You're really talking about parallel processing in this case - the most basic environment of which is called "OpenMP".
To make use of more than one CPU core on a node for your job, use the OpenMP parallel environment by adding the following option to your qsub command:
-pe OpenMP 4
where 4 is the number of CPU cores you need. Remember that phase 1 hardware only has 4 cpu cores, and phase 2 has 8. If you request more than this, your job cannot run.