[Mlir-commits] [mlir] [MLIR][NVVM] Fix the datatype error for nvvm.mma.sync when the operand is bf16 (PR #122664)

[Durgadoss R]

================
@@ -1699,8 +1699,8 @@ def NVVM_MmaOp : NVVM_Op<"mma.sync", [AttrSizedOperandSegments]> {
     | f16      | .m8n8k4   | row/col | row/col | 2x f16x2 | 2x f16x2 | 4x f16x2 or 8xf32 |
     |          | .m16n8k8  | row     | col     | 2x f16x2 | 1x f16x2 | 2x f16x2 or 4 f32 |
     |          | .m16n8k16 | row     | col     | 4x f16x2 | 2x f16x2 | 2x f16x2 or 4 f32 |
-    | bf16     | .m16n8k8  | row     | col     | 2x f16x2 | 1x f16x2 | 2x f16x2 or 4 f32 |
-    |          | .m16n8k16 | row     | col     | 4x f16x2 | 2x f16x2 | 2x f16x2 or 4 f32 |
+    | bf16     | .m16n8k8  | row     | col     | 2x i32   | 1x i32   | 4x f32            |
----------------
durga4github wrote:

One question:
So, the "num items" is not changed since the previous type was "f16x2" (i.e. also a 32-bit entity)?

https://github.com/llvm/llvm-project/pull/122664