Policies of the High Performance Computing Center at WSU have been established to give access to all users to the resources of the center. Details are outlined below in the following categories.
Any WSU faculty, student or post-doc that requests an account will be given time on our systems. See our User Information page for more details.
For jobs that require large amounts of memory or cpu time, users should run jobs in batch mode with Portable Batch System (PBS).
Users will access cronus only. From cronus, users will submit jobs to run on aries cluster.
In order to manage resources we have set up the following run limits. There
are the batch queues:
para: 72 hour limit, 2-16 cpu's
icing: Restricted to icing group
compile: 48 hour limit, 1 cpu
cobalt: Restricted to cobalt group
benchmark: Restricted to administrator
stdin: 72 hour limit, 2-16 cpu's, must use -I -V on qsub
special: no limits but only 1 job per user
pellet: Restricted to pellet group
ema3d: Restricted to ema3d group
To view the current limits of the various queues, run qstat -q.
The system administrator may adjust the queues as needed to allow for maximum utilization while avoiding oversubscription (too many jobs active). In addition, the system administrator may release queued jobs when other queues are underutilized. Thus, the qstat report may show more jobs running than allowed.
Please do not run single process serial jobs in the para queue.
Please do not leave interative PBS sessions (-I parameter) inactive for an extended period of time. . Do not start an interactive session just to reserve a slot and not use it. If there are jobs waiting in the same queue, then an idle job is subject to cancellation
We do back up the user data in /user.
We have tied job submission to licenses for the following applications. Thus, the job will not start until licenses are available.
abaqus Use abaqus_ncpus
fluent Use fluent_ncpus