Yale University researchers have simulated the power of a supercomputer by linking up smaller computers in a development that could help small organizations solve big problems, the school said.
Scientists at the Ivy League school joined with researchers from Sandia National Labs to develop computer language to link up computers giving them the power of multi-million dollar supercomputers."For the first time, scientists will have the computing power to keep detailed computer simulations of entire worlds in a box on their desks," David Gelerntere, an associate professor of computer science at Yale said last week.
Gelerntere created a computer language called "Linda" to link VAX computers and allow them to solve problems that can be broken up into separate steps, each of which is assigned to different computers which contribute to the final answer.
The language allows the smaller computers to perform parallel processing, much like supercomputers, and reach answers more quickly.
The development allowed a scientist at Sandia National Labs to create a complex simulation of a rocket exhaust plume using a network of 14 VAX computers at two locations more than 1,000 miles apart.
The simulation took 143 minutes to produce, and the same problem took the lab's Cray-1 supercomputer, valued at several million dollars, twice as long.
VAX computers, made by Digital Equipment Corp., of Maynard, Mass., are one step up from personal computers and many universities and companies have dozens of them.
"This is an idea that people have talked about for 10 or 15 years, linking existing computers to work on huge computational problems," Gelernter said.
Researchers cautioned supercomputers still do some jobs better than the linked-up smaller computers.
But they said the new developments will make computer systems more efficient by allowing computers to be used on large problems during times when they would have been idle.
Researchers have begun developing applications for the system, including a monitoring system for patients in cardiac intensive care units and faster searches through huge computer data banks.