[Mlir-commits] [mlir] [mlir][vector] add tensor.concat, bitcast, expand_shape, collapse_shape vectorization support (PR #97297)

[Oleksandr Alex Zinenko]

================
@@ -1931,6 +2134,108 @@ vectorizePadOpPrecondition(tensor::PadOp padOp,
   return success();
 }
 
+static LogicalResult
+lowerExpandOpPrecondition(tensor::ExpandShapeOp expandOp,
+                          ArrayRef<int64_t> inputVectorSizes) {
+  auto resultType = expandOp->getResultTypes()[0];
+  auto resultShape = mlir::dyn_cast<ShapedType>(resultType);
+  // check reassociation
+  llvm::SmallVector<int64_t> associateIndices;
+  for (auto &attr : expandOp.getReassociation()) {
----------------
ftynse wrote:

If `auto` had been expanded, it would have been clear that a reference is not necessary here.

https://github.com/llvm/llvm-project/pull/97297