Cornell University's research center for advanced computing said on Monday it will get $60-million (U.S.) to develop supercomputers using technology from Dell Computer Corp., Intel Corp. and Microsoft Corp.
Cornell Theory Center said the new agreement to get resources from the companies will allow it to double the size of its supercomputing "cluster" system, which already runs on servers, chips and software made by the three, and develop more affordable supercomputing systems.
High performance computing is no longer limited to huge computers built by companies such as Cray Inc. and NEC Corp., but now includes systems built with off-the-shelf components arranged in "clusters" that share the large computations by dividing up the workload.
Thomas Coleman, director of the Cornell Theory Center in Ithaca, N.Y., said the agreement was a "strong statement in support of the potential for supercomputing using industry components."
"There are several benefits to using industry standard technology," he told a telephone conference, adding that cost reductions by factors of five to 10 were possible through advanced clustered computing.
Dell, Intel and Microsoft - corporate empires built on the personal computer - are seeking new sources of growth as the PC industry matures and growth slows.
"We want to deliver Windows systems at the extremes and at all points in between," said Cliff Reeves, Microsoft's vice-president for Windows Server Product Management.
Cornell's clustered computing systems run on Microsoft's Windows operating systems geared for corporate users and systems.
Earlier this year, Cray agreed with Dell to resell the PC maker's computers to its clients. Dell would supply the hardware, and Cray would provide supercomputing know-how, the companies said.
Intel has said that its Xeon and Itanium chips would work just as well for supercomputing tasks as they would for high-end corporate servers.
Cornell's Mr. Coleman said that its high performance computing system would be used for university researchers and will also help speed deployment of clustered computing to businesses.