[PATCH] D62341: [DAGCombine][X86][AArch64][AMDGPU][MIPS][PPC] (sub x, c) -> (add x, -c) vector edition.

[Roman Lebedev via Phabricator via llvm-commits]

lebedev.ri added a comment.
Herald added a subscriber: ychen.

@nemanjai / @hfinkel ping? any advice on how to proceed here?
It seems PPC is the only regression remaining here.

In D62341#1649938 <https://reviews.llvm.org/D62341#1649938>, @lebedev.ri wrote:

> In D62341#1649924 <https://reviews.llvm.org/D62341#1649924>, @hfinkel wrote:
>
> > In D62341#1647808 <https://reviews.llvm.org/D62341#1647808>, @lebedev.ri wrote:
> >
> > > @hfinkel should i be worried about PPC regressions here?
> > >  Those are there regardless of the patch (instcombine already did this fold).
> > >  I'm not yet sure how to handle them, in some of these patterns the constant
> > >  is already constant-pool load, in some cases it's bitcasted, etc.
> >
> >
> > Yeah, those don't look good. @nemanjai , are we just missing some patterns?
>
>
> I believe you need something similar to https://reviews.llvm.org/D66805#change-JSdQ9NCYAb2i / D63558 <https://reviews.llvm.org/D63558>,
>  but i'm not sure how to do that for PPC, would be great if someone actually familiar with that backend could handle that..





Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D62341/new/

https://reviews.llvm.org/D62341