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

[Krzysztof Drewniak]

================
@@ -833,6 +833,14 @@ mfmaOpToScaledIntrinsic(MFMAOp mfma, Chipset chipset) {
       mfma.getBlocks(), chipset);
 }
 
+static std::optional<std::tuple<StringRef, uint32_t, uint32_t>>
+mfmaOpToScaledIntrinsic(ScaledMFMAOp smfma, Chipset chipset) {
+  return mfmaOpToScaledIntrinsic(smfma.getSourceA().getType(),
----------------
krzysz00 wrote:

Yeah, the reason this isn't a pure `template<typename MfmaOp>` is because the arguments don't match (MFMA has a getBlocks(), ScalingMfma doesn't, for example)

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