Date: Thu, 15 Apr 1999 02:09:05 +0100 From: Mark Brown To: Bill Thorson Cc: pgcc AT delorie DOT com Subject: Re: performance question Message-ID: <19990415020905.B231@tardis.ed.ac.uk> Mail-Followup-To: Bill Thorson , pgcc AT delorie DOT com References: <3714B295 DOT DBFA9425 AT atmos DOT colostate DOT edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-Mailer: Mutt 0.95.3i In-Reply-To: <3714B295.DBFA9425@atmos.colostate.edu>; from Bill Thorson on Wed, Apr 14, 1999 at 09:21:58AM -0600 X-Cookie: IOT trap -- core dumped X-WWW-Homepage: http://www.tardis.ed.ac.uk/~broonie/ Reply-To: pgcc AT delorie DOT com X-Mailing-List: pgcc AT delorie DOT com X-Unsubscribes-To: listserv AT delorie DOT com Precedence: bulk On Wed, Apr 14, 1999 at 09:21:58AM -0600, Bill Thorson wrote: > I'M NOT A MEMBER OF THIS LIST TO PLEASE REPLY TO ME AND NOT THE LIST. Unfortunately, it has Reply-To: set to the list, so some replies may not make it to you. > I am involved with numerical weather modeling and we sometimes use > Pentium class computers running Linux. Does the PGCC compiler patch > also provide speed improvements for g77? Is this something that I > should try? When you have a model that may runs for 7 days a 5-10% > speed-up IS significant. The optimizations are in the back end, so g77 gets the improvements. Exactly what you're going to see can only be found by benchmarking and fine-tuning the options. You may also encounter bugs, but if you're running compute-intensive jobs on Pentium hardware it's probably worth trying. The FAQ contains a number of suggestions for optimizations to try. There have been a number of problems reported in the 1.1.2 release - you might want to try 1.1.1 or the latest development sources (avalible in CVS). -- Mark Brown mailto:broonie AT tardis DOT ed DOT ac DOT uk (Trying to avoid grumpiness) http://www.tardis.ed.ac.uk/~broonie/ EUFS http://www.eusa.ed.ac.uk/societies/filmsoc/