Science and Medicine News

Better Ways to Test Chips   As computer chips grow larger and more complex and are driven at ever higher speeds, manufacturers face increasing difficulty in identifying defective chips. Now, Stanford’s Center for Reliable Computing and the specialty chip maker LSI Logic Corp. have embarked on a three-year, $400,000 effort to improve chip testing methods. In the past, manufacturers have approached the test problem by building bigger and faster electronic testers. But this strategy is becoming prohibitively expensive, says E. J. McCluskey, director of the reliable computing center and professor of electrical engineering and computer science. McCluskey and his colleagues have opted instead to develop improved testing methods, and will evaluate the effectiveness of more than two dozen test techniques. “We have designed and built an integrated circuit whose only point in life is to be tested. We will test it in lots of different ways,” McCluskey said. The study will evaluate methods that test all aspects of state-of-the-art integrated circuit performance including logic, memory and other mixed signal circuits.

Novel 3-D Display   Elizabeth Downing (left), a graduate student working with electrical engineering Professor Lambertus Hesselink, has developed a prototype video display that can produce 3-D images in a whole new way ­ by creating actual three-dimensional color images inside a solid cube of fluorescent glass. The technology, although still rudimentary, has a number of potential applications, including medical imaging and air traffic control. The technology is unique in that “it doesn’t create an image that appears to be three dimensional, it actually produces an image that is drawn in three dimensions,” Downing says. “As a result, there are few restrictions on the viewing angle and a number of people can view the images at the same time Also, the images are emissive ­ they glow ­ rather than reflective, so they can be seen easily in ordinary room light.” The technology also has limitations. The objects that it forms are transparent, not opaque, so additional processing would be required before it would be suitable for entertainment purposes. It also takes 500 times as much data to construct a three-dimensional object as it does to draw the same object in only two dimensions. The research is reported in the Aug. 30 issue of the journal Science. ST

Previous | Next