Closed skywolf829 closed 7 months ago
torch.cuda.is_available()
else cpu
.Still need to test that model.render
calls (which try to invoke the CUDA backend) will work with the stand-in code I added for when the CUDA backend isn't available. Similar situation with settings.device
.
Seems to work, have tested frontend with the DEBUG
mode in order to get frames (random noise) from backend.
To debug, we may want to run the backend locally even if we dont have CUDA installed. Without CUDA, we cannot install the pip libraries for the rendering and knn packages.
Requirements