ATC命令及结果如下

(Ascend) root@is-dau5hkvezctnl5fz-devmachine-0:/mnt/public/yixiaojie/DINOv# atc --framework=5 \ 

    --model=/mnt/public/yixiaojie/dinov/onnx/dinov_swinL1_fixed.onnx \ 

    --output=/mnt/public/yixiaojie/dinov/om/dinov_swinL1-310B1 \ 

    --input_shape="tgt_image:1,3,672,672;rand_shape:1,1,672,672" \ 

    --soc_version=Ascend310B1 \ 

    --host_env_os=linux \ 

    --host_env_cpu=aarch64   

ATC start working now, please wait for a moment. 

Warning: tiling offset out of range, index: 32 

Warning: tiling offset out of range, index: 32 

Warning: tiling offset out of range, index: 32 

Warning: tiling offset out of range, index: 32 

Warning: tiling offset out of range, index: 32 

Warning: tiling offset out of range, index: 32 

Warning: tiling offset out of range, index: 32 

Warning: tiling offset out of range, index: 32 

Warning: tiling offset out of range, index: 32 

Warning: tiling offset out of range, index: 32 

Warning: set_tiling_params does not take effect because get_op_mode is dynamic 

... 

ATC run success, welcome to the next use.

在310B上推理如下：

(base) root@orangepiaipro-20t:/home/yixiaojie/dinov# python -m ais_bench --model ./dinov_swinL1-310B1_linux_aarch64.om --output ./ --output_dirname resultdinov --outfmt TXT --batchsize 1 --warmup_count 1 --loop 10 --profiler 1 

[INFO] msprof cmd:/usr/local/Ascend/ascend-toolkit/latest/bin/msprof --output=.//profiler --application="/usr/local/miniconda3/bin/python /usr/local/miniconda3/lib/python3.9/site-packages/ais_bench/__main__.py --model ./dinov_swinL1-310B1_linux_aarch64.om --output ./ --output_dirname resultdinov --outfmt TXT --batchsize 1 --warmup_count 1 --loop 10 --profiler 1 --profiler=0 --warmup_count=0" --model-execution=on --sys-hardware-mem=on --sys-cpu-profiling=off --sys-profiling=off --sys-pid-profiling=off --dvpp-profiling=on --runtime-api=on --task-time=on --aicpu=on begin run 

[INFO] Start profiling... 

[INFO] acl init success 

[INFO] open device 0 success 

E19999: Inner Error! 

E19999: 2025-06-12-14:46:19.695.051  Failed to execute init graph[FUNC:Load][FILE:model_v2_executor.cc][LINE:171] 

        TraceBack (most recent call last): 

        [Model][FromData]call load executorV2 failed, ge result[1343225857][FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161] 

 

[ERROR] load model from file failed, model file is /home/yixiaojie/dinov/dinov_swinL1-310B1_linux_aarch64.om 

[WARN] Check failed:processModel->LoadModelFromFile(modelPath), ret:1 

 

[WARN] no model had been loaded, unload failed 

Traceback (most recent call last): 

  File "/usr/local/miniconda3/lib/python3.9/site-packages/ais_bench/__main__.py", line 3, in <module> 

    exec(open(os.path.join(cur_path, "infer/__main__.py")).read()) 

  File "<string>", line 448, in <module> 

  File "<string>", line 285, in main 

  File "<string>", line 53, in init_inference_session 

  File "/usr/local/miniconda3/lib/python3.9/site-packages/ais_bench/infer/interface.py", line 26, in __init__ 

    self.session = aclruntime.InferenceSession(self.model_path, self.device_id, options) 

RuntimeError: [1][ACL: invalid parameter] 

[INFO] end to destroy context 

[INFO] end to reset device is 0 

不知道是什么原因，应该是我的onnx文件的问题，还是ATC的问题呢