From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-patches-return-333768-listarch-gcc-patches=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 26751 invoked by alias); 7 Dec 2012 16:52:10 -0000
Received: (qmail 26741 invoked by uid 22791); 7 Dec 2012 16:52:07 -0000
X-SWARE-Spam-Status: No, hits=-6.0 required=5.0	tests=AWL,BAYES_00,KHOP_RCVD_UNTRUST,KHOP_THREADED,RCVD_IN_DNSWL_HI,RP_MATCHES_RCVD
X-Spam-Check-By: sourceware.org
Received: from cantor2.suse.de (HELO mx2.suse.de) (195.135.220.15)    by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Fri, 07 Dec 2012 16:52:00 +0000
Received: from relay1.suse.de (unknown [195.135.220.254])	(using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits))	(No client certificate requested)	by mx2.suse.de (Postfix) with ESMTP id B830DA4EB7;	Fri,  7 Dec 2012 17:51:59 +0100 (CET)
Date: Fri, 07 Dec 2012 16:52:00 -0000
From: Michael Matz <matz@suse.de>
To: Jan Hubicka <hubicka@ucw.cz>
Cc: Richard Biener <richard.guenther@gmail.com>,	Pat Haugen <pthaugen@linux.vnet.ibm.com>,	GCC Patches <gcc-patches@gcc.gnu.org>,	David Edelsohn <dje.gcc@gmail.com>
Subject: Re: [PATCH, rs6000] Update default loop peel limits
In-Reply-To: <20121207164300.GA14220@kam.mff.cuni.cz>
Message-ID: <alpine.LNX.2.00.1212071745300.21868@wotan.suse.de>
References: <50C0FCDE.5010508@linux.vnet.ibm.com> <20121206214313.GG671@atrey.karlin.mff.cuni.cz> <CAFiYyc20F4eKv0zBc22zEjrQehA5y5umYcyeY51Bj7V3fZY36A@mail.gmail.com> <20121207164300.GA14220@kam.mff.cuni.cz>
User-Agent: Alpine 2.00 (LNX 1167 2008-08-23)
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-IsSubscribed: yes
Mailing-List: contact gcc-patches-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Id: <gcc-patches.gcc.gnu.org>
List-Archive: <http://gcc.gnu.org/ml/gcc-patches/>
List-Post: <mailto:gcc-patches@gcc.gnu.org>
List-Help: <mailto:gcc-patches-help@gcc.gnu.org>
Sender: gcc-patches-owner@gcc.gnu.org
X-SW-Source: 2012-12/txt/msg00510.txt.bz2

Hi,

On Fri, 7 Dec 2012, Jan Hubicka wrote:

> > > Actually the calculix regression is also seen on core.
> > > Igor was looking into what loops got slower and why. Either we can fix that partiuclar
> > > loop or revert to the old default (that sadly causes quite a lot of code bloat)
> > 
> > Well, as the patch regressed the testcase in put in the testsuite for calculix
> > that is no surprise... gfortran.dg/reassoc_4.f.  The fix for the testcase was to
> > increase the limit with a --param :/
> 
> I would not care too much about limits for gfortran.dg/reassoc_4.f, it 
> is an artificail testcase.

Ehm, no it's not.  It _specifically_ is the important inner loop nest of 
calculix and it should be compiled to exactly the expected form (i.e. 22 
multiplications and unrolled) without any changes to params.  At least 
when you expect that -Ofast is giving good results on 454.calculix.

I.e. this testcase breaking automatically implies slower calculix results.


Ciao,
Michael.