October 08, 2007
Cappuccino & Creativity: Sangiovanni-Vincentelli, Sentovich, and Szymanski
Please note that contributed articles, blog entries, and comments posted on EDACafe.com are the views and opinion of the author and do not necessarily represent the views and opinions of the management and staff of Internet Business Systems and its subsidiary web-sites.
There are multiple parts to this week’s newsletter. If you don’t have a lot of time, or are not in the mood, just read Parts 2, 3, 4, or 5. Click on “Print Article” up there on the right, and you’ll see all of the text without the click-throughs.
However, if you do have some time and are in an expansive mood, please don’t miss Part 1 just below. It’s a conversation over cappuccino with Alberto Sangiovanni-Vincentelli, Ellen Sentovich, and Nancy Szymanski, and it’s really something special. I hope you’ll agree.
1 - Cappuccino & Creativity
2 – Coming soon to a theater near you
3 – New business cards
4 – News that’s noteworthy
5 – News that’s cool
A million years ago when I was an undergraduate at Berkeley, I used to stay up late into the night discussing everything under the sun with friends. We would argue about politics and philosophy, majors and careers, and whether there was life on other planets. There’s something about university life that fosters this kind of freewheeling conversation and there’s something about moving away from that environment, and the accumulated responsibilities of age, that precludes it ever happening ever again.
Unless you’re lucky and get invited to come for coffee at the home of U.C. Berkeley professor,
Dr. Alberto Sangiovanni-Vincentelli. If you’re doubly lucky, Cadence Research Scientist,
Dr. Ellen Sentovich will be there, and Cadence’s
Nancy Szymanski, as well. If you’re triply lucky it will be one of those lovely, wistful autumn days that Berkeley does better than most places on earth, and Dr. Sangiovanni-Vincentelli will make cappuccinos. Then you’ll sit and sip your coffee and examine great ideas and the exquisite architecture of his circa 1901 home simultaneously. And you’ll think for a minute that the last million years haven’t come and gone, and you’re back in the days of yore when you could sit and talk for hours on end, and every idea that could be imagined was open to discussion and debate.
1 – Cappuccino & Creativity
Peggy – Is there a future in EDA for young technologists?
Ellen – Certainly there’s a future for students in EDA. True, it’s changed a lot since I first began. When I first came to U.C., it was about synthesis – there was such a large group at that time. Now it’s [moved] out to the system level and I think it will move quite a bit more. In the EDA industry, we have a lot of skills and expertise that can be applied in many areas. Yes, EDA has a lot of possibility for young technologists.
got Bob to come to Berkeley instead.
Note: Dr. Brayton joined the faculty at Cal in 1987 and has just been named the 2007 recipient of the Phil Kaufman Award.]
Our approach to synthesis at that time was based on an attempt to be very mathematically rigorous and do global optimization. The other approach was a compiler technology, but the people optimizing in our group were getting better results. [This was the main emphasis] of my work from 1980 to 1988, and then we starting thinking about formal verification and how to innovate there.
In 1988, I started working with Magneti-Marelli, one of the largest companies in automobile subsystems. They were coding a new power train control for Formula 1 autos. In the old days there was a manual clutch, but the stick was very short so the poor drivers had a lot of work changing gears and their hands were bloodied from the stress. The company asked, why not do the shifting algorithmically and defined an automatic control system that would optimize the process. But the people at Magneti-Marelli were worried that their system might actually cause the car to explode. They heard there was an Italian professor at Berkeley, so they came to Cal to see me and asked us to verify that
their system worked.
We tried to do that by using formal verification, but finding a formal representation of the algorithm was not easy. The only thing we had was the actual code – assembly code for a tiny 8-bit microprocessor – so we reverse-engineered the algorithm. It turned out to be a finite state machine, and we found an error – a corner condition – that we were able to correct. It would have been much easier to have had the algorithm to start with, but this story [illustrates] why verification is so important. In engineering, there is no separation between what you want to do and how you go about accomplishing it.
This project gave us the inspiration to start working in system-level design in 1988 at Cal. At the time, we were the only ones in the world [working in this area]. It illustrates the point that because I talked to industry, we had the opportunity to address a very advanced problem in our research.
Peggy – Ellen, was Alberto your PhD advisor?
Ellen – No, I did my master’s with Richard Newton and my PhD with Bob Brayton. The first time I met Alberto, I was preparing to give a paper at ICCAD on the IBM-versus-Berkeley methods. It was [a Berkeley tradition], that the professors would rehearse the students to death who were making presentations at a conference. I had to give my ICCAD talk before a group of professors, including Alberto. It was on Boolean networks and edges, and after I was done Alberto said the first part of my talk was a clear as mud.
I was devastated, but by the time I got to ICCAD I was really prepared. During the Q&A after my talk at the conference, a professor in the audience from another university presented a [very tough question]. I was so glad at that moment that I had already been raked over the coals by Alberto during my rehearsal. I was totally un-phased by the question. [Ellen laughed.]
Alberto – This is what we called the Berkeley stamp. Our students were always rehearsed to death before a conference.
Ellen – There really is no substitute for rehearsal.
Alberto – We won several Best Paper awards this year at DAC. One of my students gave his talk 3 times before going to DAC, and lo and behold, the third time we decided the way the paper was put together was just wrong. I told him to reorganize the talk and he did it at the last minute and still won Best Paper in San Diego.
I remember in 1982, it was my first DAC ever. We were presenting two papers. One was being presented by one of my students, and I was presenting the other – a paper co-authored with Richard Newton, Bob Brayton, and Gary Hachtel. I had done the bulk of the work, so I was the one to present, but I didn’t realize that at DAC everyone had slides. I only had handwritten foils. I think the organizing committee thought I was crazy, considering the quality of the conference.
But I rehearsed and rehearsed, and was so happy to win Best Paper. The other paper from our group also won Best Paper. Afterwards, we went out to celebrate in Las Vegas, so people had to come find us to let us know that we had not only won Best Paper, but we’d won Best Presentation as well. Since that time, things have changed. Now at DAC we only give out Best Paper awards before the conference. It is too difficult and there are too many presenters to be able to evaluate all of the presentations during the conference.
You can find the full EDACafe event calendar here.
To read more news, click here.
-- Peggy Aycinena, EDACafe.com Contributing Editor.
Be the first to review this article