Re: AMBER: maximum number of processors

From: C J Kenneth Tan -- OptimaNumerics <cjtan.OptimaNumerics.com>
Date: Sat, 14 May 2005 14:57:22 +0000 (UTC)

Murat and Abd,

What you are observing is just simply that the application (AMBER)
doesn't scale in the environment that you tried. This can be due to
many reasons, including: libraries, interconnect, memory, etc..

'top' may give some indication, but it will not give the full picture
(eg: is the CPU stalling, or is the interconnect is stalling?)

You need to run a profiler on the job to get a better idea of the root
of scalability issue.


Kenneth Tan
--------------------------------------------------------------------------
News: OptimaNumerics Enhances Performance on SGI Altix Supercomputers
--------------------------------------------------------------------------
C. J. Kenneth Tan, Ph.D.
OptimaNumerics Ltd. Telephone: +44 798 941 7838
E-mail: cjtan.OptimaNumerics.com Telephone: +44 871 504 3328
Web: http://www.OptimaNumerics.com Facsimile: +44 289 066 3015
--------------------------------------------------------------------------

On 2005-05-13 23:42 -0700 Abd Ghani Abd Aziz (ghaniee.yahoo.com) wrote:

> Date: Fri, 13 May 2005 23:42:49 -0700 (PDT)
> From: Abd Ghani Abd Aziz <ghaniee.yahoo.com>
> Reply-To: amber.scripps.edu
> To: amber.scripps.edu
> Subject: Re: AMBER: maximum number of processors
>
> Dear MURAT CETINKAYA,
>
> Dear MURAT CETINKAYA,
>
> I think that your connection already achieved maximum
> peak. you can rsh to your nodes and do "top" to see
> whether your sander run using 100% of your cpu speed.
>
> Abd Ghani Abd Aziz
> Unievrsiti Putra Malaysia
>
> --- MURAT CETINKAYA <cetinkayam.psu.edu> wrote:
> > Hi Amber users,
> >
> > I normally run sander with 16 nodes (dual processor
> > in each) and I know the
> > speed of the run. This time I tried using 32 nodes,
> > but the speed seems just
> > the same as 16 node runs. The system I simulate is
> > completely the same as
> > before. Is there a limit for the number of nodes
> > that AMBER can use? Or is
> > there another point I cannot figure out?
> >
> > Thanks.
> >
> > PS: I use amber7 and all nodes are functioning well.
> > Murat CETINKAYA
> > Biomolecular Materials Lab,
> > Dept. of Engr. Science and Mechanics,
> > The Pennsylvania State University,
> > University Park, PA 16802
> > office: (814) 863 9967
> > web: www.personal.psu.edu/muc176
> >
> -----------------------------------------------------------------------
> > The AMBER Mail Reflector
> > To post, send mail to amber.scripps.edu
> > To unsubscribe, send "unsubscribe amber" to
> > majordomo.scripps.edu
> >
>
>
>
> __________________________________
> Do you Yahoo!?
> Yahoo! Small Business - Try our new resources site!
> http://smallbusiness.yahoo.com/resources/
> -----------------------------------------------------------------------
> 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 Sat May 14 2005 - 16:53:00 PDT
Custom Search