[gmx-users] old patch for slow networks

[Carsten Kutzner]

Jones de Andrade wrote:
> Hi all.
> 
> I'm writing this message to ask a simple and direct question: the 
> available patch of gmx3 for slow networks, that includes the 
> ordered-all-to-all procedures, seems fo be indicated for version 3.1.
Hi Jones,

this is actually a patch for gromacs 3.3.1.

> It's still "necessary" for slow networks using gmx3.3?

Whether it is necessary depends on your setup. In the patched version 
the all-to-all communication happens in an orderly fashion such that TCP 
packet drops are avoided. So the patch will speed up your parallel 
performance if you have packet drops in the first place. An indication 
for such drops is when you make speedup measurements on 1, 2, 3, ... 
nodes (each node may have multiple CPUs) and your performance curve is 
not smooth but abruptly drops down when you go to more than two nodes.

With the patch comes a test program for all-to-all problems.

Hope that helps,
   Carsten

> 
> Thanks a lot in advance...
> 
> Sincerally yours,
> 
> Jones
> 
> 
> ------------------------------------------------------------------------
> 
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please search the archive at http://www.gromacs.org/search before posting!
> Please don't post (un)subscribe requests to the list. Use the 
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php

-- 
Dr. Carsten Kutzner
Max Planck Institute for Biophysical Chemistry
Theoretical and Computational Biophysics Department
Am Fassberg 11
37077 Goettingen, Germany
Tel. +49-551-2012313, Fax: +49-551-2012302
http://www.mpibpc.mpg.de/research/dep/grubmueller/
http://www.gwdg.de/~ckutzne