This is an archive of the discontinued LLVM Phabricator instance.

[AMDGPU] Aggressively schedule to reduce RP in occupancy limited regions
ClosedPublic

Authored by kerbowa on Jul 21 2022, 8:38 PM.

Details

Summary

By not clustering loads and adjusting heuristics to more aggressively reduce
register pressure we may be able to increase occupancy for the function if it
was dropped in a first pass scheduling.

Similarly, try to reduce spilling if register usage exceeds lower bound
occupancy.

Diff Detail

Event Timeline

kerbowa created this revision.Jul 21 2022, 8:38 PM
Herald added a project: Restricted Project. · View Herald TranscriptJul 21 2022, 8:38 PM
kerbowa requested review of this revision.Jul 21 2022, 8:38 PM
Herald added a project: Restricted Project. · View Herald TranscriptJul 21 2022, 8:38 PM
kerbowa updated this revision to Diff 446877.Jul 22 2022, 9:41 AM

Fix not running unclustred pass on non-excess RP regions.

This revision is now accepted and ready to land.Jul 22 2022, 10:07 AM
This revision was landed with ongoing or failed builds.Jul 27 2022, 10:42 PM
This revision was automatically updated to reflect the committed changes.