diff options
| author | Matt Arsenault <Matthew.Arsenault@amd.com> | 2015-12-19 01:46:41 +0000 |
|---|---|---|
| committer | Matt Arsenault <Matthew.Arsenault@amd.com> | 2015-12-19 01:46:41 +0000 |
| commit | 2aed6ca1d33468f7e6f8509213a1b9dcd8026ec1 (patch) | |
| tree | a4575401ecb619e067cc2b4fed2a83056b80163c /llvm/test/CodeGen/AMDGPU/schedule-vs-if-nested-loop-failure.ll | |
| parent | 10a509292cda51aabf3630bf30ceb990d3bf3108 (diff) | |
| download | bcm5719-llvm-2aed6ca1d33468f7e6f8509213a1b9dcd8026ec1.tar.gz bcm5719-llvm-2aed6ca1d33468f7e6f8509213a1b9dcd8026ec1.zip | |
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.
llvm-svn: 256075
Diffstat (limited to 'llvm/test/CodeGen/AMDGPU/schedule-vs-if-nested-loop-failure.ll')
| -rw-r--r-- | llvm/test/CodeGen/AMDGPU/schedule-vs-if-nested-loop-failure.ll | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/llvm/test/CodeGen/AMDGPU/schedule-vs-if-nested-loop-failure.ll b/llvm/test/CodeGen/AMDGPU/schedule-vs-if-nested-loop-failure.ll index 3863afda5dd..e4b16c0a165 100644 --- a/llvm/test/CodeGen/AMDGPU/schedule-vs-if-nested-loop-failure.ll +++ b/llvm/test/CodeGen/AMDGPU/schedule-vs-if-nested-loop-failure.ll @@ -3,7 +3,7 @@ ; RUN: llc -O0 -march=amdgcn -mcpu=SI -verify-machineinstrs< %s | FileCheck %s -check-prefix=SI ; RUN: llc -O0 -march=amdgcn -mcpu=tonga -verify-machineinstrs< %s | FileCheck %s -check-prefix=SI -declare void @llvm.AMDGPU.barrier.local() nounwind noduplicate +declare void @llvm.AMDGPU.barrier.local() nounwind convergent ; SI-LABEL: {{^}}main( |
