Dear siesta list,

I have compiled siesta-3.2-pl-4 on the RWTH Bull computer cluster:
https://sharepoint.campus.rwth-aachen.de/units/rz/HPC/public/Lists/Bull%20Cluster%20Configuration%20Phase%202%20October%202011/AllItems.aspx

I have used intel compiler 14.0 and  openMPI 1.6.5.
I have linked using the
http://software.intel.com/en-us/articles/intel-mkl-link-line-advisor/

After submitting a parallel job the code start running but
it stops at the moment of initializing the DM with the error:

*** An error occurred in MPI_Recv
*** on communicator MPI COMMUNICATOR 10 DUP FROM 9
*** MPI_ERR_TRUNCATE: message truncated
*** MPI_ERRORS_ARE_FATAL: your MPI job will now abort

The very same input file works well on another machine (with intel MPI).
However, on the Bull cluster a compilation with intel MPI produces an executable
whose output file 'freezes' for incompatibility between Intel MPI and
the LSF batch system.

I have contacted the cluster support center and they suggest that the
problem is related to the SIESTA code.

Any hint on how to solve this?

Thanks in advance,
   Zeila


---------------------------------------------------------------
Dr. Zeila Zanolli

Marie Curie Fellow
Peter Grünberg Institut (PGI)    &
Institute for Advanced Simulation (IAS)
Forschungszentrum Jülich
D-52425 Jülich, Germany

ETSF Research Team Leader (http://www.etsf.eu/)

Mobile (Belgian): +32 487 556699
Landline (Germany).: +49 (0)2461 615193
e-mail: [email protected]
web: http://zeilazanolli.wordpress.com/home
ResearcherID: http://www.researcherid.com/rid/F-9568-2010
----------------------------------------------------------------

Responder a