[gmx-users] New single thread performance numbers

[Florian Haberl]

Hi,

On Monday, 17. September 2007 18:25, Jones de Andrade wrote:
> Hi Erik.
>
> Still under NDA? Haven't it been lift last sunday?

If its pre released hardware than it will never expire.

>
> Anyway: I don't know exactly the terms of the NDA, So if you can't answer
> me the following question, please say "no, I can't answer". I'm in touch
> with people having some compability issues with their Barcellonas and MBs,
> so could you tell us what are the configs you are dealing with?
>
> If that information is also under the NDA, really sorry for the question.
>
> Sincerally yours,
>
> Jones
>
> On 9/17/07, Erik Lindahl <lindahl at cbr.su.se> wrote:
> > Hi,
> >
> > Yes, the SPEC numbers aren't representative for typical gromacs
> > performance.
> >
> > Technically I'm still on NDA for the benchmarks I ran on Barcelona so
> > I can't share the exact numbers with you, but on average the
> > performance will be pretty much the same _per_clock_ as Intel
> > Clovertown. AMD scales a little bit better, but Intel benefits from a
> > larger cache.
> >
> > Still, it's very nice to see that AMD is at least back on the field
> > again. Let's hope for an aggressive frequency ramp-up and a price
> > war ;-)
> >
> > Cheers,
> >
> > Erik
> >
> > On Sep 17, 2007, at 2:58 PM, Mark Abraham wrote:
> > > Florian Haberl wrote:
> > >> Hi,
> > >> Barcelona is out now and SPEC CPU2006 (www.spec.org) has released
> > >> new numbers (I have taken them from the german c't magazin
> > >> www.heise.de):
> > >> Single core performance:
> > >> 435.gromacs
> > >> Opteron 2350 74.4
> > >> Opteron 2222 51.1
> > >> Xeon X5365   143.3
> > >
> > > Caveat that these numbers are for simulations without assembly
> > > optimization - i.e. C nonbonded loops. The assembly ones are
> > > considerably faster. Ref - GROMACS website.
> > >
> > > Mark
> > > _______________________________________________
> > > 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
> >
> > _______________________________________________
> > 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

-- 
-------------------------------------------------------------------------------
 Florian Haberl                        
 Computer-Chemie-Centrum   
 Universitaet Erlangen/ Nuernberg
 Naegelsbachstr 25
 D-91052 Erlangen
 Telephone:  	+49(0) − 9131 − 85 26581
 Mailto: florian.haberl AT chemie.uni-erlangen.de
-------------------------------------------------------------------------------