[Mlir-commits] [mlir] [mlir][Vectorizer] Added support to Vectorize tensor.unpack (PR #76087)

[Balaji V. Iyer.]

================
@@ -73,33 +73,52 @@ mlir::tensor::computeTransposedType(RankedTensorType rankedTensorType,
   return transposedTensorType;
 }
 
-SmallVector<int64_t>
-mlir::tensor::getPackInverseDestPermutation(PackOp packOp) {
+SmallVector<int64_t> mlir::tensor::getPackUnPackInverseDestPerm(
----------------
bviyer wrote:

I was thinking about it, but thefunction is used several places and in almost all the places they don't care about getting the PackedMetadata back (only the unpack case wanted it). This is why I created another function that doesn't this parameter and modifed the original function to have this parameter.

https://github.com/llvm/llvm-project/pull/76087