 |  |  |  |  |  |  |  |  |  |  |  |  |  |
Stanford: Silicon Valley incubatorby Bernard Cole Founded in the early 1890s, Stanford University has always existed in two worlds: one evokes a past that mirrors its bucolic and rather rural origins, while the other is well grounded in the present, erecting building blocks for the future. This split view of the world is reflected in the university's physical surroundings.Stanford's past and that of what is now called Silicon Valley can still be seen in some of the land the university received from railroad tycoon Leland Stanford. Off to the west, toward the Santa Cruz Mountains, a vista unfolds of pastures and farmland that are similar to what must have existed in the early part of the 20th century. But for a view that reflects both Stanford's immediate past and its future as the incubator from which high-tech companies grow and thrive, one has to go to the Serra Mall just west of the Main Quad and to the Oval in the center of the main campus. On the north side of the mall is the Gates Building where the Computer Science department is housed. If one walks south across the mall, the Center for Integrated Systems is to the west and the Electronics Research Laboratory building, where the department of electrical engineering is located, is to the east. Much of what defines the electronics and computer industry at the beginning of the 21st century has been born in these facilities: reduced instruction set computing, local-area networking and the beginnings of the Internet and World Wide Web, among others. From Stanford have come many of the engineers, computer scientists and executives who work at the tech com- panies that now dot land once dominated by farms, orange groves and wineries. William Hewlett and David Packard of the now-legendary Hewlett-Packard Co.; T.J. Rogers, founder, president and chief executive officer of Cypress Semiconductor Corp.; and Jim Clark, a Stanford EE professor who went on to found Silicon Graphics Inc., all attended Stanford. Cisco, Imagen, Intellicorp, PMC Sierra, Rambus, Teknowledge and Yahoo! are other companies that have benefited from the talents of Stanford alumni, students and faculty. Many of the projects that originated at Stanford have also served as the seeds from which major companies and technologies have evolved. Sun Microsystem's first product, a low-cost graphics terminal board and network computer, was based on a Stanford research prototype. Cisco's first network product grew from a Stanford prototype router. And when Vinton Cerf was a faculty member at Stanford, he designed many of the key protocols now used on the Internet. One man who has lived through much of Stanford's — and Silicon Valley's — recent academic and business history is John Hennessy, Stanford University's provost. His past lives include stints as a computer researcher; as one of the original developers of the Stanford MIPS processor-one of the three or four seminal RISC architectures now in wide use; as chairman of Stanford's computer science department; and as a founder of MIPS Technologies Inc. As provost and earlier as chairman of the computer science department, Hennessy's aim has been to support and promote Stanford's role as an informal research arm of surrounding corporations and as a practitioner of basic research that is fundamental to all key technology developments.
According to Hennessy, the ingredients for integrating the research activities with the commercial interests of the surrounding Silicon Valley are similar to those formulated by Frederick E. Terman, who, after doing radar work during WWII, returned to Stanford as dean of the department of electrical engineering. Impressed by the way industry, universities and the government had worked together during the war, he looked for a way to create that same environment. "His initial motivation was to build a West Coast industry for Stanford engineering graduates," Hennessy said. First, he had to convince the university to establish an industrial research park where companies could get seeded, and then he had to encourage the faculty to work with the industry and train engineers in what is now called entrepreneurship; that is, taking an idea and not only developing it into a viable technology, but establishing companies to commercialize it. The trick was how to aggressively focus the electrical engineering and computer science departments on leading-edge research that could be developed into new industries, without micromanaging it and killing whatever you're trying to cultivate. "The key thing is to set an agenda of strategic goals of what areas of research look interesting, have some fundamental problems to be solved and secrets to be found," Hennessy said. "The hope is that in finding the answers something useful will also be found." Then let the research itself dictate the direction rather than direct the research in directions you think most useful, Hennessy said. Once the strategic direction is cemented, finding funding sources that allow that kind of flexibility is a problem, he said. "Ironically, it was in the early days when nearly all of our research budget came from government sources, such as Darpa, that there was the least pressure to focus research in particular directions," Hennessy noted. When he was working with the MIPS architecture, Hennessy found that the Darpa funders pretty much kept their hands to themselves once they determined that his research would probably result in something that would be of benefit to them. Currently, with federal funding declining and uneven year to year, about one-third of the University's research fund- ing comes from the private sector and corporations. "Companies put pressure to micromanage, to direct the research," Hennessy said. "Directed research does indeed yield results in that you get the answers you were looking for, but that is just the point. However, the most interesting results, the research from which the most significant developments occur, is basic; the answers often end up being much different than either the funder or the researcher expected." As we move into the 21st century, Hennessy said there are several areas of research that deserve investigation and to which Stanford is devoting considerable effort. One focus will be to investigate techniques and mechanisms for squeezing more performance from existing RISC architectures as well as newer, more highly parallelized concepts that allow multiprocessing on a single chip. Stanford researchers are also looking at ways to break the bottleneck that architects face with higher clock rates, higher levels of integration and more complex pipelines: the almost asymptotic increase in design time and test and validation. But at the same time, an effort will be made to investigate the changes needed to reflect the new kinds of applications and computing environments, namely, an emphasis on application specificity that is diametrically opposite to the standardized architectural approaches of the past. In addition, new systems will require architectural adaptations that reflect an increased emphasis on availability and maintainability in the form of better reliability and better failure containment. "It is not just a matter of more MIPS any more," he said. "We have some ideas about what is needed and the solutions that may emerge. But it is likely that we will be surprised. That is the nature of academic research." The Century of the Engineer: Companies that made a difference
|
| ||||||||||||||||||
