The real power of the Grid Engine is to process batch jobs. You have to create a job file and submit it to the batch scheduler. A simple example:

$ cat example.sub

#!/bin/sh

#$ -N test

#$ -S /bin/sh

#$ -cwd

#$ -j y

your_executable_name

The lines starting with #$ are parameters for the Grid Engine. You find all of them in the manpageof qsub. We use the following in our example:

-N test is the name of the job

-S /bin/sh selects the shell used to execute the jobfile, default is csh which is probably not the one you like

-cwd means that the job is started in the current working directory, default is the home

-j y merges STDOUT and STDERR in the same file

Submit this job using qsub:

$ qsub example.sub

Your job 9 (“test”) has been submitted

With few serial jobs send them to the same node with -q option (node name is specify in):

qstat -f

qsub -q all.q@node05.cluster example.sub

Each node has 8 cpus so you can send 8 one cpu jobs per node