When you obtain a new HPC account, you will be provided with storage.  The shared storage (/home, /groups, /xdisk) is accessible from any of the three production clusters: Puma, Ocelote and ElGato. The temporary (/tmp) space is unique to each compute node.

CPU Time Allocations

Overview

All University of Arizona Principal Investigators (PIs; aka Faculty) that register for access to the UA High Performance Computing (HPC) receive these free allocations on the HPC machines which is shared among all members of their team. Currently all PIs receive:

Best practices

1. Use your standard allocation first! The standard allocation is guaranteed time on the HPC. It refreshes monthly and does not accrue (if a month's allocation isn't used it is lost).
2. Use the windfall queue when your standard allocation is exhausted. Windfall provides unlimited CPU-hours, but jobs in this queue can be stopped and restarted (pre-empted) by standard jobs.
3. If your group consistently needs more time than the free allocations, consider the HPC buy-in program.
4. Last resort for tight deadlines: PIs can request a special project allocation once per year (https://portal.hpc.arizona.edu/portal/; under the Support tab). Requesting a special project will provide qualified hours which are effectively the same as standard hours.
5. For several reasons we do not offer checkpointing.  It may be desirable to have this capability in your code.

How Allocations are Charged

The number of CPU hours a job consumes is determined by the number of CPUs it is allocated multiplied by its requested walltime. When a job is submitted, the CPU hours it requires are automatically deducted from the account. If the job ends early, the unused hours are automatically refunded.

For example, a job requesting 50 CPUs for 10 hours will be charged 500 CPU hours. When the job is submitted, all 500 CPU hours are deducted from the user's account, however, if the job only runs for 5 hours and then completes, the unused 250 hours would be refunded.

This accounting is the same regardless of which type of node you request. Standard, GPU, and high memory nodes are all charged using the same model and use the same allocation pool. If you find you are being charged for more CPUs that you are specifying in your submission script, it may be an issue with your job's memory request.

Allocations are refreshed on the first day of each month. Unused hours from the previous month do not roll over.

How to Use Your Allocation

To use your allocation, you will include your account and partition information as a SLURM directive in your batch script. The formatting for this can be found in our Running Jobs with SLURM documentation.

How to Find Your Remaining Allocation

To view your remaining allocation, use the command va in a terminal. For example:

(elgato) [user@gpu5 ~]$ va
Windfall: Unlimited

PI: parent_974 Total time: 7000:00:00
	Total used*: 1306:39:00
	Total encumbered: 92:49:00
	Total remaining: 5600:32:00
	Group: group1 Time used: 862:08:00 Time encumbered: 92:49:00
	Group: group2 Time used: 0:00:00 Time encumbered: 0:00:00

*Usage includes all subgroups, some of which may not be displayed here

SLURM Batch Queues

The batch queues, also known as partitions, on the different systems are the following:

Job Limits

To check group, user, and job limitations on resource usage, use the command job-limits $YOUR_GROUP in the terminal.

Special Allocations

Sometimes you may need an extra allocation for a conference deadline or paper submission.  Or something else.  We can offer a temporary allocation according to the guidelines here: Special Projects