[gmx-users] Experiences with Gromacs scaling on US supercomputer centers?

[David van der Spoel]

On 2012-09-26 16:24, Michael Shirts wrote:
> Hi all,
>
> I'd be interested to know about people's experiences with Gromacs on
> US national computing centers.  Which machines have it set up to scale
> the
> best?  We're putting in an XSEDE request soon, and I'm trying to
> figure out which resource to request.  Our system is
> semi-coarse-grained, using
> reaction field electrostatics, so we don't have to worry about PME.
> Of course, couldn't hurt to know about PME scaling as well.   I'm
> interesting
> scalings with 100K - 300K atoms.
>
> Of course, best performance will probably change with 4.6 because of
> all the setup tweaks, but let's start with 4.5 scaling info!

I guess you're aware of Roland Schulz' paper on cellulose systems on 
Jaguar? Over 100 million particles on 100,000 cores. We have done virus 
particles (1.2 million particles) on 2000 Cray XE6 cores (in Sweden).

You are probably aware that scaling is different for different systems, 
in particular if you don't have waters. Your systems are relatively 
small for the big US computers of course, but it is often allowed to do 
ensemble calculations.

I would suggest you get a test account for scaling tests for your 
particular system.

>
> Best,
> ~~~~~~~~~~~~
> Michael Shirts
> Assistant Professor
> Department of Chemical Engineering
> University of Virginia
> michael.shirts at virginia.edu
> (434)-243-1821
>


-- 
David van der Spoel, Ph.D., Professor of Biology
Dept. of Cell & Molec. Biol., Uppsala University.
Box 596, 75124 Uppsala, Sweden. Phone:	+46184714205.
spoel at xray.bmc.uu.se    http://folding.bmc.uu.se