Share
Blog
Cite
Print
Email
BookmarkScienceDaily (Feb. 23, 1999) — Dinesh Bhatia, associate professor of electrical and computer engineering at the University of Cincinnati, calls it "hand-crafted hardware." It's a technique dubbed RACE that Bhatia and his graduate students use to make desktop computers run more like supercomputers, and it will be explained during a presentation at the Association for Computing Machinery Symposium on Field Programmable Gate Arrays in Monterey, California, Feb. 21-23.
See also:
RACE stands for reconfigurable and adaptive computing environment. Bhatia's approach goes beyond improving software code or microchip design. It redesigns the hardware components of a computer so they can adapt as the demands of the user changes.
"The hardware itself can be made more intelligent," said Bhatia. "It can self-evolve in the best executing form."
Although that requires manipulating the hardware components and their connections by hand in the research lab, the ultimate goal is to "teach" the computer to reconfigure itself to operate in the most efficient manner possible.
"You could do it by hand, but we are trying to automate that process as much as possible," explained Bhatia. "Our eventual goal is that the users should never know if their application is executing in the traditional mode or if it is executing on our special-purpose hardware. It will just execute."
Bhatia and his graduate students began working with applications such as image-processing that are known to run slowly on traditional desktop computers. They were able to get those same applications to run five to eight times faster using RACE technology which was developed in Bhatia's Design Automation Laboratory.
Bhatia foresees a time when a desktop computer could approach supercomputer speed if properly configured for a particular application. That's one reason he's already working with molecular biology researchers at the UC Medical Center who need souped-up computing power to handle DNA sequencing, gene mapping, and other genetics problems.
"We always try to go for the real-life examples with real data, and I think that's what put us so far ahead of other people. We're able to concentrate on real applications," said Bhatia.
Additional information: http://www.ece.uc.edu/~dinesh/
Adapted from materials provided by University Of Cincinnati.
