Differences between revisions 15 and 17 (spanning 2 versions)
Revision 15 as of 2008-10-30 11:40:15
Size: 2248
Editor: localhost
Comment: converted to 1.6 markup
Revision 17 as of 2009-04-28 15:51:54
Size: 1569
Editor: GötzWaschk
Comment: openmpi-Pakete von SL
Deletions are marked like this. Additions are marked like this.
Line 3: Line 3:
At Zeuthen, two clusters are available, one with 16 dual Opteron machines connected by Infiniband and one with 8 dual Xeons and Myrinet. They are integrated into the SGE batch system. The documentation in [[Batch_System_Usage]] applies to them.  There are no dedicated parallel clusters available at the moment, but you can run parallel MPI jobs in the SGE farm. The documentation in [[Batch_System_Usage]] applies there.
Line 6: Line 6:
Since the upgrade to SL5, both clusters use the openmpi implementation of the MPI standard.

There are MPI versions for the GCC, Intel and PGI compilers installed:

/opt/openmpi/gcc/bin/mpicc

/opt/openmpi/intel/bin/mpicc

/opt/openmpi/pgi/bin/mpicc

Compilers for C++ and Fortran are available as well.
Since SL5, all batch worker nodes have the openmpi implementation of the MPI standard installed. Recently the machines were upgraded to the default SL5.3 packages of openmpi. For 64 bit applications use the installation in /usr/lib64/openmpi/1.2.7-gcc/bin, for 32 bit use
the binaries from /usr/lib/openmpi/1.2.7-gcc/bin .
Line 19: Line 10:
=== Infiniband ===
Applications for the plejade cluster must be compiled on a 64 bit SL5 machine, e.g. sl5-64.ifh.de.
Line 22: Line 11:
=== Myrinet ===
Applications for the geminide cluster must be compiled on a 32 bit SL5 machine, e.g. sl5.ifh.de.
=== Building applications ===
64 bit MPI Applications can be compiled on any 64 bit SL5 machine, e.g. sl5-64.ifh.de.
Line 27: Line 16:
A job script designated for a parallel job needs to specify the parallel environment and the number of required CPUs. The parameter looks like this for the Infiniband cluster: A job script designated for a parallel job needs to specify the parallel environment and the number of required CPUs. The parameter looks like this:
Line 29: Line 18:
#$ -pe mpich-ppn2 4 #$ -pe multicore-mpi 4
Line 31: Line 20:
On the Myrinet cluster, it is similar: Be sure to call the right mpirun version for your architecture. If you application was compiled for 64 bit, use
Line 33: Line 22:
#$ -pe mpichgm-ppn2 4

It is important to request the right limit for memory with the parameter h_vmem.

The Opteron machines have 3.3G of RAM and by default two jobs are executed on one node, so the maximal amount of memory is 1650M per process:

#$ -l h_vmem=1650M

The Xeons have 922.5M of RAM.

If your application is using threads, it is recommended to set the value of h_stack (by default the same as h_vmem) to a sane value, e.g. 10M.

Be sure to call the right mpirun version for your compiler. If you application was compiled with GCC, use

/opt/openmpi/gcc/bin/mpirun -np $NSLOTS yourapp
/usr/lib64/openmpi/1.2.7-gcc/bin/mpirun -np $NSLOTS yourapp

Usage of the Linux Clusters at DESY Zeuthen

There are no dedicated parallel clusters available at the moment, but you can run parallel MPI jobs in the SGE farm. The documentation in Batch_System_Usage applies there.

Building Applications

Since SL5, all batch worker nodes have the openmpi implementation of the MPI standard installed. Recently the machines were upgraded to the default SL5.3 packages of openmpi. For 64 bit applications use the installation in /usr/lib64/openmpi/1.2.7-gcc/bin, for 32 bit use the binaries from /usr/lib/openmpi/1.2.7-gcc/bin .

Building applications

64 bit MPI Applications can be compiled on any 64 bit SL5 machine, e.g. sl5-64.ifh.de.

Batch System Access

A job script designated for a parallel job needs to specify the parallel environment and the number of required CPUs. The parameter looks like this:

#$ -pe multicore-mpi 4

Be sure to call the right mpirun version for your architecture. If you application was compiled for 64 bit, use

/usr/lib64/openmpi/1.2.7-gcc/bin/mpirun -np $NSLOTS yourapp

AFS Access

The application binary must be available to all nodes, that's why it should be placed in an AFS directory.

BLAS library

Both ATLAS ans GotoBLAS are available.

  • ATLAS is in /opt/products/atlas
  • libgoto is in /usr/lib or /usr/lib64 respectively.

Further documentation

HPC-Clusters at DESY Zeuthen , 11/22/06, technical seminar

Cluster (last edited 2024-05-10 10:44:44 by GötzWaschk)