This is an archive of the discontinued LLVM Phabricator instance.

[AMDGPU] add fcopysign(f64, f32) pattern
ClosedPublic

Authored by vpykhtin on Oct 20 2016, 8:24 AM.

Details

Summary

DAGCombiner::visitFCOPYSIGN does

// copysign(x, fp_extend(y)) -> copysign(x, y)

when x has f64 type and y - f32 the backend cannot select fcopysign. Adding the pattern fixes rodinia\myocite test failure.

Diff Detail

Repository
rL LLVM

Event Timeline

vpykhtin updated this revision to Diff 75301.Oct 20 2016, 8:24 AM
vpykhtin retitled this revision from to [AMDGPU] add fcopysign(f64, f32) pattern.
vpykhtin updated this object.
vpykhtin added reviewers: arsenm, tstellarAMD.
vpykhtin set the repository for this revision to rL LLVM.
vpykhtin added a subscriber: nhaustov.
tstellarAMD accepted this revision.Oct 20 2016, 8:34 AM
tstellarAMD edited edge metadata.

LGTM.

This revision is now accepted and ready to land.Oct 20 2016, 8:34 AM
This revision was automatically updated to reflect the committed changes.