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

[llvmlistbot at llvm.org]

================
@@ -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(),
----------------
Muzammiluddin-Syed-ECE wrote:

IIUC, templating doesn't remove the need for branching since I would still be branching off the class of the input op to pass in the appropriate arguments. So, I opted for the branching on op type which I find clearer.

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