This is an archive of the discontinued LLVM Phabricator instance.

Enabled CUDA offloading on aarch64 host
ClosedPublic

Authored by tianshilei1992 on Mar 19 2020, 6:31 PM.

Details

Summary

I've tried on ORNL Wombat cluster which is aarch64 host with NVIDIA GPUs. Offloading works fine. All 19 tests in libomptarget passed.

Diff Detail

Event Timeline

tianshilei1992 created this revision.Mar 19 2020, 6:31 PM

Sounds good. Thanks for testing it.

grokos accepted this revision.Mar 20 2020, 11:58 AM

Building the CUDA plugin on aarch64 hosts was enabled in the ykt branch ages ago, so the configuration has been tested extensively. LGTM.

This revision is now accepted and ready to land.Mar 20 2020, 11:58 AM

I didn't have privilege to merge. Could someone do it for me?

This revision was automatically updated to reflect the committed changes.