High performance computing

High performance computing (HPC) enables researchers from diverse disciplines to perform complex computing and data analysis tasks that standard desktop machines are unable to support.

High performance computing is useful for researchers who need large amounts of computing power to perform mathematical modelling, analyse complex statistics, or search vast databases. Using HPC resources, researchers can understand or predict the behaviour of complex systems in work as diverse as molecular-level drug design, global warming research, and physical simulations.

High performance computing is enabled by the power of multiple processors operating together. Single machines containing multiple processors (eg Aquila), or multiple machines operating in a cluster (eg Corvus) can be accessed individually or can be linked together in a grid to form a distributed computing resource.

By sharing the HPC infrastructure that individual institutions might otherwise be unable to afford, more researchers can gain access to advanced data processing facilities.

Supercomputers

'Supercomputers' have signficantly more processing power and are much faster than 'standard' computers. The definition of a supercomputer shifts over time as technology evolves and what is considered 'standard' changes.

Supercomputers are typically used in scientific computing for complex calculation-intensive tasks. Universities, the military, and scientific research laboratories use supercomputers to perform calculations, render complex formulas, and perform tasks that require a large amount of computer power.

Supercomputers can be tailored for specific uses—like code breaking or even playing chess (as in the case of Deep Blue)—and are often custom-made using elements from different computer manufacturers.

eResearch SA has 3 supercomputers that are freely available for use by the South Australian research community.

Grid computing

Grid computing is the coordinated sharing of distributed computing resources over the internet. In the grid, computing resources such as eResearch SA's 3 supercomputers are connected via a standard interface. This effectively creates a single unified computing resource, granting users and applications access to vast computing and data storage capabilities.

Nodes within the grid—the individual supercomputers or data storage facilities that form part of the network—are specialised, dedicated components. Often they are customised with specific applications to meet some local need, which researchers from anywhere the grid can then make use of.

Using grid portals or custom desktop tools, researchers can easily submit tasks to the grid gateway—the standard interface that connects all the disparate elements of the grid together. The gateway manages the tasks submitted by researchers from across the grid by placing jobs in a queue and allocating them to the most appropriate nodes.

eResearch SA has grid computing facilities that form part of the national grid. We can help researchers access the grid to make use of its distributed computing power and any specialised resources that may not be available locally in South Australia.