The BBVA Foundation Frontiers of Knowledge Awards in Information and Communication Technologies has gone in this fifteenth edition to Professor Alberto Sangiovanni Vincentelli (University of California, Berkeley) for “radically transforming” the design of the chips that power today’s electronic devices, giving rise to “the modern semiconductor industry,” said the committee in its citation.
By providing software tools to ease the creation of complex chips, he made possible a worldwide explosion of integrated circuit design spanning research, industry and academia,” the citation continues. “Professor Sangiovanni Vincentelli created a rich ecosystem of electronic design automation (EDA) techniques that revolutionized – and remain at the core of – how we build computer systems.”
To add to his transformative scientific contributions, which have opened up whole new lines of research, Sangiovanni Vincentelli – Professor of Electrical Engineering and Computer Sciences at Berkeley – co-founded two companies, Cadence and Synopsys, which “collectively drive the entire semiconductor industry,” in the words of the committee, producing “design tools that are used in every single computer chip built today.” His candidature for the Frontiers of Knowledge Awards was backed by a total of 28 nominators, both institutional and individual, among them the 2017 Nobel Physics laureate Barry Barish, and 2007 Turing Award winner Joseph Sifakis.
The awardee has transformed chip design in three fundamental ways. Firstly, he proposed a method to speed up electronic circuit simulations, making them a truly practical solution. He then developed a system to automatically generate such circuits by means of hardware description languages. And lastly, he devised a program to place and connect each circuit’s multiple components for optimal performance and minimizing power consumption.
Ronald Ho, Senior Director of Silicon Engineering at Meta and secretary of the award committee, defines Sangiovanni Vincentelli as the “enabler of the chip industry, whose contributions have transformed the world of electronics” that provides us with the tech devices and systems omnipresent in our lives, from personal computers to mobile phones right up to the microprocessors installed in automobiles, planes and electrical appliances: “You can see the durability of his ideas because the two companies he founded to allow their industrial implementation are now driving the semiconductor industry. The chips we use today were built from the ideas that Sangiovanni Vincentelli put together.”
“When we think back to the beginning of the computer industry,” says Ho, “silicon chips were relatively simple, since they had few transistors and could be designed by hand. Today, in a modern smartphone, chips have billions of transistors, so, how did we get from doing chips by hand with a few thousand transistors to today’s complex chips? Well the answer is through the work of the man receiving the award today.”
Jordi Cortadella, Professor of Computer Science at BarcelonaTech (UPC), was taught by Sangiovanni Vincentelli in the late 1980s during a postdoctoral stay at Berkeley, and went on to work alongside him, co-signing various scientific papers. “The chips in today’s electronic devices”, he points out, “contain billions of transistors of nanometric size [one nanometer equaling one billionth of a meter], putting them beyond the scope of manual design. The tools developed by Sangiovanni Vincentelli enabled automated design of circuits, triggering exponential growth in chip production worldwide.”
Circuit simulation to “test” chips pre-fabrication
On completing his electrical engineering degree at Politecnico di Milano in 1971, Sangiovanni Vincentelli stayed on in a research post. Not long after, in 1975, he made the move to Berkeley, where he quickly turned his attention to the more theoretical end of electronics, like the design of numerical analysis algorithms. His colleagues there encouraged him to explore whether his algorithms might find application in circuit design.
A chip is a silicon wafer on which a large number of transistors and other miniature electronic components are laid out to form a circuit. It is these electronic circuits that run the complex calculations of our home computers, display the TV channel we press on the remote control, or turn on a warning light when our car’s tire pressure is low.
To perform all these functions on the electronic devices we use each day requires a specific set of transistors arranged in just the right way. When Sangiovanni Vincentelli began his research, this layout process was done by hand, selecting each component, connecting them together to form a circuit to the right specifications and checking that they did what they were meant to do.
Chips, even then, had hundreds of transistors so the work was slow and laborious – “very, very boring” according to the awardee – with the risk of error looming at every stage. It was this process, “more art than science” as the committee describes it, that Sangiovanni Vincentelli would end up transforming out of all recognition.
“We are all familiar with the historical miniaturization of chip features articulated by Moore’s Law,” the committee points out, “but the resulting increase in circuit density quickly led to sprawling chip layouts as complex as a large city and just as impossible to manually design.”
A program called SPICE was already available to simulate component behavior, enabling a circuit’s operation to be verified without the need to physically build it. To do so, however, the program had to solve a series of complex mathematical equations. And for the most elaborate circuits, it could take so long to arrive at a solution that its practical utility was limited in the extreme.
Sangiovanni Vincentelli realized that his numerical analysis skills could help accelerate the simulation, and designed a new set of algorithms which, he says, “provided a new way of looking at the solution of the equations.” Now, it was possible to simulate even a complex circuit in a reasonable timeframe, verifying its correct operation in a matter of seconds. This also meant a huge cost saving, as the circuit’s integrity could be checked before proceeding to its physical fabrication.”
Ho reflects on what Sangiovanni Vincentelli’s advance meant: “Before then, to check that your ideas worked, you would have to build the circuit, and it would take months. When you were finally done and could measure it, if you realized something wasn’t right, you had to start over. Instead of wasting months, now you just run a simulation for about 20 seconds and you know immediately whether or not it does what it should.”
“Hardware is not like software, which allows you to check though the program when you’ve made a mistake and send a new revised version,” adds Cortadella. “In the case of the hardware, it costs millions of dollars to design a chip. And if you get even one transistor wrong, it will all be wrong. So before you commit to fabrication, you must be sure that it will work, and that is the beauty of the simulation tools devised by Sangiovanni Vincentelli.”
Circuits designed using simple programming commands
The awardee also greatly simplified the creation of the circuits themselves. To conduct an operation as basic as adding two numbers, a circuit may require millions of components. And choosing which to use and how to arrange them is complicated work. At the time when he was starting out in electronics, the job of selecting components and forming a target circuit was reserved for specialists. Sangiovanni Vincentelli set out to alleviate this arduous task with the aid of his mathematical knowledge.
“The process of circuit design involves a series of steps starting with its ideation, what I want to achieve, and ends with its implementation in a tiny piece of silicon,” the new laureate explains. “Once you define precisely, mathematically, what these steps look like, you can build software tools that actually automate them.”
What he is describing here is his second highlighted contribution, the invention of a program that can generate circuits through the expression of simple programming commands. Thanks to this insight, it is possible to design a circuit simply by formulating the task to be performed in a language understood by a computer, which then automatically generates the set of components needed for its execution.
For further assurance, he developed ways to automatically verify that the components generated for the chip design corresponded to the circuit function formulated by the code. Without this check, errors might get overlooked among the profusion of circuit components.
Optimization of chip design with minimum human intervention
Finally, Sangiovanni Vincentelli noticed that the geometrical arrangement of components could be key to delivering higher performance or lower energy consumption. With this in mind, he also automated this stage of the circuit design process, developing algorithms that proposed an optimal placement for best performance or energy efficiency.
With these three contributions, the awardee enabled what he describes as “very large designs with minimal intervention from humans,” so team members could concentrate on the creative side of the design process without having to bother with the more tedious, and highly error-prone, details.
The result is that, thanks to Sangiovanni Vincentelli, today’s manufacturers can turn out chips vastly more complex than in the past. “In 1975,” he recalls, “the largest chip had about 2,000 transistors, and now they are packed with more than two billion.”
The three contributions that have marked his career may seem like “a straightforward set of steps,” says Ho, “but each one of them was foundational and started an entire area of research in the silicon industry.”
A leader in the world electronics industry
Sangiovanni Vincentelli is not just a “prolific inventor” and “outstanding educator,” in the words of the committee, who founded a worldwide school of thought from his base at Berkeley, his influence has extended beyond the university into the leading edge of industry.
He refuses, however, to see himself as an entrepreneur: “An entrepreneur is a guy who has this brilliant idea, and then goes off and makes this wonderful company. But it didn’t happen that way in my case.” In fact many firms had been using the tools he developed since as early as the 1980s, when they were made available in the public domain. And it was the owners of these firms that persuaded Sangiovanni Vincentelli he should move into business in order to market his circuit design solutions.
His first venture was Cadence, set up in 1983, followed in 1987 by Synopsys. From an initial staff of just three people, both have since grown into multinational companies with offices around the world employing upwards of 10,000 people.
“At first they were complementary,” he recalls, with each one dealing in a different side of the design process. “But by 1991 they started competing, so I had to choose the company I would continue to work with. After a lot of pain, I decided to continue with Cadence.” He retains the connection to this day, serving on the board and acting as a technical advisor… “It’s like one of my children,” he reflects.
Both companies remain at the forefront of the silicon industry, providing technology to clients ranging from Apple to Intel, Tesla or Boeing. “Every single digital chip, everyone in the industry, uses tools from either Cadence or Synopsys, usually both,” says Ronald Ho. The awardee has also advised General Motors on its technological base, particularly the design of its electronic auto components, and helped Pirelli with the invention of chip-in-tire technology that monitors road grip in real time.
But Sangiovanni Vincentelli is also a firm believer in making his discoveries available to whoever wishes to use and build on them, putting the advancement of knowledge before any economic interest. “We are a university, our mission is to teach people, but also to advance science. And how can you advance science if you don’t let people use your work?”
As such, the tools he has developed are open source, meaning anyone can use them to improve their chip design. The success of his companies relies on offering solutions based on university research, but geared specifically to the client’s needs.
Future applications in synthetic biology and drug discovery
Sangiovanni Vincentelli is now looking at other areas where design automation could usefully contribute, among them automobiles, planes and buildings. “Since we have been able to master the complexity of designing electronic systems, we are now thinking about how to extend the tools and algorithms that we develop to different fields.”
Although some advocate for applying artificial intelligence in this kind of project, he believes its use may prove less than transformative. “You cannot do everything with machine learning,” he reasons, and it is wrong to assume it will provide better results than existing, tried and trusted methods. “We know the physical principles that underlie the design, and by using them we can guarantee the performance of our tools.”
The scientist singles out biology as one area that, in his view, would especially benefit from design automation. His ambition is to design effective drugs with minimal side effects and even to create synthetic life forms that can be engineered for specific tasks in the treatment of disease. “Now anything is possible”, he affirms.