Job queues on the PPPL cluster
Job queues on the PPPL cluster
A submitted job will be automatically sent to the
correct queue by the job scheduler. The scheduler will determine, from
the number of nodes requested, the processors per node, and requested memory,
where the job should be queued for execution.
For special purpose jobs only (like those requiring Infiniband),
the queue must be specified in
your job script.
See usage examples for more information.
There are several batch queues on the PPPL cluster:
dawson
(the default queue for parallel jobs)
ellis (the default queue for small jobs)
kruskal (the Infiniband cluster)
mque (for large memory jobs)
gque (for jobs using GPU cores)
There are also application specific queues, like jassby for ITER work,
and mccune for TRANSP jobs.
To see jobs running on a specific queue, use the qstat -a 'queuename' command.