Dr. Robert Brayton – 2007 Kaufman Award Winner
By Peggy Aycinena
Editor of EDA Confidential and a Contributing Editor to EDA Weekly
Robert Brayton has been named the 2007 recipient of the Phil Kaufman Award for outstanding contributions to electronic design automation. Brayton spent more than a quarter of a century at IBM in Yorktown Heights working in the areas of synthesis and simulation before starting a second, lengthy career in academia as a faculty member at U.C. Berkeley, where he is currently the Cadence Distinguished Professor of Electrical Engineering and Computer Science. With 10 books, 450 papers, and a lengthy list of honors on his CV, it’s probably safe to say that in both his industrial and academic careers, Bob Brayton has excelled. His designation as this year’s Kaufman Award winner is the ultimate comment on his accomplishments in EDA.
I had a chance to speak at length recently with Dr. Brayton. We began by discussing the role his colleague, Alberto Sangiovanni-Vincentelli, played in Bob’s decision to come to Cal: “I was at IBM for 26 years, during which time I struck up a collaboration with Alberto, starting when he first came to IBM in 1979 for a year-long sabbatical. Even after he returned to Berkeley, our collaboration continued, particularly during my own sabbatical at Berkeley in 1985. In 1987, I retired from IBM and at Alberto’s urging, applied for and accepted a faculty position at Cal.”
Was the transition to California a big change for him? Bob said, “Yes, in some ways. But everything in life is an adjustment and I never regretted the move. Of course, teaching and guiding students was new to me, but the research component was very similar to what my work had been at IBM.”
Brayton noted that although all three of his children were in college or beyond by 1987, the move did have a major impact on his wife: “We moved away from a neighborhood in New York that had been our home for over 20 years. It’s always hard in those circumstances to leave family and friends behind. I’m originally from Iowa, but my wife is from Boston, so moving to the West Coast really was a change for her.”
Reared and educated in the Midwest – Brayton has a BSEE from Iowa State University – Bob has roots of his own on the East Coast having earned his Ph.D. in mathematics at MIT. It was during those years in Cambridge that Brayton developed his interest in computing. He told me, “I had a job on campus working as a research assistant in the Artificial Intelligence Lab in the Computer Science Department. I found myself working on the LISP programming language, and [in fact] wrote what I think was the first compiler for the language. It was definitely hard, working in the AI lab, but it was also a lot of fun and intellectually broadening.
“Back in those days, both John McCarthy and Marvin Minsky were in the lab. But although I really liked computers and programming, I didn’t see much of a career path in AI, so I continued on towards my degree in mathematics. Also, when I had studied engineering as an undergraduate, I realized there were some things I couldn’t understand because I couldn’t read the math. That was the original motivation for my area of study at MIT, and I still preferred that over computer science.”
After he finished his Ph.D., Brayton went to work at IBM, a decision based on a summer internship he had had at the company while still in grad school: “The summer before I finished my Ph.D., I had an internship in what turned out to be the beginnings of the IBM Research Division. They had set up a Mathematics and Science Department in six or seven old buildings out in the trees in Westchester County, New York – a lovely setting. The summer I was there, one of my math professors from MIT was also working there, Jurgen Moser. We started working on some research that involved formulating circuit equations, and ultimately published a couple of papers on our work. That was the summer of 1960. After I graduated the following year, IBM hired me back full time.” He added with a laugh, “They had just opened up the new Yorktown Heights Research Center, and I ended up spending 26 years in the same building!”
Obviously when Brayton moved to Berkeley in 1987 it was a big deal, but Bob already knew the players on campus. “Not only was Alberto at Cal,” he said, “But Richard Newton was there, and I knew him quite well. He had been a colleague for a long time. Of course, Ernie Kuh and Don Pederson were also there and among the principal names in EDA. Remember, Don Pederson helped put Berkeley on the map with the SPICE program.”
With that, our discussion turned to the real reason for Brayton’s Kaufman Award. At around the same time that the Berkeley group was developing SPICE, Brayton was similarly occupied at IBM. He said, “One of the things I had done at IBM was to help create a program that was a kind of parallel development to SPICE. We called it ASTAP [note: for Advanced STatistical Analysis Program] and it was also a circuit simulation program – the simulation program used internally at IBM for almost 25 years.”
ASTAP wasn’t the only internally developed CAD tool in use at IBM. I asked Bob why the company ever turned to third-party tools to meet their design needs. He said, “The commercially-provided design tools were catching up [with our in-house tools], and had the capability to meet some of IBM’s design needs. So it was natural for the company to go outside to save money, particularly when specialized tools could be obtained less expensively than trying to develop everything internally. It had reached a point where it was impossible, even for the big companies, to reproduce the tools available [from the vendors].”
There was another issue in that equation as well. Brayton noted that, early on, IBM had actually wanted to give their internally developed CAD tools away for free, but the U.S. Government said no. “IBM was leading the way in the [CAD tools] area and we wanted to give away the tools for free, but there were monopoly considerations for IBM at the time and we were not allowed to do so. From my point of view, it would have been great. We would have [encouraged feedback] from a number of expert users, and the company had already given away computer programs in other areas to stimulate interest in computers. In [the CAD tool area], however, the government constrained the company for anti-trust reasons.”
I asked Bob to remember back when IBM was viewed as the Microsoft of its day, the 800-pound gorilla in the industry. He agreed, “Yes, that’s true. But IBM has also always been a great company with truly great management in place – a very well run place.” He added, “And, they were able to morph into something even greater because of their management. If I were starting out in my career today, I would definitely go there again.
“Of course, some things have changed. Back when I started in 1961, the idea of basic research was very important. And IBM had such a large share of the market, they could afford to support it. As competition changed, however, things got tighter financially for the company and they moved to doing more development than research. There are still policies in place at the company, however, that are holdovers from the early years when basic research was the thing to do and it’s still a great place to work at. There is still a great deal of freedom for people there compared to working for smaller companies.”
What about the urban legend that innovation only happens in small companies? Bob responded, “I know a lot about small companies, and certainly they do innovate. But large companies do a lot of innovation in-house, while smaller companies usually start with an already developed idea and run with the product. You can’t afford to start a small company and then spend years doing research.”
So, if IBM had been able to give away their circuit analysis program for free, would it have prevailed over SPICE as the industry standard? Brayton said, “The two programs were near enough to each other [in concept] to be very competitive. We had very sophisticated ways to solve the linear equations, but SPICE had some advantages with respect to ruggedness. There were subtle advantages to each program, and I think if we could have given our program away there would have always been both SPICE and ASTAP. The fame that SPICE eventually developed would have been shared with ASTAP. [As it turned out], SPICE development has continued on, with the equations being solved now so much faster than anyone could ever have conceived of at the beginning. It has evolved significantly over the years with sophisticated new equations and multiple versions out in industry. There are so many commercial versions, in fact, I don’t even know them all.”
Given the success of U.C. Berkeley’s SPICE, I asked Bob if professors sometimes lament that their ideas are commercialized by others who then reap the financial rewards. Bob said, “It’s actually the other way around. At universities, professors love to see someone take their ideas and run with them. A lot of university folks don’t want to do that kind of work, or they wouldn’t be in academia in the first place. Also, a lot of times the people [commercializing the ideas] employ the academic as a consultant, so it’s not like you’re completely giving away the concept. That people like your ideas and want to do something with them, is really a marvelous situation.”
What have been Dr. Brayton’s principal ideas, the core of his research at the university? He told me, “The central area of our research at Berkeley has been in logic synthesis. I initially started doing work there with Alberto and Gary Hachtel, another long-term colleague of mine who went on to the University of Colorado. In 1985, while I was on sabbatical at Cal, Alberto and I developed a multi-level interactive synthesis program for logic. That was the beginning of a whole series of programs that do logic synthesis, which has continued up to the present time.
“In the middle of that effort, we also saw the connection to formal verification, so my work has also gone off in that direction as well. And yes, that supports the argument that work in synthesis naturally evolves into work in verification. There are, of course, people working in verification that didn’t come from synthesis, but it is very natural for people working in synthesis to look at verification. I remember in 1990, a research fellow from Bell Labs gave a series of lectures at Berkeley on verification. Hearing those lectures, I had a Eureka moment – We should get a project going in this area! – and since then I’ve had a number of graduate students working on verification.”
Bob reminisced, “When we first started working on synthesis there was no Synopsys, although some of our early grad students went on to work at the company. A lot of the algorithms we developed were used in the commercial area – in Synopsys and elsewhere. Since that time, Synopsys has gone on to become the leader in the field, but even with synthesis you can always improve on things. The algorithms continue to get more efficient and there are always new and interesting things to be found. We continue to be funded in our research by various agencies, including NSF, SEMI, and the California MICRO program. Plus, there are seven different industrial companies supporting our work at Berkeley. [Naturally], the research is all open and available on the web – the algorithms we’re developing can be downloaded for free.
“Berkeley has had a long tradition in this area. Don Pederson established a long time ago that the programs developed at the university would be free and easily available. I completely oppose academics going out and patenting stuff and trying to tie people’s hands in how it is used. People appreciate our work, and the openness, and they come back with additional money to support our research.”
I asked Bob if additional, massive government funding is needed to keep the U.S. competitive in technology. He said, “I think at this point, the semiconductor industry continues to be very healthy in this country. [Of course], it takes a lot of money to continue to develop the manufacturing capability, and there are concerns about offshore manufacturing – especially in Taiwan – but if you look at Intel and IBM, they have competitive, or better, manufacturing and technology here and hopefully they can keep it up. Just the existence of those two giants in this country keeps the competitive edge here. Yes, as Thomas Friedman [The World is Flat] says, you have to acknowledge the global economy and compete the best way you can.”
Given the current state of affairs in technology within this global economy, what does Dr. Brayton advise a young engineering student to study? More crucially, how can they possibly learn everything they need to know to stay ahead? Bob answered easily, “Yes, there is so much engineering to be learned. We’re constantly debating at the university how much we should require of students in the field – only one in a million can actually master everything. But remember, very little in engineering is forever. I was trained as a mathematician at MIT. When you prove a theorem in mathematics, it remains forever. But in engineering, it’s possible to bet your money on the wrong horse.
“If, on the other hand, you truly love the work you’re doing, it’s the journey and not the end result that matters. Lots of engineers want the end result to justify their work, but that’s not enough. I tell my students to study the things that really turn them on. I ask them, what are you worried about? A job? I assure them, you’re going to get a job – there are no problems getting jobs in EDA. So work on something that’s really exciting to you, something that piques your interest and utilizes your capabilities. Something that’s fun! Useful, but fun! Find that area and be very enthusiastic!”
In closing, I asked Robert Brayton if he is aware that folks in the industry like to compare the Phil Kaufman Award to the Nobel Prize in EDA. Ever self-effacing, he laughed: “Yes, it’s a very great pleasure to be receiving the Kaufman Award, but I’m not sure about that Nobel Prize thing.”