[Mlir-commits] [mlir] [mlir][amdgpu] Define an amdgpu.scaling_mfma wrapper (PR #137498)

[Krzysztof Drewniak]

================
@@ -826,11 +827,20 @@ mfmaOpToScaledIntrinsic(Type aType, Type bType, Type destType, uint32_t m,
 }
 
 static std::optional<std::tuple<StringRef, uint32_t, uint32_t>>
-mfmaOpToScaledIntrinsic(MFMAOp mfma, Chipset chipset) {
-  return mfmaOpToScaledIntrinsic(
-      mfma.getSourceA().getType(), mfma.getSourceB().getType(),
-      mfma.getDestC().getType(), mfma.getM(), mfma.getN(), mfma.getK(),
-      mfma.getBlocks(), chipset);
+mfmaOpToScaledIntrinsic(Operation *op, Chipset chipset) {
+  if (auto mfma = llvm::dyn_cast_or_null<MFMAOp>(op)) {
----------------
krzysz00 wrote:

I had this is two functions that differed by type for a reason - if nothing else, the ScalingMfma version always has blocks == 1

https://github.com/llvm/llvm-project/pull/137498