| commit | 62458bf56e88102317b255372459ccf14c0bced1 | [log] [tgz] |
|---|---|---|
| author | Craig Topper <craig.topper@gmail.com> | Tue Jun 14 03:13:00 2016 +0000 |
| committer | Craig Topper <craig.topper@gmail.com> | Tue Jun 14 03:13:00 2016 +0000 |
| tree | 9048d8a8a0b7de690d5de7ce94e1c6970a3552cb | |
| parent | 51ad7064a4a4d50d8773068be3a29ed0af5823a0 [diff] |
[AVX512] Use MOVZX32 instead of MOVZ16 for loading single v8/v4/v2/v1 masks when KMOVB is not available. This has better behavior with respect to partial register stalls since it won't need to preserve the upper 16-bits of the GPR. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@272626 91177308-0d34-0410-b5e6-96231b3b80d8