3 ms·
It's very much nvidia specific, not just using CUDA but the backing nvidia-specific management libraries. Though I don't think there's anything particularly de
by kimixa 10mo ago
It's very much nvidia specific, not just using CUDA but the backing nvidia-specific management libraries.
Though I don't think there's anything particularly device-specific they're measuring, they're using the private nvidia interfaces to do so.
- cr125rider 10mo agoOP you should call that out a little more clearly that this is Nvidia only.