espressomd-users
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ESPResSo] Hardware Requirements for Espresso and Other Things


From: Olaf Lenz
Subject: Re: [ESPResSo] Hardware Requirements for Espresso and Other Things
Date: Tue, 20 May 2008 10:18:20 +0200
User-agent: Thunderbird 2.0.0.14 (X11/20080505)

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi!

Here comes an answer to your first question.

Lorenzo Isella wrote:
> My institute would like to invest some money into computer hardware.
> In the future, I may want/need to run rather intensive simulations
> (using Espresso) which go well beyond what can be achieved by a single
> desktop pc.
> I am sure that many people on this list run their simulations on a
> cluster, so I wonder if there are some specific hardware requirements
> I should point out (RAM memory for each node, CPU, pitfalls for
> running parallel jobs etc... ) for Espresso.

Depending on the size of your jobs, recent multicore-CPUs and boards
with multiple CPUs are well suited for the use with ESPResSo. In
principle, the more cores the better, as ESPResSo can make good use of
them (e.g. Dual Quad cores). Therefore, a cluster of loosely coupled
(i.e. via Ethernet) nodes that have multi-CPU boards of multicore CPUs
should be enough for solving most problems in ESPResSo.

- From the price-performance-ratio point of view, current x86-64-Hardware
is unbeatable. However, while Opterons offer better memory-performance,
Intel-CPUs offer larger caches. Therefore, if you plan to study mainly
systems with less than 5000 particles, an Intel-CPU with 4-8MB cache is
ideal, otherwise an Opteron-System is the better choice.

If you want to do really huge simulations that require more than the
available number of cores, you will also need a fast network between
them (Infiniband or the like). However, this very much increases the
cost of a cluster.

The even more expensive big shared memory machines are not really worth
it, as ESPResSo only uses MPI parallelization, so the shared memory can
not be effectively used.

Best regards
        Olaf and Axel

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD4DBQFIMolMtQ3riQ3oo/oRArSDAJYh1jPrgvFvOXepG5KE6iJXoA7dAJoDV4kV
QkU9N+AckQJLx2lzq3Y90Q==
=1kb8
-----END PGP SIGNATURE-----



reply via email to

[Prev in Thread] Current Thread [Next in Thread]