[Mlir-commits] [mlir] [mlir][vector] Use inferRankReducedResultType for subview type inference. (PR #84395)

[Andrzej Warzyński]

================
@@ -169,3 +169,27 @@ func.func @non_unit_strides(%arg0: memref<512x16x1xf32, strided<[8192, 16, 4], o
 // The inner most unit dims can not be dropped if the strides are not ones.
 // CHECK:     func.func @non_unit_strides
 // CHECK-NOT:   memref.subview
+
+// -----
+
+func.func @contiguous_outer_dyn_inner_most_view2(%a: index, %b: index, %memref: memref<?x?x8x1xf32>) -> vector<8x1xf32> {
----------------
banach-space wrote:

It's not obvious what makes this case unique - perhaps just update `contiguous_outer_dyn_inner_most_view` instead? Alternatively, 
* `contiguous_outer_dyn_inner_most_view2` -> `contiguous_outer_dyn_inner_most_view_2_dyn_dims` (to highlight makes this unique),
* move near `contiguous_outer_dyn_inner_most_view` so that it's easy to contrast both cases (and find tests for similar scenarios).

This is really just a nit, feel free to ignore. 

https://github.com/llvm/llvm-project/pull/84395