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

[Krzysztof Drewniak]

================
@@ -687,6 +687,11 @@ def MFMAOutTypes : AnyTypeOf<[F64,
                               VectorOfLengthAndType<[4, 16, 32], [F32]>,
                               VectorOfLengthAndType<[4, 16, 32], [I32]>,
                               VectorOfLengthAndType<[4], [F64]>]>;
+// scaled_mfma
+def ScaledMFMAInTypes : AnyTypeOf<[VectorOfLengthAndType<[8], [F8E5M2FNUZ, F8E4M3FNUZ]>,
+                             VectorOfLengthAndType<[8, 32], [F8E5M2, F8E4M3FN]>,
+                             VectorOfLengthAndType<[32], [F6E2M3FN, F6E3M2FN, F4E2M1FN]>]>;
+def ScaledMFMAOutTypes : AnyTypeOf<[VectorOfLengthAndType<[4, 16, 32], [F32]>]>;
----------------
krzysz00 wrote:

I don't think there's a 32xf32 output?

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