Sign In

Communications of the ACM

ACM TechNews

Next Generation of Supercomputers Requires Radical Redesign

View as: Print Mobile App Share:
Jill Pipher

Computational and Experimental Research in Mathematics director Jill Pipher

Credit: Brown University

The next generation of exascale supercomputers could complete one billion billion calculations per second, which would be 1,000 times faster than today's most powerful supercomputers. However, just one exascale system would require the power equivalent to the maximum output of the Hoover Dam.

Researchers recently gathered to discuss the challenges of supercomputing energy efficiency during a workshop held by the Institute for Computational and Experimental Research in Mathematics (ICERM) at Brown University. "We've been increasing computing power by 1,000 fold every few years for a while now, but now we've reached the limits," says ICERM director Jill Pipher.

The U.S. Department of Energy wants to develop an exascale supercomputer that would use less than 20 megawatts of power by 2020, which would require a drastic change in computer architecture. Changing computer architecture would also require a rewrite of the software programs that run on conventional computers. "Now, if you're building these new machines, you're going to have to try writing programs in different ways," Pipher says.

One promising solution is using graphics processing units (GPUs) instead of central processing units (CPUs). GPUs use almost eight times less energy than a CPU per computer calculation.

From Mother Nature Network
View Full Article

Abstracts Copyright © 2012 Information Inc. External Link, Bethesda, Maryland, USA 


No entries found