When should I not use HPC?#

Low volume workloads#

Learning to use the HPC clusters will take both time, training and effort. If you have a low volume of work then investing time in learning to use Linux, shell scripting and other skills for using HPC clusters may be better spent elsewhere.

Low volume, low memory serial workloads#

Using a HPC cluster is not a magic bullet that will make any workload run faster. Any workflows that only run on a single core (serial processing) and do not need large amounts of memory are likely to run slower on the HPC than on most modern desktops and laptop computers.

If you run a low volume of serial jobs you will likely find your own computer would have completed these quicker.

For training purposes#

University HPC clusters are used to facilitate large computational workloads and are not usually used as a training aid / facility. Exceptions may be made for HPC specific training with prior engagement with the HPC staff.

HPC staff can help you optimise your workflows and software for use on the HPC clusters, but they cannot teach you how to use your program in great detail nor train you on the basic usage of a program.

For non-legitimate or non-research purposes#

University HPC clusters are provided to facilitate legitimate research workloads. Inappropriate usage of cluster resources, e.g. mining crypto-currency, hosting web services, abusing file storage for personal files, accessing files or software to which a user is not entitled or other non-legitimate usage will likely result in an investigation and action.

Account sharing is also not permitted and any users/parties caught sharing accounts will also likely result in an investigation and action.