[llvm-dev] [RFC] How to manifest information in LLVM-IR, or, revisiting llvm.assume

[Doerfert, Johannes via llvm-dev]

On 12/18, Doerfert, Johannes wrote:
> On 12/18, John McCall wrote:
> > On 18 Dec 2019, at 14:18, Doerfert, Johannes wrote:
> > > Hi John,
> > > 
> > > Is it correct to assume that you are in favor of
> > >   - changing llvm.assume to be more expressive
> > >   - operand bundle uses, especially w/o outlining
> > >   - outlining, if we can show it plays well with transformations, e.g.
> > >     the binding to the code is "weak"
> > 
> > Yes, this is all correct.  I’m excited that we’re looking into making this
> > less useless.
> 
> Perfect! I'll go ahead and prototype something then so we can collect
> data and experience.

First prototype: https://reviews.llvm.org/D71692

From what I've seen, we can the proposed changes work if we want to.
Outlining comes with the most overhead when we want to keep the same
amount of information but we can gradually work on this and adopt the
"native" operand bundle encodings first.

I haven't observed a problem wrt. outlining preventing anything but some
information is not properly propagated back yet.

Cheers,
  Johannes
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 228 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20191219/8bc3d429/attachment.sig>