From NJIT-ARCS HPC Wiki
There is no limit to how many jobs you can enqueue to run, but in some cases scheduler will limit how many jobs can be run concurrently on compute nodes. The SGE documentation refers to this as a "slot limit".
There is no permanent limit, but a temporary limit of 50 is imposed if researchers complain that one or more of their peers are utilizaing all the compute nodes.
The standard limit is 50 concurrently jobs on compute nodes. If you temporarily require more please email email@example.com for an exception to be placed. If a researcher floods the cluster with hundreds of jobs their limit will be set to 20.