Re: AMBER: NEC and Regatta

From: Robert Duke <rduke.email.unc.edu>
Date: Mon, 28 Jun 2004 08:18:24 -0400

There has been lots of work with pmemd on both the p690 and the
alpha/quadrics (see amber.scripps.edu website for links to pmemd resource
pre-amber8 releases of pmemd (3.0, 3.03, 3.1), pmemd 8 is part of the amber
8 release). There were lots of benchmarks published for the 3/3.03/3.1
releases for both of these platforms, and a lot of numbers are included for
the various versions of sander too. For amber 8, there are benchmarks for
these platforms on the website, generally referencing pmemd or my name.
Sorry, but I have no numbers on the NEC.
Regards - Bob Duke

----- Original Message -----
From: "Filip Lankas" <filip.lankas.epfl.ch>
To: <amber.scripps.edu>
Sent: Monday, June 28, 2004 7:47 AM
Subject: AMBER: NEC and Regatta


> Dear Amber users,
>
> does anyone have an experience with running Amber 7/8 on
>
> 1. NEC SX-5
>
> 2. IBM Regatta p690 (with Power4 CPUs)
>
> What speed and what scaling (with no. of atoms and no. of processors)
> can one expect? How would it compare with, say, Alpha ES45 connected
> with Quadrix? Are there some peculiarities to be aware of?
>
> I'm particularly interested in running explicit-solvent MD with PME.
>
> Thanks a lot in advance,
>
> Filip
>
> --
> ---------------------------------------------
> Dr. Filip Lankas
> Bernoulli Institute for Mathematics
> EPFL (Swiss Federal Institute of Technology)
> CH-1015 Lausanne
> Tel: +41 21 693 2593 Fax: +41 21 693 5530
> ---------------------------------------------
>
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>


-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Mon Jun 28 2004 - 13:53:00 PDT
Custom Search