Projects and Accounting for Fox

Projects

All jobs are run in a project, typically named ec<N> where N is a number. Use --account in job scripts to select which project the job should run in. (The queue system calls projects accounts.) Some projects are billed for their CPU hour usage on Fox.

To see which projects you have access to on a cluster, run

$ projects

Accounting

The term "CPU hour" above is an over-simplification. Jobs are accounted for both CPU and memory usage, as well as usage of GPUs.

Accounting is done in terms of billing units. Each job is assigned a number of billing units based on the requested CPUs, memory and GPUs. The usage that some projects are billed for is the number of billing units multiplied with the (actual) wall time of the job, i.e., the billing unit hours.

The number billing units of a job is calculated like this:

1. Each requested CPU is given a cost of 1.
2. The requested memory is given a cost based on a memory cost factor (see below).
3. Each requested GPU is given a cost based on a GPU cost factor (see below).
4. The number of billing units is the maximum of the CPU cost, memory cost and GPU cost.

The memory cost factor and GPU cost factor vary between the partitions on the cluster:

• The normal partition: memory factor is 0.255489 units per GiB. Thus the memory cost of a job asking for all memory on a node will be 128, i.e., the number of CPUs on the node.

• On the accel partition, the memory factor is 0.09542744 units per GiB, and the GPU factor is 24. This means that a job asking for all memory on a node, or all GPUs on a node, gets a cost of 96, the number of CPUs on the node.

^{CC Attribution: This page is maintained by the University of Oslo IT FFU-BT group. It has either been modified from, or is a derivative of, "Projects and accounting" by NRIS under CC-BY-4.0. Changes: Removed sections "List available quota", "Saga" and "Betzy".}