Menu
Swinburne Uni studies universe's secrets with new GPU-based supercomputer

Swinburne Uni studies universe's secrets with new GPU-based supercomputer

gSTAR supercomputer's predecessor now busy rendering 3D movie

Swinburne University has launched a new supercomputer dubbed gSTAR — "GPU Supercomputer for Theoretical Astrophysics". The new 120-teraflop computer will be used by students and staff at the university's Centre for Astrophysics and Supercomputing for processing the vast amount of data produced by telescopes and for modelling.

The computer, which cost $3 million with hardware supplied by SGI, supersedes the Centre's previous computer (known as the 'Green Machine'). It uses 636 Intel CPUs and 121 NVIDIA GPUs.

gSTAR will help process data gathered by the SkyMapper telescope in NSW and CSIRO's Australian Square Kilometre Array Pathfinder (ASKAP) telescope in Murchison, Western Australia. Data will also come from the Compact Array at Narrabri.

Director of the Swinburne Centre for Astrophysics and Supercomputing, Professor Warrick Couch, said that gSTAR will also process data from the Parkes telescope ('The Dish'), which will aid the search for pulsars (fast rotating neutron stars). "You have this enormous amount of data you've got to sift through; it's like a needle in a haystack exercise of sifting through all the radio data getting rid of all the noise and interference to find this regular pulse coming from these neutron stars," Couch said.

What went wrong with the Hubble Space Telescope (and what managers can learn from it)

Yabi: Bringing drag-and-drop to supercomputers

OpenStack: Building a more open Cloud

The current generation of radiotelescopes produce enormous volumes of data in the region of terabytes per day, Couch said. "That data needs to be processed and so gSTAR will be good for that task — taking data from telescopes and processing it quickly."

In addition to processing data, gSTAR will be employed for modelling. "We have a number of researchers here in our centre and elsewhere in Australia who model a whole range of things from individual stars to systems of stars to individual galaxies and groups and clusters of galaxies, to whole universes," Couch said. "gSTAR will be very useful and powerful for that as well."

Although the focus is on astronomy, Couch said scientists in other areas will also be able to utilise gSTAR.

The Green Machine that gSTAR replaces was installed at the university in 2007. It was based on 1950 nodes each running two quad-core processors and was capable of 12 teraflops. Like gSTAR, the Green Machine was used for data processing and modelling, though Couch said the balance was "more tilted towards the modelling side". "It's only really now that telescopes like SkyMapper and ASKAP are coming online and will start to produce data so there's going to be a greater demand in that area," he said

Although gSTAR has taken the place of the Green Machine, there is still some life left in the older supercomputer, Couch revealed. "We have a group here in our centre that makes 3D astronomy movies which we sell and provide to clients around the world who have 3D movie theatres," Couch said.

"It's essentially an outreach thing: We're trying to educate people and turn them on to astronomy and we just signed a contract with a film company here in Melbourne to produce an IMAX [astronomy] movie in 3D. For the next six or seven months the Green Machine is largely going to be used to do the animation and rendering for that movie."

Rohan Pearce is the editor of Techworld Australia. Contact him at rohan_pearce at idg.com.au.

Follow Rohan on Twitter: @rohan_p

Join the CIO Australia group on LinkedIn. The group is open to CIOs, IT Directors, COOs, CTOs and senior IT managers.

Join the newsletter!

Or

Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

More about CSIROCSIROIntelPathfinder HoldingsSGI AustraliaSwinburne University

Show Comments
[]