Call it Blue Gene Lite. A scaled-down version of IBM’s well-known Blue Gene effort was finally acknowledged by the company, and despite rumors of “Baby Blue Gene” and “Blue Light,” it’s officially called Blue Gene/L.
In a partnership with the Lawrence Livermore National Laboratory, IBM is developing the new supercomputer based on the still-in-progress Blue Gene technology aimed at computational problems that can be divided to run on tens of thousands of processors. The machine will be 15 times faster, 15 times more power-efficient, and take up 50 times less space per computation than current supercomputers, IBM says. The company adds that Baby Blue should operate at about 200 teraflops, topping the total compute power of the world’s 500 fastest supercomputers, with a target completion date of 2005.
Blue Gene believers, never fear. IBM says it will continue work on the petaflop-scale machine. “[Blue Gene/L] represents a major step along the path to getting to our eventual goal of building the even larger computer as part of the same project,” says Bob Germain, who heads up the science and application portion of the Blue Gene project. Still, Germain declines to pin down a date for the supercomputer, which was originally slated to be built by 2004. “We still intend to get to the petaflop, but it may be that we stretch out the schedule somewhat,” he says.
Blue Gene/L will be not only less specific than its predecessor (or successor, as the case may be), which is being designed for protein folding, but will also be more easily adaptable to commercial applications for which Blue Gene was not intended.
IBM is working on the hardware angle and will create the single-purpose protein-folding applications for the petaflop computer, while LLNL is in charge of broader applications it plans to run on Blue Gene/L.
— Bernadette Toner