[PATCH] D92553: [ARM][X86] Match dual lane vmovs from insert_vector_elt

[Dave Green via Phabricator via llvm-commits]

dmgreen added a comment.

Now that I think about it more, perhaps the X86 code changes should be handled differently. There were three tests that changed, two of which (avx512-insert-extract.ll and haddsub-2.ll) are the same instructions in a different order. The other (avx512-mask-op.ll:test18) has a different series of kshiftl/kshiftr's with a single extra instruction.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D92553/new/

https://reviews.llvm.org/D92553