Page MenuHomePhabricator
Differential D27905

AMDGPU: Swap order of operands in fadd/fsub combine
ClosedPublic
Actions

Authored by arsenm on Dec 18 2016, 1:51 PM.

Details

Reviewers
• tstellarAMD
nhaehnle
Summary

FMA is canonicalized to constant in the middle operand. Do
the same so fmad matches and avoid an extra combine step.

Diff Detail

Event Timeline

arsenm updated this revision to Diff 81892.Dec 18 2016, 1:51 PM
arsenm retitled this revision from to AMDGPU: Swap order of operands in fadd/fsub combine .
arsenm updated this object.
arsenm added a reviewer: nhaehnle.
arsenm added a subscriber: llvm-commits.
Herald added a reviewer: • tstellarAMD. · View Herald TranscriptDec 18 2016, 1:51 PM
Herald added subscribers: tony-tye, yaxunl, wdng, kzhuravl. · View Herald Transcript
• tstellarAMD accepted this revision.Dec 19 2016, 2:21 PM
• tstellarAMD edited edge metadata.

LGTM.

This revision is now accepted and ready to land.Dec 19 2016, 2:21 PM
arsenm closed this revision.Dec 21 2016, 8:14 PM

r290313

Revision Contents
Changeset List

PathSize
lib/
Target/
AMDGPU/
8 lines
test/
CodeGen/
AMDGPU/
16 lines
4 lines
4 lines

Diff 81892

View Options

lib/Target/AMDGPU/SIISelLowering.cpp

Loading...
View Options

test/CodeGen/AMDGPU/fmul-2-combine-multi-use.ll

Loading...
View Options

test/CodeGen/AMDGPU/fmuladd.f16.ll

Loading...
View Options

test/CodeGen/AMDGPU/fmuladd.f32.ll

Loading...