[Mlir-commits] [mlir] [mlir][linalg] Enable masked vectorisation for depthwise convolutions (PR #81625)

[Cullen Rhodes]

================
@@ -300,3 +300,20 @@ vector::createUnrollIterator(VectorType vType, int64_t targetRank) {
   shapeToUnroll = shapeToUnroll.slice(0, firstScalableDim);
   return StaticTileOffsetRange(shapeToUnroll, /*unrollStep=*/1);
 }
+
+SmallVector<OpFoldResult> vector::getMixedSizesXfer(bool hasTensorSemantics,
+                                                    Operation *xfer,
+                                                    RewriterBase &rewriter) {
+  auto loc = xfer->getLoc();
+
+  Value blah = TypeSwitch<Operation *, Value>(xfer)
----------------
c-rhodes wrote:

use more descriptive name
```suggestion
  Value base = TypeSwitch<Operation *, Value>(xfer)
```

https://github.com/llvm/llvm-project/pull/81625