qsub

the qsub command sends a job to the queue system (eg. qsub script), which receives a unique number (the so-called job-id).
Sample form of the script file:

#PBS -N name
#PBS -q queue
#PBS -l resources
#PBS -j oe
job
All information passed to the queue system in the script must start with #PBS characters, after which you can use different options:

• -N defines the name of the task; lack of this option causes the name of the task to be the initial string of characters from the script,
• -q defines the queue under which the task is to be executed; if this option is missing, the task will be started as part of the default queue,
• -l defines the use of different resources:
• nodes, e.g. nodes=4 means the use of 4 nodes on which it will be used 1 core, nodes=2:ppn=5 means the use of two nodes on which 5 cores will be used,
• walltime, np walltime=2:15:40 means to complete the task in real time in 2h 15m 40s,
• mem, np mem=320kb means the use of 320kB memory,
• -j connects the output stream and errors into one file with the standard name: name.ojob-id; lack of this option will create two files: name.ojob-id and name.ejob-id,
• about other options you can learn from the command help (man qsub).

The following example of a script sent to the queue:
#PBS -N test-1
#PBS -q batch
#PBS -l nodes=3:ppn=4, walltime=2:00:00
#PBS -j oe
#PBS -o test.o
job
defines the following conditions of the job:

1. the job will have a name test-1
2. the job will be running in the queue batch
3. three nodes will be needed for execution, 4 cores will be occupied on each of them
4. the job will be performed for a maximum of 2 hours
5. the error stream will be attached to the output stream
6. the result of the script will be saved in the file test.o