eResearch South Australia selects SGI to supply HPC system

eResearch technologies in South Australia has selected SGI to provide infrastructure for its new high-performance computing (HPC) system with the Tizard machine.

Tizard will be the state’s most powerful public research HPC system and it will be used to support researchers in many different fields including climate change, evolutionary biology, high-energy physics, applied mathematics, chemistry, medicine, genomics and agriculture. The SGI HPC system is a mix of different computing systems, all optimised to complete specific tasks. The combination of computing systems has been purchased specifically to support the computational demands of eResearch technologies.

“The Tizard machine represents an exciting new era in South Australia’s eResearch capability,” said Dr Paul Coddington, deputy director of eResearch SA. “The Tizard will be of immense value to researchers at the University of Adelaide, University of South Australia, Flinders University and State Government research facilities and groups, providing a much wider range of computing modes and overall power than eRSA’s existing supercomputer.”

The SGI HPC Cluster system provides 2304 cores delivering 24 Teraflops. The system has 48 compute nodes each with 128 GB of memory. There are an additional 17 nodes utilising 68 GPUs including NVIDIA M2090 Tesla cards and two large memory nodes with 512 and 1024 GB of memory.

The new research infrastructure was purchased with funds from a Linkage Infrastructure, Equipment and Facilities scheme grant. The chief investigator on the grant, Professor Derek Leinweber, Professor of Physics and Head of the School of Chemistry & Physics, will be one of the first users of the HPC machine. His research into quantum chromodynamics (QCD), the theory of the strong interaction, requires the use of the most powerful supercomputers that we can build.

“The installation of the Tizard machine will transform the way we perform these computations as we harness the power of dedicated graphics hardware, or GPUs,” said Professor Leinweber. Leading this transformation is Senior Research Associate Dr Waseem Kamleh. He said, “The massively parallel architecture of GPUs is not just suited for playing the latest video games, but can also be used to run our calculations up to 100 times faster when compared to running on standard CPUs alone. Exploiting this speed we will be able to acquire the massive statistics needed to explore aspects of QCD that are otherwise unknown. This is crucial to discovering the dynamics of the basic building blocks of the universe.”