14 DAY TRIAL // Although initially designed to render 3D computer-game scenes, graphics processing units have reached a stage where they are capable of high-speed computing tasks, even being able to take upon themselves part of the CPU's workload. To take advantage of this, the Australian Commonwealth Scientific and Research Organization (CSIRO) has put together a cluster that will be able to divide tasks between individual GPUs. The cluster (about as large as six refrigerators put together) is expected to compute at speeds thousands of times higher than those attainable by normal PCs.
"This cluster will be part of our family of high-end computers in CSIRO and important to our e-Research Strategy. It will enable CSIRO to, in a cost effective way, be globally competitive in addressing computational challenges for 'big science,'" CSIRO Information Sciences Group Executive Dr. Alex Zelinsky said.
The cluster will be used for a number of tasks, such as analyzing the placement of genetic fragments on the genomes, constructing virtual, 3D medical images from the Australian Synchrotron and simulating the interactions between oceanic nutrients and the plankton. The machinery will do this using 61,440 computer cores combined with powerful GPU units for efficiency.
"GPUs have been around a while, hidden in your computer game console but now we're seeing them in scientific computing. GPUs speed up data processing by allowing a computer to massively multi-task through parallel processing." Dr. Taylor, CSIRO Computational and Simulation Science leader, explained.
GPUs not only manage to yield processing speeds 30 to 70 times faster than CPUs, but they also consume much less power than central processing units do. Nevertheless, there is a certain complication brought on by this new type of computing conglomerate, namely the fact that it requires a different type of coding, a coding that most scientists are not trained in.
"It requires a new approach to coding," Dr. Taylor added. "You have to divide up each task in a way that makes best use of the extra processors. It's like having to give instructions to 10 bricklayers building a wall, instead of one. We will be training others in the scientific community in how to use our facility."
This new GPU cluster will help assist the work of all the other supercomputers belonging to CSIRO, such as the recently installed NCI facility at the Australian National University. The cluster was inaugurated earlier today at CSIRO Discovery, Clunies Ross St. Acton, Canberra.