This is an archive of the discontinued LLVM Phabricator instance.

[AMDGPU] Divergence-driven instruction selection for bitreverse
ClosedPublic

Authored by foad on Feb 14 2022, 2:36 AM.

Diff Detail

Event Timeline

foad created this revision.Feb 14 2022, 2:36 AM
foad requested review of this revision.Feb 14 2022, 2:36 AM
Herald added a project: Restricted Project. · View Herald TranscriptFeb 14 2022, 2:36 AM

Looks good but I'd like to see the selection-only LIT test:
uniform bitreverse i32,
divergent bitreverse i32,
uniform bitreverse i64,
divergent bitreverse i64
with -stop-after=amdgpu-isel
The purpose of the selection-only tests is to control the divergence-driven selection not broken.
Otherwise, it could be broken but full-stack tests are still matching because of the moveToVALU hackery.

foad updated this revision to Diff 411083.Feb 24 2022, 4:54 AM

Add isel tests.

foad edited the summary of this revision. (Show Details)Feb 24 2022, 4:54 AM
arsenm accepted this revision.Feb 24 2022, 10:14 AM
This revision is now accepted and ready to land.Feb 24 2022, 10:14 AM
This revision was landed with ongoing or failed builds.Feb 24 2022, 12:22 PM
This revision was automatically updated to reflect the committed changes.