This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [WIP PATCH] Re: Inefficient end-of-loop value computation - missed optimization somewhere?
- From: "Ulrich Weigand" <uweigand at de dot ibm dot com>
- To: richard dot guenther at gmail dot com (Richard Guenther)
- Cc: gcc-patches at gcc dot gnu dot org, pinskia at gcc dot gnu dot org (Andrew Pinski)
- Date: Thu, 15 Mar 2012 17:50:44 +0100 (CET)
- Subject: Re: [WIP PATCH] Re: Inefficient end-of-loop value computation - missed optimization somewhere?
Richard Guenter wrote:
> On Thu, Mar 8, 2012 at 3:29 PM, Ulrich Weigand <uweigand@de.ibm.com> wrote:
> > I'm wondering where to go from there:
> >
> > - Why are those special re-association cases handled only in forwprop,
> > and not in fold-cost? I would have expected forwprop to just propagate
> > subexpressions into a tree and then use fold-const to actually simplify ...
>
> We are moving away from GENERIC expression simplification towards
> simplifications/combines on GIMPLE SSA (Andrew Pinski is working
> on this a lot on his branch at the moment).
OK, I see.
> > - Should I try to improve forwprop to handle casts and additional re-
> > association cases until it handles the above expression?
>
> Yes, ideally by trying to sub-divide this task into separate profitable
> transforms. Maybe Andrew can chime in here?
Thanks, I'll try to have a look into this direction ...
Bye,
Ulrich
--
Dr. Ulrich Weigand
GNU Toolchain for Linux on System z and Cell BE
Ulrich.Weigand@de.ibm.com