[gmx-users] Barcelona vs Xeon

[Berk Hess]

Hi,

For Gromacs 4.0, which will come out in a few months, the situation will be
quite different. The parallelization has been much improved and automatic
sorting of atoms optimizes the memory access a lot.
Both these things have much more effect on the Core2 CPUs which don't
have hypertransport or a built in memory controller.
So both the scaling and the absolute will go up significantly for Core2
and only slightly for AMD CPUs.

Berk.

>From: Carsten Kutzner <ckutzne at gwdg.de>
>Reply-To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>To: Discussion list for GROMACS users <gmx-users at gromacs.org>
>Subject: Re: [gmx-users] Barcelona vs Xeon
>Date: Wed, 28 Nov 2007 09:55:05 +0100
>
>Hi Martin,
>
>we have carried out benchmarks on a Barcelona 2.0 GHz versus a 2.3 GHz
>Clovertown. Here are numbers for an 80000 atom system with PME
>electrostatics and Gromacs 3.3.1.
>
>On the Barcelona we get a performance of 0.38, 0.71, 1.27 and 2.40
>ns/day for 1, 2, 4, and 8 CPUs, respectively. On the Clovertown we get
>0.52, 0.97, 1.75, and 2.74 ns/day. So indeed the scaling on the AMDs is
>much better on 8 CPUs (80% compared to 66%), but the absolute
>performance is better on the Intels.
>
>For a 2.0 GHz Clovertown the absolute performance on 8 CPUs is probably
>nearly the same as on a Barcelona, but the single-CPU performance will
>still be better on the Clovertown. One has to keep in mind, however,
>that this is just a single benchmark and other MD systems might show a
>different scaling behaviour.
>
>Carsten
>
>
>
>Martin Höfling wrote:
> > Hi all,
> >
> > Barcelona 1.9GHz vs Xeon 2.0GHz
> >
> > has anyone here benchmarked these two available 8-core solutions? 
>Speed/core
> > wins probably the intel side but what about scaling on 8 cores? Has the
> > NUMA/Hypertransport architecture an advantage here?
> >
> > In price, they are very similar. Xeon is of course available with higher
> > frequencies but also increases in cost/cpu very fast.
> >
> > Best
> > 	Martin
> > _______________________________________________
> > gmx-users mailing list    gmx-users at gromacs.org
> > http://www.gromacs.org/mailman/listinfo/gmx-users
> > Please search the archive at http://www.gromacs.org/search before 
>posting!
> > Please don't post (un)subscribe requests to the list. Use the
> > www interface or send it to gmx-users-request at gromacs.org.
> > Can't post? Read http://www.gromacs.org/mailing_lists/users.php
>
>--
>Dr. Carsten Kutzner
>Max Planck Institute for Biophysical Chemistry
>Theoretical and Computational Biophysics Department
>Am Fassberg 11
>37077 Goettingen, Germany
>Tel. +49-551-2012313, Fax: +49-551-2012302
>http://www.mpibpc.mpg.de/research/dep/grubmueller/
>http://www.gwdg.de/~ckutzne
>_______________________________________________
>gmx-users mailing list    gmx-users at gromacs.org
>http://www.gromacs.org/mailman/listinfo/gmx-users
>Please search the archive at http://www.gromacs.org/search before posting!
>Please don't post (un)subscribe requests to the list. Use the
>www interface or send it to gmx-users-request at gromacs.org.
>Can't post? Read http://www.gromacs.org/mailing_lists/users.php

_________________________________________________________________
Play online games with your friends with Messenger 
http://www.join.msn.com/messenger/overview