[clang] [CIR][AMDGPU] Support packed v2f16 and v2bf16 atomic fadd builtins (PR #231555)

Arseniy Obolenskiy via cfe-commits cfe-commits at lists.llvm.org
Thu Nov 5 21:28:48 PST 2026


================
@@ -1587,7 +1568,7 @@ mlir::LogicalResult CIRToLLVMAtomicFetchOpLowering::matchAndRewrite(
   if (auto intTy = mlir::dyn_cast<cir::IntType>(op.getVal().getType())) {
     isInt = true;
     isSignedInt = intTy.isSigned();
-  } else if (mlir::isa<cir::SingleType, cir::DoubleType>(
+  } else if (mlir::isa<cir::SingleType, cir::DoubleType, cir::VectorType>(
----------------
aobolensk wrote:

Switched to cir::isFPOrVectorOfFPType

https://github.com/llvm/llvm-project/pull/231555


More information about the cfe-commits mailing list