A researcher at the University of Maryland’s engineering school has created a prototype of a parallel processor that could represent a way around the speed barriers that conventional serial processors have recently hit. Uzi Vishkin claims his net of 64 processors cranking away at 75Mhz individually is capable of computing speeds 100 times faster than current desktops.
Parallel processing is a way of breaking down a task so that many processors can work on it simultaneously. Although it has been in use in supercomputers for years, writing programs to control such an unwieldy piece of hardware has never been practical, and the flexibility of such computers was severely limited. Vishkin says he has developed algorithms that make his license-plate-sized bank of processors easy to program for.
He compares the challenge of processing a chunk of data to cleaning a house. “Suppose you hire one person to clean your home, and it takes five hours, or 300 minutes, for the person to perform each task, one after the other,” Vishkin said in a statement. “That’s analogous to the current serial processing method. Now imagine that you have 100 cleaning people who can work on your home at the same time! That’s the parallel processing method.”
The challenge, of course, is getting 100 people to collaboratively work on your house without any of them performing the same work twice or getting in one another’s way. Vishkin claims his algorithms make this complicated division of work practical for the first time.
Vishkin is in the process of demonstrating his technology to leaders in government and industry, and has allowed local high school students access to his prototype to prove how easy it is to program for. In the future, he foresees 1,000 processors on the same chip the size of a fingernail. He has also announced a contest to name the new computer, with a $500 cash prize for the winner.