Re: performance loss on GPUs

From: Aron Broom (broomsday_at_gmail.com)
Date: Wed Nov 28 2012 - 09:57:06 CST

Just checked some simulations, and I'm only really seeing an ~5% speed drop
over 50,000,000 steps with implicit solvent. That is also using
metadynamics, which could slow things down over time also.

~Aron

On Wed, Nov 28, 2012 at 3:21 AM, Axel Kohlmeyer <akohlmey_at_gmail.com> wrote:

> On Wed, Nov 28, 2012 at 8:13 AM, Thomas Evangelidis <tevang3_at_gmail.com>
> wrote:
> > Greetings,
> >
> > Is it normal to see a performance loss on GPUs after a few hours of
> > simulation. I am running an implicit solvent simulation on 2 Tesla M2070
> and
> > 2 CPUs. At the beginning I got 0.0126392 s/step, but after 9 hours I get
> > 0.0248662/step. Is this normal?
>
> there are two possible explanations.
>
> - the system becomes more disordered
> over time and hence memory accesses
> become more scattered and thus caching
> is less efficient. many MD programs now
> have an option to sort atoms regularly
> according to their domains (or patches in
> NAMD lingo). the most extreme effect has
> been reported for HOOMD, which uses only
> an single GPU and sorts atom according
> to a space filling hilbert curve. this is more
> likely to happen for a dense system, i.e.
> with explicit liquid
>
> - the average number of neighbors is changing,
> e.g. when the system shrinks or aggregates.
> more neighbors within the cutoff means more
> work to compute and hence slower execution.
> this is more likely for an implicit solvent system,
> but you are in the best position to tell.
>
>
> ciao,
> axel.
>
>
> >
> > thanks,
> > Thomas
> >
> >
> > --
> >
> > ======================================================================
> >
> > Thomas Evangelidis
> >
> > PhD student
> >
> > University of Athens
> > Faculty of Pharmacy
> > Department of Pharmaceutical Chemistry
> > Panepistimioupoli-Zografou
> > 157 71 Athens
> > GREECE
> >
> > email: tevang_at_pharm.uoa.gr
> >
> > tevang3_at_gmail.com
> >
> >
> > website: https://sites.google.com/site/thomasevangelidishomepage/
> >
> >
> >
>
>
>
> --
> Dr. Axel Kohlmeyer akohlmey_at_gmail.com http://goo.gl/1wk0
> International Centre for Theoretical Physics, Trieste. Italy.
>
>

-- 
Aron Broom M.Sc
PhD Student
Department of Chemistry
University of Waterloo

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:22:18 CST