[Mlir-commits] [mlir] [mlir][vector] Update `CastAwayInsertLeadingOneDim` (PR #211777)

[Federico Bruzzone]

================
@@ -534,10 +534,10 @@ func.func @cast_away_insert_leading_one_dims_scalar_scalable(%s: f32, %v: vector
 // CHECK-SAME:    %[[S:.*]]: f32,
 // CHECK-SAME:    %[[V:.*]]: vector<1x[1]x4xf32>) -> vector<1x[1]x4xf32> {
 func.func @cast_away_insert_leading_one_dims_scalar_skip_scalable_dim(%s: f32, %v: vector<1x[1]x4xf32>) -> vector<1x[1]x4xf32> {
-// CHECK:           %[[EXTRACT:.*]] = vector.extract %[[V]][0] : vector<[1]x4xf32> from vector<1x[1]x4xf32>
-// CHECK:           %[[INSERT:.*]] = vector.insert %[[S]], %[[EXTRACT]] [0, 0] : f32 into vector<[1]x4xf32>
-// CHECK:           %[[BCAST:.*]] = vector.broadcast %[[INSERT]] : vector<[1]x4xf32> to vector<1x[1]x4xf32>
-// CHECK:           return %[[BCAST]] : vector<1x[1]x4xf32>
+// CHECK:           %[[SC:.*]] = vector.shape_cast %[[V]] : vector<1x[1]x4xf32> to  vector<[1]x4xf32>
----------------
FedericoBruzzone wrote:

```suggestion
// CHECK:           %[[SC:.*]] = vector.shape_cast %[[V]] : vector<1x[1]x4xf32> to vector<[1]x4xf32>
```

https://github.com/llvm/llvm-project/pull/211777