Jobs can be queued using the PBS queuing system. The basic idea is to create a PBS job file (see below) then enter qsub pbs_job_file. Related commands that may be of interest are qstat, qdel, qmgr and xpbs. Try pbsjobs (located in /home/Tools/bin) for additional detail about queued and running jobs.
Available queues include: default (limit 48 hours), fast (limit 2 hours), long (limit 7 days), long-long (limit 14 days). For current information on queue capacities and other properties run qmgr. Available special resources are: CPU types xeon28 (2.8GHz Xeon), P4D930 (Pentium 4 D 930), x5160 (Xeon 5160); Memory GB2, GB4, GB8. For a list of nodes and their properties, enter pbsnodes -a. Also, see the hardware description for the cluster.
For example, to request two nodes each with two 2.8GHz Xeons and 4 GB of
memory on the fast queue, enter
qsub -l nodes=2:ppn=2:xeon28:GB4 -q fast pbs_job_file
Sample pbs_job_file:
#!/bin/csh
#PBS -j oe
cd $PBS_O_WORKDIR
date > RunAt
hostname >> RunAt
kvasp > output
echo Done >> RunAt
date >> RunAt
Sample pbs_job_file for MPI:
#!/bin/csh
#PBS -j oe
#PBS -l nodes=1:ppn=2
cd $PBS_O_WORKDIR
date > RunAt
echo MPI Job-ID $PBS_JOBID >> RunAt
echo hostname `hostname` >> RunAt
cat $PBS_NODEFILE > node-file
echo PBS_NODEFILE >> RunAt
cat $PBS_NODEFILE >> RunAt
mpirun -np 2 -machinefile node-file pkvasp > output
echo Done >> RunAt
date >> RunAt