : 0.000002s : 17: predicate.transpose_eliminate 1.56% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.56% : 0.000004s : 25: predicate.tuple_list_get_item_const_eliminator 1.48% : 0.000003s : 25: predicate.tuple_list_get_item_depend_reorder 3.27% : 0.000008s : 39: predicate.tuple_list_get_item_eliminator 1.49% : 0.000003s : 25: predicate.tuple_list_get_set_item_eliminator 2.05% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 1.85% : 0.000004s : 31: predicate.tuple_to_list_eliminator_ 2.64% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.36% : 0.000008s : 56: predicate.updatestate_useless_node_eliminater 0.43% : 0.000001s : 4: predicate.value_based_eliminate 0.63% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.49% : 0.000001s : 8: predicate.virtual_output_eliminate 0.26% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.30% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000801 15 56.91% : 0.000456s : 7: func_graph_cloner_run.FuncGraphClonerGraph 43.09% : 0.000345s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.043134 192 0.01% : 0.000004s : 1: ForceFp32Comm 7.89% : 0.003405s : 1: add_attr 7.87% : 0.003394s : 1: add_attr_with_inline 0.01% : 0.000004s : 1: add_comm_op_reuse_tag 0.11% : 0.000049s : 1: add_recomputation 0.01% : 0.000005s : 1: assign_add_opt 0.19% : 0.000080s : 1: auto_monad 0.05% : 0.000020s : 1: auto_monad_reorder 0.01% : 0.000004s : 1: begin_end_overlap_inline 0.01% : 0.000006s : 1: bias_add_comm_swap 1.51% : 0.000651s : 1: bootstrap 0.06% : 0.000028s : 1: cconv 0.01% : 0.000004s : 1: comm_op_add_attrs 0.04% : 0.000016s : 1: control_data_broadcast_order 0.02% : 0.000010s : 1: convert_after_rewriter 0.05% : 0.000023s : 1: cse_after_recomputation 0.01% : 0.000005s : 1: dataset_repeat_opt 0.01% : 0.000005s : 1: detach_backward 0.02% : 0.000008s : 1: environ_conv 0.14% : 0.000061s : 1: event_method 0.01% : 0.000005s : 1: full_micro_interleaved_order_control 0.01% : 0.000004s : 1: get_jit_bprop_graph 0.03% : 0.000011s : 1: graph_reusing 0.01% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.01% : 0.000004s : 1: handle_group_info 0.01% : 0.000006s : 1: inline 0.01% : 0.000006s : 1: insert-virtual-dataset 0.05% : 0.000021s : 1: interleave_parallel_branches 0.01% : 0.000004s : 1: interleave_split_concat_branches 0.01% : 0.000006s : 1: label_fine_grained_interleaved_index 0.02% : 0.000007s : 1: label_micro_interleaved_index 1.00% : 0.000432s : 1: loop_unroll 0.01% : 0.000005s : 1: merge_cast_opt 0.01% : 0.000005s : 1: micro_interleaved_order_control 1.10% : 0.000474s : 1: mutable_eliminate 0.02% : 0.000008s : 1: offloading_packed_experts 0.03% : 0.000014s : 1: opt.transform.loop_unroll_optimizer 0.03% : 0.000013s : 1: opt.transform.mutable_eliminate 3.32% : 0.001430s : 78: opt.transform.opt_a 0.07% : 0.000029s : 1: opt.transform.opt_after_cconv 0.05% : 0.000023s : 1: opt.transform.opt_after_jit_grad 0.22% : 0.000096s : 28: opt.transform.opt_b 0.11% : 0.000046s : 2: opt.transform.opt_trans_graph 0.10% : 0.000043s : 4: opt.transform.symbol_engine_opt 6.93% : 0.002989s : 1: opt_a 0.24% : 0.000102s : 1: opt_after_cconv 1.10% : 0.000473s : 1: opt_after_jit_grad 0.46% : 0.000196s : 1: opt_b 11.78% : 0.005081s : 1: optimize 0.05% : 0.000020s : 1: optimize_parallel_all_gather_comm 0.02% : 0.000009s : 1: order_py_execute_after_rewriter 0.05% : 0.000021s : 1: overlap_grad_flash_sp 0.01% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.02% : 0.000007s : 1: overlap_grad_ring_attention 0.01% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.01% : 0.000005s : 1: overlap_opt_shard_in_pipeline 0.01% : 0.000005s : 1: overlap_param_gather 0.01% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.02% : 0.000007s : 1: overlap_recompute_and_grad_model_parallel 0.01% : 0.000005s : 1: overlap_recompute_comm 0.02% : 0.000007s : 1: parallel-infer-symbol 0.01% : 0.000004s : 1: parallel-infer-symbol-second 0.01% : 0.000005s : 1: partial_unused_args_eliminate 0.01% : 0.000005s : 1: pipeline_parallel_scheduler 0.01% : 0.000005s : 1: pipeline_split 0.11% : 0.000046s : 1: pre_auto_parallel 0.08% : 0.000034s : 1: py_interpret_to_execute 0.03% : 0.000013s : 1: py_interpret_to_execute_after_opt_a 0.01% : 0.000004s : 1: remove_cast_before_assign_add 0.04% : 0.000016s : 1: remove_dup_value 0.91% : 0.000394s : 1: renormalize.infer 0.83% : 0.000360s : 1: renormalize.specialize 0.01% : 0.000005s : 1: reorder_send_recv_between_fp_bp 0.02% : 0.000007s : 1: rewriter_after_jit_bprop_graph 0.08% : 0.000036s : 1: rewriter_after_opt_a 0.25% : 0.000107s : 1: rewriter_before_opt_a 0.01% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.01% : 0.000005s : 1: slice_recompute_activation 0.01% : 0.000004s : 1: split_layernorm_comm 0.01% : 0.000005s : 1: split_matmul_comm_elemetwise 0.02% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.44% : 0.000189s : 1: symbol_engine_optimizer 0.17% : 0.000075s : 1: tuple_transform 51.88% : 0.022379s : 1: type_inference TotalTime = 0.0919273, [21] [bootstrap]: 0.0004431 [type_inference]: 0.016899 [event_method]: 8.757e-05 [auto_monad]: 0.0201711 [graph_reusing]: 1.704e-05 [inline]: 3.83001e-06 [add_attr]: 0.00354256, [1] [add_attr_with_inline]: 0.0035331, [1] [Cycle 1]: 0.00010041, [2] [tag_attr]: 5.236e-05 [meta_addattr_fg_expand]: 1.384e-05 [parallel-infer-symbol]: 3.38e-06 [pre_auto_parallel]: 6.977e-05 [insert-virtual-dataset]: 2.63e-06 [parallel-infer-symbol-second]: 8.2e-07 [dataset_repeat_opt]: 2.19001e-06 [pipeline_split]: 1.73002e-06 [optimize]: 0.0497274, [53] [py_interpret_to_execute]: 5.134e-05 [rewriter_before_opt_a]: 0.00022937 [opt_a]: 0.0452729, [3] [Cycle 1]: 0.0097999, [45] [expand_dump_flag]: 5.61e-06 [switch_simplify]: 0.00010391 [loop_unroll]: 8.752e-05 [a_1]: 0.00195912 [with_stream_mark]: 2.851e-05 [recompute_prepare]: 2.651e-05 [updatestate_depend_eliminate]: 1.102e-05 [updatestate_assign_eliminate]: 9.42999e-06 [updatestate_loads_eliminate]: 8.92e-06 [parameter_eliminate]: 2.73e-06 [a_2]: 0.00027621 [accelerated_algorithm]: 3.602e-05 [shard]: 2.01998e-06 [meta_shard_fg_expand]: 4.82e-06 [shard_inline]: 1.804e-05 [merge_send_recv]: 1.887e-05 [auto_parallel]: 1.388e-05 [parallel]: 1.885e-05 [flash_sp]: 1.057e-05 [merge_comm]: 1.224e-05 [allreduce_fusion]: 1.072e-05 [matmul_add_comm_reduction]: 2.943e-05 [allreduce_slice_to_reducescatter]: 7.00005e-07 [virtual_shard_identity]: 1.993e-05 [virtual_dataset]: 1.767e-05 [get_grad_eliminate_]: 1.717e-05 [virtual_output]: 1.713e-05 [merge_forward]: 1.131e-05 [cell_reuse_recompute_pass]: 1.20999e-06 [offload_activation]: 2.05e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.353e-05 [merge_recompute_call_nodes]: 1.40001e-06 [before_grad]: 3.241e-05 [set_forward_comm_id_for_comm_node_pass]: 1.148e-05 [meta_fg_expand]: 0.00182165 [flash_sp_send_recv_attached]: 6.36998e-06 [receive_attached]: 2.53998e-06 [after_resolve]: 6.874e-05 [a_after_grad]: 9.262e-05 [renormalize]: 0.00383322 [add_forward_monad_depend]: 1.253e-05 [auto_monad_grad]: 6.51999e-06 [auto_monad_eliminator]: 7.837e-05 [cse]: 0.00024238 [a_3]: 0.0004339 [Cycle 2]: 0.00532204, [45] [expand_dump_flag]: 3.06001e-06 [switch_simplify]: 5.992e-05 [loop_unroll]: 5.535e-05 [a_1]: 0.00210218 [with_stream_mark]: 2.971e-05 [recompute_prepare]: 2.508e-05 [updatestate_depend_eliminate]: 1.302e-05 [updatestate_assign_eliminate]: 1.138e-05 [updatestate_loads_eliminate]: 1.004e-05 [parameter_eliminate]: 2.19999e-06 [a_2]: 0.00029381 [accelerated_algorithm]: 2.524e-05 [shard]: 2.21998e-06 [meta_shard_fg_expand]: 6.44001e-06 [shard_inline]: 2.076e-05 [merge_send_recv]: 1.742e-05 [auto_parallel]: 1.858e-05 [parallel]: 8.80001e-06 [flash_sp]: 4.02e-06 [merge_comm]: 1.282e-05 [allreduce_fusion]: 1.207e-05 [matmul_add_comm_reduction]: 2.257e-05 [allreduce_slice_to_reducescatter]: 7.10017e-07 [virtual_shard_identity]: 2.426e-05 [virtual_dataset]: 1.8e-05 [get_grad_eliminate_]: 1.821e-05 [virtual_output]: 1.867e-05 [merge_forward]: 1.201e-05 [cell_reuse_recompute_pass]: 1.54998e-06 [offload_activation]: 2.113e-05 [cell_reuse_handle_not_recompute_node_pass]: 4.305e-05 [merge_recompute_call_nodes]: 1.67001e-06 [before_grad]: 3.907e-05 [set_forward_comm_id_for_comm_node_pass]: 1.423e-05 [meta_fg_expand]: 0.00019255 [flash_sp_send_recv_attached]: 2.56e-06 [receive_attached]: 2.71999e-06 [after_resolve]: 3.437e-05 [a_after_grad]: 3.098e-05 [renormalize]: 0.00147442 [add_forward_monad_depend]: 6.53998e-06 [auto_monad_grad]: 2.66e-06 [auto_monad_eliminator]: 3.673e-05 [cse]: 0.00011078 [a_3]: 0.00013976 [Cycle 3]: 0.0301312, [45] [expand_dump_flag]: 2.29001e-06 [switch_simplify]: 2.16e-05 [loop_unroll]: 1.832e-05 [a_1]: 0.0005896 [with_stream_mark]: 2.426e-05 [recompute_prepare]: 2.233e-05 [updatestate_depend_eliminate]: 1.224e-05 [updatestate_assign_eliminate]: 1.104e-05 [updatestate_loads_eliminate]: 9.68002e-06 [parameter_eliminate]: 1.54e-06 [a_2]: 0.0283873 [accelerated_algorithm]: 3.144e-05 [shard]: 3.12002e-06 [meta_shard_fg_expand]: 8.74e-06 [shard_inline]: 2.058e-05 [merge_send_recv]: 2.559e-05 [auto_parallel]: 2.265e-05 [parallel]: 9.87001e-06 [flash_sp]: 2.16998e-06 [merge_comm]: 2.774e-05 [allreduce_fusion]: 1.329e-05 [matmul_add_comm_reduction]: 3.019e-05 [allreduce_slice_to_reducescatter]: 7.10017e-07 [virtual_shard_identity]: 2.393e-05 [virtual_dataset]: 1.97e-05 [get_grad_eliminate_]: 1.906e-05 [virtual_output]: 1.866e-05 [merge_forward]: 1.226e-05 [cell_reuse_recompute_pass]: 3.56999e-06 [offload_activation]: 2.384e-05 [cell_reuse_handle_not_recompute_node_pass]: 4.587e-05 [merge_recompute_call_nodes]: 1.74e-06 [before_grad]: 3.54e-05 [set_forward_comm_id_for_comm_node_pass]: 1.386e-05 [meta_fg_expand]: 9.38002e-06 [flash_sp_send_recv_attached]: 2.18998e-06 [receive_attached]: 2.47001e-06 [after_resolve]: 2.685e-05 [a_after_grad]: 3.067e-05 [renormalize]: 8.00064e-08 [add_forward_monad_depend]: 4.35e-06 [auto_monad_grad]: 2.52001e-06 [auto_monad_eliminator]: 3.629e-05 [cse]: 8.346e-05 [a_3]: 0.00012993 [py_interpret_to_execute_after_opt_a]: 3.218e-05 [slice_cell_reuse_recomputed_activation]: 2.14e-06 [rewriter_after_opt_a]: 0.00017517 [convert_after_rewriter]: 1.946e-05 [order_py_execute_after_rewriter]: 1.27e-05 [mutable_eliminate]: 0.00076502 [opt_b]: 0.00068847, [1] [Cycle 1]: 0.00067769, [7] [b_1]: 0.00047188 [b_2]: 2.592e-05 [updatestate_depend_eliminate]: 1.818e-05 [updatestate_assign_eliminate]: 1.074e-05 [updatestate_loads_eliminate]: 1.104e-05 [renormalize]: 7.00005e-07 [cse]: 8.763e-05 [optimize_parallel_all_gather_comm]: 4.134e-05 [overlap_param_gather]: 2.00002e-06 [cconv]: 3.926e-05 [loop_unroll]: 0.00051287 [opt_after_cconv]: 0.00026591, [1] [Cycle 1]: 0.0002585, [7] [c_1]: 0.00010184 [parameter_eliminate]: 4.62e-06 [updatestate_depend_eliminate]: 1.787e-05 [updatestate_assign_eliminate]: 1.073e-05 [updatestate_loads_eliminate]: 1.059e-05 [cse]: 7.211e-05 [renormalize]: 3.50003e-07 [remove_dup_value]: 0.00010474 [tuple_transform]: 0.00025443, [1] [Cycle 1]: 0.00024835, [4] [d_1]: 0.0001964 [none_parameter_eliminate]: 2.49001e-06 [renormalize]: 3.59985e-07 [switch_simplify]: 2.211e-05 [partial_unused_args_eliminate]: 2.44001e-06 [add_recomputation]: 0.0001362 [cse_after_recomputation]: 7.034e-05, [1] [Cycle 1]: 6.396e-05, [1] [cse]: 5.659e-05 [environ_conv]: 1.918e-05 [swap_dp_allreduce_reducescatter]: 1.833e-05 [bias_add_comm_swap]: 2.88e-06 [label_micro_interleaved_index]: 5.22e-06 [label_fine_grained_interleaved_index]: 3.08e-06 [merge_cast_opt]: 1.25001e-06 [slice_recompute_activation]: 2.01998e-06 [micro_interleaved_order_control]: 2.39001e-06 [assign_add_opt]: 1.40001e-06 [ForceFp32Comm]: 8.80013e-07 [remove_cast_before_assign_add]: 1.03001e-06 [full_micro_interleaved_order_control]: 2.34001e-06 [reorder_send_recv_between_fp_bp]: 2.56998e-06 [comm_op_add_attrs]: 1.10999e-06 [add_comm_op_reuse_tag]: 9.80013e-07 [interleave_split_concat_branches]: 1.12999e-06 [interleave_parallel_branches]: 1.23002e-06 [overlap_opt_shard_in_pipeline]: 1.24998e-06 [overlap_opt_shard_grad_in_pipeline]: 1.72001e-06 [control_data_broadcast_order]: 3.707e-05 [grouped_pairwise_exchange_alltoall]: 1.91e-06 [offloading_packed_experts]: 1.073e-05 [overlap_recompute_and_grad_model_parallel]: 1.02e-05 [overlap_grad_matmul_and_grad_allreduce]: 1.58002e-06 [overlap_recompute_allgather_and_fa_grad]: 1.45999e-06 [overlap_recompute_comm]: 2.63e-06 [overlap_grad_ring_attention]: 9.25001e-06 [overlap_grad_flash_sp]: 5.154e-05 [begin_end_overlap_inline]: 7.39994e-07 [split_matmul_comm_elemetwise]: 2.49001e-06 [split_layernorm_comm]: 1.74e-06 [handle_group_info]: 1.00999e-06 [symbol_engine_optimizer]: 0.00058307, [1] [Cycle 1]: 0.00057659, [6] [build]: 0.00032023 [elim_shapecalc]: 6.857e-05 [elim_not_effective]: 4.994e-05 [opt_reshape]: 3.031e-05 [fold_const_symbol]: 5.422e-05 [renormalize]: 3.10014e-07 [detach_backward]: 3.36001e-06 [pipeline_parallel_scheduler]: 1.74998e-06 [auto_monad_reorder]: 4.794e-05 [get_jit_bprop_graph]: 2.24999e-06 [rewriter_after_jit_bprop_graph]: 5.24e-06 [opt_after_jit_grad]: 0.00060887 [validate]: 7.764e-05 Sums bootstrap : 0.000443s : 0.51% type_inference : 0.016899s : 19.47% event_method : 0.000088s : 0.10% auto_monad : 0.020171s : 23.23% graph_reusing : 0.000017s : 0.02% inline : 0.000004s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000052s : 0.06% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000014s : 0.02% parallel-infer-symbol : 0.000003s : 0.00% pre_auto_parallel : 0.000070s : 0.08% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000051s : 0.06% optimize.rewriter_before_opt_a : 0.000229s : 0.26% optimize.opt_a.expand_dump_flag : 0.000011s : 0.01% optimize.opt_a.switch_simplify : 0.000185s : 0.21% optimize.opt_a.loop_unroll : 0.000161s : 0.19% optimize.opt_a.a_1 : 0.004651s : 5.36% optimize.opt_a.with_stream_mark : 0.000082s : 0.10% optimize.opt_a.recompute_prepare : 0.000074s : 0.09% optimize.opt_a.updatestate_depend_eliminate : 0.000036s : 0.04% optimize.opt_a.updatestate_assign_eliminate : 0.000032s : 0.04% optimize.opt_a.updatestate_loads_eliminate : 0.000029s : 0.03% optimize.opt_a.parameter_eliminate : 0.000006s : 0.01% optimize.opt_a.a_2 : 0.028957s : 33.36% optimize.opt_a.accelerated_algorithm : 0.000093s : 0.11% optimize.opt_a.shard : 0.000007s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000020s : 0.02% optimize.opt_a.shard_inline : 0.000059s : 0.07% optimize.opt_a.merge_send_recv : 0.000062s : 0.07% optimize.opt_a.auto_parallel : 0.000055s : 0.06% optimize.opt_a.parallel : 0.000038s : 0.04% optimize.opt_a.flash_sp : 0.000017s : 0.02% optimize.opt_a.merge_comm : 0.000053s : 0.06% optimize.opt_a.allreduce_fusion : 0.000036s : 0.04% optimize.opt_a.matmul_add_comm_reduction : 0.000082s : 0.09% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000068s : 0.08% optimize.opt_a.virtual_dataset : 0.000055s : 0.06% optimize.opt_a.get_grad_eliminate_ : 0.000054s : 0.06% optimize.opt_a.virtual_output : 0.000054s : 0.06% optimize.opt_a.merge_forward : 0.000036s : 0.04% optimize.opt_a.cell_reuse_recompute_pass : 0.000006s : 0.01% optimize.opt_a.offload_activation : 0.000065s : 0.08% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000122s : 0.14% optimize.opt_a.merge_recompute_call_nodes : 0.000005s : 0.01% optimize.opt_a.before_grad : 0.000107s : 0.12% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000040s : 0.05% optimize.opt_a.meta_fg_expand : 0.002024s : 2.33% optimize.opt_a.flash_sp_send_recv_attached : 0.000011s : 0.01% optimize.opt_a.receive_attached : 0.000008s : 0.01% optimize.opt_a.after_resolve : 0.000130s : 0.15% optimize.opt_a.a_after_grad : 0.000154s : 0.18% optimize.opt_a.renormalize : 0.005308s : 6.11% optimize.opt_a.add_forward_monad_depend : 0.000023s : 0.03% optimize.opt_a.auto_monad_grad : 0.000012s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000151s : 0.17% optimize.opt_a.cse : 0.000437s : 0.50% optimize.opt_a.a_3 : 0.000704s : 0.81% optimize.py_interpret_to_execute_after_opt_a : 0.000032s : 0.04% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000175s : 0.20% optimize.convert_after_rewriter : 0.000019s : 0.02% optimize.order_py_execute_after_rewriter : 0.000013s : 0.01% optimize.mutable_eliminate : 0.000765s : 0.88% optimize.opt_b.b_1 : 0.000472s : 0.54% optimize.opt_b.b_2 : 0.000026s : 0.03% optimize.opt_b.updatestate_depend_eliminate : 0.000018s : 0.02% optimize.opt_b.updatestate_assign_eliminate : 0.000011s : 0.01% optimize.opt_b.updatestate_loads_eliminate : 0.000011s : 0.01% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000088s : 0.10% optimize.optimize_parallel_all_gather_comm : 0.000041s : 0.05% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000039s : 0.05% optimize.loop_unroll : 0.000513s : 0.59% optimize.opt_after_cconv.c_1 : 0.000102s : 0.12% optimize.opt_after_cconv.parameter_eliminate : 0.000005s : 0.01% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000018s : 0.02% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000011s : 0.01% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000011s : 0.01% optimize.opt_after_cconv.cse : 0.000072s : 0.08% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000105s : 0.12% optimize.tuple_transform.d_1 : 0.000196s : 0.23% optimize.tuple_transform.none_parameter_eliminate : 0.000002s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000022s : 0.03% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000136s : 0.16% optimize.cse_after_recomputation.cse : 0.000057s : 0.07% optimize.environ_conv : 0.000019s : 0.02% optimize.swap_dp_allreduce_reducescatter : 0.000018s : 0.02% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000005s : 0.01% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000001s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000002s : 0.00% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000037s : 0.04% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000011s : 0.01% optimize.overlap_recompute_and_grad_model_parallel : 0.000010s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000002s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000003s : 0.00% optimize.overlap_grad_ring_attention : 0.000009s : 0.01% optimize.overlap_grad_flash_sp : 0.000052s : 0.06% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000320s : 0.37% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000069s : 0.08% optimize.symbol_engine_optimizer.elim_not_effective : 0.000050s : 0.06% optimize.symbol_engine_optimizer.opt_reshape : 0.000030s : 0.03% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000054s : 0.06% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000048s : 0.06% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000005s : 0.01% opt_after_jit_grad : 0.000609s : 0.70% validate : 0.000078s : 0.09% Time group info: ------[substitution.] 0.001394 367 5.97% : 0.000083s : 16: substitution.arithmetic_simplify 1.02% : 0.000014s : 15: substitution.elim_not_effective 1.93% : 0.000027s : 2: substitution.elim_shapecalc_of_broadcastargs 0.40% : 0.000006s : 6: substitution.float_depend_g_call 1.31% : 0.000018s : 10: substitution.float_tuple_getitem_switch 1.70% : 0.000024s : 15: substitution.fold_const_symbol 1.12% : 0.000016s : 17: substitution.graph_param_transform 0.22% : 0.000003s : 2: substitution.incorporate_call 0.14% : 0.000002s : 2: substitution.incorporate_call_switch 49.84% : 0.000695s : 20: substitution.inline 1.40% : 0.000020s : 2: substitution.inline_without_move 1.53% : 0.000021s : 42: substitution.j_node_and_user_rematch 1.57% : 0.000022s : 3: substitution.less_batch_normalization 1.59% : 0.000022s : 14: substitution.minmaximum_grad 0.41% : 0.000006s : 2: substitution.opt_reshape 0.52% : 0.000007s : 6: substitution.partial_eliminate 0.69% : 0.000010s : 4: substitution.reduce_eliminate 2.10% : 0.000029s : 42: substitution.remove_not_recompute_node 2.17% : 0.000030s : 10: substitution.replace_applicator 0.95% : 0.000013s : 12: substitution.replace_old_param 1.29% : 0.000018s : 6: substitution.reshape_eliminate 0.21% : 0.000003s : 1: substitution.set_cell_output_no_recompute 4.31% : 0.000060s : 18: substitution.tuple_list_convert_item_index_to_positive 1.94% : 0.000027s : 18: substitution.tuple_list_get_item_const_eliminator 2.84% : 0.000040s : 18: substitution.tuple_list_get_item_depend_reorder 9.27% : 0.000129s : 46: substitution.tuple_list_get_item_eliminator 3.56% : 0.000050s : 18: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.016799 2 87.14% : 0.014638s : 1: type_inference.infer 12.86% : 0.002160s : 1: type_inference.specialize ------[replace.] 0.000321 40 56.74% : 0.000182s : 20: replace.inline 43.26% : 0.000139s : 20: replace.tuple_list_get_item_eliminator ------[match.] 0.000727 40 93.82% : 0.000683s : 20: match.inline 6.18% : 0.000045s : 20: match.tuple_list_get_item_eliminator ------[predicate.] 0.001214 8180 0.96% : 0.000012s : 89: predicate.accumulaten_eliminater 0.46% : 0.000006s : 17: predicate.ad_related_special_op_eliminate 0.59% : 0.000007s : 51: predicate.addn_check_dump 1.02% : 0.000012s : 89: predicate.addn_zero_filter 0.97% : 0.000012s : 89: predicate.adjust_all_reduce_mul_add 2.34% : 0.000028s : 140: predicate.arithmetic_simplify 1.02% : 0.000012s : 89: predicate.cast_eliminate 1.12% : 0.000014s : 96: predicate.check_bprop_eliminate 0.60% : 0.000007s : 51: predicate.compare_switch_simplify 0.11% : 0.000001s : 17: predicate.const_output_eliminate 0.59% : 0.000007s : 51: predicate.depend_value_elim 1.06% : 0.000013s : 89: predicate.dict_get_item_const_eliminator 1.10% : 0.000013s : 89: predicate.dict_get_item_eliminator 0.98% : 0.000012s : 89: predicate.dict_set_item_eliminator 0.49% : 0.000006s : 34: predicate.dumpgradient_eliminate 0.12% : 0.000001s : 17: predicate.elim_not_effective 0.30% : 0.000004s : 17: predicate.elim_shapecalc_of_broadcastargs 1.20% : 0.000015s : 106: predicate.environ_add_const_eliminate 1.20% : 0.000015s : 106: predicate.environ_get_add_eliminate 1.15% : 0.000014s : 106: predicate.environ_get_depend_swap 1.81% : 0.000022s : 157: predicate.environ_get_eliminate 1.16% : 0.000014s : 106: predicate.environ_get_set_eliminate 1.45% : 0.000018s : 129: predicate.exchange_switch_depend_value 2.01% : 0.000024s : 129: predicate.float_depend_g_call 0.59% : 0.000007s : 51: predicate.float_environ_get_switch 0.86% : 0.000010s : 68: predicate.float_tuple_getitem_switch 0.10% : 0.000001s : 17: predicate.fold_const_symbol 0.62% : 0.000008s : 51: predicate.get_grad_eliminate 0.13% : 0.000002s : 17: predicate.graph_param_transform 0.62% : 0.000008s : 51: predicate.incorporate_call 0.55% : 0.000007s : 51: predicate.incorporate_call_switch 5.49% : 0.000067s : 354: predicate.inline 1.19% : 0.000014s : 74: predicate.inline_without_move 0.30% : 0.000004s : 51: predicate.j_node_and_user_rematch 0.78% : 0.000009s : 51: predicate.less_batch_normalization 1.63% : 0.000020s : 143: predicate.list_to_tuple_eliminator_ 2.51% : 0.000031s : 232: predicate.load_eliminater 0.47% : 0.000006s : 17: predicate.loop_unroll_after_grad 2.04% : 0.000025s : 179: predicate.loop_unroll_before_grad 1.45% : 0.000018s : 123: predicate.make_slice_get_slice_eliminator 0.89% : 0.000011s : 51: predicate.merge_addn 1.06% : 0.000013s : 96: predicate.micro_step_allgather_replace 1.06% : 0.000013s : 96: predicate.mini_step_allgather_replace 1.01% : 0.000012s : 89: predicate.minmaximum_grad 0.50% : 0.000006s : 17: predicate.mutable_eliminate 0.23% : 0.000003s : 17: predicate.opt_reshape 0.24% : 0.000003s : 17: predicate.parallel_virtual_node 1.83% : 0.000022s : 129: predicate.partial_defer_inline 1.59% : 0.000019s : 126: predicate.partial_eliminate 0.98% : 0.000012s : 89: predicate.print_const_string_wrapper 0.59% : 0.000007s : 51: predicate.reduce_all_const_elim 1.30% : 0.000016s : 89: predicate.reduce_eliminate 2.54% : 0.000031s : 232: predicate.redundant_stop_gradient_eliminater 0.33% : 0.000004s : 51: predicate.remove_not_recompute_node 1.59% : 0.000019s : 205: predicate.replace_applicator 0.51% : 0.000006s : 74: predicate.replace_old_param 0.12% : 0.000001s : 17: predicate.reset_defer_inline 1.09% : 0.000013s : 89: predicate.reshape_eliminate 1.10% : 0.000013s : 96: predicate.row_tensor_add_zeros_like 0.23% : 0.000003s : 17: predicate.row_tensor_eliminate 1.19% : 0.000014s : 96: predicate.same_eliminate 0.37% : 0.000005s : 51: predicate.set_cell_output_no_recompute 0.70% : 0.000008s : 51: predicate.shard_identity_eliminate 0.47% : 0.000006s : 34: predicate.special_op_eliminate 1.13% : 0.000014s : 51: predicate.specialize_transform 1.38% : 0.000017s : 96: predicate.split_environ_get_set_with_tuple_value 1.06% : 0.000013s : 74: predicate.stack_unstack_eliminate 0.22% : 0.000003s : 17: predicate.switch_call_monad_eliminater 1.56% : 0.000019s : 129: predicate.switch_defer_inline 2.61% : 0.000032s : 225: predicate.switch_layer_defer_inline 4.49% : 0.000055s : 376: predicate.switch_simplify 0.96% : 0.000012s : 89: predicate.tile_eliminate 0.98% : 0.000012s : 89: predicate.transpose_eliminate 1.66% : 0.000020s : 123: predicate.tuple_list_convert_item_index_to_positive 1.67% : 0.000020s : 123: predicate.tuple_list_get_item_const_eliminator 1.51% : 0.000018s : 123: predicate.tuple_list_get_item_depend_reorder 3.11% : 0.000038s : 194: predicate.tuple_list_get_item_eliminator 1.56% : 0.000019s : 123: predicate.tuple_list_get_set_item_eliminator 2.19% : 0.000027s : 174: predicate.tuple_list_set_item_eliminator 1.61% : 0.000020s : 143: predicate.tuple_to_list_eliminator_ 2.53% : 0.000031s : 232: predicate.updatestate_pure_node_eliminater 3.17% : 0.000038s : 283: predicate.updatestate_useless_node_eliminater 0.22% : 0.000003s : 17: predicate.value_based_eliminate 0.63% : 0.000008s : 51: predicate.virtual_dataset_eliminate 0.64% : 0.000008s : 51: predicate.virtual_output_eliminate 0.22% : 0.000003s : 17: predicate.virtual_view_grad_eliminate 0.23% : 0.000003s : 17: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002355 43 57.47% : 0.001353s : 19: func_graph_cloner_run.FuncGraphClonerGraph 42.53% : 0.001001s : 24: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.186857 233 0.00% : 0.000004s : 1: ForceFp32Comm 1.90% : 0.003548s : 1: add_attr 1.89% : 0.003537s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.08% : 0.000142s : 1: add_recomputation 0.00% : 0.000005s : 1: assign_add_opt 10.81% : 0.020199s : 1: auto_monad 0.03% : 0.000053s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 0.25% : 0.000471s : 1: bootstrap 0.02% : 0.000043s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.02% : 0.000042s : 1: control_data_broadcast_order 0.01% : 0.000024s : 1: convert_after_rewriter 0.04% : 0.000075s : 1: cse_after_recomputation 0.00% : 0.000005s : 1: dataset_repeat_opt 0.00% : 0.000009s : 1: detach_backward 0.01% : 0.000023s : 1: environ_conv 0.05% : 0.000096s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000006s : 1: get_jit_bprop_graph 0.01% : 0.000024s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.00% : 0.000007s : 1: inline 0.00% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000006s : 1: label_fine_grained_interleaved_index 0.00% : 0.000008s : 1: label_micro_interleaved_index 0.28% : 0.000523s : 1: loop_unroll 0.00% : 0.000004s : 1: merge_cast_opt 0.00% : 0.000005s : 1: micro_interleaved_order_control 0.41% : 0.000774s : 1: mutable_eliminate 0.01% : 0.000014s : 1: offloading_packed_experts 0.02% : 0.000035s : 1: opt.transform.loop_unroll_optimizer 0.02% : 0.000036s : 1: opt.transform.mutable_eliminate 18.97% : 0.035439s : 117: opt.transform.opt_a 0.05% : 0.000100s : 1: opt.transform.opt_after_cconv 0.04% : 0.000071s : 1: opt.transform.opt_after_jit_grad 0.25% : 0.000463s : 28: opt.transform.opt_b 0.11% : 0.000214s : 2: opt.transform.opt_trans_graph 0.11% : 0.000197s : 4: opt.transform.symbol_engine_opt 24.23% : 0.045277s : 1: opt_a 0.14% : 0.000270s : 1: opt_after_cconv 0.33% : 0.000620s : 1: opt_after_jit_grad 0.37% : 0.000693s : 1: opt_b 26.62% : 0.049733s : 1: optimize 0.02% : 0.000046s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000016s : 1: order_py_execute_after_rewriter 0.03% : 0.000056s : 1: overlap_grad_flash_sp 0.00% : 0.000005s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000012s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000005s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000013s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000006s : 1: overlap_recompute_comm 0.00% : 0.000007s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000006s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.04% : 0.000075s : 1: pre_auto_parallel 0.03% : 0.000056s : 1: py_interpret_to_execute 0.02% : 0.000037s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.06% : 0.000112s : 1: remove_dup_value 1.64% : 0.003073s : 2: renormalize.infer 1.19% : 0.002215s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000009s : 1: rewriter_after_jit_bprop_graph 0.10% : 0.000182s : 1: rewriter_after_opt_a 0.13% : 0.000235s : 1: rewriter_before_opt_a 0.00% : 0.000006s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000005s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000005s : 1: split_matmul_comm_elemetwise 0.01% : 0.000022s : 1: swap_dp_allreduce_reducescatter 0.31% : 0.000586s : 1: symbol_engine_optimizer 0.14% : 0.000258s : 1: tuple_transform 9.06% : 0.016922s : 1: type_inference [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:45:38.641.46 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. [WARNING] ME(165502:281472900173616,MainProcess):2026-01-29-17:45:38.644.66 [mindspore/common/tensor.py:142] For 'Tensor', if 'dtype' is not None, 'input_data', 'shape' or 'init' must not be None. TotalTime = 0.0543085, [21] [bootstrap]: 0.0004029 [type_inference]: 0.035751 [event_method]: 4.996e-05 [auto_monad]: 7.539e-05 [graph_reusing]: 6.73e-06 [inline]: 2.83e-06 [add_attr]: 0.00337048, [1] [add_attr_with_inline]: 0.00336148, [1] [Cycle 1]: 6.403e-05, [2] [tag_attr]: 2.767e-05 [meta_addattr_fg_expand]: 6.98e-06 [parallel-infer-symbol]: 3.93001e-06 [pre_auto_parallel]: 4.228e-05 [insert-virtual-dataset]: 2.96999e-06 [parallel-infer-symbol-second]: 7.89994e-07 [dataset_repeat_opt]: 2.12999e-06 [pipeline_split]: 2.06e-06 [optimize]: 0.0138634, [53] [py_interpret_to_execute]: 3.002e-05 [rewriter_before_opt_a]: 0.00010541 [opt_a]: 0.011639, [2] [Cycle 1]: 0.0109359, [45] [expand_dump_flag]: 3.08e-06 [switch_simplify]: 4.765e-05 [loop_unroll]: 4.487e-05 [a_1]: 0.00088638 [with_stream_mark]: 1.566e-05 [recompute_prepare]: 1.066e-05 [updatestate_depend_eliminate]: 4.31002e-06 [updatestate_assign_eliminate]: 3.2e-06 [updatestate_loads_eliminate]: 2.73998e-06 [parameter_eliminate]: 1.62999e-06 [a_2]: 8.228e-05 [accelerated_algorithm]: 7.55998e-06 [shard]: 2.21998e-06 [meta_shard_fg_expand]: 2.09999e-06 [shard_inline]: 6.22001e-06 [merge_send_recv]: 8.35001e-06 [auto_parallel]: 7.7e-06 [parallel]: 1.828e-05 [flash_sp]: 9.02e-06 [merge_comm]: 3.56001e-06 [allreduce_fusion]: 3.41999e-06 [matmul_add_comm_reduction]: 7.66999e-06 [allreduce_slice_to_reducescatter]: 6.30011e-07 [virtual_shard_identity]: 7.92998e-06 [virtual_dataset]: 6.23002e-06 [get_grad_eliminate_]: 5.56002e-06 [virtual_output]: 6.76e-06 [merge_forward]: 4.43001e-06 [cell_reuse_recompute_pass]: 1.10001e-06 [offload_activation]: 9.69999e-06 [cell_reuse_handle_not_recompute_node_pass]: 1.283e-05 [merge_recompute_call_nodes]: 1.72001e-06 [before_grad]: 1.05e-05 [set_forward_comm_id_for_comm_node_pass]: 4.03999e-06 [meta_fg_expand]: 2.96999e-06 [flash_sp_send_recv_attached]: 2.79999e-06 [receive_attached]: 2.27999e-06 [after_resolve]: 1.117e-05 [a_after_grad]: 9.42001e-06 [renormalize]: 0.00930151 [add_forward_monad_depend]: 6.73e-06 [auto_monad_grad]: 2.09e-06 [auto_monad_eliminator]: 1.465e-05 [cse]: 2.16e-05 [a_3]: 5.46e-05 [Cycle 2]: 0.00069115, [45] [expand_dump_flag]: 2.73998e-06 [switch_simplify]: 7.73001e-06 [loop_unroll]: 6.26998e-06 [a_1]: 0.00014918 [with_stream_mark]: 1.398e-05 [recompute_prepare]: 6.68e-06 [updatestate_depend_eliminate]: 3.24001e-06 [updatestate_assign_eliminate]: 2.29999e-06 [updatestate_loads_eliminate]: 3.31999e-06 [parameter_eliminate]: 1.67001e-06 [a_2]: 7.295e-05 [accelerated_algorithm]: 6.56999e-06 [shard]: 2.09999e-06 [meta_shard_fg_expand]: 1.64998e-06 [shard_inline]: 6.33e-06 [merge_send_recv]: 7.62002e-06 [auto_parallel]: 8.84e-06 [parallel]: 9.56e-06 [flash_sp]: 3.66001e-06 [merge_comm]: 3.66999e-06 [allreduce_fusion]: 3.02002e-06 [matmul_add_comm_reduction]: 6.84001e-06 [allreduce_slice_to_reducescatter]: 5.8001e-07 [virtual_shard_identity]: 6.88e-06 [virtual_dataset]: 5.76998e-06 [get_grad_eliminate_]: 5.92001e-06 [virtual_output]: 5.66998e-06 [merge_forward]: 3.98001e-06 [cell_reuse_recompute_pass]: 2.19001e-06 [offload_activation]: 1.004e-05 [cell_reuse_handle_not_recompute_node_pass]: 1.269e-05 [merge_recompute_call_nodes]: 1.39e-06 [before_grad]: 9.15001e-06 [set_forward_comm_id_for_comm_node_pass]: 4.10998e-06 [meta_fg_expand]: 3.04999e-06 [flash_sp_send_recv_attached]: 1.81e-06 [receive_attached]: 2.48e-06 [after_resolve]: 1.196e-05 [a_after_grad]: 9.79999e-06 [renormalize]: 1.20024e-07 [add_forward_monad_depend]: 1.74e-06 [auto_monad_grad]: 1.30999e-06 [auto_monad_eliminator]: 9.01002e-06 [cse]: 1.428e-05 [a_3]: 3.425e-05 [py_interpret_to_execute_after_opt_a]: 1.209e-05 [slice_cell_reuse_recomputed_activation]: 2.66e-06 [rewriter_after_opt_a]: 3.178e-05 [convert_after_rewriter]: 6.26e-06 [order_py_execute_after_rewriter]: 5.07999e-06 [mutable_eliminate]: 0.00059513 [opt_b]: 0.00019747, [1] [Cycle 1]: 0.00019068, [7] [b_1]: 0.00011672 [b_2]: 7.90998e-06 [updatestate_depend_eliminate]: 5.71003e-06 [updatestate_assign_eliminate]: 2.53e-06 [updatestate_loads_eliminate]: 2.53e-06 [renormalize]: 4.39992e-07 [cse]: 2.027e-05 [optimize_parallel_all_gather_comm]: 1.561e-05 [overlap_param_gather]: 2.09999e-06 [cconv]: 2.448e-05 [loop_unroll]: 0.00044457 [opt_after_cconv]: 9.953e-05, [1] [Cycle 1]: 9.364e-05, [7] [c_1]: 2.94e-05 [parameter_eliminate]: 2.37001e-06 [updatestate_depend_eliminate]: 6.01e-06 [updatestate_assign_eliminate]: 2.35002e-06 [updatestate_loads_eliminate]: 2.83e-06 [cse]: 1.764e-05 [renormalize]: 2.00002e-07 [remove_dup_value]: 1.187e-05 [tuple_transform]: 7.174e-05, [1] [Cycle 1]: 6.667e-05, [4] [d_1]: 4.033e-05 [none_parameter_eliminate]: 1.29e-06 [renormalize]: 2.19996e-07 [switch_simplify]: 6.60002e-06 [partial_unused_args_eliminate]: 1.84e-06 [add_recomputation]: 4.265e-05 [cse_after_recomputation]: 2.043e-05, [1] [Cycle 1]: 1.629e-05, [1] [cse]: 1.073e-05 [environ_conv]: 4.68999e-06 [swap_dp_allreduce_reducescatter]: 5.25001e-06 [bias_add_comm_swap]: 2.51998e-06 [label_micro_interleaved_index]: 4e-06 [label_fine_grained_interleaved_index]: 2.77002e-06 [merge_cast_opt]: 1.59998e-06 [slice_recompute_activation]: 2.36998e-06 [micro_interleaved_order_control]: 2.72001e-06 [assign_add_opt]: 1.16997e-06 [ForceFp32Comm]: 8.50006e-07 [remove_cast_before_assign_add]: 1.17e-06 [full_micro_interleaved_order_control]: 2.34001e-06 [reorder_send_recv_between_fp_bp]: 2.88e-06 [comm_op_add_attrs]: 1.12e-06 [add_comm_op_reuse_tag]: 1.12e-06 [interleave_split_concat_branches]: 1.35999e-06 [interleave_parallel_branches]: 1.05001e-06 [overlap_opt_shard_in_pipeline]: 1.32999e-06 [overlap_opt_shard_grad_in_pipeline]: 1.99e-06 [control_data_broadcast_order]: 1.351e-05 [grouped_pairwise_exchange_alltoall]: 1.89999e-06 [offloading_packed_experts]: 4.12998e-06 [overlap_recompute_and_grad_model_parallel]: 4.60001e-06 [overlap_grad_matmul_and_grad_allreduce]: 1.32e-06 [overlap_recompute_allgather_and_fa_grad]: 1.54e-06 [overlap_recompute_comm]: 2.26e-06 [overlap_grad_ring_attention]: 4.37e-06 [overlap_grad_flash_sp]: 1.902e-05 [begin_end_overlap_inline]: 5.19998e-07 [split_matmul_comm_elemetwise]: 2.12001e-06 [split_layernorm_comm]: 1.89e-06 [handle_group_info]: 1.11997e-06 [symbol_engine_optimizer]: 0.00015571, [1] [Cycle 1]: 0.00015047, [6] [build]: 6.941e-05 [elim_shapecalc]: 1.164e-05 [elim_not_effective]: 1.513e-05 [opt_reshape]: 6.84001e-06 [fold_const_symbol]: 1.412e-05 [renormalize]: 2.19996e-07 [detach_backward]: 2.36e-06 [pipeline_parallel_scheduler]: 1.60999e-06 [auto_monad_reorder]: 1.748e-05 [get_jit_bprop_graph]: 1.65001e-06 [rewriter_after_jit_bprop_graph]: 4.23999e-06 [opt_after_jit_grad]: 0.00049708 [validate]: 5.189e-05 Sums bootstrap : 0.000403s : 0.81% type_inference : 0.035751s : 71.58% event_method : 0.000050s : 0.10% auto_monad : 0.000075s : 0.15% graph_reusing : 0.000007s : 0.01% inline : 0.000003s : 0.01% add_attr.add_attr_with_inline.tag_attr : 0.000028s : 0.06% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000007s : 0.01% parallel-infer-symbol : 0.000004s : 0.01% pre_auto_parallel : 0.000042s : 0.08% insert-virtual-dataset : 0.000003s : 0.01% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000030s : 0.06% optimize.rewriter_before_opt_a : 0.000105s : 0.21% optimize.opt_a.expand_dump_flag : 0.000006s : 0.01% optimize.opt_a.switch_simplify : 0.000055s : 0.11% optimize.opt_a.loop_unroll : 0.000051s : 0.10% optimize.opt_a.a_1 : 0.001036s : 2.07% optimize.opt_a.with_stream_mark : 0.000030s : 0.06% optimize.opt_a.recompute_prepare : 0.000017s : 0.03% optimize.opt_a.updatestate_depend_eliminate : 0.000008s : 0.02% optimize.opt_a.updatestate_assign_eliminate : 0.000005s : 0.01% optimize.opt_a.updatestate_loads_eliminate : 0.000006s : 0.01% optimize.opt_a.parameter_eliminate : 0.000003s : 0.01% optimize.opt_a.a_2 : 0.000155s : 0.31% optimize.opt_a.accelerated_algorithm : 0.000014s : 0.03% optimize.opt_a.shard : 0.000004s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000004s : 0.01% optimize.opt_a.shard_inline : 0.000013s : 0.03% optimize.opt_a.merge_send_recv : 0.000016s : 0.03% optimize.opt_a.auto_parallel : 0.000017s : 0.03% optimize.opt_a.parallel : 0.000028s : 0.06% optimize.opt_a.flash_sp : 0.000013s : 0.03% optimize.opt_a.merge_comm : 0.000007s : 0.01% optimize.opt_a.allreduce_fusion : 0.000006s : 0.01% optimize.opt_a.matmul_add_comm_reduction : 0.000015s : 0.03% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000001s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000015s : 0.03% optimize.opt_a.virtual_dataset : 0.000012s : 0.02% optimize.opt_a.get_grad_eliminate_ : 0.000011s : 0.02% optimize.opt_a.virtual_output : 0.000012s : 0.02% optimize.opt_a.merge_forward : 0.000008s : 0.02% optimize.opt_a.cell_reuse_recompute_pass : 0.000003s : 0.01% optimize.opt_a.offload_activation : 0.000020s : 0.04% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000026s : 0.05% optimize.opt_a.merge_recompute_call_nodes : 0.000003s : 0.01% optimize.opt_a.before_grad : 0.000020s : 0.04% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000008s : 0.02% optimize.opt_a.meta_fg_expand : 0.000006s : 0.01% optimize.opt_a.flash_sp_send_recv_attached : 0.000005s : 0.01% optimize.opt_a.receive_attached : 0.000005s : 0.01% optimize.opt_a.after_resolve : 0.000023s : 0.05% optimize.opt_a.a_after_grad : 0.000019s : 0.04% optimize.opt_a.renormalize : 0.009302s : 18.62% optimize.opt_a.add_forward_monad_depend : 0.000008s : 0.02% optimize.opt_a.auto_monad_grad : 0.000003s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000024s : 0.05% optimize.opt_a.cse : 0.000036s : 0.07% optimize.opt_a.a_3 : 0.000089s : 0.18% optimize.py_interpret_to_execute_after_opt_a : 0.000012s : 0.02% optimize.slice_cell_reuse_recomputed_activation : 0.000003s : 0.01% optimize.rewriter_after_opt_a : 0.000032s : 0.06% optimize.convert_after_rewriter : 0.000006s : 0.01% optimize.order_py_execute_after_rewriter : 0.000005s : 0.01% optimize.mutable_eliminate : 0.000595s : 1.19% optimize.opt_b.b_1 : 0.000117s : 0.23% optimize.opt_b.b_2 : 0.000008s : 0.02% optimize.opt_b.updatestate_depend_eliminate : 0.000006s : 0.01% optimize.opt_b.updatestate_assign_eliminate : 0.000003s : 0.01% optimize.opt_b.updatestate_loads_eliminate : 0.000003s : 0.01% optimize.opt_b.renormalize : 0.000000s : 0.00% optimize.opt_b.cse : 0.000020s : 0.04% optimize.optimize_parallel_all_gather_comm : 0.000016s : 0.03% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000024s : 0.05% optimize.loop_unroll : 0.000445s : 0.89% optimize.opt_after_cconv.c_1 : 0.000029s : 0.06% optimize.opt_after_cconv.parameter_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000006s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000002s : 0.00% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000003s : 0.01% optimize.opt_after_cconv.cse : 0.000018s : 0.04% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000012s : 0.02% optimize.tuple_transform.d_1 : 0.000040s : 0.08% optimize.tuple_transform.none_parameter_eliminate : 0.000001s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000007s : 0.01% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000043s : 0.09% optimize.cse_after_recomputation.cse : 0.000011s : 0.02% optimize.environ_conv : 0.000005s : 0.01% optimize.swap_dp_allreduce_reducescatter : 0.000005s : 0.01% optimize.bias_add_comm_swap : 0.000003s : 0.01% optimize.label_micro_interleaved_index : 0.000004s : 0.01% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.01% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.01% optimize.assign_add_opt : 0.000001s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.01% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000001s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000001s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000014s : 0.03% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000004s : 0.01% optimize.overlap_recompute_and_grad_model_parallel : 0.000005s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000002s : 0.00% optimize.overlap_recompute_comm : 0.000002s : 0.00% optimize.overlap_grad_ring_attention : 0.000004s : 0.01% optimize.overlap_grad_flash_sp : 0.000019s : 0.04% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000002s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000069s : 0.14% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000012s : 0.02% optimize.symbol_engine_optimizer.elim_not_effective : 0.000015s : 0.03% optimize.symbol_engine_optimizer.opt_reshape : 0.000007s : 0.01% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000014s : 0.03% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000002s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000017s : 0.03% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000004s : 0.01% opt_after_jit_grad : 0.000497s : 1.00% validate : 0.000052s : 0.10% Time group info: ------[substitution.] 0.000292 40 14.24% : 0.000042s : 8: substitution.arithmetic_simplify 1.15% : 0.000003s : 2: substitution.elim_not_effective 1.90% : 0.000006s : 2: substitution.fold_const_symbol 1.91% : 0.000006s : 4: substitution.graph_param_transform 70.10% : 0.000205s : 6: substitution.inline 1.15% : 0.000003s : 4: substitution.j_node_and_user_rematch 1.67% : 0.000005s : 4: substitution.remove_not_recompute_node 1.74% : 0.000005s : 4: substitution.replace_old_param 6.13% : 0.000018s : 6: substitution.tuple_list_get_item_eliminator ------[type_inference.] 0.035680 2 96.80% : 0.034538s : 1: type_inference.infer 3.20% : 0.001142s : 1: type_inference.specialize ------[replace.] 0.000092 12 59.14% : 0.000054s : 6: replace.inline 40.86% : 0.000038s : 6: replace.tuple_list_get_item_eliminator ------[match.] 0.000216 12 92.94% : 0.000201s : 6: match.inline 7.06% : 0.000015s : 6: match.tuple_list_get_item_eliminator ------[predicate.] 0.000237 1534 0.97% : 0.000002s : 17: predicate.accumulaten_eliminater 0.68% : 0.000002s : 4: predicate.ad_related_special_op_eliminate 0.44% : 0.000001s : 8: predicate.addn_check_dump 0.98% : 0.000002s : 17: predicate.addn_zero_filter 0.89% : 0.000002s : 17: predicate.adjust_all_reduce_mul_add 2.32% : 0.000006s : 25: predicate.arithmetic_simplify 1.08% : 0.000003s : 17: predicate.cast_eliminate 0.50% : 0.000001s : 8: predicate.check_bprop_eliminate 0.40% : 0.000001s : 8: predicate.compare_switch_simplify 0.16% : 0.000000s : 4: predicate.const_output_eliminate 0.55% : 0.000001s : 8: predicate.depend_value_elim 0.99% : 0.000002s : 17: predicate.dict_get_item_const_eliminator 1.16% : 0.000003s : 17: predicate.dict_get_item_eliminator 1.05% : 0.000003s : 17: predicate.dict_set_item_eliminator 0.80% : 0.000002s : 8: predicate.dumpgradient_eliminate 0.22% : 0.000001s : 4: predicate.elim_not_effective 0.34% : 0.000001s : 4: predicate.elim_shapecalc_of_broadcastargs 1.21% : 0.000003s : 21: predicate.environ_add_const_eliminate 1.09% : 0.000003s : 21: predicate.environ_get_add_eliminate 1.14% : 0.000003s : 21: predicate.environ_get_depend_swap 1.69% : 0.000004s : 29: predicate.environ_get_eliminate 1.16% : 0.000003s : 21: predicate.environ_get_set_eliminate 1.65% : 0.000004s : 29: predicate.exchange_switch_depend_value 2.39% : 0.000006s : 29: predicate.float_depend_g_call 0.46% : 0.000001s : 8: predicate.float_environ_get_switch 0.61% : 0.000001s : 12: predicate.float_tuple_getitem_switch 0.16% : 0.000000s : 4: predicate.fold_const_symbol 0.71% : 0.000002s : 8: predicate.get_grad_eliminate 0.16% : 0.000000s : 4: predicate.graph_param_transform 0.45% : 0.000001s : 8: predicate.incorporate_call 0.39% : 0.000001s : 8: predicate.incorporate_call_switch 5.74% : 0.000014s : 70: predicate.inline 0.59% : 0.000001s : 8: predicate.inline_without_move 0.28% : 0.000001s : 8: predicate.j_node_and_user_rematch 0.90% : 0.000002s : 8: predicate.less_batch_normalization 1.96% : 0.000005s : 31: predicate.list_to_tuple_eliminator_ 2.82% : 0.000007s : 48: predicate.load_eliminater 0.61% : 0.000001s : 4: predicate.loop_unroll_after_grad 3.00% : 0.000007s : 47: predicate.loop_unroll_before_grad 1.66% : 0.000004s : 25: predicate.make_slice_get_slice_eliminator 0.49% : 0.000001s : 8: predicate.merge_addn 0.49% : 0.000001s : 8: predicate.micro_step_allgather_replace 0.67% : 0.000002s : 8: predicate.mini_step_allgather_replace 0.97% : 0.000002s : 17: predicate.minmaximum_grad 0.86% : 0.000002s : 4: predicate.mutable_eliminate 0.27% : 0.000001s : 4: predicate.opt_reshape 0.33% : 0.000001s : 4: predicate.parallel_virtual_node 2.24% : 0.000005s : 29: predicate.partial_defer_inline 1.65% : 0.000004s : 27: predicate.partial_eliminate 0.95% : 0.000002s : 17: predicate.print_const_string_wrapper 0.50% : 0.000001s : 8: predicate.reduce_all_const_elim 1.31% : 0.000003s : 17: predicate.reduce_eliminate 2.78% : 0.000007s : 48: predicate.redundant_stop_gradient_eliminater 0.50% : 0.000001s : 8: predicate.remove_not_recompute_node 1.49% : 0.000004s : 31: predicate.replace_applicator 0.39% : 0.000001s : 8: predicate.replace_old_param 0.20% : 0.000000s : 4: predicate.reset_defer_inline 1.15% : 0.000003s : 17: predicate.reshape_eliminate 0.59% : 0.000001s : 8: predicate.row_tensor_add_zeros_like 0.35% : 0.000001s : 4: predicate.row_tensor_eliminate 0.73% : 0.000002s : 8: predicate.same_eliminate 0.35% : 0.000001s : 8: predicate.set_cell_output_no_recompute 0.65% : 0.000002s : 8: predicate.shard_identity_eliminate 0.70% : 0.000002s : 8: predicate.special_op_eliminate 0.59% : 0.000001s : 8: predicate.specialize_transform 0.87% : 0.000002s : 8: predicate.split_environ_get_set_with_tuple_value 0.63% : 0.000001s : 8: predicate.stack_unstack_eliminate 0.27% : 0.000001s : 4: predicate.switch_call_monad_eliminater 1.81% : 0.000004s : 29: predicate.switch_defer_inline 2.24% : 0.000005s : 37: predicate.switch_layer_defer_inline 5.38% : 0.000013s : 88: predicate.switch_simplify 1.13% : 0.000003s : 17: predicate.tile_eliminate 0.99% : 0.000002s : 17: predicate.transpose_eliminate 1.62% : 0.000004s : 25: predicate.tuple_list_convert_item_index_to_positive 1.46% : 0.000003s : 25: predicate.tuple_list_get_item_const_eliminator 1.43% : 0.000003s : 25: predicate.tuple_list_get_item_depend_reorder 3.24% : 0.000008s : 39: predicate.tuple_list_get_item_eliminator 1.54% : 0.000004s : 25: predicate.tuple_list_get_set_item_eliminator 2.21% : 0.000005s : 33: predicate.tuple_list_set_item_eliminator 1.89% : 0.000004s : 31: predicate.tuple_to_list_eliminator_ 2.52% : 0.000006s : 48: predicate.updatestate_pure_node_eliminater 3.08% : 0.000007s : 56: predicate.updatestate_useless_node_eliminater 0.35% : 0.000001s : 4: predicate.value_based_eliminate 0.57% : 0.000001s : 8: predicate.virtual_dataset_eliminate 0.58% : 0.000001s : 8: predicate.virtual_output_eliminate 0.21% : 0.000001s : 4: predicate.virtual_view_grad_eliminate 0.41% : 0.000001s : 4: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.000969 15 48.71% : 0.000472s : 7: func_graph_cloner_run.FuncGraphClonerGraph 51.29% : 0.000497s : 8: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.082437 192 0.03% : 0.000022s : 1: ForceFp32Comm 4.10% : 0.003376s : 1: add_attr 4.08% : 0.003365s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.06% : 0.000047s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.10% : 0.000081s : 1: auto_monad 0.03% : 0.000021s : 1: auto_monad_reorder 0.00% : 0.000003s : 1: begin_end_overlap_inline 0.01% : 0.000006s : 1: bias_add_comm_swap 0.52% : 0.000426s : 1: bootstrap 0.03% : 0.000028s : 1: cconv 0.01% : 0.000005s : 1: comm_op_add_attrs 0.02% : 0.000017s : 1: control_data_broadcast_order 0.01% : 0.000010s : 1: convert_after_rewriter 0.03% : 0.000024s : 1: cse_after_recomputation 0.01% : 0.000005s : 1: dataset_repeat_opt 0.01% : 0.000006s : 1: detach_backward 0.01% : 0.000008s : 1: environ_conv 0.07% : 0.000057s : 1: event_method 0.01% : 0.000005s : 1: full_micro_interleaved_order_control 0.01% : 0.000006s : 1: get_jit_bprop_graph 0.01% : 0.000010s : 1: graph_reusing 0.01% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000004s : 1: handle_group_info 0.01% : 0.000006s : 1: inline 0.01% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.01% : 0.000004s : 1: interleave_split_concat_branches 0.01% : 0.000006s : 1: label_fine_grained_interleaved_index 0.01% : 0.000007s : 1: label_micro_interleaved_index 0.55% : 0.000453s : 1: loop_unroll 0.01% : 0.000005s : 1: merge_cast_opt 0.01% : 0.000005s : 1: micro_interleaved_order_control 0.73% : 0.000604s : 1: mutable_eliminate 0.01% : 0.000008s : 1: offloading_packed_experts 0.02% : 0.000013s : 1: opt.transform.loop_unroll_optimizer 0.02% : 0.000015s : 1: opt.transform.mutable_eliminate 1.80% : 0.001480s : 78: opt.transform.opt_a 0.03% : 0.000028s : 1: opt.transform.opt_after_cconv 0.03% : 0.000023s : 1: opt.transform.opt_after_jit_grad 0.11% : 0.000095s : 28: opt.transform.opt_b 0.05% : 0.000045s : 2: opt.transform.opt_trans_graph 0.05% : 0.000044s : 4: opt.transform.symbol_engine_opt 14.12% : 0.011642s : 1: opt_a 0.12% : 0.000103s : 1: opt_after_cconv 0.61% : 0.000506s : 1: opt_after_jit_grad 0.24% : 0.000202s : 1: opt_b 16.83% : 0.013870s : 1: optimize 0.02% : 0.000019s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000008s : 1: order_py_execute_after_rewriter 0.03% : 0.000022s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000007s : 1: overlap_grad_ring_attention 0.01% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.01% : 0.000005s : 1: overlap_param_gather 0.01% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000008s : 1: overlap_recompute_and_grad_model_parallel 0.01% : 0.000005s : 1: overlap_recompute_comm 0.01% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.01% : 0.000005s : 1: partial_unused_args_eliminate 0.01% : 0.000005s : 1: pipeline_parallel_scheduler 0.01% : 0.000006s : 1: pipeline_split 0.06% : 0.000047s : 1: pre_auto_parallel 0.04% : 0.000034s : 1: py_interpret_to_execute 0.02% : 0.000016s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.02% : 0.000016s : 1: remove_dup_value 10.58% : 0.008723s : 1: renormalize.infer 0.69% : 0.000567s : 1: renormalize.specialize 0.01% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.01% : 0.000007s : 1: rewriter_after_jit_bprop_graph 0.04% : 0.000036s : 1: rewriter_after_opt_a 0.13% : 0.000110s : 1: rewriter_before_opt_a 0.01% : 0.000005s : 1: slice_cell_reuse_recomputed_activation 0.01% : 0.000005s : 1: slice_recompute_activation 0.01% : 0.000005s : 1: split_layernorm_comm 0.01% : 0.000005s : 1: split_matmul_comm_elemetwise 0.01% : 0.000008s : 1: swap_dp_allreduce_reducescatter 0.19% : 0.000159s : 1: symbol_engine_optimizer 0.09% : 0.000075s : 1: tuple_transform 43.39% : 0.035771s : 1: type_inference TotalTime = 0.138662, [21] [bootstrap]: 0.00041805 [type_inference]: 0.0965158 [event_method]: 9.729e-05 [auto_monad]: 0.00014775 [graph_reusing]: 1.013e-05 [inline]: 3.01999e-06 [add_attr]: 0.00372839, [1] [add_attr_with_inline]: 0.00371595, [1] [Cycle 1]: 0.00011432, [2] [tag_attr]: 5.764e-05 [meta_addattr_fg_expand]: 1.353e-05 [parallel-infer-symbol]: 3.90998e-06 [pre_auto_parallel]: 7.524e-05 [insert-virtual-dataset]: 2.63e-06 [parallel-infer-symbol-second]: 8.89995e-07 [dataset_repeat_opt]: 2.41998e-06 [pipeline_split]: 1.71e-06 [optimize]: 0.0364777, [53] [py_interpret_to_execute]: 5.283e-05 [rewriter_before_opt_a]: 0.00021625 [opt_a]: 0.0195685, [3] [Cycle 1]: 0.0119633, [45] [expand_dump_flag]: 5.78002e-06 [switch_simplify]: 9.952e-05 [loop_unroll]: 8.505e-05 [a_1]: 0.00292022 [with_stream_mark]: 3.808e-05 [recompute_prepare]: 3.461e-05 [updatestate_depend_eliminate]: 1.268e-05 [updatestate_assign_eliminate]: 9.20001e-06 [updatestate_loads_eliminate]: 8.89e-06 [parameter_eliminate]: 3.44001e-06 [a_2]: 0.000287 [accelerated_algorithm]: 4.345e-05 [shard]: 2.31998e-06 [meta_shard_fg_expand]: 8.50001e-06 [shard_inline]: 1.897e-05 [merge_send_recv]: 2.052e-05 [auto_parallel]: 1.84e-05 [parallel]: 2.129e-05 [flash_sp]: 1.294e-05 [merge_comm]: 1.18e-05 [allreduce_fusion]: 1.078e-05 [matmul_add_comm_reduction]: 3.161e-05 [allreduce_slice_to_reducescatter]: 6.50005e-07 [virtual_shard_identity]: 2.275e-05 [virtual_dataset]: 1.85e-05 [get_grad_eliminate_]: 1.741e-05 [virtual_output]: 1.797e-05 [merge_forward]: 1.226e-05 [cell_reuse_recompute_pass]: 1.71002e-06 [offload_activation]: 2.165e-05 [cell_reuse_handle_not_recompute_node_pass]: 3.894e-05 [merge_recompute_call_nodes]: 1.72001e-06 [before_grad]: 3.595e-05 [set_forward_comm_id_for_comm_node_pass]: 1.485e-05 [meta_fg_expand]: 0.00206469 [flash_sp_send_recv_attached]: 7.66999e-06 [receive_attached]: 3.19001e-06 [after_resolve]: 8.575e-05 [a_after_grad]: 0.00010352 [renormalize]: 0.00456784 [add_forward_monad_depend]: 1.396e-05 [auto_monad_grad]: 7.48e-06 [auto_monad_eliminator]: 8.533e-05 [cse]: 0.00027147 [a_3]: 0.00044467 [Cycle 2]: 0.00562429, [45] [expand_dump_flag]: 3.63999e-06 [switch_simplify]: 6.118e-05 [loop_unroll]: 5.643e-05 [a_1]: 0.00213717 [with_stream_mark]: 2.962e-05 [recompute_prepare]: 2.253e-05 [updatestate_depend_eliminate]: 1.303e-05 [updatestate_assign_eliminate]: 1.122e-05 [updatestate_loads_eliminate]: 1.017e-05 [parameter_eliminate]: 1.24e-06 [a_2]: 0.00029588 [accelerated_algorithm]: 2.61e-05 [shard]: 2.79999e-06 [meta_shard_fg_expand]: 7.25e-06 [shard_inline]: 2.04e-05 [merge_send_recv]: 1.91e-05 [auto_parallel]: 1.895e-05 [parallel]: 9.97999e-06 [flash_sp]: 4.05e-06 [merge_comm]: 1.246e-05 [allreduce_fusion]: 1.241e-05 [matmul_add_comm_reduction]: 4.194e-05 [allreduce_slice_to_reducescatter]: 9.5999e-07 [virtual_shard_identity]: 2.287e-05 [virtual_dataset]: 1.886e-05 [get_grad_eliminate_]: 1.854e-05 [virtual_output]: 1.782e-05 [merge_forward]: 1.191e-05 [cell_reuse_recompute_pass]: 2.07999e-06 [offload_activation]: 2.342e-05 [cell_reuse_handle_not_recompute_node_pass]: 4.052e-05 [merge_recompute_call_nodes]: 1.42e-06 [before_grad]: 3.525e-05 [set_forward_comm_id_for_comm_node_pass]: 1.347e-05 [meta_fg_expand]: 0.00020718 [flash_sp_send_recv_attached]: 2.18002e-06 [receive_attached]: 3.14999e-06 [after_resolve]: 3.084e-05 [a_after_grad]: 3.144e-05 [renormalize]: 0.00170214 [add_forward_monad_depend]: 7.94002e-06 [auto_monad_grad]: 2.62001e-06 [auto_monad_eliminator]: 3.813e-05 [cse]: 0.00011868 [a_3]: 0.00014634 [Cycle 3]: 0.00194045, [45] [expand_dump_flag]: 2.96999e-06 [switch_simplify]: 2.144e-05 [loop_unroll]: 1.88e-05 [a_1]: 0.00061411 [with_stream_mark]: 2.567e-05 [recompute_prepare]: 2.019e-05 [updatestate_depend_eliminate]: 1.326e-05 [updatestate_assign_eliminate]: 1.083e-05 [updatestate_loads_eliminate]: 1.004e-05 [parameter_eliminate]: 2.01e-06 [a_2]: 0.00028376 [accelerated_algorithm]: 2.52e-05 [shard]: 2.22001e-06 [meta_shard_fg_expand]: 3.245e-05 [shard_inline]: 2.15e-05 [merge_send_recv]: 1.854e-05 [auto_parallel]: 1.892e-05 [parallel]: 8.13001e-06 [flash_sp]: 1.91e-06 [merge_comm]: 1.25e-05 [allreduce_fusion]: 1.24e-05 [matmul_add_comm_reduction]: 2.036e-05 [allreduce_slice_to_reducescatter]: 7.39994e-07 [virtual_shard_identity]: 2.081e-05 [virtual_dataset]: 1.83e-05 [get_grad_eliminate_]: 1.777e-05 [virtual_output]: 1.774e-05 [merge_forward]: 1.162e-05 [cell_reuse_recompute_pass]: 3.29001e-06 [offload_activation]: 2.337e-05 [cell_reuse_handle_not_recompute_node_pass]: 4.396e-05 [merge_recompute_call_nodes]: 1.77001e-06 [before_grad]: 3.558e-05 [set_forward_comm_id_for_comm_node_pass]: 1.294e-05 [meta_fg_expand]: 9.24e-06 [flash_sp_send_recv_attached]: 1.69e-06 [receive_attached]: 2.16998e-06 [after_resolve]: 2.378e-05 [a_after_grad]: 3.067e-05 [renormalize]: 1.00001e-07 [add_forward_monad_depend]: 3.08998e-06 [auto_monad_grad]: 2.94999e-06 [auto_monad_eliminator]: 3.056e-05 [cse]: 7.209e-05 [a_3]: 0.00012552 [py_interpret_to_execute_after_opt_a]: 2.69e-05 [slice_cell_reuse_recomputed_activation]: 2.15002e-06 [rewriter_after_opt_a]: 0.00014936 [convert_after_rewriter]: 2.131e-05 [order_py_execute_after_rewriter]: 1.58e-05 [mutable_eliminate]: 0.0129133 [opt_b]: 0.00073774, [1] [Cycle 1]: 0.00072566, [7] [b_1]: 0.00049022 [b_2]: 2.381e-05 [updatestate_depend_eliminate]: 2.103e-05 [updatestate_assign_eliminate]: 1.077e-05 [updatestate_loads_eliminate]: 1.09e-05 [renormalize]: 7.7e-07 [cse]: 0.00011639 [optimize_parallel_all_gather_comm]: 4.598e-05 [overlap_param_gather]: 2.31e-06 [cconv]: 3.647e-05 [loop_unroll]: 0.00080981 [opt_after_cconv]: 0.00028407, [1] [Cycle 1]: 0.00027512, [7] [c_1]: 0.00010302 [parameter_eliminate]: 4.47e-06 [updatestate_depend_eliminate]: 1.772e-05 [updatestate_assign_eliminate]: 1.036e-05 [updatestate_loads_eliminate]: 1.082e-05 [cse]: 8.793e-05 [renormalize]: 3.89991e-07 [remove_dup_value]: 0.00010924 [tuple_transform]: 0.00027869, [1] [Cycle 1]: 0.00027177, [4] [d_1]: 0.00022329 [none_parameter_eliminate]: 2.60002e-06 [renormalize]: 2.50002e-07 [switch_simplify]: 2.146e-05 [partial_unused_args_eliminate]: 1.99999e-06 [add_recomputation]: 0.00013166 [cse_after_recomputation]: 7.17e-05, [1] [Cycle 1]: 6.472e-05, [1] [cse]: 5.67e-05 [environ_conv]: 2.149e-05 [swap_dp_allreduce_reducescatter]: 1.656e-05 [bias_add_comm_swap]: 3.18e-06 [label_micro_interleaved_index]: 8.08001e-06 [label_fine_grained_interleaved_index]: 3.4e-06 [merge_cast_opt]: 1.52001e-06 [slice_recompute_activation]: 2.34001e-06 [micro_interleaved_order_control]: 2.74999e-06 [assign_add_opt]: 1.52001e-06 [ForceFp32Comm]: 9.70002e-07 [remove_cast_before_assign_add]: 1.48002e-06 [full_micro_interleaved_order_control]: 2.31998e-06 [reorder_send_recv_between_fp_bp]: 2.73e-06 [comm_op_add_attrs]: 1.25001e-06 [add_comm_op_reuse_tag]: 1.02e-06 [interleave_split_concat_branches]: 1.17e-06 [interleave_parallel_branches]: 1.67999e-06 [overlap_opt_shard_in_pipeline]: 1.64e-06 [overlap_opt_shard_grad_in_pipeline]: 1.92999e-06 [control_data_broadcast_order]: 4e-05 [grouped_pairwise_exchange_alltoall]: 1.70001e-06 [offloading_packed_experts]: 9.99001e-06 [overlap_recompute_and_grad_model_parallel]: 1.126e-05 [overlap_grad_matmul_and_grad_allreduce]: 1.35001e-06 [overlap_recompute_allgather_and_fa_grad]: 1.40001e-06 [overlap_recompute_comm]: 2.66e-06 [overlap_grad_ring_attention]: 9.37001e-06 [overlap_grad_flash_sp]: 5.146e-05 [begin_end_overlap_inline]: 7.29982e-07 [split_matmul_comm_elemetwise]: 2.62001e-06 [split_layernorm_comm]: 1.72001e-06 [handle_group_info]: 1.04e-06 [symbol_engine_optimizer]: 0.00050304, [1] [Cycle 1]: 0.00049659, [6] [build]: 0.00024738 [elim_shapecalc]: 7.283e-05 [elim_not_effective]: 4.696e-05 [opt_reshape]: 2.72e-05 [fold_const_symbol]: 5.462e-05 [renormalize]: 3.7998e-07 [detach_backward]: 3.26001e-06 [pipeline_parallel_scheduler]: 1.50999e-06 [auto_monad_reorder]: 4.68e-05 [get_jit_bprop_graph]: 1.59e-06 [rewriter_after_jit_bprop_graph]: 6.17999e-06 [opt_after_jit_grad]: 0.00085417 [validate]: 8.532e-05 Sums bootstrap : 0.000418s : 0.31% type_inference : 0.096516s : 72.38% event_method : 0.000097s : 0.07% auto_monad : 0.000148s : 0.11% graph_reusing : 0.000010s : 0.01% inline : 0.000003s : 0.00% add_attr.add_attr_with_inline.tag_attr : 0.000058s : 0.04% add_attr.add_attr_with_inline.meta_addattr_fg_expand : 0.000014s : 0.01% parallel-infer-symbol : 0.000004s : 0.00% pre_auto_parallel : 0.000075s : 0.06% insert-virtual-dataset : 0.000003s : 0.00% parallel-infer-symbol-second : 0.000001s : 0.00% dataset_repeat_opt : 0.000002s : 0.00% pipeline_split : 0.000002s : 0.00% optimize.py_interpret_to_execute : 0.000053s : 0.04% optimize.rewriter_before_opt_a : 0.000216s : 0.16% optimize.opt_a.expand_dump_flag : 0.000012s : 0.01% optimize.opt_a.switch_simplify : 0.000182s : 0.14% optimize.opt_a.loop_unroll : 0.000160s : 0.12% optimize.opt_a.a_1 : 0.005671s : 4.25% optimize.opt_a.with_stream_mark : 0.000093s : 0.07% optimize.opt_a.recompute_prepare : 0.000077s : 0.06% optimize.opt_a.updatestate_depend_eliminate : 0.000039s : 0.03% optimize.opt_a.updatestate_assign_eliminate : 0.000031s : 0.02% optimize.opt_a.updatestate_loads_eliminate : 0.000029s : 0.02% optimize.opt_a.parameter_eliminate : 0.000007s : 0.01% optimize.opt_a.a_2 : 0.000867s : 0.65% optimize.opt_a.accelerated_algorithm : 0.000095s : 0.07% optimize.opt_a.shard : 0.000007s : 0.01% optimize.opt_a.meta_shard_fg_expand : 0.000048s : 0.04% optimize.opt_a.shard_inline : 0.000061s : 0.05% optimize.opt_a.merge_send_recv : 0.000058s : 0.04% optimize.opt_a.auto_parallel : 0.000056s : 0.04% optimize.opt_a.parallel : 0.000039s : 0.03% optimize.opt_a.flash_sp : 0.000019s : 0.01% optimize.opt_a.merge_comm : 0.000037s : 0.03% optimize.opt_a.allreduce_fusion : 0.000036s : 0.03% optimize.opt_a.matmul_add_comm_reduction : 0.000094s : 0.07% optimize.opt_a.allreduce_slice_to_reducescatter : 0.000002s : 0.00% optimize.opt_a.virtual_shard_identity : 0.000066s : 0.05% optimize.opt_a.virtual_dataset : 0.000056s : 0.04% optimize.opt_a.get_grad_eliminate_ : 0.000054s : 0.04% optimize.opt_a.virtual_output : 0.000054s : 0.04% optimize.opt_a.merge_forward : 0.000036s : 0.03% optimize.opt_a.cell_reuse_recompute_pass : 0.000007s : 0.01% optimize.opt_a.offload_activation : 0.000068s : 0.05% optimize.opt_a.cell_reuse_handle_not_recompute_node_pass : 0.000123s : 0.09% optimize.opt_a.merge_recompute_call_nodes : 0.000005s : 0.00% optimize.opt_a.before_grad : 0.000107s : 0.08% optimize.opt_a.set_forward_comm_id_for_comm_node_pass : 0.000041s : 0.03% optimize.opt_a.meta_fg_expand : 0.002281s : 1.71% optimize.opt_a.flash_sp_send_recv_attached : 0.000012s : 0.01% optimize.opt_a.receive_attached : 0.000009s : 0.01% optimize.opt_a.after_resolve : 0.000140s : 0.11% optimize.opt_a.a_after_grad : 0.000166s : 0.12% optimize.opt_a.renormalize : 0.006270s : 4.70% optimize.opt_a.add_forward_monad_depend : 0.000025s : 0.02% optimize.opt_a.auto_monad_grad : 0.000013s : 0.01% optimize.opt_a.auto_monad_eliminator : 0.000154s : 0.12% optimize.opt_a.cse : 0.000462s : 0.35% optimize.opt_a.a_3 : 0.000717s : 0.54% optimize.py_interpret_to_execute_after_opt_a : 0.000027s : 0.02% optimize.slice_cell_reuse_recomputed_activation : 0.000002s : 0.00% optimize.rewriter_after_opt_a : 0.000149s : 0.11% optimize.convert_after_rewriter : 0.000021s : 0.02% optimize.order_py_execute_after_rewriter : 0.000016s : 0.01% optimize.mutable_eliminate : 0.012913s : 9.68% optimize.opt_b.b_1 : 0.000490s : 0.37% optimize.opt_b.b_2 : 0.000024s : 0.02% optimize.opt_b.updatestate_depend_eliminate : 0.000021s : 0.02% optimize.opt_b.updatestate_assign_eliminate : 0.000011s : 0.01% optimize.opt_b.updatestate_loads_eliminate : 0.000011s : 0.01% optimize.opt_b.renormalize : 0.000001s : 0.00% optimize.opt_b.cse : 0.000116s : 0.09% optimize.optimize_parallel_all_gather_comm : 0.000046s : 0.03% optimize.overlap_param_gather : 0.000002s : 0.00% optimize.cconv : 0.000036s : 0.03% optimize.loop_unroll : 0.000810s : 0.61% optimize.opt_after_cconv.c_1 : 0.000103s : 0.08% optimize.opt_after_cconv.parameter_eliminate : 0.000004s : 0.00% optimize.opt_after_cconv.updatestate_depend_eliminate : 0.000018s : 0.01% optimize.opt_after_cconv.updatestate_assign_eliminate : 0.000010s : 0.01% optimize.opt_after_cconv.updatestate_loads_eliminate : 0.000011s : 0.01% optimize.opt_after_cconv.cse : 0.000088s : 0.07% optimize.opt_after_cconv.renormalize : 0.000000s : 0.00% optimize.remove_dup_value : 0.000109s : 0.08% optimize.tuple_transform.d_1 : 0.000223s : 0.17% optimize.tuple_transform.none_parameter_eliminate : 0.000003s : 0.00% optimize.tuple_transform.renormalize : 0.000000s : 0.00% optimize.tuple_transform.switch_simplify : 0.000021s : 0.02% optimize.partial_unused_args_eliminate : 0.000002s : 0.00% optimize.add_recomputation : 0.000132s : 0.10% optimize.cse_after_recomputation.cse : 0.000057s : 0.04% optimize.environ_conv : 0.000021s : 0.02% optimize.swap_dp_allreduce_reducescatter : 0.000017s : 0.01% optimize.bias_add_comm_swap : 0.000003s : 0.00% optimize.label_micro_interleaved_index : 0.000008s : 0.01% optimize.label_fine_grained_interleaved_index : 0.000003s : 0.00% optimize.merge_cast_opt : 0.000002s : 0.00% optimize.slice_recompute_activation : 0.000002s : 0.00% optimize.micro_interleaved_order_control : 0.000003s : 0.00% optimize.assign_add_opt : 0.000002s : 0.00% optimize.ForceFp32Comm : 0.000001s : 0.00% optimize.remove_cast_before_assign_add : 0.000001s : 0.00% optimize.full_micro_interleaved_order_control : 0.000002s : 0.00% optimize.reorder_send_recv_between_fp_bp : 0.000003s : 0.00% optimize.comm_op_add_attrs : 0.000001s : 0.00% optimize.add_comm_op_reuse_tag : 0.000001s : 0.00% optimize.interleave_split_concat_branches : 0.000001s : 0.00% optimize.interleave_parallel_branches : 0.000002s : 0.00% optimize.overlap_opt_shard_in_pipeline : 0.000002s : 0.00% optimize.overlap_opt_shard_grad_in_pipeline : 0.000002s : 0.00% optimize.control_data_broadcast_order : 0.000040s : 0.03% optimize.grouped_pairwise_exchange_alltoall : 0.000002s : 0.00% optimize.offloading_packed_experts : 0.000010s : 0.01% optimize.overlap_recompute_and_grad_model_parallel : 0.000011s : 0.01% optimize.overlap_grad_matmul_and_grad_allreduce : 0.000001s : 0.00% optimize.overlap_recompute_allgather_and_fa_grad : 0.000001s : 0.00% optimize.overlap_recompute_comm : 0.000003s : 0.00% optimize.overlap_grad_ring_attention : 0.000009s : 0.01% optimize.overlap_grad_flash_sp : 0.000051s : 0.04% optimize.begin_end_overlap_inline : 0.000001s : 0.00% optimize.split_matmul_comm_elemetwise : 0.000003s : 0.00% optimize.split_layernorm_comm : 0.000002s : 0.00% optimize.handle_group_info : 0.000001s : 0.00% optimize.symbol_engine_optimizer.build : 0.000247s : 0.19% optimize.symbol_engine_optimizer.elim_shapecalc : 0.000073s : 0.05% optimize.symbol_engine_optimizer.elim_not_effective : 0.000047s : 0.04% optimize.symbol_engine_optimizer.opt_reshape : 0.000027s : 0.02% optimize.symbol_engine_optimizer.fold_const_symbol : 0.000055s : 0.04% optimize.symbol_engine_optimizer.renormalize : 0.000000s : 0.00% detach_backward : 0.000003s : 0.00% pipeline_parallel_scheduler : 0.000002s : 0.00% auto_monad_reorder : 0.000047s : 0.04% get_jit_bprop_graph : 0.000002s : 0.00% rewriter_after_jit_bprop_graph : 0.000006s : 0.00% opt_after_jit_grad : 0.000854s : 0.64% validate : 0.000085s : 0.06% Time group info: ------[substitution.] 0.001592 367 5.36% : 0.000085s : 16: substitution.arithmetic_simplify 0.79% : 0.000013s : 15: substitution.elim_not_effective 1.81% : 0.000029s : 2: substitution.elim_shapecalc_of_broadcastargs 0.38% : 0.000006s : 6: substitution.float_depend_g_call 1.11% : 0.000018s : 10: substitution.float_tuple_getitem_switch 1.52% : 0.000024s : 15: substitution.fold_const_symbol 0.97% : 0.000015s : 17: substitution.graph_param_transform 0.17% : 0.000003s : 2: substitution.incorporate_call 0.16% : 0.000003s : 2: substitution.incorporate_call_switch 54.25% : 0.000864s : 20: substitution.inline 1.83% : 0.000029s : 2: substitution.inline_without_move 1.23% : 0.000020s : 42: substitution.j_node_and_user_rematch 1.47% : 0.000023s : 3: substitution.less_batch_normalization 1.48% : 0.000024s : 14: substitution.minmaximum_grad 0.24% : 0.000004s : 2: substitution.opt_reshape 0.46% : 0.000007s : 6: substitution.partial_eliminate 0.61% : 0.000010s : 4: substitution.reduce_eliminate 1.84% : 0.000029s : 42: substitution.remove_not_recompute_node 2.17% : 0.000035s : 10: substitution.replace_applicator 0.82% : 0.000013s : 12: substitution.replace_old_param 1.38% : 0.000022s : 6: substitution.reshape_eliminate 0.34% : 0.000005s : 1: substitution.set_cell_output_no_recompute 3.97% : 0.000063s : 18: substitution.tuple_list_convert_item_index_to_positive 1.84% : 0.000029s : 18: substitution.tuple_list_get_item_const_eliminator 2.59% : 0.000041s : 18: substitution.tuple_list_get_item_depend_reorder 8.70% : 0.000139s : 46: substitution.tuple_list_get_item_eliminator 2.51% : 0.000040s : 18: substitution.tuple_list_get_set_item_eliminator ------[type_inference.] 0.096388 2 97.08% : 0.093575s : 1: type_inference.infer 2.92% : 0.002813s : 1: type_inference.specialize ------[replace.] 0.000381 40 59.65% : 0.000227s : 20: replace.inline 40.35% : 0.000154s : 20: replace.tuple_list_get_item_eliminator ------[match.] 0.000899 40 94.58% : 0.000850s : 20: match.inline 5.42% : 0.000049s : 20: match.tuple_list_get_item_eliminator ------[predicate.] 0.013359 8180 0.09% : 0.000012s : 89: predicate.accumulaten_eliminater 0.04% : 0.000006s : 17: predicate.ad_related_special_op_eliminate 0.05% : 0.000007s : 51: predicate.addn_check_dump 0.09% : 0.000012s : 89: predicate.addn_zero_filter 0.09% : 0.000012s : 89: predicate.adjust_all_reduce_mul_add 0.20% : 0.000027s : 140: predicate.arithmetic_simplify 0.10% : 0.000013s : 89: predicate.cast_eliminate 0.10% : 0.000014s : 96: predicate.check_bprop_eliminate 0.05% : 0.000007s : 51: predicate.compare_switch_simplify 0.01% : 0.000001s : 17: predicate.const_output_eliminate 0.05% : 0.000007s : 51: predicate.depend_value_elim 0.10% : 0.000013s : 89: predicate.dict_get_item_const_eliminator 0.10% : 0.000014s : 89: predicate.dict_get_item_eliminator 0.09% : 0.000012s : 89: predicate.dict_set_item_eliminator 0.05% : 0.000006s : 34: predicate.dumpgradient_eliminate 0.01% : 0.000002s : 17: predicate.elim_not_effective 0.03% : 0.000004s : 17: predicate.elim_shapecalc_of_broadcastargs 0.11% : 0.000015s : 106: predicate.environ_add_const_eliminate 0.11% : 0.000014s : 106: predicate.environ_get_add_eliminate 0.11% : 0.000015s : 106: predicate.environ_get_depend_swap 0.16% : 0.000022s : 157: predicate.environ_get_eliminate 0.11% : 0.000015s : 106: predicate.environ_get_set_eliminate 0.14% : 0.000018s : 129: predicate.exchange_switch_depend_value 0.20% : 0.000026s : 129: predicate.float_depend_g_call 0.06% : 0.000007s : 51: predicate.float_environ_get_switch 0.08% : 0.000010s : 68: predicate.float_tuple_getitem_switch 0.01% : 0.000001s : 17: predicate.fold_const_symbol 0.06% : 0.000008s : 51: predicate.get_grad_eliminate 0.01% : 0.000002s : 17: predicate.graph_param_transform 0.06% : 0.000008s : 51: predicate.incorporate_call 0.05% : 0.000007s : 51: predicate.incorporate_call_switch 0.51% : 0.000068s : 354: predicate.inline 0.11% : 0.000014s : 74: predicate.inline_without_move 0.03% : 0.000003s : 51: predicate.j_node_and_user_rematch 0.08% : 0.000010s : 51: predicate.less_batch_normalization 0.16% : 0.000021s : 143: predicate.list_to_tuple_eliminator_ 0.23% : 0.000031s : 232: predicate.load_eliminater 0.06% : 0.000008s : 17: predicate.loop_unroll_after_grad 0.18% : 0.000024s : 179: predicate.loop_unroll_before_grad 0.13% : 0.000018s : 123: predicate.make_slice_get_slice_eliminator 0.06% : 0.000007s : 51: predicate.merge_addn 0.10% : 0.000013s : 96: predicate.micro_step_allgather_replace 0.10% : 0.000013s : 96: predicate.mini_step_allgather_replace 0.10% : 0.000013s : 89: predicate.minmaximum_grad 90.66% : 0.012111s : 17: predicate.mutable_eliminate 0.02% : 0.000003s : 17: predicate.opt_reshape 0.03% : 0.000004s : 17: predicate.parallel_virtual_node 0.20% : 0.000027s : 129: predicate.partial_defer_inline 0.14% : 0.000019s : 126: predicate.partial_eliminate 0.09% : 0.000012s : 89: predicate.print_const_string_wrapper 0.05% : 0.000007s : 51: predicate.reduce_all_const_elim 0.12% : 0.000017s : 89: predicate.reduce_eliminate 0.23% : 0.000031s : 232: predicate.redundant_stop_gradient_eliminater 0.03% : 0.000005s : 51: predicate.remove_not_recompute_node 0.15% : 0.000020s : 205: predicate.replace_applicator 0.05% : 0.000007s : 74: predicate.replace_old_param 0.01% : 0.000001s : 17: predicate.reset_defer_inline 0.10% : 0.000013s : 89: predicate.reshape_eliminate 0.10% : 0.000014s : 96: predicate.row_tensor_add_zeros_like 0.02% : 0.000003s : 17: predicate.row_tensor_eliminate 0.12% : 0.000015s : 96: predicate.same_eliminate 0.03% : 0.000004s : 51: predicate.set_cell_output_no_recompute 0.07% : 0.000009s : 51: predicate.shard_identity_eliminate 0.04% : 0.000006s : 34: predicate.special_op_eliminate 0.06% : 0.000008s : 51: predicate.specialize_transform 0.12% : 0.000016s : 96: predicate.split_environ_get_set_with_tuple_value 0.10% : 0.000013s : 74: predicate.stack_unstack_eliminate 0.02% : 0.000003s : 17: predicate.switch_call_monad_eliminater 0.14% : 0.000019s : 129: predicate.switch_defer_inline 0.25% : 0.000034s : 225: predicate.switch_layer_defer_inline 0.41% : 0.000054s : 376: predicate.switch_simplify 0.10% : 0.000013s : 89: predicate.tile_eliminate 0.09% : 0.000012s : 89: predicate.transpose_eliminate 0.32% : 0.000043s : 123: predicate.tuple_list_convert_item_index_to_positive 0.15% : 0.000020s : 123: predicate.tuple_list_get_item_const_eliminator 0.13% : 0.000018s : 123: predicate.tuple_list_get_item_depend_reorder 0.29% : 0.000039s : 194: predicate.tuple_list_get_item_eliminator 0.14% : 0.000019s : 123: predicate.tuple_list_get_set_item_eliminator 0.21% : 0.000027s : 174: predicate.tuple_list_set_item_eliminator 0.15% : 0.000020s : 143: predicate.tuple_to_list_eliminator_ 0.22% : 0.000030s : 232: predicate.updatestate_pure_node_eliminater 0.29% : 0.000038s : 283: predicate.updatestate_useless_node_eliminater 0.02% : 0.000003s : 17: predicate.value_based_eliminate 0.06% : 0.000008s : 51: predicate.virtual_dataset_eliminate 0.06% : 0.000008s : 51: predicate.virtual_output_eliminate 0.02% : 0.000002s : 17: predicate.virtual_view_grad_eliminate 0.03% : 0.000004s : 17: predicate.zero_like_fill_zero ------[func_graph_cloner_run.] 0.002861 43 52.92% : 0.001514s : 19: func_graph_cloner_run.FuncGraphClonerGraph 47.08% : 0.001347s : 24: func_graph_cloner_run.FuncGraphSpecializer ------[meta_graph.] 0.000000 0 ------[manager.] 0.000000 0 ------[pynative] 0.000000 0 ------[others.] 0.206620 233 0.00% : 0.000004s : 1: ForceFp32Comm 1.81% : 0.003736s : 1: add_attr 1.80% : 0.003720s : 1: add_attr_with_inline 0.00% : 0.000004s : 1: add_comm_op_reuse_tag 0.07% : 0.000138s : 1: add_recomputation 0.00% : 0.000004s : 1: assign_add_opt 0.08% : 0.000157s : 1: auto_monad 0.02% : 0.000051s : 1: auto_monad_reorder 0.00% : 0.000004s : 1: begin_end_overlap_inline 0.00% : 0.000006s : 1: bias_add_comm_swap 0.21% : 0.000440s : 1: bootstrap 0.02% : 0.000040s : 1: cconv 0.00% : 0.000004s : 1: comm_op_add_attrs 0.02% : 0.000044s : 1: control_data_broadcast_order 0.01% : 0.000025s : 1: convert_after_rewriter 0.04% : 0.000075s : 1: cse_after_recomputation 0.00% : 0.000006s : 1: dataset_repeat_opt 0.00% : 0.000007s : 1: detach_backward 0.01% : 0.000025s : 1: environ_conv 0.05% : 0.000107s : 1: event_method 0.00% : 0.000005s : 1: full_micro_interleaved_order_control 0.00% : 0.000005s : 1: get_jit_bprop_graph 0.01% : 0.000015s : 1: graph_reusing 0.00% : 0.000005s : 1: grouped_pairwise_exchange_alltoall 0.00% : 0.000005s : 1: handle_group_info 0.00% : 0.000006s : 1: inline 0.00% : 0.000007s : 1: insert-virtual-dataset 0.00% : 0.000004s : 1: interleave_parallel_branches 0.00% : 0.000004s : 1: interleave_split_concat_branches 0.00% : 0.000007s : 1: label_fine_grained_interleaved_index 0.01% : 0.000011s : 1: label_micro_interleaved_index 0.40% : 0.000822s : 1: loop_unroll 0.00% : 0.000005s : 1: merge_cast_opt 0.00% : 0.000006s : 1: micro_interleaved_order_control 6.26% : 0.012936s : 1: mutable_eliminate 0.01% : 0.000013s : 1: offloading_packed_experts 0.02% : 0.000039s : 1: opt.transform.loop_unroll_optimizer 5.89% : 0.012163s : 1: opt.transform.mutable_eliminate 4.07% : 0.008414s : 117: opt.transform.opt_a 0.05% : 0.000101s : 1: opt.transform.opt_after_cconv 0.04% : 0.000073s : 1: opt.transform.opt_after_jit_grad 0.23% : 0.000472s : 28: opt.transform.opt_b 0.12% : 0.000242s : 2: opt.transform.opt_trans_graph 0.09% : 0.000195s : 4: opt.transform.symbol_engine_opt 9.47% : 0.019573s : 1: opt_a 0.14% : 0.000288s : 1: opt_after_cconv 0.42% : 0.000865s : 1: opt_after_jit_grad 0.36% : 0.000745s : 1: opt_b 17.66% : 0.036485s : 1: optimize 0.02% : 0.000050s : 1: optimize_parallel_all_gather_comm 0.01% : 0.000019s : 1: order_py_execute_after_rewriter 0.03% : 0.000056s : 1: overlap_grad_flash_sp 0.00% : 0.000004s : 1: overlap_grad_matmul_and_grad_allreduce 0.01% : 0.000012s : 1: overlap_grad_ring_attention 0.00% : 0.000005s : 1: overlap_opt_shard_grad_in_pipeline 0.00% : 0.000004s : 1: overlap_opt_shard_in_pipeline 0.00% : 0.000006s : 1: overlap_param_gather 0.00% : 0.000004s : 1: overlap_recompute_allgather_and_fa_grad 0.01% : 0.000014s : 1: overlap_recompute_and_grad_model_parallel 0.00% : 0.000006s : 1: overlap_recompute_comm 0.00% : 0.000008s : 1: parallel-infer-symbol 0.00% : 0.000004s : 1: parallel-infer-symbol-second 0.00% : 0.000005s : 1: partial_unused_args_eliminate 0.00% : 0.000005s : 1: pipeline_parallel_scheduler 0.00% : 0.000005s : 1: pipeline_split 0.04% : 0.000081s : 1: pre_auto_parallel 0.03% : 0.000057s : 1: py_interpret_to_execute 0.02% : 0.000032s : 1: py_interpret_to_execute_after_opt_a 0.00% : 0.000004s : 1: remove_cast_before_assign_add 0.06% : 0.000115s : 1: remove_dup_value 1.70% : 0.003514s : 2: renormalize.infer 1.32% : 0.002734s : 2: renormalize.specialize 0.00% : 0.000006s : 1: reorder_send_recv_between_fp_bp 0.00% : 0.000010s : 1: rewriter_after_jit_bprop_graph 0.08% : 0.000156s : 1: rewriter_after_opt_a 0.11% : 0.000222s : 1: rewriter_before_opt_a 0.00% : 0.000006s : 1: slice_cell_reuse_recomputed_activation 0.00% : 0.000006s : 1: slice_recompute_activation 0.00% : 0.000005s : 1: split_layernorm_comm 0.00% : 0.000006s : 1: split_matmul_comm_elemetwise 0.01% : 0.000020s : 1: swap_dp_allreduce_reducescatter 0.25% : 0.000506s : 1: symbol_engine_optimizer 0.14% : 0.000283s : 1: tuple_transform 46.72% : 0.096541s : 1: type_inference Start to compare view out with no view out Compare ['Success'] for 0'th output. *****************End Generalize test GRAPH_MODE_O0 ViewTensor case***************** *****************Start Generalize test GRAPH_MODE_O0 EmptyTensor case***************** [WARNING]: TEST_OP_GENERALIZE GRAPH_MODE_O0 'EmptyTensor' in 'disable_case', EmptyTensor case is skipped. *****************End Generalize test GRAPH_MODE_O0 EmptyTensor case***************** *****************Start Generalize test GRAPH_MODE_O0 ScalarTensor case***************** [INFO]: TEST_OP_GENERALIZE Start to test scalar tensor [INFO]: TEST_OP_GENERALIZE Start to test 1d tensor [INFO]: TEST_OP_GENERALIZE Start to compare scalar tensor out with 1d tensor out Compare ['Success'] for 0'th output. *****************End Generalize test GRAPH_MODE_O0 ScalarTensor case***************** *****************Start Generalize test GRAPH_MODE_O0 GradByRequirement case***************** [INFO]: TEST_OP_GENERALIZE Start to test all grads [INFO]: TEST_OP_GENERALIZE Start to test grad by requirement on position 0 [INFO]: TEST_OP_GENERALIZE Start to test grad by requirement on position 1 [INFO]: TEST_OP_GENERALIZE Start to compare grad by requirement with all grads Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. *****************End Generalize test GRAPH_MODE_O0 GradByRequirement case***************** *****************Start Generalize test GRAPH_MODE_O0 Deterministic case***************** [INFO]: TEST_OP_GENERALIZE test deterministic case loop 0 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 1 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 2 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 3 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE test deterministic case loop 4 [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func forward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare forward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic first time. [INFO]: TEST_OP_GENERALIZE Start to test func backward with deterministic second time. [INFO]: TEST_OP_GENERALIZE Start to compare backward second time out with first time out Compare ['Success'] for 0'th output. Compare ['Success'] for 1'th output. *****************End Generalize test GRAPH_MODE_O0 Deterministic case***************** [WARNING]: TEST_OP_GENERALIZE GRAPH_MODE_GE is skipped. ************************************************************************************************** **************************************TEST_OP_GENERALIZE End************************************** ************************************************************************************************** ************************************************************************************************* **************************************TEST_OP_DYNAMIC Start************************************** ************************************************************************************************* [WARNING]: TEST_DYNAMIC GRAPH_MODE_GE is skipped, Ge backend don't support dynamic situations any longer. *********************Begin DynamicTest for with PYNATIVE_MODE ********************* Start DynamicTest running on PYNATIVE_MODE with static shape with first inputs... End Start DynamicTest running on PYNATIVE_MODE with static shape with second inputs... End Start DynamicTest testing with [PYNATIVE_MODE] [Resize]... Compare ['Success'] for 0'th output of Resize. End Start DynamicTest testing with [PYNATIVE_MODE] [DYNAMIC_SHAPE] [None]... Compare ['Success'] for 0'th output of Forward. Compare ['Success'] for 0'th output of Grad. Compare ['Success'] for 1'th output of Grad. End Start DynamicTest testing with [PYNATIVE_MODE] [DYNAMIC_RANK] [None]... Compare ['Success'] for 0'th output of Forward. Compare ['Success'] for 0'th output of Grad. Compare ['Success'] for 1'th output of Grad. End *********************End DynamicTest for with PYNATIVE_MODE ********************* *********************Begin DynamicTest for with GRAPH_MODE_O0 ********************* Start DynamicTest running on GRAPH_MODE_O0 with static shape with first inputs... End Start DynamicTest running on GRAPH_MODE_O0 with static shape with second inputs... End Start DynamicTest testing with [GRAPH_MODE_O0] [Resize]... Compare ['Success'] for 0'th output of Resize. End Start DynamicTest testing with [GRAPH_MODE_O0] [DYNAMIC_SHAPE] [None]... Compare ['Success'] for 0'th output of Forward. Compare ['Success'] for 0'th output of Grad. Compare ['Success'] for 1'th output of Grad. End Start DynamicTest testing with [GRAPH_MODE_O0] [DYNAMIC_RANK] [None]... Compare ['Success'] for 0'th output of Forward. Compare ['Success'] for 0'th output of Grad. Compare ['Success'] for 1'th output of Grad. End *********************End DynamicTest for with GRAPH_MODE_O0 ********************* ************************************************************************************************* ***************************************TEST_OP_DYNAMIC End*************************************** ************************************************************************************************* . [hook] pytest_runtest_teardown:test_mint_mul_dynamic_shape_test_op[kbk] tests/st/mint/test_functional_mul.py::test_mint_mul_dynamic_shape_test_op[kbk],max_mem:4.0M =============================== warnings summary =============================== ../../../../../../../../usr/local/Ascend/cann-8.5.0/python/site-packages/tbe/dsl/classifier/transdata/transdata_classifier.py:222 /usr/local/Ascend/cann-8.5.0/python/site-packages/tbe/dsl/classifier/transdata/transdata_classifier.py:222: DeprecationWarning: invalid escape sequence \B """ ../../../../../../../../usr/local/Ascend/cann-8.5.0/python/site-packages/tbe/dsl/unify_schedule/vector/transdata/common/graph/transdata_graph_info.py:143 /usr/local/Ascend/cann-8.5.0/python/site-packages/tbe/dsl/unify_schedule/vector/transdata/common/graph/transdata_graph_info.py:143: DeprecationWarning: invalid escape sequence \c """ ../../../../../../../../usr/local/Ascend/cann-8.5.0/python/site-packages/tbe/dsl/unify_schedule/vector/transdata/common/graph/transdata_graph_info.py:170 /usr/local/Ascend/cann-8.5.0/python/site-packages/tbe/dsl/unify_schedule/vector/transdata/common/graph/transdata_graph_info.py:170: DeprecationWarning: invalid escape sequence \c """ ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/numpy/core/getlimits.py:549 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/numpy/core/getlimits.py:549: UserWarning: The value of the smallest subnormal for type is zero. setattr(self, word, getattr(machar, word).flat[0]) ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/numpy/core/getlimits.py:89 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/numpy/core/getlimits.py:89: UserWarning: The value of the smallest subnormal for type is zero. return self._float_to_str(self.smallest_subnormal) ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/numpy/core/getlimits.py:549 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/numpy/core/getlimits.py:549: UserWarning: The value of the smallest subnormal for type is zero. setattr(self, word, getattr(machar, word).flat[0]) ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/numpy/core/getlimits.py:89 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/numpy/core/getlimits.py:89: UserWarning: The value of the smallest subnormal for type is zero. return self._float_to_str(self.smallest_subnormal) ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/batchnorm_fold2.py:57 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/batchnorm_fold2.py:57: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("batchnorm_fold2") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/batchnorm_fold2_grad.py:56 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/batchnorm_fold2_grad.py:56: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("batchnorm_fold2_grad") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/batchnorm_fold2_grad_reduce.py:48 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/batchnorm_fold2_grad_reduce.py:48: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("batchnorm_fold2_grad_reduce") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/correction_mul.py:51 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/correction_mul.py:51: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("correction_mul") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/correction_mul_grad.py:51 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/correction_mul_grad.py:51: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("correction_mul_grad") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/correction_mul_grad.py:143 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/correction_mul_grad.py:143: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("correction_mul_grad_reduce") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perlayer.py:50 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perlayer.py:50: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("fake_learned_scale_quant_perlayer") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perlayer_grad.py:92 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perlayer_grad.py:92: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("fake_learned_scale_quant_perlayer_grad_d") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perlayer_grad_reduce.py:49 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perlayer_grad_reduce.py:49: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("fake_learned_scale_quant_perlayer_grad_d_reduce") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perchannel.py:50 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perchannel.py:50: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("fake_learned_scale_quant_perchannel") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perchannel_grad.py:91 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perchannel_grad.py:91: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("fake_learned_scale_quant_perchannel_grad_d") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perchannel_grad_reduce.py:48 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_learned_scale_quant_perchannel_grad_reduce.py:48: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("fake_learned_scale_quant_perchannel_grad_d_reduce") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_quant_perchannel.py:52 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_quant_perchannel.py:52: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("fake_quant_perchannel") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_quant_perchannel_grad.py:81 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_quant_perchannel_grad.py:81: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("fake_quant_perchannel_grad") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_quant_perlayer.py:54 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_quant_perlayer.py:54: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("fake_quant_per_layer") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_quant_perlayer_grad.py:81 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/fake_quant_perlayer_grad.py:81: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("fake_quant_per_layer_grad") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/minmax_update_perchannel.py:50 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/minmax_update_perchannel.py:50: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("minmax_update_perchannel") ../../../../../../anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/minmax_update_perlayer.py:50 /home/jenkins/anaconda3/envs/ci39/lib/python3.9/site-packages/mindspore/ops/_op_impl/_custom_op/minmax_update_perlayer.py:50: DeprecationWarning: te_fusion.fusion_manager.fusion_manager.register is deprecated,please replace it with tbe.common.register.register_op_compute @fusion_manager.register("minmax_update_perlayer") test_functional_mul.py::test_mint_mul_dynamic_shape_test_op[pynative] /usr/local/Ascend/cann-8.5.0/python/site-packages/asc_op_compile_base/asc_op_compiler/ascendc_compile_gen_code.py:161: DeprecationWarning: invalid escape sequence \w match = re.search(f'{option}=(\w+)', ' '.join(compile_options)) -- Docs: https://docs.pytest.org/en/stable/warnings.html ================== 2 passed, 26 warnings in 524.85s (0:08:44) ==================