From owner-freebsd-amd64@FreeBSD.ORG Fri Dec 3 06:02:57 2004 Return-Path: Delivered-To: freebsd-amd64@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 7A58016A4CE for ; Fri, 3 Dec 2004 06:02:57 +0000 (GMT) Received: from troutmask.apl.washington.edu (troutmask.apl.washington.edu [128.208.78.105]) by mx1.FreeBSD.org (Postfix) with ESMTP id 5008743D2F for ; Fri, 3 Dec 2004 06:02:57 +0000 (GMT) (envelope-from sgk@troutmask.apl.washington.edu) Received: from troutmask.apl.washington.edu (localhost [127.0.0.1]) iB362vmH016236; Thu, 2 Dec 2004 22:02:57 -0800 (PST) (envelope-from sgk@troutmask.apl.washington.edu) Received: (from sgk@localhost)iB362veW016235; Thu, 2 Dec 2004 22:02:57 -0800 (PST) (envelope-from sgk) Date: Thu, 2 Dec 2004 22:02:56 -0800 From: Steve Kargl To: Tommi L?tti Message-ID: <20041203060256.GA16165@troutmask.apl.washington.edu> References: <20041203011408.163FE106C28@FreeBSD.csie.NCTU.edu.tw> <20041203045849.GA15850@troutmask.apl.washington.edu> <41AFF4BF.7070503@iki.fi> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <41AFF4BF.7070503@iki.fi> User-Agent: Mutt/1.4.1i cc: FreeBSD-amd64@freebsd.org Subject: Re: OT Re: AMD64- Solutions X-BeenThere: freebsd-amd64@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Porting FreeBSD to the AMD64 platform List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 03 Dec 2004 06:02:57 -0000 On Fri, Dec 03, 2004 at 02:08:15PM +0900, Tommi L?tti wrote: > Steve Kargl wrote: > >As far as my jobs, I run numerical simulations of heat > >transfer in biological tissues and nonlinear acoustics. > >The numerical techniques are standard and hybrid finite > >difference solutions to coupled PDEs. I can easily > >consume the 16 GB if not more memory as I add more > >realistic physics to the problem. > > Interesting. Are you using a home-brewn software for analysis or are you > running a commercial package? I use homebrewed in that I write all my own numerical codes in Fortran 90/95. I can profile the code and identify numerical and time consuming bottlenecks where I can spend my time optimizing the algorithms. I have found that I simply cannot get any performance out of any matlab solution (not to mention that matlab restricts one to at most 2 GB of memory). > I'm doing some electrophysiology research (in vivo and in vitro) and > we're contemplating currently using some Matlab and/or Labview for > analyzis. The Lynx produces some 64 channels of data and one run can > easily amount a couple of gigs worth of files... which would easily > warrant the use of some 2-way+ configurations. Although matlab doesn't > thread... If you already have matlab code in house, then check out octave. You can find it in ports/math/octave. I have been able to run most matlab codes given to me by others with little or no changes. octave actually uses the system's memory better than matlab can. -- Steve