RE: [AMBER] PROBLEM WITH PMEMD PERFORMANCE

From: Ross Walker <ross.rosswalker.co.uk>
Date: Thu, 15 Apr 2010 23:38:11 -0700

Hi Juan,

> When I run my job on 8 processors it goes forward perfectly, but
> when I run it on 32 processors the job stops at the
> first step and I get the following error :
>
> vlimit exceeded for step 1; vmax = 48.6876
> rank 31 in job 1 cn040.null_58505 caused collective abort of all
> ranks exit status of rank 31: return code 1
>
> How is it possible that my system keeps stable when usign 8 processors
> and unstable when using 32? I guess this has to be a problem of
> the compilation or configuration of PMEMD in the HPC Im working on.
> Could you give me your opinion about it?

This should not be happening, the results on 32 processors should be the
same as on 8 processors, at least for the first few hundred steps before
rounding differences start to cause divergence. However a big initial
difference like this suggests something very wrong. Can you send me your
structure and input files along with the mpi version and compiler version
that was used and I will see if I can replicate this.

All the best
Ross


/\
\/
|\oss Walker

| Assistant Research Professor |
| San Diego Supercomputer Center |
| Tel: +1 858 822 0854 | EMail:- ross.rosswalker.co.uk |
| http://www.rosswalker.co.uk | http://www.wmd-lab.org/ |

Note: Electronic Mail is not secure, has no guarantee of delivery, may not
be read every day, and should not be used for urgent or sensitive issues.






_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Apr 16 2010 - 00:00:04 PDT
Custom Search