This is an archive of the discontinued LLVM Phabricator instance.

[InstCombine] Teach canEvaluateTruncated to handle arithmetic shift (including those with vector splat shift amount)
ClosedPublic

Authored by aaboud on Aug 16 2017, 12:49 AM.

Details

Diff Detail

Repository
rL LLVM

Event Timeline

spatel accepted this revision.Aug 16 2017, 11:42 AM

LGTM.

Just to make sure I'm understanding: the multiply patch (D36679) is independent of this?

This revision is now accepted and ready to land.Aug 16 2017, 11:42 AM

LGTM.

Just to make sure I'm understanding: the multiply patch (D36679) is independent of this?

Yes, I managed to find an example where I am using OR instruction and that AShr instruction cannot be replaced by LShr instruction.
So, we can commit D36679 separately after this one.

This revision was automatically updated to reflect the committed changes.