This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [graphite] Remove checks for flag_loop_strip_mine, flag_loop_block, flag_loop_interchange in graphite_trans_bb_block.
- From: Jack Howarth <howarth at bromo dot msbb dot uc dot edu>
- To: Tobias Grosser <grosser at fim dot uni-passau dot de>
- Cc: Sebastian Pop <sebpop at gmail dot com>, gcc-patches at gcc dot gnu dot org, rajagopal dwarak <dwarak dot rajagopal at amd dot com>, Jagasia Harsha <harsha dot jagasia at amd dot com>, Konrad Trifunovic <konrad dot trifunovic at gmail dot com>, louis-noel dot pouchet at inria dot fr, Cédric Bastoul <cedric dot bastoul at inria dot fr>, Karthik A <chill dot shanky at gmail dot com>, Harle Christophe <christophe dot harle at amd dot com>, Albert Cohen <Albert dot Cohen at inria dot fr>, Adrien ELICHE <aeliche at isty dot uvsq dot fr>, Sjodin Jan <Jan dot Sjodin at amd dot com>
- Date: Thu, 14 Aug 2008 16:04:58 -0400
- Subject: Re: [graphite] Remove checks for flag_loop_strip_mine, flag_loop_block, flag_loop_interchange in graphite_trans_bb_block.
- References: <1218738013.1353.52.camel@tobilaptop>
On Thu, Aug 14, 2008 at 03:20:13PM -0300, Tobias Grosser wrote:
> Hi graphities, hi world,
>
> as we are discussing about merging graphite to trunk, I think more people
> will be interested in graphite changes. So I will send all my patches
> to gcc-patches.
>
> Have fun and follow the graphite development!
>
> I worked on the last remaining polyhedron problems reported by Jack. Now
> graphite compiles all polyhedron test cases.
> (BTW: Jack, thanks for testing graphite so much)
>
Tobias,
Thanks. I'll try another build tonight. Do we have any particular
expections on what sort of speed improvements all of the loop optimizations
might yield on the polyhedron benchmarks? So far I don't think I've seem
more than a few percent.
Jack
ps Dorit Nuzman apparently will try to fix PR33113 for gcc 4.4 so
that the vectorizor can handle the symbolic stride information that secv
returns. This should allow more of the polyhedron benchmarks to be
vectorized.