At first, I used nvcc(version 10.0) and changed the CUDA_PATH in lib/make.sh to the cuda lib in anaconda(version 8.0) to run lib/make.sh, but I met undefined symbol: __cudaPopCallConfiguration problem. It seemed that I need to use nvcc 8.0 to compile the code.
Then, I installed CUDA 8.0 and use its nvcc 8.0 to compile the code again and it works.
So, I think in order to run the code, we need CUDA 8.0 toolkit not only in anaconda.
At first, I used nvcc(version 10.0) and changed the CUDA_PATH in
lib/make.sh
to the cuda lib in anaconda(version 8.0) to runlib/make.sh
, but I metundefined symbol: __cudaPopCallConfiguration
problem. It seemed that I need to use nvcc 8.0 to compile the code. Then, I installed CUDA 8.0 and use its nvcc 8.0 to compile the code again and it works. So, I think in order to run the code, we need CUDA 8.0 toolkit not only in anaconda.