Hi 这边用ATC工具转换onnx模型,显示LayerNorm op转换失败,cann版本6.0.1,麻烦帮忙看下,感谢!
模型文件:模型文件:https://bj.bcebos.com/v1/aipe-easyedge-test/wanglichang/test/params_converNext-T.onnx?authorization=bce-auth-v1%2F50c8bb753dcb4e1d8646bb1ffefd3503%2F2023-05-30T12%3A05%3A07Z%2F1800%2Fhost%2F180a9f521f6d4f228845c45201171f0e03ede89190948634c65b0d851f441e64
转换命令:
atc --model=/home/work/params --framework=5 --output=/tmp/output_0/params --precision_mode=allow_mix_precision --soc_version=Ascend310 --input_format=NCHW --output_type=FP32 --input_shape=input:1,3,224,224
错误日志:
ATC start working now, please wait for a moment.
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - ATC run failed, Please check the detail log, Try 'atc --help' for more information
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - E49999: Inner Error!
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - E49999 Pre-compile op[/downsample_layers.0/downsample_layers.0.1/Div/Mul/Add/LayerNorm] failed, oppath[/usr/local/ascend-toolkit/6.0.RC1/opp/op_impl/built-in/ai_core/tbe/impl/layer_norm.py], optype[LayerNorm], taskID[10]. Please check op's compilation error message.[FUNC:ReportBuildErrMessage][FILE:fusion_op.cc][LINE:856]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - Pre-compile op[/downsample_layers.2/downsample_layers.2.0/Div/Mul/Add/LayerNorm] failed, oppath[/usr/local/ascend-toolkit/6.0.RC1/opp/op_impl/built-in/ai_core/tbe/impl/layer_norm.py], optype[LayerNorm], taskID[93]. Please check op's compilation error message.[FUNC:ReportBuildErrMessage][FILE:fusion_op.cc][LINE:856]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - Pre-compile op[/downsample_layers.3/downsample_layers.3.0/Div/Mul/Add/LayerNorm] failed, oppath[/usr/local/ascend-toolkit/6.0.RC1/opp/op_impl/built-in/ai_core/tbe/impl/layer_norm.py], optype[LayerNorm], taskID[207]. Please check op's compilation error message.[FUNC:ReportBuildErrMessage][FILE:fusion_op.cc][LINE:856]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp][Node /downsample_layers.2/downsample_layers.2.0/Div/Mul/Add/LayerNorm] Failed to pre-compile. Tid is [140155845338880], TaskId is [93] [FUNC:ProcessFailPreCompTask][FILE:tbe_op_store_adapter.cc][LINE:171]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp][Node /downsample_layers.3/downsample_layers.3.0/Div/Mul/Add/LayerNorm] Failed to pre-compile. Tid is [140155845338880], TaskId is [207] [FUNC:ProcessFailPreCompTask][FILE:tbe_op_store_adapter.cc][LINE:171]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp][Node /downsample_layers.1/downsample_layers.1.0/Div/Mul/Add/LayerNorm] Failed to pre-compile. Tid is [140155845338880], TaskId is [51] [FUNC:ProcessFailPreCompTask][FILE:tbe_op_store_adapter.cc][LINE:171]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp][Node /downsample_layers.0/downsample_layers.0.1/Div/Mul/Add/LayerNorm] Failed to pre-compile. Tid is [140155845338880], TaskId is [10] [FUNC:ProcessFailPreCompTask][FILE:tbe_op_store_adapter.cc][LINE:171]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp]Failed to process failed task. Thread_id is [140155845338880].[FUNC:ParallelPreCompileOp][FILE:tbe_op_store_adapter.cc][LINE:445]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp]Failed to pre-compile graph [partition0_rank1_new_sub_graph2][FUNC:PreCompileOp][FILE:op_compiler.cc][LINE:607]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - Call OptimizeFusedGraph failed, ret:-1, engine_name:AIcoreEngine, graph_name:partition0_rank1_new_sub_graph2[FUNC:OptimizeSubGraph][FILE:graph_optimize.cc][LINE:130]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - subgraph 0 optimize failed[FUNC:OptimizeSubGraphWithMultiThreads][FILE:graph_manager.cc][LINE:714]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - build graph failed, graph id:0, ret:-1[FUNC:BuildModel][FILE:ge_generator.cc][LINE:1496]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - Solution: Please contact support engineer.
模型结构:

Hi 这边用ATC工具转换onnx模型,显示LayerNorm op转换失败,cann版本6.0.1,麻烦帮忙看下,感谢!
模型文件:模型文件:https://bj.bcebos.com/v1/aipe-easyedge-test/wanglichang/test/params_converNext-T.onnx?authorization=bce-auth-v1%2F50c8bb753dcb4e1d8646bb1ffefd3503%2F2023-05-30T12%3A05%3A07Z%2F1800%2Fhost%2F180a9f521f6d4f228845c45201171f0e03ede89190948634c65b0d851f441e64
转换命令:
atc --model=/home/work/params --framework=5 --output=/tmp/output_0/params --precision_mode=allow_mix_precision --soc_version=Ascend310 --input_format=NCHW --output_type=FP32 --input_shape=input:1,3,224,224
错误日志:
ATC start working now, please wait for a moment.
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - ATC run failed, Please check the detail log, Try 'atc --help' for more information
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - E49999: Inner Error!
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - E49999 Pre-compile op[/downsample_layers.0/downsample_layers.0.1/Div/Mul/Add/LayerNorm] failed, oppath[/usr/local/ascend-toolkit/6.0.RC1/opp/op_impl/built-in/ai_core/tbe/impl/layer_norm.py], optype[LayerNorm], taskID[10]. Please check op's compilation error message.[FUNC:ReportBuildErrMessage][FILE:fusion_op.cc][LINE:856]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - Pre-compile op[/downsample_layers.2/downsample_layers.2.0/Div/Mul/Add/LayerNorm] failed, oppath[/usr/local/ascend-toolkit/6.0.RC1/opp/op_impl/built-in/ai_core/tbe/impl/layer_norm.py], optype[LayerNorm], taskID[93]. Please check op's compilation error message.[FUNC:ReportBuildErrMessage][FILE:fusion_op.cc][LINE:856]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - Pre-compile op[/downsample_layers.3/downsample_layers.3.0/Div/Mul/Add/LayerNorm] failed, oppath[/usr/local/ascend-toolkit/6.0.RC1/opp/op_impl/built-in/ai_core/tbe/impl/layer_norm.py], optype[LayerNorm], taskID[207]. Please check op's compilation error message.[FUNC:ReportBuildErrMessage][FILE:fusion_op.cc][LINE:856]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp][Node /downsample_layers.2/downsample_layers.2.0/Div/Mul/Add/LayerNorm] Failed to pre-compile. Tid is [140155845338880], TaskId is [93] [FUNC:ProcessFailPreCompTask][FILE:tbe_op_store_adapter.cc][LINE:171]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp][Node /downsample_layers.3/downsample_layers.3.0/Div/Mul/Add/LayerNorm] Failed to pre-compile. Tid is [140155845338880], TaskId is [207] [FUNC:ProcessFailPreCompTask][FILE:tbe_op_store_adapter.cc][LINE:171]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp][Node /downsample_layers.1/downsample_layers.1.0/Div/Mul/Add/LayerNorm] Failed to pre-compile. Tid is [140155845338880], TaskId is [51] [FUNC:ProcessFailPreCompTask][FILE:tbe_op_store_adapter.cc][LINE:171]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp][Node /downsample_layers.0/downsample_layers.0.1/Div/Mul/Add/LayerNorm] Failed to pre-compile. Tid is [140155845338880], TaskId is [10] [FUNC:ProcessFailPreCompTask][FILE:tbe_op_store_adapter.cc][LINE:171]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp]Failed to process failed task. Thread_id is [140155845338880].[FUNC:ParallelPreCompileOp][FILE:tbe_op_store_adapter.cc][LINE:445]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - [SubGraphOpt][Pre-Comp]Failed to pre-compile graph [partition0_rank1_new_sub_graph2][FUNC:PreCompileOp][FILE:op_compiler.cc][LINE:607]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - Call OptimizeFusedGraph failed, ret:-1, engine_name:AIcoreEngine, graph_name:partition0_rank1_new_sub_graph2[FUNC:OptimizeSubGraph][FILE:graph_optimize.cc][LINE:130]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - subgraph 0 optimize failed[FUNC:OptimizeSubGraphWithMultiThreads][FILE:graph_manager.cc][LINE:714]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - build graph failed, graph id:0, ret:-1[FUNC:BuildModel][FILE:ge_generator.cc][LINE:1496]
[2023-05-30, 23:00:22 CST] {pod_manager.py:226} INFO - Solution: Please contact support engineer.
模型结构: