ACM TechNews

Berkeley Lab Researchers Propose a New Breed of Supercomputers for Improving Global Climate Predictions

Lawrence Berkeley National Laboratory (05/05/08) Wang, Ucilia

Researchers at the U.S. Department of Energy's Lawrence Berkeley National Laboratory have proposed a new way to improve global climate change predictions through the use of a supercomputer with low-power embedded microprocessors. Michael Wehner and Lenny Oliker of the Berkeley Lab's Computational Research Division, and John Shalf of the National Energy Research Scientific Computing Center, say a new class of supercomputers that uses embedded processor technology could create a cost-effective machine for modeling climate conditions to better understand climate change. In their paper, "Towards Ultra-High Resolution Models of Climate and Weather," the researchers conclude that a supercomputer using about 20 million embedded microprocessors would cost $75 million to construct, but would consume less than 4 megawatts of power and have a peak performance of 200 petaflops. Wehner, Oliker, and Shalf, along with researchers from University of California, Berkeley and Colorado State University, are now working to build a prototype system capable of running a new global atmospheric model developed at Colorado State. "What we have demonstrated is that in the exascale computing regime, it makes more sense to target machine design for specific applications," Wehner says. "It will be impractical from a cost and power perspective to build general-purpose machines like today's supercomputers."

http://www.lbl.gov/Science-Articles/Archive/NE-climate-predictions.html

---

© Copyright 2008 Information, Inc. This service may be reproduced for internal distribution.