ralna / spral

Sparse Parallel Robust Algorithms Library
https://ralna.github.io/spral/
Other
106 stars 26 forks source link

Support newer NVIDIA GPUs #87

Closed jfowkes closed 2 years ago

jfowkes commented 2 years ago

Add support for newer NVIDIA GPUs and drop unsupported one, for details see: https://arnon.dk/matching-sm-architectures-arch-and-gencode-for-various-nvidia-cards/

This should hopefully help resolve #84

jfowkes commented 2 years ago

@AndrewLister-STFC I've now added in the missing repeated compute gencode, the flags should now match the linked article for the following:

Sample flags for generation on CUDA 11.7 for maximum compatibility with V100 and T4 Turing cards, but also support newer RTX 3080, and Drive AGX Orin

This seems reasonable to me as the RTX 4080 and H100 are yet to ship (so we don't support those) and we have only dropped support for the ancient deprecated sm_50.