Differential D8063
Cycles: Preliminary Cuda 11 build support Authored by Brecht Van Lommel (brecht) on Jun 17 2020, 11:00 PM.
Details
Basic cmake changes, there's nothing surprising in here. Cuda 11 dropped sm_30 and added sm_80 the Given there has been little to no testing of
Diff Detail
Event TimelineComment Actions
Comment Actions During final testing bumped into some optix issues, it does not like sm_30 anymore, and anything below 52 gives deprecation warnings, given we only support optix on the 900+ cards i think 52 should be alright? @Patrick Mours (pmoursnv) / @Brecht Van Lommel (brecht) do either of you have a strong opinions here? Comment Actions I think we should bump the OptiX target to at least sm_50, if not sm_52 (for the purpose of CUDA 11) in general. It cannot run on earlier cards anyway and having it at sm_30 misses some (albeit small) optimization potential (since most optimization is handled in the PTX->binary driver step). I wanted to do that a while ago, but seamingly forgot about it. Comment Actions Background information: I abandoned this due to lack of time to land it, the patch seems 'ok', I still needed to check if msvc 2017/2019 , nvcc/cubincc , cuda10/11 all build without issues. Also did not validate the Registers/Blocks/Threads settings for SM_8X yet |