[Mlir-commits] [mlir] [mlir][AMDGPU] Add scaled wmma ops for gfx1250 (PR #169854)

[Justin Rosner]

================
@@ -1218,6 +1227,56 @@ def AMDGPU_ScaledMFMAOp :
   let hasCanonicalizer = 1;
 }
 
+def AMDGPU_ScaledWMMAOp
+    : AMDGPU_Op<"scaled_wmma", [AllTypesMatch<["destC", "destD"]>, Pure]>,
+      Arguments<(ins ConfinedAttr<I32Attr, [IntIsOneOf<[16, 32]>]>:$m,
+          ConfinedAttr<I32Attr, [IntIsOneOf<[16]>]>:$n,
+          ConfinedAttr<I32Attr, [IntIsOneOf<[128]>]>:$k,
+          ScaledWMMAInTypes:$sourceA, ScaledWMMAInTypes:$sourceB,
+          ScaledWMMAOutTypes:$destC,
+          VectorOfLengthAndType<[4, 8], [F8E8M0FNU, F8E4M3FN]>:$scaleA,
+          ConfinedAttr<I32Attr, [IntIsOneOf<[0, 1]>]>:$scaleAIdx,
+          VectorOfLengthAndType<[4, 8], [F8E8M0FNU, F8E4M3FN]>:$scaleB,
+          ConfinedAttr<I32Attr, [IntIsOneOf<[0, 1]>]>:$scaleBIdx)>,
----------------
justinrosner wrote:

Updated to use `a_first_scale_lane` and `b_first_scale_lane`. Also, I saw that you updated https://github.com/llvm/llvm-project/pull/170718 already.

https://github.com/llvm/llvm-project/pull/169854