Science Live recently hosted an online chat with University of Tennessee professor Jack Dongarra and Lawrence Berkeley National Laboratory deputy director Horst Simon. Dongarra and Simon discussed the coming class of exascale systems and the research into developing supercomputers that are hundreds of times faster than today's state-of-the-art systems.
Dongarra says the benefits of exascale computing reach into every segment of technology and science, including energy research, life science, manufacturing, and entertainment. Simon says the major barrier to developing exascale systems is power consumption. "Extrapolation from today's technology to the exascale would lead to systems with 100 MW or more power requirements," he says. Breaking the exascale barrier will require a major effort on multiple fronts, in addition to drastic changes to hardware, software, algorithms, and applications, say Dongarra and Simon.
Another hurdle will be the financial cost it will take to develop exascale systems; Dongarra notes that building an exascale-class system by 2020 would cost about $200 million, not including the cost of the research that comes before that.
From HPC Wire
View Full Article
Abstracts Copyright © 2013 Information Inc., Bethesda, Maryland, USA
No entries found