[PATCH] D57862: [AArch64] Fix condition for "high-vector" DUP optimizations.

[Eli Friedman via Phabricator via llvm-commits]

efriedma added a comment.

Yes, it probably makes sense to high vector extracts in some cases.  Actually, for some loops, it might even make sense to introduce an extra shuffle outside the loop to avoid a high-vector extract inside the loop.


Repository:
  rL LLVM

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D57862/new/

https://reviews.llvm.org/D57862