[PATCH] D145301: Add more efficient vector bitcast for AArch64

[Dave Green via Phabricator via llvm-commits]

dmgreen added a comment.

I haven't looks into the details here, just high level. If we can custom legalize the bitcast then that sounds like it could handle quite a few cases. Although it might not automatically optimize quite so nicely in all of them.

One thing worth remembering is that bitcast work a bit funny under bigendian. They are defined as storing the value then loading it again, which can result in a different lane order.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D145301/new/

https://reviews.llvm.org/D145301