This is an archive of the discontinued LLVM Phabricator instance.

[InstCombine] Gating select arithmetic optimization
ClosedPublic

Authored by mcberg2017 on Sep 25 2017, 2:40 PM.

Details

Summary

These changes faciliate positive behavior for arithmetic based select expressions that match its translation criteria, keeping code size gated to neutral or improved scenarios.

Diff Detail

Repository
rL LLVM

Event Timeline

mcberg2017 created this revision.Sep 25 2017, 2:40 PM

Please use the -U option to include full context in your diffs (e.g., git show HEAD -U999999 > mypatch.patch).

Updated for full context diff, the patch file now contains all the code.

mcrosier retitled this revision from Gating select arithmetic optimization to [InstCombine] Gating select arithmetic optimization.
mcberg2017 added a comment.EditedSep 27 2017, 9:28 AM

Chad, would like me to amend the test file to include the changes you and
Balaram made?

This revision is now accepted and ready to land.Sep 27 2017, 9:30 AM
mcrosier edited edge metadata.Sep 27 2017, 9:37 AM

Chad, would like me to amend the test file to include the changes you and
Balaram made?

I think it would be a good idea to add both test cases as it makes it easier to understand this change, IMHO.

mcrosier accepted this revision.Sep 27 2017, 9:38 AM

Do you have commit access, btw? I can commit if not.

This revision was automatically updated to reflect the committed changes.

Do you have commit access, btw? I can commit if not.

Nevermind, I just went ahead and committed the change. Thanks for the fix!