Spring 1999 Newsletter

The Future of VLSI Education

An Inteview with Professor Thomas Lee

Joanna Evans

In February, Professor Thomas Lee chaired an ISSCC panel discussion on the current state of VLSI education, entitled "They Don't Make Engineers Like They Used To...?" The CIS Newsletter sat down with him recently to hear his words of wisdom on the subject and discover what the ISSCC discussion panel had to say about it.

JE: Tell me a little about the panel, its members and its purpose.

TL: The purpose of the panel meeting was to provide a forum for faculty, students and industry members to voice their concerns -- a sort of public ranting and whining for engineers and engineering education professionals. The panel was nice mix of folks from academia and industry. We had Bob Pease, a well-known iconoclast from National Semiconductor who questions constantly, in print and in person, the use of Computer Aided Design (CAD) tools. We knew he'd be a good person to have on the panel and he certainly stirred things up with his controversial viewpoints. We also had Hugo De Man, a professor from Katholieke University in Belgium, Professor Behzad Razavi from UCLA Ñ a graduate of Stanford Ñ and Professor Charles Sodini from MIT. Eric Swanson, from Crystal Semiconductor (now part of Cirrus Logic in Texas), was also a very entertaining, opinionated participant. Nicky Lu, from ETRON Technology, rounded out the industrial perspective.

JE: What was the most constructive result of the discussion?

TL: One of the most interesting points brought up, in my opinion, was Professor De Man's observation that the scope of engineering projects is much larger today than it was 10 or 20 years ago. Where we used to have one engineer -- or two or three -- completing a project, we now have projects with a hundred million transistor circuits. So it's impossible for one single person to take care of the whole thing. Engineering has become much more of a group activity, and engineers really haven't been trained in teamwork. That's not something we've focused on, certainly not at Stanford. It happens incidentally, but there isn't this overarching philosophy that we must instill teamwork skills into the students. And Hugo argued that we ought to spend a little more time thinking about this issue. He's noticed in the companies he's worked for that engineers tend to be these very solitary creatures and that the projects suffer because of it. I found that very thought provoking.

JE: Was the panel able to arrive at some sort of consensus on any of the issues facing VLSI educators and students?

TL: I'd say the discussion went in several different phases. In the first phase, we expressed concerns over the current status of engineering education and engineering just as a profession. Then the question was, "Given all these concerns, what should we do about it?" And there was much more uniformity here than in the first phase of discussion. The prevailing opinion was that actually, we're doing a pretty decent job in the area of engineering education, about as well as can be expected given how fast things change in our business. Universities are balancing a certain fundamental skill set with giving students enough working knowledge and experience that they aren't totally useless the day they graduate.

JE: Bob Pease mentioned in his ISSCC statement that "We're doing a lousy job of training engineers," while Eric Swanson felt that "the education of integrated circuit engineers has historically been excellent." Where do you stand on this?

TL: I think universities, by and large, have done a pretty impressive job of equilibrating what I consider to be an impossible set of constraints. On the one hand we have a fixed four-year undergraduate window in which to educate students enough to be useful engineers. Yet the amount of knowledge they're expected to carry with them upon graduation is growing at an exponential pace. So how can you possibly reconcile a fixed four-year window with an ever-increasing amount of material? It's clearly not possible to do this. You have to either evolve a race of students that just becomes genetically more brilliant with each generation, or say, "Look, the best we can do is give them this core set of abilities, and expect that they'll teach themselves to a certain extent." Whatever we teach them today, that's not fundamental, may not be applicable 10 years from now, so if they're going to remain relevant as engineers, they're going to have to become self-taught. Realistically, that's the only way to keep them in the business.

JE: Professor Razavi suggested that this becomes somewhat of a challenge in the "fast-paced, deadline-ridden" environment of industry. Do you think corporations should take a stronger role in encouraging continuing education?

TL: Yes, and it's in industry's best interest to do so, though there hasn't been much of a formal program initiated industry-wide yet. I think one of the most important things I took away from this panel discussion was that industry is probably going to have to formalize continuing education as it becomes increasingly critical. And I think a conscious appreciation of that fact is going to become more widespread as the shortage of adequately trained engineers grows by the day.

JE: In his statement, Nicky Lu brought up the need to expose students to a broader base of disciplines earlier on in their education. Did the panel share his concern?

TL: The panel was definitely split on this issue. Professor Razavi felt that given the choice between broad and shallow or narrow and deep, depth was preferable to breadth. Of course what everyone wants is breadth and depth, but since that's not possible, Professor Razavi argued, quite convincingly in my opinion, that a shallow knowledge is not really of much use to anyone.

"Engineering has become much more of a group activity, and engineers really haven't been trained in teamwork"

---

JE: How do students feel about VLSI education and circuit simulators?

TL: The students who spoke at the session, by and large, were pretty satisfied with their education. In fact they thought some of the educators on the panel were being harsher on themselves than the students would be, so that was nice to hear.

JE: Are other disciplines facing similar issues with simulators or theoretical application rather than real "hands-on" learning? And if so, is their approach applicable to engineering?

TL: I have thought about that and I really can't identify another profession that's in quite the same fix. Simply because engineering is a discipline that lies at the intersection of theory and practice. The theoretical principles don't change, of course, but the way we practice the art does change from generation to generation. In fact, today a generation is as short as 3-6 years. So trying to keep pace with that is unique to engineering.

JE: What is the greatest concern you have about VLSI education and circuit simulators?

TL: Some students do trust the simulators a little too much. I try to discourage that. Students who don't question the results of a computer, even when those results are clearly absurd, get a "special" grade.

JE: What direction is VLSI education heading in and how will it affect industry?

TL: As circuits increase in complexity, CAD tools become more necessary and I think it's fair to say that some of these extremely large and complex circuits simply can't be built without them. There are some who believe CAD tools cause "brain rot," but I think the productivity gains they provide cannot be ignored. There's a way to balance these concerns against the demands of high-productivity engineers.

JE: Are there any plans for the panel to reconvene?

TL: No, this was sort of an ad hoc, one-shot event. But I do think ISSCC revisits the issue every four to six years. As complexity continues to increase, it's always a good idea to look around and see if you're responding to that stress in an appropriate way.

Professor Thomas Lee received his degrees from MIT. He has designed PLLs and various analog circuits for Analog Devices, Rambus, DEC and AMD. His research focus at Stanford has been on CMOS ICs for communications. He has twice won the "Best Paper" award at ISSCC, was awarded a Packard Fellowship in 1997, and is the author of "The Design of CMOS Radio-Frequency Integrated Circuits." He was recently named recipient of a Phi Beta Kappa "Excellence in Teaching" award.