The Parallel Universe Grows

If two heads are better than one, why not a dozen? Or 50, or 100? That's the latest thinking in the commercial data processing market. For 40 years, mainframers such as IBM and Unisys have struggled to build the fastest single computer possible. But now they're bumping into fundamental physical limits, making each new speedup terribly expensive. So, the commercial mainframe industry, like supercomputer makers before it, is turning to parallel processing, ganging together scores of cheap microprocessors to attack big computing problems en masse.

To continue reading this article you must be a Bloomberg Professional Service Subscriber.