To submit a job, users have to create a submission script like the ones below. There are three utilities that you will need for submitting, monitoring, and deleting jobs. These are:

       qsub32 --- submit jobs to the queueing system.

       qstat32 --- check the status of your job(s).

       qdel32 --- delete a job from the queueing system

Users can use "qsub32 script_name" to submit a job and use "qdel32 job_number" to delete a job. The current configuration does not have any complex name. All of the nodes are available for both serial and parallel jobs. Users have to use  #$ -l  (options) in their DQS script for their parallel jobs. For example:

    #!/bin/bash
    #$ -cwd
    #$ -l qty.eq.4
    #$ -N Your_Job_Name
    #$ -A Your_Account_Name
    mpirun -np $NUM_HOSTS -machinefile $HOSTS_FILE ./a.out

will run a.out on 4 CPUs by using the MPICH parallel protocol. For users who prefer to use the LAM MPI protocol, the script would be:

    #!/bin/bash
    #$ -cwd
    #$ -l qty.eq.4
    #$ -N Your_Job_Name
    #$ -A Your_Account_Name
     lamboot $HOSTS_FILE
     lmpirun -c $NUM_HOSTS -c2c ~/a.out
     wipe $HOSTS_FILE

For a serial job, the DQS script would be:

    #!/bin/csh
    #$ -l qty.eq.1
    #$ -N Your_Job_Name
    #$ -A Your_Account_Name
    #$ -cwd
     ./a.out

  Please read the DQS User's Guide for more information.

  You can learn more about DQS at Brookhaven National Lab and University of Minnesota Supercomputing Institute. You can also visit the DQS homepage at Florida State University.

If the maximum cpu time is exceeded, the system will send a SIGXCPU to your process. If you choose to trap and ignore the signal, the operating system will kill your process.

Attempts to exceed memory limits will simply result in denial of memory allocation to whichever system call requests it. Your process will not be explicitly killed, but it may die, depending on how you've written your code.