“We’re used to hearing about how game-changing technology makes whole new ways of living and working possible. But what makes the game-changing technologies possible? We’re going to talk about Berkeley’s contribution in this domain, a bit upstream from the technology we all know.”
This season of the Berkeley Remix we’re bringing to life stories about our home — UC Berkeley — from our collection of thousands of oral histories. Please join us for our fourth season, Let There Be Light: 150 Years at UC Berkeley, inspired by the University’s motto, Fiat Lux. Our episodes this season explore issues of identity — where we’ve been, who we are now, the powerful impact Berkeley’s identity as a public institution has had on student and academic life, and the intertwined history of campus and community.
The three-episode season explores how housing has been on the front lines of the battle for student welfare throughout the University’s history; how UC Berkeley created a culture of innovation that made game-changing technologies possible; and how political activism on campus was a motivator for the farm-to-table food scene in the city of Berkeley. All episodes include audio from interviews from the Oral History Center of The Bancroft Library.
Episode 2, “Berkeley Lightning: A Public University’s Role in the Rise of Silicon Valley,” is about the contributions of UC Berkeley Engineering to the rise of the semiconductor industry in what became known as Silicon Valley in the 1960s and 70s. In contrast to the influential entrepreneurial spirit of a private university like Stanford, Berkeley’s status as a public institution had a different impact on Silicon Valley. We focus on the development of the first widely used design program for prototyping microchips. Originally designed by and for students, the software spread like lightning in part because Berkeley, as a public institution, made it available free of charge. The world has not been the same since.
This episode uses audio from the Oral History Center of The Bancroft Library, including interviews with Paul R. Gray, Professor of Engineering Emeritus, Department of Electrical Engineering and Computer Science and Dr. Laurence Nagel, CEO of Omega Enterprises, former senior manager at Bell Laboratories, with a PhD from UC Berkeley EECS (oral history forthcoming).
“Berkeley Lightning” was produced, written, narrated, and edited by Oral History Center historian Paul Burnett.
The following is a written version of the The Berkeley Remix Podcast Season 4, Episode 2, “Berkeley Lightning: A Public University’s Role in the Rise of Silicon Valley. ”
Narration: Silicon Valley. It’s a real place, the valley roughly encompassed by Santa Clara County at the southern end of San Francisco Bay. But it’s also a mythic place, with just the right combination of top universities, electronics firms, defense dollars, and a concentration of rare, plucky college-dropout geniuses who would go on to hatch world-changing technologies in suburban garages.
We want to take that apart a bit. The university that looms largest in nearly every story of the rise of Silicon Valley is near the heart of that actual valley, Stanford University. But about 30 miles north, on the eastern edge of the Bay, lies the University of California, Berkeley, a longstanding rival to Stanford, if football is your game.
Our story here focuses on this other university, a public, state university, that established institutions and teams to develop innovations that would make the culture of innovation possible. There are many different stories we could tell about Berkeley’s role in the rise of Silicon Valley, from specific technologies such as flash memory to digital-to-analog conversion, aka the hardware and software that make it possible for you to listen to me right now.
We’re used to hearing about how game-changing technology makes whole new ways of living and working possible. But what makes the game-changing technologies possible? We’re going to talk about Berkeley’s contribution in this domain, a bit upstream from the technology we all know.
The centerpiece of just about any discussion of Silicon Valley is the development of its namesake, the silicon microchip, a tiny wafer packed with an ever-growing number of all the components that make up an electronic circuit in this microscopic space, what comes to be called an integrated circuit. Chief among these components is the transistor. Transistors do many things, but among them is to act as a switch, which allows them to process digital information, zeros or ones, much more efficiently and cheaply than tube-based mainframe computers, the ones that used to fill up entire basements of office buildings. When the transistor was invented, the race was on to increase the density and number of transistors. There are a few reasons why you want to make these integrated circuits smaller and denser. For one thing, they work better and more efficiently. And, you can cram them into small spaces, such as at the tip of a missile, for example.
But imagine how tricky they are to make. Here is Berkeley engineering professor Paul Gray explaining the dimensions of the microchip.
Gray: It was probably maybe fifty mils by a hundred mils. A mil is a thousandth of an inch. So that would be a tenth of an inch in one dimension and one-twentieth of an inch in the other dimension.
Narration: And here is Larry Nagel, who was a student at Berkeley’s microelectronics lab in the 1960s, on making chip prototypes.
Nagel: And being a little bit clumsy we had to make up for a couple of times when things got dropped and things got otherwise messed up. So I guess I was probably working maybe probably four weeks at that before I had a chip that actually worked.
Narration: Now this was not normal, routine work for an electrical engineering student in the 1960s. Just a handful of universities that had links the electronics industry and to military research had founded specialized microelectronics labs by the early 1960s. Here’s Dr. Gray again:
Gray: Berkeley had started the country’s first laboratory in which you could fabricate an integrated circuit. And that was in the about ’63, ’64 timeframe…Don Pederson was the faculty member here who really spearheaded the establishment of that laboratory. In that era Stanford and MIT also were starting labs—I think Berkeley was the first and then Stanford and MIT, some a year or two or three later, also got on the same track. … Berkeley and Stanford and MIT continued for the next several decades as the main institutions with this fabrication capability. And we still have a big lab, fabrication facility over there in the CITRIS Building.
Narration: So, UC Berkeley was the first university in the country to have a microchip fabrication facility. But that’s just the beginning. You have to understand that making microchips, by hand, is hard, really hard, and time-consuming.
Nagel: Well, just to build the circuit alone would probably take a couple of days for a hundred transistors for a kit. Maybe for a twenty transistor circuit a day. So one or two days, something of that order of magnitude. But actually measuring and getting the thing to work right, debugging it, could take a good deal longer than that. That could take weeks.
Narration: But from the time the microelectronics lab was founded until the end of the 1960s, the number of transistors on the same space on a chip went way up.
Gray: We were building chips that had on the order of a hundred transistors on them. It’s very difficult to predict by building a physical breadboard or a prototype out of discrete components how a chip like that’s going to behave electrically. You really needed circuit simulation even at that point, a program that would simulate the electrical behavior of a circuit.
Narration: Now if a computer program could do the work of prototyping a circuit, you wouldn’t have to waste time building dud after dud. As a student at the University of Arizona, Paul Gray then learned about the work at Berkeley from his mentor, David Holland. And when Gray went off to work down in the valley at Fairchild Semiconductor, right around the time that Gordon Moore and others split off to found Intel, he recognized the importance of the work that Berkeley researchers were doing.
Gray: I do remember having a lot of connections with universities in general, Berkeley and Stanford, on various topics in those years. Of course, we were in an industrial park created by Fred Terman, owned by Stanford. We were on their land. But I don’t remember having a sense of a Stanford dominance of the landscape in terms of university engagements. We had a lot of interaction with the Berkeley people because of the computer-aided design activity. Don and his group here at Berkeley were developing that kind of simulator. We needed that. So we got a connection going and we got one of the early versions of SPICE, I think it was called something else at that point, and used that.
Burnett: CANCER, I think.
Gray: Yes, correct.
Narration: SPICE? CANCER? These are strange if cool names for software, but this will be explained later. As a result of his reaching out to Berkeley to get a hold of this computer program called CANCER, Paul Gray was invited to Berkeley to teach for a year, and then joined the faculty, where he got the back story on what this software was all about.
First of all, Berkeley engineering was structured for this interaction between computing and electronics. At the time, only Berkeley and MIT had electrical engineering and computer science in the same department. Second, there was a leading light of the Electrical Engineering Department named Donald Pederson, who made computer-aided design a priority.
Gray: Somewhere in the early sixties, Don had recognized this need for computer simulation. … back in those days you could still try out your design by building what they called a breadboard and you plugged discrete devices in and it mimicked the chip, how the chip was going to behave. That was pretty ineffective anyway. But once you got bigger than a hundred transistors or so it became completely impossible to do that. Don recognized very early, the way things were going, it was going to be essential. And it was one of those early interdisciplinary things. To build an effective simulator of electronic circuits you have to have someone that knows devices and models the behavior of the transistors electrically. You have to have somebody that understands computer numerical analysis and how you actually solve differential equations on a computer on a large scale. And you have to have circuit people who understand what’s needed.
Narration: The other piece of the story is the fact that UC Berkeley is an institution of higher education, and students need to be taught in an efficient manner. So Don Pederson asked Ron Rohrer to develop a graduate course where the challenge was to have the students build design software. Here’s Dr. Nagel again:
Nagel: But I would say actually the major emphasis of the simulation programs that were developed at Berkeley were more as teaching tools, so that students could actually get a first-hand—again, Don was an intuitive guy. By running circuits on a computer you could get an intuitive feel for how the circuit would work, something that would take hours and hours and hours if you were to do it in the lab. Because in one night you could build five different variations of some particular circuit, simulate it, and have your results of which variation worked the best. That would be hours and maybe days of laboratory work. By doing it on a computer the entire class—it was no longer just a graduate exercise. Undergraduates could also enjoy this thing.
Narration: But the first graduate class to get this program built was intense, to say the least.
Nagel: And, of course, Ron [Rohrer] came in the first day and said, “Well, for those of you who think this is going to be a course on circuit synthesis, you’re in for a shock because this is going to be a course on circuit simulation and you guys are going to learn all about circuit simulation by writing a circuit simulator. The judge for how well you do will be Don Pederson. If he likes the program that you write, you’ll all get As. If he doesn’t like the program you write, you’ll all fail.” Ron was a very brash guy. So immediately half the class turned white as a sheet and left the room and were gone. [laughter]
Narration: Building on a number of earlier versions, Larry Nagel’s CANCER program was the result of that class, [computer analysis of non-linear circuits, excluding radiation]. Here’s Larry Nagel explaining the significance:
Nagel: “CANCER: Computer Analysis on Non–Linear Circuits…Excluding Radiation!” Because we were very proud of the fact that we’d developed this program with no money from the government at a time when the government was heavily funding research into radiation effects on circuits…
[background audio of student protests, chanting]
…because we were at the time very much worried about some kind of a nuclear event disabling our missiles. … So that’s how we got the “excluding radiation,” because we weren’t doing radiation. We were Berkeley at the time, right? This was Berkeley. This was not MIT. This was Berkeley. So that’s how CANCER got started.
Narration: So, you had institutional innovations, a microelectronics lab and a teaching approach that focused on computer-aided circuit design, resulting in this technical innovation. But a crucial innovation was not technical at all; it was legal, and social. Larry Nagel’s SPICE, or Simulation Program with Integrated Circuit Emphasis, was central to this part of the story, as was UC Berkeley’s status as a public university:
Nagel: I like to think that SPICE was actually the first open-source project way back before there was such a thing as open-source. But Don Pederson had a very strong belief that anything that was developed at a public institution should be in the public domain. So all of the Berkeley programs were available free of charge, or basically for whatever it cost to load the program onto a tape. So the fact that this program was for free had an enormous impact in a lot of ways. First of all, the program was made available to anybody so it diffused very quickly out to various different universities. And all the students that learned to use SPICE took it with them to industry. So it wasn’t at all long after the original release of SPICE in 1971 that basically every major integrated circuit manufacturer had their own version of SPICE. There was a TI SPICE at Texas Instruments. There was an ADI SPICE at Analogue Devices. After I graduated there was a program called ADVICE, which was developed at Bell Laboratories and used at Bell Laboratories. …
Don’s deal was that you can have the program for free but if you find a bug in it and fix it you have to give it back to us. You have to tell us what the bug is and how you fixed it. There were several generations of students that kept improving SPICE. SPICE improved because you had this entire base of industry feeding information back.
Nagel: I think for Don it was really a matter of principle. He didn’t do a cost-benefit analysis. He just said, “This is how it has to be. We’re a public institution. We have to make it publicly available.” But if you look in hindsight, the reason that the program became as widely accepted as it did was largely because it was freely available to everyone. Anybody could walk out of Cory Hall with a tape and they had their version of SPICE. That process went on for a long time. I think the last version of SPICE that was released was SPICE III and that was in 1980s. So we’re talking about Berkeley being pretty much a SPICE factory for fifteen, maybe even twenty years.
Narration: Here’s Paul Gray again:
Gray: Long story short, within ten or fifteen years of that point in time, every circuit design engineer in the world was using some flavor of a derivative of SPICE. It became an industry standard. The industry wouldn’t really exist without that kind of simulation capability. Once you get [the scale] to the thousands and tens of thousands and hundreds of thousands of transistors, it’s the only way you can do design. … So it was a huge innovation and had an incredible impact and also pioneered a great software dissemination model for universities. Many others emulated that public domain model of dissemination. … It has been a big part of the landscape here at Berkeley for many years.
Narration: And so, is the Silicon Valley phenomenon really just about Stanford, chip companies in the Valley, and amateur hobbyists?
Gray:
If you asked the random citizen on the street anywhere in the Bay Area they would probably say that Stanford was way more important than Berkeley. I think “way more important” is not correct. … judging by the number of companies started by faculty or former students, or by the numbers of alumni employed in the valley, one could argue that Berkeley and Stanford are comparable contributors. In terms of innovations that have mattered, there are many.
Narration: So, what does this mean? The development of computer-aided design greatly facilitated experimental research in the design of new microchips. But there is something even more fundamental going on here. First, easy availability of quasi-open source software encouraged commercial development of new chip designs without making it prohibitively expensive to do so through proprietary licensing, copyright, and other legal devices. But Berkeley’s model still allowed companies to make a version of the software they could own.
UC Berkeley deliberately went down a different road, as a direct consequence of its hard-wiring as a public university. The second pivot point here is the separation of design from manufacturing. This software, and all its descendants, allowed companies to focus on the more-profitable end of microchip design, and to outsource the much-more-expensive part of the process, manufacturing, to other companies, and, eventually, to other countries. Now there are consequences to this change, environmental, socioeconomics and otherwise that we can’t cover here. But suffice it to say that in 2015, there were just a handful of major semiconductor fabrication facilities in the United States, with exports totaling $40 billion. Compare that to the fab-less semiconductor design companies, many of which are located in Silicon Valley: total exports $166 billion.
This is just one example of the influence that UC Berkeley’s Department of Electrical Engineering and Computer Science has had on the size, character, and global leadership of Silicon Valley. It was a combination of social, legal, and technical innovation that struck the Valley with lightning force, and in turn accelerated change and growth in the semiconductor industry to this very day. For more about these stories, visit the Oral History Center website at ucblib.link/OHC and search across our entire collection or through the full interviews with Paul Gray and Laurence Nagel.