Hi,
I tried to install cudnn 6 for Cuda 8.0 in order to install tensorflow 1.4. When I try to check the installation by doing :
make clean && make in a copy version of the cudnn_samples_v6/mnistCUDNN folder I got this error message :
rm -rf *o
rm -rf mnistCUDNN
/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
fp16_dev.cu(23): error: identifier “__float2half_rn” is undefined
detected during instantiation of “void gpu_float2half_rn(int, const value_type *, half1 *) [with value_type=float]”
(36): here
1 error detected in the compilation of “/tmp/tmpxft_00004c6a_00000000-9_fp16_dev.compute_53.cpp1.ii”.
Makefile:203 : la recette pour la cible « fp16_dev.o » a échouée
make: *** [fp16_dev.o] Erreur 1
Do you have any idea of the origin of this error ?
Moreover it seems that the installation is correct because I can run GPU version of tensorflow but I don’t know if there is another way to check the cudnn installation.
By the way, I also have CUDA 9.0 and CUDA 9.1 install on my machine and cudnn 7.1.
Best,