Brett Viren <brett.viren@gmail.com> writes: > I think Debian is already doing a good job with CUDA, at least as it > pertains to my work with Python+GPU. My thanks and please keep it up! I'd very much like to second this sentiment! Thanks especially to Mo's amazing work with PyTorch, my personal workflow has improved tremendously over the years. Yes, almost all the compute hardware I have access to (through no choice of my own) is CUDA-based, but at least now I can do prototyping and experimentation with the CPU-based PyTorch on my Debian laptop before deploying. This is already hugely more convenient than just a few years ago. As for the actual poll: I honestly don't know. I try as much as possible to rely on OpenCL when I can, but the overwhelming ubiquity of CUDA in deep learning in particular may be putting Debian at a disadvantage if we choose to ignore it. But then again, it's also pretty terrible to help its dominance along. Best, Gard
Attachment:
signature.asc
Description: PGP signature