Added a function to retrieve the compute capability of the host machine to compile for that specific architecture.
Checking if a specific CUDA version is installed doesn't mean a certain compute capability is available, so instead I think we should just directly get the compute capability to append the correct _ccflag. See this post
MzeroMiko
commented
1 month ago
Added a function to retrieve the compute capability of the host machine to compile for that specific architecture.
Checking if a specific CUDA version is installed doesn't mean a certain compute capability is available, so instead I think we should just directly get the compute capability to append the correct _ccflag. See this post