Communications of the ACM
New Computer Cluster Gets Its Grunt From Games
"GPUs speed up data processing by allowing a computer to massively multi-task through parallel processing," says CSIRO leader John Taylor.
Credit: eResearch Australasia
A new graphics processing units (GPUs) computer cluster will process CSIRO research data thousands of times faster and more efficiently than a desktop PC as a complementary system to the supercomputing resources available to CSIRO researchers. The new GPU cluster is the first of its kind in Australia, and is about the size of six large refrigerators, containing 61,440 compute cores. CSIRO's John Taylor says the new cluster combines central processing units (CPUs) and GPUs to make the system more efficient.
"GPUs speed up data processing by allowing a computer to massively multitask through parallel processing," Taylor says. GPUs are generally less expensive, per unit of processing power, and more energy efficient than a CPU-based supercomputer, and GPUs can be 30 to 70 times faster than CPUs. "This cluster will be part of our family of high-end computers in CSIRO and important to our e-Research Strategy," says CSIRO's Alex Zelinsky. "It will enable CSIRO to, in a cost-effective way, be globally competitive in addressing computational challenges for 'big science.' "
Taylor notes that running experiments on the GPU cluster requires a new approach to coding, specifically dividing up each task to make the best use of the extra processors.
From CSIRO (Australia)
View Full Article
Abstracts Copyright © 2009 Information Inc., Bethesda, Maryland, USA 
No entries found