This is an archive of the discontinued LLVM Phabricator instance.

[AMDGPU] Fix regression in AMDGPULibCalls allowing native for doubles
ClosedPublic

Authored by rampitec on Aug 28 2017, 10:51 AM.

Details

Summary

Under -cl-fast-relaxed-math we could use native_sqrt, but f64 was
allowed to produce HSAIL's nsqrt instruction. HSAIL is not here
and we stick with non-existing native_sqrt(double) as a result.

Add check for f64 to not return native functions and also remove
handling of f64 case for fold_sqrt.

Diff Detail

Repository
rL LLVM

Event Timeline

rampitec created this revision.Aug 28 2017, 10:51 AM
This revision is now accepted and ready to land.Aug 28 2017, 10:53 AM
This revision was automatically updated to reflect the committed changes.