[PATCH] D136432: [AMDGPU] Combine BFI instructions.

[Thomas Symalla via Phabricator via llvm-commits]

tsymalla added a comment.

In D136432#3873879 <https://reviews.llvm.org/D136432#3873879>, @jsilvanus wrote:

>> As such case cannot be combined into a BFI instruction, I guess it's relatively straight-forwardly translated into something like this:
>
> Why can't we, isn't this essentially the definition of BFI?

I think I misunderstood your question. If you are asking how generic BFIs are generated:
Such cases where we can prove that Y is used on both sides of the OR expression, will be currently transformed by InstCombineAndOrXor to the canonical form after the non-canonical form is created in the frontend. The sequence is then directly picked up and transformed during ISel into a single V_BFI. There's a pattern match for it.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D136432/new/

https://reviews.llvm.org/D136432