Usage on HECC compute systems is charged using a standard billing unit (SBU). SBUs are charged to each PBS job running on the compute nodes according to the following formula:

SBUs charged = wall-clock hours used * number of MAUs * SBU rate

where MAU represents the minimum allocatable unit of resources available through PBS.

On Athena, a MAU is a node with 256 cores. On Aitken, a MAU is a node with 40 cores (Cascade Lake) or 128 cores (Rome or Milan). On Electra, a MAU is a node with 28 cores (Broadwell) or 40 cores (Skylake). On Endeavour, a MAU is a socket with 28 cores (Cascade Lake).

For Cabeus GPU nodes, a MAU is one full node.

Charges are based on the number of MAUs allocated to a job, not how many cores are actually used during runtime. Once a user is allocated the resources, that user has exclusive access to those resources until the user's job completes or exceeds its requested wall-clock time.

Note: Usage of the front-end systems and the Lou data analysis nodes (LDANs) is not charged.

The following table shows the SBU rates for each processor type:

On Endeavour, charges apply both to jobs that run successfully and those that are interrupted. Interrupted job charges are calculated by subtracting 1 hour from the elapsed job time, then multiplying the total by the number of MAUs used. Users are encouraged to have their applications checkpoint roughly every hour.

For example, if you submit a 24-hour job on Endeavour that requires 16 MAUs (448 cores) and the job stops unexpectedly after running 12 hours, your allocation will be charged. The accounting system will subtract 1 hour from the 12 hours the job ran, and compute the SBUs as follows: 11 hours x 16 MAUs x 1.31 = 230.56 SBUs.