[PATCH] D73135: [AArch64][ARM] Always expand ordered vector reductions (PR44600)

[Dave Green via Phabricator via llvm-commits]

dmgreen accepted this revision.
dmgreen added a comment.
This revision is now accepted and ready to land.

Ah, I had presumed that the SVE instructions were pairwise. Nice.

I think we may run into something similar in MVE for nonan min/max reductions, where we do not know how to lower fmaximum. And we may not be testing NoNan flags correctly. I will attempt to go through and make sure we have tests and that it's working OK.

LGTM for this patch. Thanks.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D73135/new/

https://reviews.llvm.org/D73135