RPCSX/llvm
1
0
Fork 0
mirror of https://github.com/RPCSX/llvm.git synced 2024-11-30 15:10:33 +00:00
Code Issues Actions Packages Projects Releases Wiki Activity

AMDGPU: mark atomic instructions as sources of divergence

Browse Source 
Summary:
As explained by the comment, threads will typically see different values
returned by atomic instructions even if the arguments are equal.

Reviewers: arsenm, tstellarAMD

Subscribers: arsenm, llvm-commits

Differential Revision: http://reviews.llvm.org/D18156

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@263719 91177308-0d34-0410-b5e6-96231b3b80d8
This commit is contained in:
Nicolai Haehnle 2016-03-17 16:21:59 +00:00
parent 96ac27b20b
commit d3939c80f8
2 changed files with 22 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
lib/Target/AMDGPU/AMDGPUTargetTransformInfo.cpp
View File

@ -186,6 +186,13 @@ bool AMDGPUTTIImpl::isSourceOfDivergence(const Value *V) const {
if (const LoadInst *Load = dyn_cast<LoadInst>(V))
return Load->getPointerAddressSpace() == AMDGPUAS::PRIVATE_ADDRESS;
// Atomics are divergent because they are executed sequentially: when an
// atomic operation refers to the same address in each thread, then each
// thread after the first sees the value written by the previous thread as
// original value.
if (isa<AtomicRMWInst>(V) || isa<AtomicCmpXchgInst>(V))
return true;
if (const IntrinsicInst *Intrinsic = dyn_cast<IntrinsicInst>(V)) {
const TargetMachine &TM = getTLI()->getTargetMachine();
return isIntrinsicSourceOfDivergence(TM.getIntrinsicInfo(), Intrinsic);

15
test/Analysis/DivergenceAnalysis/AMDGPU/atomics.ll Normal file
View File

@ -0,0 +1,15 @@
; RUN: opt -mtriple=amdgcn-- -analyze -divergence %s | FileCheck %s
; CHECK: DIVERGENT: %orig = atomicrmw xchg i32* %ptr, i32 %val seq_cst
define i32 @test1(i32* %ptr, i32 %val) #0 {
%orig = atomicrmw xchg i32* %ptr, i32 %val seq_cst
ret i32 %orig
}
; CHECK: DIVERGENT: %orig = cmpxchg i32* %ptr, i32 %cmp, i32 %new seq_cst seq_cst
define {i32, i1} @test2(i32* %ptr, i32 %cmp, i32 %new) {
%orig = cmpxchg i32* %ptr, i32 %cmp, i32 %new seq_cst seq_cst
ret {i32, i1} %orig
}
attributes #0 = { "ShaderType"="0" }