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

[Krzysztof Drewniak]

================
@@ -506,6 +506,40 @@ LogicalResult GatherToLDSOp::verify() {
   return success();
 }
 
+LogicalResult ScaledMFMAOp::verify() {
+  unsigned opselA = getOpselA() >> 8;
+  unsigned opselB = getOpselB() >> 8;
+
+  if (opselA != 0)
+    return emitOpError("Opsel A must be a zero extended 8 bit value");
+
+  if (opselB != 0)
+    return emitOpError("Opsel B must be a zero extended 8 bit value");
+
+  auto isValidType =
----------------
krzysz00 wrote:

This should be handled with a variant of `MFMAInTypes`

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