Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Check failed: registry_.count(op_type) != 0 Conv2D operation is not registered. op_type=Conv2D #782

Open
Augenstern-create opened this issue Jun 21, 2023 · 0 comments

Comments

@Augenstern-create
Copy link

System information

-**环境配置docker pull registry.cn-hangzhou.aliyuncs.com/xiaomimace/mace-dev

  • OS Platform and Distribution:20.04.1-Ubuntu
  • NDK version:android-ndk-r18b
  • GCC version:gcc version 5.4.0 20160609 (Ubuntu 5.4.0-6ubuntu1~16.04.11)
  • MACE version:mace-master.zip/2323-6-21
  • Python version: Python 2.7.12
  • Bazel version:Build label: 0.16.0

Model deploy file (*.yml)

mace-models-master/inception-v3/inception-v3-quantize-retrain-dsp.yml

library_name: inception-v3-quantize-retrain-dsp
target_abis: [armeabi-v7a]
target_socs: [sdm660]
model_graph_format: file
model_data_format: file
models:
  inception_v3_quantize_retrain_dsp:
    platform: tensorflow
    model_file_path: https://cnbj1.fds.api.xiaomi.com/mace/miai-models/inception-v3/inception_v3_quant_frozen.pb
    model_sha256_checksum: 8247ba8c753128da5086a8a5eb85191e2b6bde59d4ba7435b9d07ae053ee3e2e
    subgraphs:
      - input_tensors:
          - input
        input_shapes:
          - 1,299,299,3
        input_ranges:
          - -1.0,1.0
        output_tensors:
          - output
        output_shapes:
          - 1,1001
        check_tensors:
          - InceptionV3/Logits/Conv2d_1c_1x1/BiasAdd:0
        check_shapes:
          - 1,1,1,1001
        validation_inputs_data:
          - https://cnbj1.fds.api.xiaomi.com/mace/inputs/keyboard_299x299.bin
    runtime: dsp
    limit_opencl_kernel_time: 0
    nnlib_graph_mode: 0
    obfuscate: 0
    winograd: 0
    quantize: 1

Describe the problem

Check failed: registry_.count(op_type) != 0 Conv2D operation is not registered. op_type=Conv2D

To Reproduce

Steps to reproduce the problem:

1.docker pull registry.cn-hangzhou.aliyuncs.com/xiaomimace/mace-dev
2.docker run -it --privileged --name MACE -v /dev/bus/usb:/dev/bus/usb --net=host -v /home/lansong/public:/public registry.cn-hangzhou.aliyuncs.com/xiaomimace/mace-dev /bin/bash
"/home/lansong/public:/public"文件夹中有mace-master与mace-models-master
3.export ANDROID_NDK_HOME=/opt/android-ndk-r18b
4.cd /public/mace-master
5.tools/clear_workspace.sh [--expunge]
6.RUNTIME=HEXAGON QUANTIZE=OFF bash tools/cmake/cmake-build-armeabi-v7a.sh
7.python tools/python/convert.py --config ../mace-models-master/inception-v3/inception-v3-quantize-retrain-dsp.yml
8.python tools/python/run_model.py --config ../mace-models-master/inception-v3/inception-v3-quantize-retrain-dsp.yml --validate vlog_level=2 --runtime=dsp

Error information / logs

Please include the full log and/or traceback here.

#python tools/python/run_model.py --config ../mace-models-master/inception-v3/inception-v3-quantize-retrain-dsp.yml --validate vlog_level=2 --runtime=dsp
CMD> bazel build //mace/proto:mace_py
CMD> bazel build //mace/proto:mace_py
WARNING: --batch mode is deprecated. Please instead explicitly shut down your Bazel server using the command "bazel shutdown".
Loading:
Loading: 0 packages loaded
INFO: Analysed target //mace/proto:mace_py (17 packages loaded).
INFO: Found 1 target...
[0 / 22] [-----] BazelWorkspaceStatusAction stable-status.txt
Target //mace/proto:mace_py up-to-date:
bazel-genfiles/mace/proto/mace_pb2.py
INFO: Elapsed time: 1.238s, Critical Path: 0.02s
INFO: 0 processes.
INFO: Build completed successfully, 1 total action
INFO: Build completed successfully, 1 total action

CMD> cp -f bazel-genfiles/mace/proto/mace_pb2.py /public/mace-master/tools/python/py_proto
CMD> cp -f bazel-genfiles/mace/proto/mace_pb2.py /public/mace-master/tools/python/py_proto

CMD> bazel build //mace/proto:micro_mem_py
CMD> bazel build //mace/proto:micro_mem_py
WARNING: --batch mode is deprecated. Please instead explicitly shut down your Bazel server using the command "bazel shutdown".
Loading:
Loading: 0 packages loaded
INFO: Analysed target //mace/proto:micro_mem_py (17 packages loaded).
INFO: Found 1 target...
[2 / 9] [-----] BazelWorkspaceStatusAction stable-status.txt
Target //mace/proto:micro_mem_py up-to-date:
bazel-genfiles/mace/proto/micro_mem_pb2.py
INFO: Elapsed time: 1.220s, Critical Path: 0.02s
INFO: 0 processes.
INFO: Build completed successfully, 1 total action
INFO: Build completed successfully, 1 total action

CMD> cp -f bazel-genfiles/mace/proto/micro_mem_pb2.py /public/mace-master/tools/python/py_proto
CMD> cp -f bazel-genfiles/mace/proto/micro_mem_pb2.py /public/mace-master/tools/python/py_proto

CMD> bazel build //third_party/caffe:caffe_py
CMD> bazel build //third_party/caffe:caffe_py
WARNING: --batch mode is deprecated. Please instead explicitly shut down your Bazel server using the command "bazel shutdown".
Loading:
Loading: 0 packages loaded
INFO: Analysed target //third_party/caffe:caffe_py (17 packages loaded).
INFO: Found 1 target...
[0 / 17] [-----] BazelWorkspaceStatusAction stable-status.txt
Target //third_party/caffe:caffe_py up-to-date:
bazel-genfiles/third_party/caffe/caffe_pb2.py
INFO: Elapsed time: 1.270s, Critical Path: 0.02s
INFO: 0 processes.
INFO: Build completed successfully, 1 total action
INFO: Build completed successfully, 1 total action

CMD> cp -f bazel-genfiles/third_party/caffe/caffe_pb2.py /public/mace-master/tools/python/py_proto
CMD> cp -f bazel-genfiles/third_party/caffe/caffe_pb2.py /public/mace-master/tools/python/py_proto

CMD> adb devices
2
tools/python/run_model.py:76: Run on devices: ['d4f9f60b']
2
tools/python/run_model.py:90: Run model inception_v3_quantize_retrain_dsp
{'model_sha256_checksum': '8247ba8c753128da5086a8a5eb85191e2b6bde59d4ba7435b9d07ae053ee3e2e', 'input_tensors': ['input'], 'data_type': 1, 'limit_opencl_kernel_time': 0, 'input_data_formats': [], 'winograd': 0, 'nnlib_graph_mode': 0, 'subgraphs': {'default_graph': {'output_data_types': [1], 'model_sha256_checksum': '8247ba8c753128da5086a8a5eb85191e2b6bde59d4ba7435b9d07ae053ee3e2e', 'data_type': 1, 'input_data_types': [1], 'output_tensors': ['output'], 'input_tensors': ['input'], 'input_shapes': [[1, 299, 299, 3]], 'quantize': 1, 'output_data_formats': [<DataFormat.NHWC: 1>], 'model_file_path': 'https://cnbj1.fds.api.xiaomi.com/mace/miai-models/inception-v3/inception_v3_quant_frozen.pb', 'check_tensors': ['InceptionV3/Logits/Conv2d_1c_1x1/BiasAdd:0'], 'input_ranges': [[-1.0, 1.0]], 'output_shapes': [[1, 1001]], 'limit_opencl_kernel_time': 0, 'input_data_formats': [<DataFormat.NHWC: 1>], 'winograd': 0, 'nnlib_graph_mode': 0, 'platform': <Platform.TENSORFLOW: 0>, 'obfuscate': 0, 'validation_inputs_data': ['https://cnbj1.fds.api.xiaomi.com/mace/inputs/keyboard_299x299.bin'], 'check_shapes': [[1, 1, 1, 1001]], 'runtime': <DeviceType.HEXAGON: 3>}}, 'quantize': 1, 'obfuscate': 0, 'output_data_formats': [], 'platform': <Platform.TENSORFLOW: 0>, 'output_tensors': ['output'], 'validation_inputs_data': ['https://cnbj1.fds.api.xiaomi.com/mace/inputs/keyboard_299x299.bin'], 'runtime': <DeviceType.HEXAGON: 3>, 'input_ranges': [], 'model_file_path': 'https://cnbj1.fds.api.xiaomi.com/mace/miai-models/inception-v3/inception_v3_quant_frozen.pb'}
CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/interior
CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/interior

CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp
CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp

CMD> adb -s d4f9f60b push build/inception_v3_quantize_retrain_dsp/model/inception_v3_quantize_retrain_dsp.pb /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp
CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp
CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp

CMD> adb -s d4f9f60b push build/inception_v3_quantize_retrain_dsp/model/inception_v3_quantize_retrain_dsp.data /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp
2
/public/mace-master/tools/python/utils/util.py:157: Downloading file https://cnbj1.fds.api.xiaomi.com/mace/inputs/keyboard_299x299.bin to /tmp/tmpuNJ9dS/inception_v3_quantize_retrain_dsp_input, please wait ...
2
/public/mace-master/tools/python/utils/util.py:167: Model downloaded successfully.
CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_in
CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_in

CMD> adb -s d4f9f60b push /tmp/tmpuNJ9dS/* /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_in
CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_out
CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_out

Install target from build/cmake-build/armeabi-v7a/install/bin/mace_run to /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp
CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp
CMD> adb -s d4f9f60b shell mkdir -p /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp

CMD> adb -s d4f9f60b push build/cmake-build/armeabi-v7a/install/bin/mace_run /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp
CMD> adb -s d4f9f60b push third_party/nnlib/armeabi-v7a/libhexagon_controller.so /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp
CMD> adb -s d4f9f60b push build/cmake-build/armeabi-v7a/install/lib/libc++_shared.so /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp
['', 'MACE_INTERNAL_STORAGE_PATH=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/interior', 'MACE_CPP_MIN_VLOG_LEVEL=0', 'LD_LIBRARY_PATH=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp', '/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/mace_run', "--input_data_format='NHWC'", "--output_node='output'", "--output_shape='1,1001'", "--model_data_file='/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/inception_v3_quantize_retrain_dsp.data'", "--input_node='input'", "--model_file='/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/inception_v3_quantize_retrain_dsp.pb'", "--model_name='inception_v3_quantize_retrain_dsp'", "--output_data_format='NHWC'", "--input_shape='1,299,299,3'", 'vlog_level=2', '--input_file=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_in/inception_v3_quantize_retrain_dsp', '--output_file=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_out/inception_v3_quantize_retrain_dsp']
 MACE_INTERNAL_STORAGE_PATH=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/interior MACE_CPP_MIN_VLOG_LEVEL=0 LD_LIBRARY_PATH=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/mace_run --input_data_format='NHWC' --output_node='output' --output_shape='1,1001' --model_data_file='/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/inception_v3_quantize_retrain_dsp.data' --input_node='input' --model_file='/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/inception_v3_quantize_retrain_dsp.pb' --model_name='inception_v3_quantize_retrain_dsp' --output_data_format='NHWC' --input_shape='1,299,299,3' vlog_level=2 --input_file=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_in/inception_v3_quantize_retrain_dsp --output_file=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_out/inception_v3_quantize_retrain_dsp
['', 'MACE_INTERNAL_STORAGE_PATH=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/interior', 'MACE_CPP_MIN_VLOG_LEVEL=0', 'LD_LIBRARY_PATH=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp', '/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/mace_run', "--input_data_format='NHWC'", "--output_node='output'", "--output_shape='1,1001'", "--model_data_file='/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/inception_v3_quantize_retrain_dsp.data'", "--input_node='input'", "--model_file='/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/inception_v3_quantize_retrain_dsp.pb'", "--model_name='inception_v3_quantize_retrain_dsp'", "--output_data_format='NHWC'", "--input_shape='1,299,299,3'", 'vlog_level=2', '--input_file=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_in/inception_v3_quantize_retrain_dsp', '--output_file=/data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_out/inception_v3_quantize_retrain_dsp']
CMD> adb -s d4f9f60b push /tmp/tmpADD9nH/cmd.sh /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp
CMD> adb -s d4f9f60b push /tmp/tmpADD9nH/cmd.sh /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp
15 KB/s (944 bytes in 0.058s)

CMD> adb -s d4f9f60b shell sh /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/cmd.sh
CMD> adb -s d4f9f60b shell sh /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/cmd.sh
I /public/mace-master/mace/tools/mace_run.cc:719] model name: inception_v3_quantize_retrain_dsp
I /public/mace-master/mace/tools/mace_run.cc:720] mace version: unknown
I /public/mace-master/mace/tools/mace_run.cc:721] input node: input
I /public/mace-master/mace/tools/mace_run.cc:722] input shape: 1,299,299,3
I /public/mace-master/mace/tools/mace_run.cc:723] input data_type: float32
I /public/mace-master/mace/tools/mace_run.cc:724] input data_format: NHWC
I /public/mace-master/mace/tools/mace_run.cc:725] output node: output
I /public/mace-master/mace/tools/mace_run.cc:726] output shape: 1,1001
I /public/mace-master/mace/tools/mace_run.cc:727] output data_format: NHWC
I /public/mace-master/mace/tools/mace_run.cc:728] input_file: /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_in/inception_v3_quantize_retrain_dsp
I /public/mace-master/mace/tools/mace_run.cc:729] output_file: /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_out/inception_v3_quantize_retrain_dsp
I /public/mace-master/mace/tools/mace_run.cc:730] input dir:
I /public/mace-master/mace/tools/mace_run.cc:731] output dir: output
I /public/mace-master/mace/tools/mace_run.cc:732] model_data_file: /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/inception_v3_quantize_retrain_dsp.data
I /public/mace-master/mace/tools/mace_run.cc:733] model_file: /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/inception_v3_quantize_retrain_dsp.pb
I /public/mace-master/mace/tools/mace_run.cc:734] accelerator_cache_policy: 0
I /public/mace-master/mace/tools/mace_run.cc:735] accelerator_binary_file:
I /public/mace-master/mace/tools/mace_run.cc:736] accelerator_storage_file:
I /public/mace-master/mace/tools/mace_run.cc:737] apu_boost_hint: 100
I /public/mace-master/mace/tools/mace_run.cc:738] apu_preference_hint: 1
I /public/mace-master/mace/tools/mace_run.cc:739] round: 1
I /public/mace-master/mace/tools/mace_run.cc:740] restart_round: 1
I /public/mace-master/mace/tools/mace_run.cc:741] gpu_perf_hint: 3
I /public/mace-master/mace/tools/mace_run.cc:742] gpu_priority_hint: 3
I /public/mace-master/mace/tools/mace_run.cc:743] num_threads: -1
I /public/mace-master/mace/tools/mace_run.cc:744] cpu_affinity_policy: 1
I /public/mace-master/mace/tools/mace_run.cc:786] raw_output_data_types[0] is float32
I /public/mace-master/mace/core/memory/rpcmem/rpcmem.cc:32] Rpcmem is supported. type: 0
I /public/mace-master/mace/rpcmems/rpcmem_factory.cc:53] Rpcmem is supported. type: 0
I /public/mace-master/mace/libmace/engines/serial_engine.cc:32] Creating SerialEngine, MACE version: unknown
F /public/mace-master/mace/core/registry/ops_registry.cc:115] Check failed: registry_.count(op_type) != 0 Conv2D operation is not registered. op_type=Conv2D
F /public/mace-master/mace/core/registry/ops_registry.cc:115] backtrace:
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x96f9562
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x96fb728
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x96fb6fa
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x96fb886
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x96fb8c2
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x971ffc4
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x97388f0
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x9737420
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x972f80a
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x9729844
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x97280de
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x9727efc
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x9702aa8
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x9702c88
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x9705288
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x9705666
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x96f5df0
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0x96f6074
F /public/mace-master/mace/core/registry/ops_registry.cc:115]  pc 0xe7f7948c __libc_init
Aborted

2
/public/mace-master/tools/python/utils/util.py:157: Downloading file https://cnbj1.fds.api.xiaomi.com/mace/miai-models/inception-v3/inception_v3_quant_frozen.pb to /tmp/tmpuNJ9dS/inception_v3_quant_frozen.pb-8247ba8c753128da5086a8a5eb85191e2b6bde59d4ba7435b9d07ae053ee3e2e.pb, please wait ...
2
/public/mace-master/tools/python/utils/util.py:167: Model downloaded successfully.
CMD> adb -s d4f9f60b pull /data/local/tmp/mace_run/inception_v3_quantize_retrain_dsp/validate_out /tmp/tmpuNJ9dS/validate_out
2023-06-21 03:46:36.566752: I tensorflow/core/platform/cpu_feature_guard.cc:140] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 AVX512F FMA
******************************************************************
          Similarity Test failed because of empty output
******************************************************************

Traceback (most recent call last):
  File "tools/python/run_model.py", line 418, in <module>
    run_models(flags, args)
  File "tools/python/run_model.py", line 83, in run_models
    run_models_for_device(flags, args, dev)
  File "tools/python/run_model.py", line 92, in run_models_for_device
    run_model_for_device(flags, args, dev, name, model_conf)
  File "tools/python/run_model.py", line 281, in run_model_for_device
    "")
  File "/public/mace-master/tools/python/validate.py", line 543, in validate
    log_file)
  File "/public/mace-master/tools/python/validate.py", line 247, in validate_tf_model
    real_out_shape, real_out_data_format)
  File "/public/mace-master/tools/python/validate.py", line 116, in compare_output
    "Similarity Test failed because of empty output"))
  File "/public/mace-master/tools/python/utils/util.py", line 71, in error
    + CMDColors.ENDC)
  File "/public/mace-master/tools/python/utils/util.py", line 45, in get_frame_info
    caller_frame = inspect.stack()[int(level)]
ValueError: invalid literal for int() with base 10: '******************************************************************\n          Similarity Test failed because of empty output          \n******************************************************************'

Additional context

mace-master/mace/flows/CMakeLists.txt中

添加
if(MACE_ENABLE_HEXAGON)
  target_link_libraries(flows PRIVATE hexagon_flow)
endif(MACE_ENABLE_HEXAGON)

[CMakeLists.txt](https://github.com/XiaoMi/mace/files/11812061/CMakeLists.txt)

mace-master/mace/runtimes/hexagon/CMakeLists.txt中

更改
target_link_libraries(hexagon_runtime core)
为
target_link_libraries(hexagon_runtime core hexagon_controller)

mace-master/mace/runtimes/CMakeLists.txt中

添加
if(MACE_ENABLE_HEXAGON)
  target_link_libraries(runtimes PRIVATE hexagon_runtime)
endif(MACE_ENABLE_HEXAGON)
[CMakeLists.txt](https://github.com/XiaoMi/mace/files/11812081/CMakeLists.txt)

mace-master/tools/cmake/cmake-build-armeabi-v7a.sh中

添加
MACE_ENABLE_HEXAGON
[cmake-build-armeabi-v7a.zip](https://github.com/XiaoMi/mace/files/11812102/cmake-build-armeabi-v7a.zip)

mace-master/tools/python/run_model.py中

parser.add_argument(
        "--target_abi",
        type=str,
        default="arm64-v8a",
        help="Target ABI: host, armeabi-v7a, arm64-v8a,"
             " arm-linux-gnueabihf, aarch64-linux-gnu"
    )
改为
parser.add_argument(
        "--target_abi",
        type=str,
        default="armeabi-v7a",
        help="Target ABI: host, armeabi-v7a, arm64-v8a,"
             " arm-linux-gnueabihf, aarch64-linux-gnu"
    )
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant