projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: 4b9d868) | patch
AMDGPU: Switch barrier intrinsics to using convergent
authorMatt Arsenault <Matthew.Arsenault@amd.com>
Sat, 19 Dec 2015 01:46:41 +0000 (01:46 +0000)
committerMatt Arsenault <Matthew.Arsenault@amd.com>
Sat, 19 Dec 2015 01:46:41 +0000 (01:46 +0000)
commit7aed0ccd46f547e87a750051265c8c3faeaf4a33
tree9a9ef3b1dd1fc9d9288339cbabd3a531e86f6369tree | snapshot
parent4b9d868cc774eba681a93a61f056105b7dfd9c8fcommit | diff
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
21 files changed:
lib/Target/AMDGPU/AMDGPUIntrinsics.td diff | blob | history
test/CodeGen/AMDGPU/addrspacecast.ll diff | blob | history
test/CodeGen/AMDGPU/array-ptr-calc-i32.ll diff | blob | history
test/CodeGen/AMDGPU/drop-mem-operand-move-smrd.ll diff | blob | history
test/CodeGen/AMDGPU/ds-negative-offset-addressing-mode-loop.ll diff | blob | history
test/CodeGen/AMDGPU/ds-sub-offset.ll diff | blob | history
test/CodeGen/AMDGPU/ds_read2.ll diff | blob | history
test/CodeGen/AMDGPU/ds_read2_superreg.ll diff | blob | history
test/CodeGen/AMDGPU/ds_read2st64.ll diff | blob | history
test/CodeGen/AMDGPU/ds_write2.ll diff | blob | history
test/CodeGen/AMDGPU/ds_write2st64.ll diff | blob | history
test/CodeGen/AMDGPU/flat-address-space.ll diff | blob | history
test/CodeGen/AMDGPU/indirect-private-64.ll diff | blob | history
test/CodeGen/AMDGPU/llvm.AMDGPU.div_fmas.ll diff | blob | history
test/CodeGen/AMDGPU/merge-stores.ll diff | blob | history
test/CodeGen/AMDGPU/schedule-vs-if-nested-loop-failure.ll diff | blob | history
test/CodeGen/AMDGPU/si-triv-disjoint-mem-access.ll diff | blob | history
test/CodeGen/AMDGPU/store-barrier.ll diff | blob | history
test/CodeGen/AMDGPU/wait.ll diff | blob | history
test/Transforms/LoopUnroll/AMDGPU/lit.local.cfg [new file with mode: 0644] blob
test/Transforms/LoopUnroll/AMDGPU/unroll-barrier.ll [new file with mode: 0644] blob
C/C++ LLVM-based compilers that forbids OOTA behaviors