[Mlir-commits] [mlir] [mlir][vector] Update tests for xfer-permute-lowering (nfc) (PR #101468)

[Andrzej Warzyński]

================
@@ -325,33 +381,46 @@ module attributes {transform.with_named_sequence} {
 
 //       CHECK:   #[[MAP:.*]] = affine_map<(d0, d1, d2, d3) -> (d1, 0, d3)>
 //       CHECK:   func.func @transfer_read_reduce_rank_scalable(
-//  CHECK-SAME:       %[[ARG_0:.*]]: memref<?x?x?x?xf32>) -> vector<8x[4]x2x3xf32> {
+//  CHECK-SAME:     %[[MEM:.*]]: memref<?x?x?x?xf32>) -> vector<8x[4]x2x3xf32> {
 //       CHECK:     %[[C0:.*]] = arith.constant 0 : index
-//       CHECK:     %[[TFR:.*]] = vector.transfer_read %arg0[%[[C0]], %[[C0]], %[[C0]], %[[C0]]]{{.*}} permutation_map = #[[MAP]]} : memref<?x?x?x?xf32>, vector<[4]x2x3xf32>
+//       CHECK:     %[[TFR:.*]] = vector.transfer_read %[[MEM]][%[[C0]], %[[C0]], %[[C0]], %[[C0]]]{{.*}} permutation_map = #[[MAP]]} : memref<?x?x?x?xf32>, vector<[4]x2x3xf32>
----------------
banach-space wrote:

`TR` tends to be used for transposes :)

On a related note, I think that some level of inconsistency is fine - it's bound to happen. It's just that what we have today is quite a chaos 😅 In my refactoring, I tend to focus on the inputs and the outputs as that's one thing that's common across all tests. But otherwise ... 🤷🏻 

But I won't mind if you prefer more rigour :)

https://github.com/llvm/llvm-project/pull/101468