I am trying to compile ESPResSo in a small cluster. But I run into some troubles.
The cluster runs rocks cluster 6.0. It is composed of one master and five slaves nodes. Each slave node has two cores.
I have two different but related problems.
i) I tried to compile ESPResSo as a root in the folder: /opt/.
bio condor eclipse fftw-3.3.3 gridengine iozone maven mpi-tests pdsh pxeflash rocks stream SUNWhpc tomcat-5.0.30 tomcat-6.0.14
chkrootkit confuse ESPResSo ganglia intel iperf mpich2 openmpi perl python sqlite sun-ct tomcat tomcat-5.5.17 tripwire
====
mpirun was unable to launch the specified application as it could not access
or execute an executable:
Executable: ./Espresso"
====
To launch my jobs via qsub, I created either a symbolic link to the ESPResSo executable or specifying the full path and used a "file.sh", that looks like:
#!/bin/bash
#$ -cwd
#$ -j y
#$ -S /bin/bash
/opt/openmpi/bin/mpirun -n 2 ./Espresso < Mixt_Colloids.tcl
ii) My second problem is related with the number of processors that ESPResSo recognizes during compilation.
Because of the problem stated above, I decided to compile ESPResSo as a normal user in my home.
I successfully compile it, and I able to run serial and parallel jobs that uses a number of processors EQUAL to the number of cores in each node.
If I try to run a single job with say 8 processors, thus ESPResSo runs in parallel in one single node!
It does not use the 8 available cores.
Checking the config.log file, I found the following lines:
========
configure:7874: checking the number of available CPUs
configure:7895: result: 2
=====
So, I was wondering if anybody has experience with rocks clusters, where is the right place to compile ESPResSo as a root to allow normal users to launch jobs? and How I can compile ESPResSo in order to run in all available nodes (and cores)?
I would appreciate any help or suggestion.
As always... Thanks a lot for your time and help!
Salvador