[Beowulf] VASP on Clusters

Peter Kjellström cap at nsc.liu.se
Mon May 9 02:46:16 PDT 2005


We run VASP on several clusters here and we have had no reports from users 
that would indicate problems when running complex jobs on many processors.

Your description below does not say how bad the results are. Running a job on 
many processors will almost never generate the exact same results as running 
it on one processor would. This is the case for many parallel applications.

The only other thing with VASP and multi processor runs that I am aware of is 
that some components will only run on 2^n number of processors. Starting them 
on anything else should (according to our VASP guy) only result in some 
idleing processors and not corrupt output.

Anyway here is what we currently run (as reference):
OS           : rh-7.3 to centos-3.4 (rhel3u4)
Compiler     : intel-8.0-039
VASP version : 4.6
CPU types    : Xeon 2.2 GHz to Pentium4 3.4 GHz
MPI libs     : mpich-1.2.5,1.2.6 on ethernet

If you could dig up a clear cut (preferably small) testcase that demonstrates 
this problem I'd be glad to try it on our systems.

/Peter K

On Thursday 05 May 2005 18.13, Daniel Majchrzak wrote:
> ...
> The problem is that when we run the program on larger jobs the results
> are inconsistent.  As I understand it, the older cluster will run small
> systems, but gives erroneous results on medium and large jobs.  The
> newer cluster gives good results on small and medium runs but fails on a
> large test case. (The results are different then those from a single
> processor job run on two different machines.)
> ...
> Thanks,
> Dan

  Peter Kjellström               |
  National Supercomputer Centre  |
  Sweden                         | http://www.nsc.liu.se
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://www.beowulf.org/pipermail/beowulf/attachments/20050509/0ffb949c/attachment.sig>

More information about the Beowulf mailing list