[Mlir-commits] [mlir] [mlir][AMDGPU] Add gfx950 MFMAs to the amdgpu.mfma op (PR #133553)

[Daniel Hernandez-Juarez]

================
@@ -687,7 +708,14 @@ static std::optional<StringRef> mfmaOpToIntrinsic(MFMAOp mfma,
       return ROCDL::mfma_f32_16x16x8bf16::getOperationName();
   }
 
-  if (isa<IntegerType>(sourceElem) && destElem.isInteger(32)) {
+  if (isa<IntegerType>(sourceElem) && sourceElem.getIntOrFloatBitWidth() >= 8 &&
----------------
dhernandez0 wrote:

sourceElem.isInteger(8)? I think all intrinsics here are i8 input?

https://github.com/llvm/llvm-project/pull/133553