From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <gcc-return-49853-listarch-gcc=gcc.gnu.org@gcc.gnu.org>
Received: (qmail 14917 invoked by alias); 12 Apr 2002 17:55:00 -0000
Mailing-List: contact gcc-help@gcc.gnu.org; run by ezmlm
Precedence: bulk
List-Archive: <http://gcc.gnu.org/ml/gcc/>
List-Post: <mailto:gcc@gcc.gnu.org>
List-Help: <http://gcc.gnu.org/ml/>
Sender: gcc-owner@gcc.gnu.org
Received: (qmail 14902 invoked from network); 12 Apr 2002 17:54:50 -0000
Received: from unknown (HELO pheriche.sun.com) (192.18.98.34)
  by sources.redhat.com with SMTP; 12 Apr 2002 17:54:50 -0000
Received: from engmail2.Eng.Sun.COM ([129.146.1.25])
	by pheriche.sun.com (8.9.3+Sun/8.9.3) with ESMTP id LAA19404;
	Fri, 12 Apr 2002 11:54:44 -0600 (MDT)
Received: from boojum.Eng.Sun.COM (boojum.Eng.Sun.COM [129.144.178.171])
	by engmail2.Eng.Sun.COM (8.9.3+Sun/8.9.3/ENSMAIL,v2.1p1) with ESMTP id KAA01597;
	Fri, 12 Apr 2002 10:54:44 -0700 (PDT)
Received: from localhost (starlex@localhost)
	by boojum.Eng.Sun.COM (8.9.3+Sun/8.9.1) with ESMTP id KAA15922;
	Fri, 12 Apr 2002 10:54:44 -0700 (PDT)
X-Authentication-Warning: boojum.Eng.Sun.COM: starlex owned process doing -bs
Date: Fri, 12 Apr 2002 11:42:00 -0000
From: Alexey Starovoytov <starlex@eng.sun.com>
X-X-Sender:  <starlex@boojum>
Reply-To: Alexey Starovoytov <stork@sun.com>
To: Andi Kleen <ak@suse.de>
cc: <gcc@gcc.gnu.org>
Subject: Re: contribution
In-Reply-To: <p73d6x5bd3c.fsf@oldwotan.suse.de>
Message-ID: <Pine.GSO.4.31.0204121042030.15886-100000@boojum>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-SW-Source: 2002-04/txt/msg00505.txt.bz2

On 12 Apr 2002, Andi Kleen wrote:

> Alexey Starovoytov <starlex@eng.sun.com> writes:
>
> > For example SPECfp benchmark 172.mgrid showed 48% improvement
> > compiled with "-O2 -Ws,-O2,-fno-strict-aliasing" vs plain GCC with "-O2"
>
> You could compiled 172.mgrid with gcc twice this way and it showed 48%
> improvement ? Or did you use a different compiler as backend?
It was gcc3.0.4 as a backend.

Environment:
US2 450Mhz Solaris 2.8

gcc_3.0.4   -O2:
Success 172.mgrid ratio=59.73, runtime=3013.766979

GCC_3.0.4 -> C -> GCC_3.0.4    -O2 -Ws,-O2,-fno-strict-aliasing
Success 172.mgrid ratio=88.55, runtime=2032.768725

GCC_3.1 -> C -> GCC_3.0.4    -O2 -Ws,-O2,-fno-strict-aliasing
Success 172.mgrid ratio=98.22, runtime=1832.557732

> If you used gcc - is it known where that huge difference comes from?
I didn't have time to check.

Best regards,
Alexey.