IBM Builds Second Fastest Computer

Big Blue has once again set a record for building one of the world's fastest super computers. IBM's Watson Blue Gene system will be second only to its own Blue Gene/L supercomputer installed at Lawrence Livermore National Labs (LLNL) that became the world's fastest system in March.

The Watson Blue Gene (BGW) system will achieve a processing speed of 91.29 trillion floating point operations per second (teraflops), nearly matching BlueGene/L, which has reached 135.3 teraflops.

BGW is installed at IBM's Thomas J. Watson Research Center in Yorktown Heights, NY for the purpose of exploring life sciences, hydrodynamics, materials sciences, quantum chemistry, molecular dynamics and fluid dynamics and even to business applications.

In comparison to today's technology, the word's first supercomputer, CRAY 1, reached a top processing speed of 133 megaflops. One megaflop is equivalent to one million instructions per second; whereas, a teraflop is equivalent to a trillion operations per second.

BGW is the size of 20 refrigerators but is less than half the size of comparable systems, IBM said. The first results obtained from the system were published in the American Chemical Society in April 2005.

"IBM researchers will use BGW to accelerate discovery in a variety of disciplines," said Tilak Agerwala, vice president, Systems at IBM Research. "Researchers, scientists, engineers and inventors can now ask more questions, test more theories, try more designs, and simulate more conditions than was ever possible before."

Aside from its own internal research projects, Big Blue will share BGW's cycles with the of Energy's Innovative and Novel Computational Impact on Theory and Experiment (INCITE). IBM's Center for Business Optimization will also use the machine to supplement the talents of its math scientists and assist in the development of algorithms that solve business related problems.

IBM has leveraged the size and power advantages of Blue Gene to establish its Capacity on Demand program where it leases out the number crunching capabilities of its super computing centers to commercial disciplines.