This is an archive of the discontinued LLVM Phabricator instance.

AMDGPU: Prepare for reducing private element size.
ClosedPublic

Authored by arsenm on Feb 10 2016, 4:09 PM.

Details

Reviewers
tstellarAMD
Summary

Tests for the new scalarize all private access options will be
included with a future commit.

The only functional change is to make the split/scalarize behavior
for private access of > 4 element vectors to be consistent
with the flat/global handling. This makes the spilling worse
in the two changed tests.

Diff Detail

Event Timeline

arsenm updated this revision to Diff 47555.Feb 10 2016, 4:09 PM
arsenm retitled this revision from to AMDGPU: Prepare for reducing private element size..
arsenm updated this object.
arsenm added a reviewer: tstellarAMD.
arsenm added a subscriber: llvm-commits.
arsenm updated this revision to Diff 47586.Feb 10 2016, 7:54 PM

Forgot to add new test file

lib/Target/AMDGPU/SIISelLowering.cpp
1661

Shouldn't this be (NumElements >= 4) ?

arsenm updated this revision to Diff 47836.Feb 12 2016, 12:48 PM

Use > 4 instead of >= 8. These should have the same effect because this should only ever be called for legal types

tstellarAMD accepted this revision.Feb 12 2016, 5:43 PM
tstellarAMD edited edge metadata.

LGTM.

This revision is now accepted and ready to land.Feb 12 2016, 5:43 PM
arsenm closed this revision.Feb 12 2016, 8:23 PM

r260804