Hi everyone!
I am trying to build torchaudio from source. In my environment, CUDA and its related libraries (e.g. cuDNN) are installed in non-default paths.
How can I pass custom build arguments to cmake via setup.py?
We do not have CUDA specific code yet, and so the torchaudio build doesn’t depend the nvidia compiler. What error are you encountering? If you do run into an error with this, I’d recommend opening an issue directly on github.com/pytorch/audio
