Skip to content

Commit 22ca6fd

Browse files
carlobertollironlieb
authored andcommitted
Emit fast FP atomics for gfx942. This should not include atomic compare.
Change-Id: I7851050301401ed24652dc975afe7066dd5f03b8
1 parent e367b9b commit 22ca6fd

File tree

1 file changed

+4
-3
lines changed

1 file changed

+4
-3
lines changed

clang/lib/CodeGen/CGOpenMPRuntimeGPU.cpp

Lines changed: 4 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -4236,9 +4236,10 @@ llvm::Value *CGOpenMPRuntimeGPU::getXteamRedSum(
42364236
bool CGOpenMPRuntimeGPU::supportFastFPAtomics() {
42374237
CudaArch Arch = getCudaArch(CGM);
42384238
switch (Arch) {
4239-
case CudaArch::GFX90a:
4240-
return true;
4241-
default:
4239+
case CudaArch::GFX90a:
4240+
case CudaArch::GFX942:
4241+
return true;
4242+
default:
42424243
break;
42434244
}
42444245
return false;

0 commit comments

Comments
 (0)