sophgo / tpu-mlir

Machine learning compiler based on MLIR for Sophgo TPU.
Other
612 stars 154 forks source link

model_deploy.py --chip cv183x 报错 #186

Open micro-step opened 3 months ago

micro-step commented 3 months ago

--chip bm1684x 可以通过,cv183x 报错 ==---------------------------== GmemAllocator use FitFirstAssign [Success]: tpuc-opt yolov5s_cv183x_f16_tpu.mlir --mlir-disable-threading --strip-io-quant="quant_input=False quant_output=False quant_input_list= quant_output_list=" --processor-tpu-optimize --dev-parallel --weight-reorder --subnet-divide="dynamic=False" --op-reorder --layer-group="opt=2 group_by_cores=auto compress_mode=none" --core-parallel --address-assign -o yolov5s_cv183x_f16_final.mlir [Running]: tpuc-opt yolov5s_cv183x_f16_final.mlir --codegen="model_file=yolov5s_1684x_f16.bmodel embed_debug_info=False model_version=latest" -o /dev/null [Success]: tpuc-opt yolov5s_cv183x_f16_final.mlir --codegen="model_file=yolov5s_1684x_f16.bmodel embed_debug_info=False model_version=latest" -o /dev/null [Running]: mv compiler_profile_0.[td][xa]t yolov5s_1684x_f16.bmodel.compiler_profile_0.txt mv: cannot stat 'compiler_profile_0.[td][xa]t': No such file or directory [CMD]: model_runner.py --input yolov5s_in_f32.npz --model yolov5s_1684x_f16.bmodel --output yolov5s_cv183x_f16_model_outputs.npz
begin to cmodel init... mult engine c_model init bmcpu init: skip cpu_user_defined Cannot open libusercpu.so, disable user cpu layer. bmcpu init: skip cpu_user_defined Cannot open libusercpu.so, disable user cpu layer. [BMODEL][ModelCtx:496] FATAL: File[yolov5s_1684x_f16.bmodel] is broken .. python3: /__w/tpu-mlir/tpu-mlir/bindings/pyruntime/pyruntime_bm.cpp:289: PythonModel::PythonModel(const std::string &, int, const std::string &): Assertion `flag == true' failed.

we11d0ne commented 2 months ago

--chip bm1684x 可以通过,cv183x 报错 ==---------------------------== GmemAllocator use FitFirstAssign [Success]: tpuc-opt yolov5s_cv183x_f16_tpu.mlir --mlir-disable-threading --strip-io-quant="quant_input=False quant_output=False quant_input_list= quant_output_list=" --processor-tpu-optimize --dev-parallel --weight-reorder --subnet-divide="dynamic=False" --op-reorder --layer-group="opt=2 group_by_cores=auto compress_mode=none" --core-parallel --address-assign -o yolov5s_cv183x_f16_final.mlir [Running]: tpuc-opt yolov5s_cv183x_f16_final.mlir --codegen="model_file=yolov5s_1684x_f16.bmodel embed_debug_info=False model_version=latest" -o /dev/null [Success]: tpuc-opt yolov5s_cv183x_f16_final.mlir --codegen="model_file=yolov5s_1684x_f16.bmodel embed_debug_info=False model_version=latest" -o /dev/null [Running]: mv compiler_profile_0.[td][xa]t yolov5s_1684x_f16.bmodel.compiler_profile_0.txt mv: cannot stat 'compiler_profile_0.[td][xa]t': No such file or directory [CMD]: model_runner.py --input yolov5s_in_f32.npz --model yolov5s_1684x_f16.bmodel --output yolov5s_cv183x_f16_model_outputs.npz begin to cmodel init... mult engine c_model init bmcpu init: skip cpu_user_defined Cannot open libusercpu.so, disable user cpu layer. bmcpu init: skip cpu_user_defined Cannot open libusercpu.so, disable user cpu layer. [BMODEL][ModelCtx:496] FATAL: File[yolov5s_1684x_f16.bmodel] is broken .. python3: /__w/tpu-mlir/tpu-mlir/bindings/pyruntime/pyruntime_bm.cpp:289: PythonModel::PythonModel(const std::string &, int, const std::string &): Assertion `flag == true' failed.

cv18xx系列好像不支持bmodel,应该可以转cvimodel