Supercomputer Centers Challenge Industry to Pledge to the Future
Larry Smarr, the director of the University of Illinois’ National Center for Supercomputing Applications, suggested that corporate America follow George Washington’s lead.
“When Washington got to the Delaware, he couldn’t wait three months for a break in the weather,” said Smarr, who has been prodding corporations to look beyond the next fiscal quarter and join in a “historic quantum jump” into the supercomputing world of the 1990s.
“It’s a time for leadership, a time for vision,” Smarr said during a recent interview. “We’re going to find out if corporate America has the stuff to meet the challenge.”
Smarr has issued a lofty challenge. He wants eight diverse industrial corporations--each with $1-million checks in hand--to join the National Science Foundation-funded center at Champaign-Urbana.
And, he wants them to make a commitment within six months: “That’s the window of opportunity for corporate America, and I hope for the good of all that we can work it out.”
Also out beating the corporate bushes is Robert Randall, manager of resource development for the San Diego Supercomputer Center, which hopes to bolster its NSF-funded research by adding $40 million during the next five years.
“There are nights when I wake up in a cold sweat, wondering where I’m going to get $40 million,” said Randall, who said he is “confident” that the “tall order” can be accomplished.
The added money would allow the San Diego center to replace its initial supercomputer. “As bigger and better supercomputers become available, we’d make them available to the research and academic communities,” Randall said.
That is the same plan that Smarr has developed at the Illinois center.
However, the centers are not merely turning to corporate coffers for a handout.
Corporations that contribute $1 million annually to the Illinois center will be invited to send their “senior researchers, the guys who might have 100 people under them and millions of dollars of computer equipment,” to the Illinois center, Smarr said. Once on campus, the corporate scientists will work shoulder-to-shoulder with university researchers who are fine-tuning one of the world’s most powerful computers.
“There is a revolution about to take place in the Fortune 1,000,” said William Schrader, executive director for the Center for Theory and Simulation in Science and Engineering at Cornell University. Schrader, a biologist by training, has been spending half of his day courting companies “like Corning, Kodak, General Electric, General Motors, Dupont and Exxon . . . the kind of people who are knowledgeable about what the computing environment of the 1990s will require.”
Those companies “all will need large-scale supercomputing to stay competitive worldwide, and while we won’t teach them how to do it, we will develop a 1990s environment ahead of time,” said Schrader, who wants 30 companies to each contribute a minimum of $100,000 annually.
Although the San Diego center has identified a minimum level of corporate involvement, it has not made that level public, said Director Sidney Karin, who added that “we would like to remain flexible” enough to encourage various kinds of research arrangements with corporations.
That arrangement would help the 19-member consortium that is helping finance the center “get closer together” with industry, said Karin, who added that “there are things that each group can learn from each other.”
Smarr is confident that industry soon will recognize that the four supercomputing centers (at Cornell University; San Diego; Princeton, N.J., and Champaign-Urbana) provide an alternative to “costly (research) duplication that is an enormous drag on the nation’s productivity,” Smarr said.
A physicist by training, Smarr explained that the same supercomputing research that gives him a better explanation of “how gas flows around a black hole” would also help answer the down-to-earth questions of industrial scientists who want to know “how gas flows generate thunderstorms . . . how gas flows inside a turbine, or how gas flows around a car. The laws of gas dynamics are constant and universal.”
Electronics companies will be able to harness the supercomputer’s lightning-like speed, and construct complex computer models that gobble up expensive computer time on a typical corporate mainframe.
The supercomputer will reach beyond industrial applications, said Alfred Brenner, president of the Princeton-based Consortium for Scientific Computing. A New York University professor is using a supercomputer to track the flow of blood through artificial hearts, a life-saving exercise that could eliminate the need to “install an inefficient heart valve that could kill the (recipient),” Brenner said. “This way, you can study a hundred or a thousand heart valves to see which flows the blood best.”
Other medical researchers are using the supercomputer’s power to model the formation of cholesterol in a human heart, and how the human lung works, “tasks that bring (today’s mainframe) computers to their knees,” said Suresh Shukla, national marketing manager of engineering and scientific services for Boeing Computer Services, a Seattle-based company that sells supercomputer time.
Supercomputers will help condense the time that pharmaceutical companies must spend to conduct complex molecular modeling that uncovers new drugs, Shukla said. Supercomputing will cut that research process from “two years on a typical mainframe to just a day or two,” Shukla said. “That means a shortening of research and development cycles, and as machines get cheaper and cheaper, more research (will get) done.”
As the cost of supercomputing time drops, Brenner said, smaller companies that have been priced out of supercomputing will start conducting research.
To that end, Illinois’ center will provide different levels of access to its supercomputer, he said, “to make sure that everyone in the country can afford to get into supercomputing.”
Not surprisingly, the nation’s computer manufacturers have made dramatic donations of equipment and services to the computer centers. Cray, the nation’s premier supercomputer manufacturer, has donated $5 million in equipment and service to the Illinois center and funded $5 million in research at the San Diego center.
That vendor interest is sparked by the realization that placing equipment in a “national showcase (such as Illinois) . . . is a clear investment in downstream profitability, “ Smarr said.
Shukla said the centers will alleviate a “tremendous shortage of talented and trained people” that the burgeoning world of supercomputing will demand. Boeing, which likely will gain customers as the supercomputing revolution spreads, is consequently “working in concert with the centers,” Shukla acknowledged.
However, links between corporate America and the four NSF-funded centers have been slow to appear. Although the four centers will be computing by early next year, none has announced a major corporate association.
Some corporate reticence stems from the fear that academicians will gain too much access to a corporation’s delicate proprietary information.
“It’s a challenge to know how to do this sort of thing because we’re breaking new ground,” said Schrader, adding that corporations want to understand “who will own the patents and copyrights” and what happens when “academic freedom is pitted against a company’s right to keep trade secrets that (are developed) in our environment.”
Consequently, the negotiations between industry and the centers are “essentially like making a commitment for any delicate involvement, be it marriage, or buying a house,” Brenner said. “All of the (parties) are interested but they have to continue to test the waters.
“Even high-tech industry is conservative (and) moves relatively slowly.”
The chances are good that “there will again be an increase in the level of interest by industry . . . once we’re up and running and talking about things we’re doing,” said Sidney Karin, director of the San Diego Supercomputer Center.
“We have a balanced budget through 1986 so our need for funds . . . really starts in 1987,” said Karin, who added: “I’m not disappointed in our results to date.”
“There’s a need for a real commitment,” said Doyle Knight, a Rutgers University professor who serves as secretary for the 19 universities that organized the Princeton-based Consortium for Scientific Computing. “Once you get involved, you have to stick with it, because it’s not just a new phone system.”
Smarr acknowledged that the centers are “going to butt heads” as they ask corporations for their support. However, he added, “There’s plenty to go around. And, besides, that’s the way you decide who dances with whom.”
