mahmoodn commented on issue #16001: Low kernel performance URL: https://github.com/apache/incubator-mxnet/issues/16001#issuecomment-524996948 @sxjscience Unfortunately, it failed with the following error ``` g++ -std=c++11 -c -DMSHADOW_FORCE_STREAM -Wall -Wsign-compare -O3 -DNDEBUG=1 -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/mshadow/ -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/dmlc-core/include -fPIC -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/nnvm/include -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/dlpack/include -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/nnvm/tvm/include -Iinclude -funroll-loops -Wno-unused-parameter -Wno-unknown-pragmas -Wno-unused-local-typedefs -msse3 -DMSHADOW_USE_F16C=0 -I/usr/local/cuda/include -DMSHADOW_USE_CBLAS=1 -DMSHADOW_USE_MKL=0 -DMSHADOW_RABIT_PS=0 -DMSHADOW_DIST_PS=0 -DMSHADOW_USE_PASCAL=0 -DMXNET_USE_OPENCV=1 -I/usr/include/opencv -fopenmp -DMXNET_USE_OPERATOR_TUNING=1 -DMSHADOW_USE_CUDNN=1 -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/cub -DMXNET_ENABLE_CUDA_RTC=1 -DMXNET_USE_NCCL=0 -DMXNET_USE_LIBJPEG_TURBO=0 -MMD -c src/operator/nn/mkldnn/mkldnn_sum.cc -o build/src/operator/nn/mkldnn/mkldnn_sum.o g++ -std=c++11 -c -DMSHADOW_FORCE_STREAM -Wall -Wsign-compare -O3 -DNDEBUG=1 -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/mshadow/ -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/dmlc-core/include -fPIC -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/nnvm/include -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/dlpack/include -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/nnvm/tvm/include -Iinclude -funroll-loops -Wno-unused-parameter -Wno-unknown-pragmas -Wno-unused-local-typedefs -msse3 -DMSHADOW_USE_F16C=0 -I/usr/local/cuda/include -DMSHADOW_USE_CBLAS=1 -DMSHADOW_USE_MKL=0 -DMSHADOW_RABIT_PS=0 -DMSHADOW_DIST_PS=0 -DMSHADOW_USE_PASCAL=0 -DMXNET_USE_OPENCV=1 -I/usr/include/opencv -fopenmp -DMXNET_USE_OPERATOR_TUNING=1 -DMSHADOW_USE_CUDNN=1 -I/home/mh.naderan/mx/incubator-mxnet/3rdparty/cub -DMXNET_ENABLE_CUDA_RTC=1 -DMXNET_USE_NCCL=0 -DMXNET_USE_LIBJPEG_TURBO=0 -MMD -c src/operator/nn/cudnn/cudnn_algoreg.cc -o build/src/operator/nn/cudnn/cudnn_algoreg.o In file included from include/mxnet/./resource.h:31:0, from include/mxnet/operator.h:39, from src/operator/nn/cudnn/./../convolution-inl.h:33, from src/operator/nn/cudnn/./cudnn_algoreg-inl.h:34, from src/operator/nn/cudnn/cudnn_algoreg.cc:26: include/mxnet/./../../src/common/random_generator.h: In static member function ‘static void mxnet::common::random::RandGenerator<mshadow::gpu, DType>::AllocState(mxnet::common::random::RandGenerator<mshadow::gpu, DType>*)’: include/mxnet/./../../src/common/random_generator.h:154:5: error: there are no arguments to ‘CUDA_CALL’ that depend on a template parameter, so a declaration of ‘CUDA_CALL’ must be available [-fpermissive] CUDA_CALL(cudaMalloc(&inst->states_, ^~~~~~~~~ include/mxnet/./../../src/common/random_generator.h:154:5: note: (if you use ‘-fpermissive’, G++ will accept your code, but allowing the use of an undeclared name is deprecated) include/mxnet/./../../src/common/random_generator.h: In static member function ‘static void mxnet::common::random::RandGenerator<mshadow::gpu, DType>::FreeState(mxnet::common::random::RandGenerator<mshadow::gpu, DType>*)’: include/mxnet/./../../src/common/random_generator.h:159:5: error: there are no arguments to ‘CUDA_CALL’ that depend on a template parameter, so a declaration of ‘CUDA_CALL’ must be available [-fpermissive] CUDA_CALL(cudaFree(inst->states_)); ^~~~~~~~~ In file included from src/operator/nn/cudnn/./../../linalg.h:31:0, from src/operator/nn/cudnn/./../convolution-inl.h:44, from src/operator/nn/cudnn/./cudnn_algoreg-inl.h:34, from src/operator/nn/cudnn/cudnn_algoreg.cc:26: src/operator/nn/cudnn/./../.././c_lapack_api.h: At global scope: src/operator/nn/cudnn/./../.././c_lapack_api.h:330:25: note: #pragma message: Warning: lapack usage not enabled, linalg-operators will not be available. Ensure that lapack library is installed and build with USE_LAPACK=1 to get lapack functionalities. " functionalities.") ^ ``` It seems that 1.2.0 is not compatible with cuda 10. I will try some alternatives, either version above 1.2.0 or cuda 9.
---------------------------------------------------------------- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. For queries about this service, please contact Infrastructure at: [email protected] With regards, Apache Git Services
