[PATCH] D111960: [X86][AVX] Prefer VINSERTF128 over VPERM2F128 for 128->256 subvector concatenations

[Simon Pilgrim via Phabricator via llvm-commits]

RKSimon added a comment.

In D111960#3069056 <https://reviews.llvm.org/D111960#3069056>, @lebedev.ri wrote:

> Speaking of later updating the costs, what do you think about marking the tuples
> needing revisiting in the cost tables for which their codegen has changed
> in `llvm/test/CodeGen/X86/vector-interleaved-*` here?

I agree - we should be trying to help maintain the link between the two.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D111960/new/

https://reviews.llvm.org/D111960