Read the Original Article at http://www.informationweek.com/news/showArticle.jhtml?articleID=6500850
A project that used the power of idle home PCs to search for anthrax cures has reached its research goal after only 24 days, storming through an analysis of 3.5 billion drugs to identify more than 300,000 candidates that may work as treatments for the deadly toxin. Project sponsors say the effort may help prove the efficacy of grid computing as a tool for enterprise computing.
The quick work was made possible by the combined computing power of more than a million PCs tackling the problem in small pieces. Each PC performed pattern matching on a potential drug to see if it would bond with the anthrax toxin, then sent the results back over the Internet. "It's just a huge, very repetitive process that calls for an enormous amount of computing power," says David Wilson, VP of marketing for distributed computing company United Devices Inc., which provided the software for the project. "People in the life sciences who deal with this kind of research would be blown away by the speed at which we got it done."
United Devices is looking to license its software to pharmaceutical companies to run on their own PCs, behind company firewalls. "Some of these companies have 25,000 or more PCs," Wilson says. "That's an enormous amount of computing power that dwarfs the power they've got in their data centers." And the software could be used by smaller companies to farm work out to PCs outside of their firms, he says. "If you're a small pharmaceutical or biotech player trying to be competitive, you can't go out and buy a sufficient amount of computing power, so one option is to go out on the Internet."
Wilson says United Devices has focused on the life sciences business so far, but sees plenty of other places the technology could be put to work, such as financial-services companies doing risk analysis and digital content creators performing complicated 3-D rendering.
Gartner analyst Rob Batchelder agrees. "If you have applications that can be sliced up into little pieces, it's the way to go," he says. "It's really well-optimized for problems that are very data light but computation intensive."