OSCER users,

DON'T MISS THIS CHANCE!
SUBMIT YOUR JOBS WHILE QUEUE PENDING TIMES ARE LOW!

On Schooner, queue pending times for jobs that are currently
running in the "public" queues:

(Public queues: normal, all, largemem, largejobs, longjobs,
gpu, 32gb_20core, 64gb_24core, 2650, 2670.)

Percentage of pending time / requested runtime

* 1 core/up to 1 GB: < 1%
(For example, if a job requested 48 hours, it pended ~half an hour.)

* 1 core/3-4 GB: 2%
(For example, if a job requested 48 hours, it pended ~an hour.)

* 1 core/5-8 GB: 1%
(For example, if a job requested 48 hours, it pended ~half an hour.)

* 1 core/9-16 GB RAM: 179%
(For example, if a job requested 48 hours, it pended ~86 hours.)
(One user submitted hundreds of these.)

* 2 cores/5-8 GB RAM: << 1%
(For example, if a job requested 48 hours, it pended ~half an hour.)

* 5-8 cores/5-8 GB RAM: 1%
(For example, if a job requested 48 hours, it pended ~half an hour.)

* 5-8 cores/9-16 GB RAM: 9%
(For example, if a job requested 48 hours, it pended ~4 hours.)

* 9-16 cores/0.5-1 GB RAM: 570%
(For example, if a job requested 48 hours, it pended ~274 hours.)

* 9-16 cores/5-8 GB RAM: 52%
(For example, if a job requested 48 hours, it pended ~25 hours.)

* 9-16 cores/9-16 GB RAM: 0%
(For example, if a job requested 48 hours, it started in seconds.)

* 9-16 cores/17-32 GB RAM: 14%
(For example, if a job requested 48 hours, it pended ~7 hours.)

* 1 full node: 61%
(For example, if a job requested 48 hours, it pended ~29 hours.)

* 2 full nodes: 17%
(For example, if a job requested 48 hours, it pended ~8 hours.)

* 3-4 full nodes: 18%
(For example, if a job requested 48 hours, it pended ~9 hours.)

* 5-8 full nodes: 12%
(For example, if a job requested 48 hours, it pended ~6 hours.)

* 33-64 full nodes: << 1%
(For example, if a job requested 48 hours, it pended ~half an hour.)

NOTE: These pending times are AVERAGES based on what happened to
jobs that are CURRENTLY RUNNING. Newly submitted jobs may
experience substantially different pending times.

Our goal is to have most jobs start after pending for
no more than the amount of runtime that was requested,
but this depends on what else is running and pending.

Many factors affect how long jobs pend in the queue
before starting, including:

* how long a job has been pending already;
* how big the job is (cores/nodes and RAM requested);
* how much runtime the job has requested ("wall clock" hours);
* how much (or how little) usage you've had recently.

---

Henry Neeman ([log in to unmask])
Director, OU Supercomputing Center for Education & Research (OSCER)
Associate Professor, Gallogly College of Engineering
Adjunct Associate Professor, School of Computer Science
OU Information Technology
The University of Oklahoma

Engineering Lab 212, 200 Felgar St, Norman OK 73019
405-325-5386 (office), 405-325-5486 (fax), 405-245-3823 (cell),
[log in to unmask] (to e-mail me a text message)
http://www.oscer.ou.edu/