IBM announced today that it will spend $29 million to launch an international collaboration to train supercomputer firepower on business problems, such as managing investment risk and scheduling airline routes. Named after Deep Blue, the chess-playing program that defeated Grand Master Gary Kasparov, the Deep Computing Institute (DCI)--a virtual entity to be run out of Yorktown Heights, New York--is now officially up and running.
For the last 2 decades, scientists have harnessed the power of parallel processing supercomputers--whose multiple processors perform thousands of sophisticated calculations simultaneously--to solve previously intractable numerical problems and create stunning images of everything from genes to galaxies. "Now we want to take the modeling that scientists are used to and extend it to business decision making," says William Pulleyblank, director of DCI.
Good business decisions, however, aren't as photogenic as spiral galaxies. Tasked with solving this visualization problem will be 120 experts in academia, government labs, and industry connected in an electronic network. They should get plenty of outside help: By making IBM's source code to Data Explorer--a supercomputer program that allows researchers to turn trillions of bits of data into understandable pictures--available freely on the Web, DCI hopes to encourage scientists and business technology experts to create and share innovative ways of displaying nonphysical data. DCI researchers will also develop software and provide consulting expertise to outside scientists.
"This project could really extend" current work in visualization and remote collaboration beyond pure science, says Robert Patterson of the National Center for Supercomputing Applications in Champaign, Illinois.