RE: [AMBER] PROBLEM WITH PMEMD PERFORMANCE

From: Ross Walker <ross.rosswalker.co.uk>
Date: Tue, 13 Apr 2010 09:00:13 -0700

Hi Juan,

Looks like you only built the serial version of PMEMD.

It would help to see the beginning of your input file, the section from the
beginning to section 4. I bet there is no MPI in there.

Recompile PMEMD in parallel and try again.

Note, from AMBER 11 parallel PMEMD will be renamed pmemd.MPI (the serial
being called pmemd) to avoid such confusion.

All the best
Ross

> -----Original Message-----
> From: amber-bounces.ambermd.org [mailto:amber-bounces.ambermd.org] On
> Behalf Of Juan Carlos
> Sent: Monday, April 12, 2010 3:24 AM
> To: amber.ambermd.org
> Subject: [AMBER] PROBLEM WITH PMEMD PERFORMANCE
>
> Hello again Amber users,
>
> I working on PMEMD molecular dynamics with several systems, and Ive
> observed
> a weird behaviour of the PMEMD performance when checking the output
> files. You
> can see an example of what Im refering to in the document attached.
> Ive always seen
> this behaviour when using PMEMD, and the problem of it is that the
> dynamics slow
> down significantly because, according to the output file, they return
> to
> previous steps
> again a again. Could it be a problem of the compilation of PMEMD in the
> high
> performance computer Im using?
>
> Im looking forward to reading your wise comments.
>
> Thank you.
> /
> Juan Carlos Muoz Garca.
> Institute of Chemical Research, Sevilla, Spain.
> Spanish Research Council./


_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Apr 13 2010 - 09:30:03 PDT
Custom Search