AMBER Archive (2007)

Subject: Re: AMBER: pmend vs sander.MPI

From: Robert Duke (rduke_at_email.unc.edu)
Date: Thu Dec 06 2007 - 09:18:20 CST


Yes. The minimum number of processors is 1 (for which you must build the
uniprocessor version in pmemd - the mpi version of pmemd requires a minimum
of 2 processors, unlike sander). There currently is no OpenMPI
configuration option, but you can probably look at the libraries and include
files being used by sander.MPI, and will get it right.
Regards - Bob Duke
----- Original Message -----
From: "Francesco Pietra" <chiendarret_at_yahoo.com>
To: "Amber" <amber_at_scripps.edu>
Sent: Thursday, December 06, 2007 10:13 AM
Subject: AMBER: pmend vs sander.MPI

> Is pmend likely to improve performance with respect to sander.MPI for a
> small
> number of processors on dualopteron amd64 Linux/OpenMPI/intel compilers?
> If
> yes, which is the minimum number of processors to try if pmend defeats
> sander.MPI in speed for a long MD (production)?
>
> Thanks
> francesco pietra
> Very likely the answer was already on the web, though I was unable to find
> it.
> I'll try to compile pmend according to the response.
>
>
>
>
> ____________________________________________________________________________________
> Never miss a thing. Make Yahoo your home page.
> http://www.yahoo.com/r/hs
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
>

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu