This is an archive of the discontinued LLVM Phabricator instance.

[NEON] Support intrinsic for scalar and vector versions of the VRINTN instruction
ClosedPublic

Authored by kosarev on Apr 11 2018, 5:37 AM.

Diff Detail

Event Timeline

kosarev created this revision.Apr 11 2018, 5:37 AM

Look like we need to add some tests too?

Can you use llvm.rint.f32, instead of adding a new intrinsic?

kosarev updated this revision to Diff 142150.Apr 12 2018, 4:50 AM

A test case added.

Eli, the patch doesn't add new intrinsics, just allows scalar arguments for an existing one. Switching from existing ARM-specific intrinsics to some more generic ones seems to be out of scope of this change.

efriedma accepted this revision.Apr 12 2018, 12:10 PM

Okay. Then LGTM.

This revision is now accepted and ready to land.Apr 12 2018, 12:10 PM
This revision was automatically updated to reflect the committed changes.