Switch the install-cuda default to True. If capable hardware is present,
CUDA should be automatically installed. Without this, users that deploy
to gpu instances have to explicitly configure install-cuda=True.
Setting this to True is a no-op on non-gpu hardware as layer-nvidia-cuda
will not do anything unless lspci detects an available gpu.
If an operator really wants GPU hardware without CUDA, they can always
configure with install-cuda=False.
... also get rid of pesky .pyc in our Makefile 'clean' target.
Switch the
install-cudadefault to True. If capable hardware is present, CUDA should be automatically installed. Without this, users that deploy to gpu instances have to explicitly configureinstall-cuda=True.Setting this to True is a no-op on non-gpu hardware as layer-nvidia-cuda will not do anything unless
lspcidetects an available gpu.If an operator really wants GPU hardware without CUDA, they can always configure with
install-cuda=False.... also get rid of pesky .pyc in our Makefile 'clean' target.