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

[Justin Rosner]

================
@@ -447,101 +447,100 @@ LogicalResult WMMAOp::verify() {
 //===----------------------------------------------------------------------===//
 
 LogicalResult ScaledWMMAOp::verify() {
+  // Helper functions for type classification.
+  auto isF8 = [](Type t) {
+    return isa<Float8E4M3FNType, Float8E5M2Type, Float8E8M0FNUType,
+               Float8E4M3FNUZType, Float8E5M2FNUZType>(t);
+  };
----------------
justinrosner wrote:

It looks like we need E5M2 and E4M3. I don't think the ocp uses the FNUZ variants, so I removed those as well

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