Submitting Jobs for Execution

The cluster uses the Torque PBS resource manager for job submission. PBS enables users to simply define the parameters of what their job needs (x number of processors, with y amount of memory, and for z amount of time), and PBS considers all of these parameters and schedules the job for execution on the appropriate node(s). With PBS, there is no need to manually seek out a node that is unused to run your computations on, or worry about other users starting a job on the same node(s) as yours.

Jobs are submitted to PBS with a job script that defines a number of parameters to inform PBS about how to schedule it. We will step through a sample PBS script and explain what each directive does.