tree 38f619b1dbc3039ed01909e73db8cd2a19eddc06
parent 51be368e3fb558e6d678db7c6e9f6840c1cf7aa5
author Matt Arsenault <Matthew.Arsenault@amd.com> 1440200991 +0000
committer Matt Arsenault <Matthew.Arsenault@amd.com> 1440200991 +0000

AMDGPU: Minor cleanups to help with f16 support

The main change is inverting the condition for the
operand class classes so that VT.Size == 16 uses VGPR_32
instead of 64.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@245764 91177308-0d34-0410-b5e6-96231b3b80d8