[Mlir-commits] [mlir] [MLIR][AMDGPU] Add OCP FP8 support for new hardware (PR #106160)

[Jakub Kuderski]

================
@@ -272,14 +272,16 @@ LogicalResult MFMAOp::verify() {
   }
 
   Type sourceBType = getSourceB().getType();
-  if (sourceElem.isFloat8E5M2FNUZ() || sourceElem.isFloat8E4M3FNUZ()) {
+  if (sourceElem.isFloat8E5M2FNUZ() || sourceElem.isFloat8E4M3FNUZ() ||
+      sourceElem.isFloat8E5M2() || sourceElem.isFloat8E4M3FN()) {
----------------
kuhar wrote:

Also here and in the other conditions below.

It may also make sense to add a helper in some core header that checks if the type is any f8 type.

https://github.com/llvm/llvm-project/pull/106160