================
@@ -3327,6 +3327,37 @@ TEST(TargetParserTest,
testAMDGPUDynamicVGPRAllocGranule) {
8u);
}
+TEST(TargetParserTest, testAMDGPUDynamicVGPRAddressableNum) {
+ for (auto Kind :
+ {AMDGPU::GK_GFX1200, AMDGPU::GK_GFX1250, AMDGPU::GK_GFX1310}) {
+ SCOPED_TRACE(AMDGPU::getArchNameAMDGCN(Kind).str());
+ for (bool IsWave32 : {false, true}) {
----------------
nkotikal wrote:
getAddressableNumVGPRs in this test shouldnt reach the IsWave32 conditional
anyway because DynamicVGPRBlockSize is nonzero. So it doesn't make sense to me
to loop over both false and true unless you believe this is an important
regression guard?
https://github.com/llvm/llvm-project/pull/226821
_______________________________________________
llvm-branch-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits
