Skip to content

Conversation

vchuravy
Copy link
Member

Motivated by a conversation with @pmccormick, @neboat, and @stelleg.

Based of the benchmark code https://gist.github.com/vchuravy/7d639aedd98c51d32541ae999504685a, it's not yet behavioral identical to the C++ code in https://github.com/pmccormick/kitsune/blob/ea8ab2916f043a8b33bc20c1c1ec4f280a8f6c1d/kitsune/experiments/raytracer/raytrace-kokkos.cpp but currently performance seems to be not matching NVCC Pat also wrote a CUDA version in https://github.com/pmccormick/kitsune/blob/dev/12.x/kitsune/experiments/raytracer/raytrace-cuda.cu as a baseline

@maleadt
Copy link
Member

maleadt commented Jan 26, 2022

The NVPTX back-end user guide instructs to optimize after linking libdevice: https://llvm.org/docs/NVPTXUsage.html#linking-with-libdevice

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants