
The hostname of the cluster located at the Informatics Facility is:
xeuser.ivec.org
You will need an ssh client to connect to the system.
Linux and MacOSX users should be able to use command-line ssh and scp.
If you are connecting from a Windows machine, a freely available ssh client is PuTTY. To transfer files to/from the system you can use the freely available WinSCP.
There are several file systems available depending on user requirements. The appropriate file system should be use whenever possible.
Small quota, backed up. Intended for source code, executables and input files
Large quota, not backed up. Intended to be used for job data that must live beyond the lifetime of the job
There is currently only one queue available on the system:
Environment modules are used to customise your shell environment, allowing for dynamic changes to your path environment variables. They work for all supported shells. This is the preferred way to use software that is installed on the system.
will list the software currently supported by modules. If you require a package that is not listed, please contact the iVEC systems administration staff to request its installation (subject to licensing restrictions).
will modify the environment settings required by package_name.
will return the environment to it's state prior to the module load command.
will show what modules you currently have loaded.
For example, the command module load intel-cc/9.0.026 will adjust your PATH environment variable to include the location of the Intel C/C++ compiler executables, your MANPATH environment variable to include the location of the Intel C/C++ compiler manpages and your LD_LIBRARY_PATH environment variable to include the location of the Intel C/C++ compiler libraries.
Module commands can be added to your shell startup scripts and to jobscripts.
Each interactive process you run or batch job you submit has imposed on it a time limit, a memory use limit, and a processor count limit.
Some commands to monitor system resource usage are:
The batch system in use on xe is Torque. You submit jobs to Torque specifying the number of CPUs, the amount of memory, and the length of time needed (and possibly other resources). Torque runs the job when the resources are available, subject to constraints on maximum resource usage.
% qsub -P aa99 -q normal -l walltime=20:00:00,vmem=300MB -w ./a.out ^D (control-D)
or via a jobscript:
% cat jobscript #!/bin/csh #PBS -P a99 #PBS -q normal #PBS -l walltime=20:00:00,vmem=300MB #PBS -w ./a.out
which can be submitted via the command qsub jobscript.
When you want to run a MPI job, you must load either the openmpi module (openmpi compiled with GCC) or the openmpi-intel module (openmpi compiled with Intel C compiler).
You don't need to specify the number of processors to run using the -np switch, you specify the number using qsub
Each node has eight cores, so if we want to run a 16 processor job, we would run on two nodes with eight cores per node. So we would submit the following:
% cat jobscript module load openmpi mpirun a.out
using qsub -l nodes=2:ppn=8 jobscript