This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Loop optimizer issues
- From: Pop Sébastian <pop at gauvain dot u-strasbg dot fr>
- To: Aldy Hernandez <aldyh at redhat dot com>
- Cc: Diego Novillo <dnovillo at redhat dot com>,Zdenek Dvorak <rakdver at atrey dot karlin dot mff dot cuni dot cz>,"gcc at gcc dot gnu dot org" <gcc at gcc dot gnu dot org>,Richard Henderson <rth at redhat dot com>,Jason Merrill <jason at redhat dot com>, Jan Hubicka <jh at suse dot cz>,Daniel Berlin <dberlin at dberlin dot org>
- Date: Mon, 2 Jun 2003 23:06:38 +0200
- Subject: Re: Loop optimizer issues
- References: <20030530183552.GA27110@atrey.karlin.mff.cuni.cz> <1054585449.9789.146.camel@frodo.toronto.redhat.com> <xvtk7c49pcw.fsf@krusty.sfbay.redhat.com>
On Mon, Jun 02, 2003 at 01:49:35PM -0700, Aldy Hernandez wrote:
> >>>>> "Diego" == Diego Novillo <dnovillo@redhat.com> writes:
>
> > To avoid duplicate effort, let's just raise our hand whenever we are
> > going to get started with something. The initial list that I had in
> > mind (at the tree level) includes:
>
I'm working on an induction variable detector based on chains of recurrences.
There is an older algorithm based on the detection of strongly connected
components in the SSA described by M.Wolfe that could be interesting to have
in the case where we're not happy with the chrecs...
> I'm starting to work on an aldy^H^H^Hutovectorization pass. We seem
Cool :-)
> to have most of the infrastructure in place except data analysis and
> the loop unroller. I can probably get away without the loop unroller
> for now by unrolling the loops myself.
>
The unroller can be recycled from the one at rtl level.
However I still don't have the number of iterations a loop will run,
and this seems to be dependent on the IV detector.
Should I prepare a patch that registers the loop unroller in the cfg hooks?