This is the mail archive of the fortran@gcc.gnu.org mailing list for the GNU Fortran project.

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]
Other format:	[Raw text]

Re: induct regression in gcc 4.3.1

From: Tim Prince <TimothyPrince at sbcglobal dot net>
To: Jack Howarth <howarth at bromo dot msbb dot uc dot edu>
Cc: fortran at gcc dot gnu dot org
Date: Fri, 20 Jun 2008 18:00:55 +0900
Subject: Re: induct regression in gcc 4.3.1
Domainkey-signature: a=rsa-sha1; q=dns; c=nofws; s=s1024; d=sbcglobal.net; h=Received:X-YMail-OSG:X-Yahoo-Newman-Property:Message-ID:Date:From:Reply-To:User-Agent:MIME-Version:To:CC:Subject:References:In-Reply-To:X-Enigmail-Version:Content-Type:Content-Transfer-Encoding; b=CSjNA8mg1d639qn0NrdsSJNOxkl7PykYqSykAsNVZbOJczhNZ0dAY0Wmt1/L4LXZanGcZ0bSXQPC9/EZQuQZEFWpQUyYLUffMshUbS/9OdQgQRjVG+AnPgh+ak6zaeIkvtQMI/lzrLNaPDCxRCy+ei7kKuuOCWTJAV0vFkG+Va8= ;
References: <20080621000141.GA25884@bromo.msbb.uc.edu>
Reply-to: tprince at computer dot org

Jack Howarth wrote:

   It appears that the performance regression which
I found in the induct polyhedron benchmark in gcc 4.3.1 is
triggered by the use of -ffast-math. Benchmarking induct
compiled with gcc 4.2.4 and 4.3.1 shows....
gcc 4.2.4 gcc 4.3.1 Ave Run Ave Run -O3 73.57 71.71 -ffast-math -O3 68.27 89.76

This seems insufficiently informative. Is it attempting to vectorize dot products of length 3, for example?

Follow-Ups:
- Re: induct regression in gcc 4.3.1
  - From: Jack Howarth
- Re: induct regression in gcc 4.3.1
  - From: Jack Howarth

References:
- induct regression in gcc 4.3.1
  - From: Jack Howarth

Index Nav:	[Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav:	[Date Prev] [Date Next]	[Thread Prev] [Thread Next]