[PATCH] D83663: [LoopRotate] Bump rotation-max-header-size to 64

Roman Lebedev via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Mon Jul 13 04:40:02 PDT 2020


lebedev.ri added a comment.

In D83663#2146895 <https://reviews.llvm.org/D83663#2146895>, @fhahn wrote:

> > - there's pretty small (+0.01 .. +0.05% compile-time regression)
> > - this causes, on average, +0.03% size-text increase So i'm wondering, what is the procedure here, would this be a reasonable change?
>
> I think the cap is mostly there to guard against unnecessary code size increases.


Aha.

> It would be interesting to see what the impact on code size would be with -Os across more than the ctmark tests. I think for -Oz we already force the header size to 0, but I am not sure if it would be great to add another -Os threshold.

I did indeed think about `-Os`/`-Oz` cases, and i indeed suspect this might not be beneficial for them.
I'm fully okay with keeping `16` for `-Os`.

But sadly i'm rather unfamiliar wit them, so i'm not sure i'm in position to actually assess the effects for `-Os`.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D83663/new/

https://reviews.llvm.org/D83663





More information about the llvm-commits mailing list