CUDNN 7.0.4: MNIST Compile Error
Hello, I'm currently running into an error when testing and verifying my CUDNN7.0.4 installation by compiling /cudnn_samples_v7/mnistCUDNN. Q: Any suggestion or idea on the error? Thank you! This is running with CUDA 8.0 on Ubuntu 16.04 Terminal Result: [code] ~/cudnn_samples_v7/mnistCUDNN$ make clean && make nvcc warning : The 'compute_20', 'sm_20', and 'sm_21' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning). rm -rf *o rm -rf mnistCUDNN nvcc warning : The 'compute_20', 'sm_20', and 'sm_21' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning). /usr/local/cuda/bin/nvcc -ccbin g++ -I/usr/local/cuda/include -IFreeImage/include -m64 -gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=sm_35 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_53,code=sm_53 -gencode arch=compute_53,code=compute_53 -o fp16_dev.o -c fp16_dev.cu g++ -I/usr/local/cuda/include -IFreeImage/include -o fp16_emu.o -c fp16_emu.cpp g++ -I/usr/local/cuda/include -IFreeImage/include -o mnistCUDNN.o -c mnistCUDNN.cpp In file included from gemv.h:17:0, from mnistCUDNN.cpp:35: gemv.h: In function ‘void gemv(cublasHandle_t, int, int, float, const half1*, const half1*, float, half1*)’: gemv.h:106:40: error: cannot convert ‘cudaDataType_t’ to ‘cublasDataType_t’ for argument ‘9’ to ‘cublasStatus_t cublasSgemmEx(cublasHandle_t, cublasOperation_t, cublasOperation_t, int, int, int, const float*, const void*, cublasDataType_t, int, const void*, cublasDataType_t, int, const float*, void*, cublasDataType_t, int)’ m) ); ^ error_util.h:82:9: note: in definition of macro ‘checkCublasErrors’ if (status != 0) { \ ^ gemv.h:106:40: error: cannot convert ‘cudaDataType_t’ to ‘cublasDataType_t’ for argument ‘9’ to ‘cublasStatus_t cublasSgemmEx(cublasHandle_t, cublasOperation_t, cublasOperation_t, int, int, int, const float*, const void*, cublasDataType_t, int, const void*, cublasDataType_t, int, const float*, void*, cublasDataType_t, int)’ m) ); ^ error_util.h:83:50: note: in definition of macro ‘checkCublasErrors’ _error << "Cublas failure\nError code " << status; \ ^ Makefile:200: recipe for target 'mnistCUDNN.o' failed make: *** [mnistCUDNN.o] Error 1 [/code]
Hello,

I'm currently running into an error when testing and verifying my CUDNN7.0.4 installation by compiling /cudnn_samples_v7/mnistCUDNN. Q: Any suggestion or idea on the error? Thank you!

This is running with CUDA 8.0 on Ubuntu 16.04

Terminal Result:
~/cudnn_samples_v7/mnistCUDNN$ make clean && make
nvcc warning : The 'compute_20', 'sm_20', and 'sm_21' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
rm -rf *o
rm -rf mnistCUDNN
nvcc warning : The 'compute_20', 'sm_20', and 'sm_21' architectures are deprecated, and may be removed in a future release (Use -Wno-deprecated-gpu-targets to suppress warning).
/usr/local/cuda/bin/nvcc -ccbin g++ -I/usr/local/cuda/include -IFreeImage/include -m64 -gencode arch=compute_30,code=sm_30 -gencode arch=compute_35,code=sm_35 -gencode arch=compute_50,code=sm_50 -gencode arch=compute_53,code=sm_53 -gencode arch=compute_53,code=compute_53 -o fp16_dev.o -c fp16_dev.cu
g++ -I/usr/local/cuda/include -IFreeImage/include -o fp16_emu.o -c fp16_emu.cpp
g++ -I/usr/local/cuda/include -IFreeImage/include -o mnistCUDNN.o -c mnistCUDNN.cpp
In file included from gemv.h:17:0,
from mnistCUDNN.cpp:35:
gemv.h: In function ‘void gemv(cublasHandle_t, int, int, float, const half1*, const half1*, float, half1*)’:
gemv.h:106:40: error: cannot convert ‘cudaDataType_t’ to ‘cublasDataType_t’ for argument ‘9’ to ‘cublasStatus_t cublasSgemmEx(cublasHandle_t, cublasOperation_t, cublasOperation_t, int, int, int, const float*, const void*, cublasDataType_t, int, const void*, cublasDataType_t, int, const float*, void*, cublasDataType_t, int)’
m) );
^
error_util.h:82:9: note: in definition of macro ‘checkCublasErrors’
if (status != 0) { \
^
gemv.h:106:40: error: cannot convert ‘cudaDataType_t’ to ‘cublasDataType_t’ for argument ‘9’ to ‘cublasStatus_t cublasSgemmEx(cublasHandle_t, cublasOperation_t, cublasOperation_t, int, int, int, const float*, const void*, cublasDataType_t, int, const void*, cublasDataType_t, int, const float*, void*, cublasDataType_t, int)’
m) );
^
error_util.h:83:50: note: in definition of macro ‘checkCublasErrors’
_error << "Cublas failure\nError code " << status; \
^
Makefile:200: recipe for target 'mnistCUDNN.o' failed
make: *** [mnistCUDNN.o] Error 1

#1
Posted 02/14/2018 07:21 PM   
Scroll To Top

Add Reply