[llvm] [mlir] [AMDGPU][MLIR]Add shmem-optimization as an op using transform dialect (PR #81550)

[Oleksandr Alex Zinenko via llvm-commits]

================
@@ -52,12 +50,12 @@ constexpr int64_t kDefaultVectorSizeBits = 64;
 static Value permuteVectorOffset(OpBuilder &b, Location loc,
                                  ArrayRef<Value> indices, MemRefType memrefTy,
                                  int64_t srcDim, int64_t tgtDim) {
-  // Adjust the src index to change how often the permutation changes
-  // if necessary.
+  /// Adjust the src index to change how often the permutation changes
+  /// if necessary.
----------------
ftynse wrote:

Doxygen-style (`///`) comments are intended for top-level entities such as classes and functions, *not* for inline comments. Please revert back to the regular `//` comments.

https://github.com/llvm/llvm-project/pull/81550