South Australia’s fledgling biotechnology industry has received the first of two shots in the arm with the commissioning of Hydra, a $1.7 million system that is Australia’s third-largest supercomputer and the largest of its type.

Organised by the recently-created South Australian Partnership for Advanced Computing (SAPAC) — a high-performance computing (HPC) oversight group linking the University of South Australia, University of Adelaide, and Flinders University — the purchase of Hydra significantly boosts computing power available to researchers working in biotechnology, environmental studies, fluid dynamics, molecular modelling, computational physics and myriad other applications.

As if this server wasn’t a big enough coup, the state government will soon bring even more number-crunching power into the state after recently announcing a $3.1 million fund that will be used, in part, to fund the purchase of a second supercomputer to join Hydra. That fund will also support interconnection of the Hydra facility into the AARNet-led GrangeNet or CSIRO-directed CeNTIE grid computing networks.

The influx of computing power into South Australia is a major improvement the only other high-performance computing previously capabilities available to researchers were through Perseus and Orion, each a cluster of several hundred standard PCs coordinated by specialised software and running the Linux and Sun Microsystems Solaris operating systems. While they were effective for many applications, their usefulness was limited by technical factors such as the high latency of the network which connected them. SAPAC director Tony Williams concedes that the 258-processor Hydra system — which has been benchmarked at around 700 gigaFLOPs (700 billion floating point operations per second) and peaks at 1200 gigaFLOPS — would have been unthinkable in a state with just 1.2 million people without the combined efforts of SAPAC members, the SA government, and the Australian Research Council, which kicked in $500,000 towards the purchase.

“It would make no sense for each university in Adelaide to try to have its own,” he says. “(The state government) is a very strong believer in building up the information economy in the state, and one way of doing that is building world-class HPC facilities. It’s a recognition of what HPC can do to stimulate research, and the community of users is growing. I’m expecting the bioinformatics and biotechnology people to be major players.”

Officially known as the IBM 1350, Hydra does away with conventional proprietary, esoteric and wildly expensive supercomputers by aggregating 129 two-processor computing nodes, each containing a pair of 2.4GHz Intel Xeon processors and 2GB of memory. The system is based on the Linux operating system and a high-speed network, based on technology from Myrinet, provides low-latency links between nodes that lets scientific applications coordinate many processors on a single task.

The SAPAC system is the third such supercomputer in Australia, with the other two installed in recent months at the University of Melbourne and VPAC (Victorian Partnership for Advanced Computing). “The last six months have seen a real take-up in the HPC marketplace for this kind of machine,” says Andrew Brockfield, senior HPC specialist at IBM.

“We got more compute power for the dollar than we would have with any other architecture,” explains Williams, who would like to see HPC administrators adopt common job dispatch standards so massive computational processes might eventually be spread across systems at different sites.