This is an archive of the discontinued LLVM Phabricator instance.

AMDGPU: Use v_accvgpr_mov_b32 when copying AGPR tuples on gfx90a
ClosedPublic

Authored by arsenm on Dec 15 2021, 9:01 AM.

Details

Reviewers
rampitec
Summary

This is an optimization, but also fixes a compile failure when no free
VGPRs are available. The problem still exists for gfx908 where a
scratch register is still required. This also still exists for the
SGPR to AGPR case.

Diff Detail

Event Timeline

arsenm created this revision.Dec 15 2021, 9:01 AM
arsenm requested review of this revision.Dec 15 2021, 9:01 AM
Herald added a project: Restricted Project. · View Herald TranscriptDec 15 2021, 9:01 AM
Herald added a subscriber: wdng. · View Herald Transcript
rampitec accepted this revision.Dec 15 2021, 9:11 AM

LGTM, thank you!

This revision is now accepted and ready to land.Dec 15 2021, 9:11 AM