High Performance Computing (HPC) is a vital tool for researchers who need to perform large calculations, or many smaller calculations simultaneously.
HPC systems (often called supercomputers) are typically a cluster of computers working together. For example, instead of taking a year to calculate an answer on your desktop PC, it may provide the answer in just a few days or even hours.
Spartan is the University's local HPC service, which is freely available to all researchers. It combines high performance bare-metal compute with flexible cloud infrastructure to suit a wide range of use-cases. Specialist high memory and GPGPU resources are also available.
We have a dedicated helpdesk to support Spartan. You can use this to request assistance, new software or a password reset. Just email email@example.com
If you need more resources, you can also access national resources through the National Computational Merit Allocation Scheme (NCMAS). This includes resources at NCI, Pawsey and MASSIVE.
Curious about how HPC can be used to accelerate your research? Here are a few examples from our existing users at the University.
Analyze Brain Function
Dr. Scott Kolbe is a NHMRC Peter Doherty Fellow in the Department of Anatomy and Neuroscience, studying brain function and structure using neuroimaging. Dr. Kolbe's group conducts much of their interactive analysis in the Nectar cloud using a customised remote desktop environment. This allows easy access to data stores, and simplifies on-boarding for new students and collaborators. ResPlat's High Performance Computing (HPC) system, Spartan, is used for batch processing of large MRI datasets that would otherwise impractical on a single computer.
Designing Better Gas Turbines
Prof. Richard Sandberg is Chair of Computational Mechanics in the Department of Mechanical Engineering, using computational fluid dynamics to simulate turbulence in gas turbines used for propulsion (aircraft) and power generation. Prof. Sandberg's group also utilises our HPC system Spartan to undertake the intensive computations necessary to accurately produce data that cannot be obtained in a laboratory in that detail. This includes use of our general-purpose GPU partition, which allows many computations to be completed in parallel, accessing details and problem domains that have previously been inaccessible.