3 Commits

Author SHA1 Message Date
Matt Arsenault
fae18e933b AMDGPU: Remove some old intrinsic uses from tests
git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@260493 91177308-0d34-0410-b5e6-96231b3b80d8
2016-02-11 06:02:01 +00:00
Matt Arsenault
7aed0ccd46 AMDGPU: Switch barrier intrinsics to using convergent
noduplicate prevents unrolling of small loops that happen to have
barriers in them. If a loop has a barrier in it, it is OK to duplicate
it for the unroll.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@256075 91177308-0d34-0410-b5e6-96231b3b80d8
2015-12-19 01:46:41 +00:00
Matt Arsenault
626338d11e AMDGPU: Handle sub of constant for DS offset folding
sub C, x - > add (sub 0, x), C for DS offsets.

This is mostly to fix regressions that show up when
SeparateConstOffsetFromGEP is enabled.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@247054 91177308-0d34-0410-b5e6-96231b3b80d8
2015-09-08 19:34:22 +00:00