[Mlir-commits] [mlir] [mlir][linalg][nfc] Fix `linalg.matmul_transpose_a` def. (PR #97690)

[Jerry Shih]

================
@@ -1336,7 +1336,7 @@ structured_op: !LinalgStructuredOpConfig
     name: C
     kind: output_tensor
     type_var: U
-    shape_map: affine_map<()[s0, s1, s2] -> (s2, s1)>
+    shape_map: affine_map<()[s0, s1, s2] -> (s1, s2)>
----------------
JerryShih wrote:

> So this PR doesn't change the `indexing_maps` - these seem to be actually correct. And `shape_map` seem to be dropped when converting to tablegen. I guess this is only fixing OpDSL?
> 

Yes, the input/output's affine_map doesn't used in mlir-linalg-ods-gen tool now. I just try to fix the weird declarion in `core_named_ops.py`.

> @JerryShih How can we test/check/verify this change?
Sorry, I don't know how to test/verify this update. It's just try to fix the matrix dim name.



https://github.com/llvm/llvm-project/pull/97690