Engineering Schools Trail Chip Design Changes

Complexity in designing chips is relatively well understood, even if it’s not easy to solve the problems and actually create the chips. But engineering schools are only beginning to grasp the enormity of the change, and their curricula are running years behind what is happening in the industry.

Corporations have spent years tearing down silos, and technology has forced the same kinds of changes for engineers. But in academia, silos continue to flourish as they have for decades. Professors don’t, as a rule, make a name for themselves in multiple disciplines. In fact, it becomes increasingly difficult to distinguish one professor’s skill set from another’s if they are both trained in more than one area.

That creates a huge challenge for combining expertise the same way chip engineers now have to combine their skills to meet challenges at 65nm and 45nm process nodes.

“Universities are quick to respond on some things, but from an infrastructure side they’re slow,” said Jan Rabaey, Donald O. Pederson distinguished professor of electrical engineering and computer science at the University of California at Berkeley. “Research is more adaptive, and you see multidisciplinary groups emerging there in fields like neuroengineering. But that’s hard to reflect in an educational program. It’s still being taught from the roots up.”

Many schools are in the process of overhauling their engineering departments—MIT, San Jose State University, Tulane, and the University of California, among others. The new buzz phrase is “top-down learning,” and it’s making inroads in areas where multidisciplinary approaches are required.

“What some schools are doing is first looking at a global design, then looking at the components, and then addressing the challenges in each area,” said Rabaey. “The idea is you teach from a problem-driven perspective. Your first attempt is high-level, then you drill down to the components. Whether that is successful is still to be seen. A second option is project-driven teaching. You build something by bringing different groups together, so if you’re going to build a robot the class consists of people from different backgrounds. You have different modules and form teams for those modules, and then you use multiple professors to teach.”

Rabaey said the upside is that it could well change the mix of students who enroll in engineering schools. He said the current demographics for engineering schools are 80% male, 20% female. But he said if the focus of study is energy efficiency and green initiatives rather than engineering, the mix of students would change dramatically.

At least part of the challenge also is retooling the ranks of working engineers, who now must deal with multidisciplinary chip design while their degrees typically are in electrical engineering. Belle Wei, dean of the College of Engineering at San Jose State University, said one solution is to work directly with companies and create curricula that meets their needs.

“The university teaches fundamentals, but we have to tie it to end applications,” Wei said. “We have a computer engineering department because it’s necessary to learn more software. We also have an embedded systems program at a graduate level in response to employers’ requests. Our curriculum has to be responsive to them.”

She noted that a rising percentage of students at SJSU comes from the ranks of corporations. “We increased the number of corporate Master’s Degrees from 5 to 14 in the past four years. We’re teaching software engineering at IBM, optoelectronics at KLA-Tencor and embedded engineering at BAE. We’re actually out in those companies.”

Sometimes those approaches are cross-disciplinary, but Wei said that’s often a function of the companies themselves. In mechanical engineering programs, for example, there are often courses in electrical and computing engineering. And in electrical engineering, the common addition these days is computer science.

—Ed Sperling