[Mlir-commits] [mlir] [MLIR][NVVM] Support for dense and sparse MMA with block scaling (PR #170566)

[Kirill Vedernikov]

================
@@ -1559,6 +1559,650 @@ LogicalResult MmaSpOp::verify() {
   return success();
 }
 
+//===----------------------------------------------------------------------===//
+// MMA Block Scale Operations - Shared Helpers
+//===----------------------------------------------------------------------===//
+
+namespace {
+// Shared structure for MMA operand fragments (A, B, C)
+struct OperandFragment {
----------------
kvederni wrote:

It was renamed from `OperandFragment` to `MMAOperandFragment` in [5febdda](https://github.com/llvm/llvm-project/pull/170566/commits/5febdda3f2742ddec22d3b299ea7e378c14e22d8).

https://github.com/llvm/llvm-project/pull/170566