3 ms·
Is that because AMD GPUs are technically inferior for ML, or because of tooling is only written for NVIDIA?
by giomasce 3y ago
Is that because AMD GPUs are technically inferior for ML, or because of tooling is only written for NVIDIA?
- jacquesm 3y agoA bit of both, NV has put a ton of work in their software libraries which makes doing some stuff on NV a viable option, you'd have to do a lot of work to get to an equivalent level on AMD. But if all you need is PyTorch support then AMD may well be a viable option.