|
|
The Silicon Gate Design of the 4004: From Idea to Reality
"Federico Faggin joined Intel to turn Hoff's vision into a silicon reality. In less than one year, Faggin and his team delivered the 4004, which was introduced in November, 1971."
Caption under Faggin's picture on the 4004 Microprocessor Display at the opening of the New Intel Museum to the public in 1992.
"…and then we hired Dr. Faggin who we had known back at Fairchild to come in and actually lead the project to realize the processor that Dr. Hoff had conceived."
Dr. Gordon Moore's words on the first microprocessor in a clip from the documentary “Silicon Valley Story,” (1996), by journalist and film director Chiara Sottocorona.
"When I joined Intel the first major challenge for me was to devise the methodology that I was going to use in the silicon gate technology which was also a new technology. That technology allowed to make complex circuits in a different way than metal gate. Because nothing like that was done before I had to figure out how to do it."
Federico Faggin describing his role as project leader and designer of the first microprocessor, in the video "The Designer Behind the World's First Microprocessor – Federico Faggin" Filmed at the Intel Museum in Santa Clara, California (2011)
"Now it’s a team effort. In 1970 Federico Faggin designed the 4004 microprocessor chip by himself at Intel in nine months; our 32-bit microprocessor took 100 man-years!"
Bob Noyce, co-founder of Intel, in the article “Silicon Valley” by Moira Johnston, page 466, National Geographic, October 1982.
The Enabling Design is Signed by Federico Faggin
The silicon-gate-based design, entirely created by Federico Faggin, was the invention that made the first microprocessor a reality in 1971. This was not a routine design since such complexity had never been integrated into a single chip before. It was only because of his innovations in MOS Silicon Gate Technology (SGT) that Faggin could design the 4004 in one chip. The bootstrap load and the buried contact - the ideas at the core of Faggin’s design methodology that he applied for the first time in the 4004 – rescued Hoff’s architecture and allowed it to be implemented in 1970. Without the bootstrap load and the buried contact, Hoff’s architecture was unfeasible because the microprocessor would have been too slow and too costly for any practical application. Therefore, the real invention of the first microprocessor was not in formulating the architecture of a simple CPU, similar to many others at that time, but in creating and implementing the technology that for the first time could fit in a single chip the entire function of a CPU.
Ted Hoff was hoping that his architectural proposal consisting of a block computer architecture and instruction set could be translated into a single chip design. However Hoff was in no position to evaluate the feasibility of the project nor to lead its development, since he was not a chip designer. It was Federico Faggin who invented how to design and layout 2300 random logic transistors into a real chip 3 mm by 4m, with low cost and with 5 times the speed and twice the density of the incumbent MOS technology. This was a feat no one had done before, and Faggin’s methodology was used for all early Intel’s microprocessors. Faggin also led the project to its successful outcome and promoted the marketing of the 4004 to management by demonstrating that the 4004 could be used for applications other than calculators.
"Well, the big problem was I was not an MOS designer."
Ted Hoff in a 1995 interview by Rob Walker (Silicon Genesis)
The MCS-4 Architecture
“The stored-program computer, which was used prominently as a calculator in the 1950s and 1960s, was one of the most lauded achievements of the post-war era and was familiar to every engineer working in the semiconductor industry." (pg. 245)
William Aspray: "The Social Construction of the Microprocessor A Japanese and American Story," in Facets: New Perspectives on the History of Semiconductors, ed. Andrew Goldstein & William Aspray (New Brunswick: IEEE Center for the History of Electrical Engineering, 1997)
In 1969, Busicom, a Japanese calculator manufacturer, approached Intel to have them convert into silicon their logic design for a family of calculators. Their approach followed the footsteps of the Olivetti’s Programma 101, the world’s first desk-top programmable calculator, unveiled at the New York World Fair in 1965 and sold the same year. The Programma 101 was designed with a CPU (central processing unit), ROM (read only memory), and serial read-write memory, all made with discrete components. Busicom had a similar architecture where they envisioned the CPU made in three MOS chips, ROM and shift register memory made in two chips, plus two other input-output chips.
Ted Hoff, head of the Application Research Department at Intel, recognized that much of the complexity of Busicom’s design was due to the use of serial memory, and since Intel was then developing their first dynamic RAM (random access memory) chip, he could see that the design could be much simplified by using a more conventional and more general-purpose computer architecture based on RAM memory. With the assistance of Stan Mazor, and through interaction with Busicom engineers, among them Masatoshi Shima, Hoff formulated the MCS-4 architecture, reducing to 4 chips the design which would have required 7 chips in Busicom’s proposal.
Hoff believed that the CPU could possibly be done in one chip but then, neither he nor Mazor were chip designers, nor did Intel have any chip designers capable of doing complex random logic chips. Therefore Hoff’s architectural proposal had been idling for about 6 months until Federico Faggin was hired to head the project in April 1970.
Invention of the MOS Silicon Gate Technology
Federico Faggin came to Intel from Fairchild Semiconductor where in 1968 he created the MOS Silicon Gate Technology (SGT) and designed the 3708, the world’s first commercial integrated circuit (IC) using SGT. The SGT was the first commercial technology for the fabrication of MOS Integrated Circuits to use self-aligned gates, which were made with poly-crystalline silicon. The SGT was adopted by Intel and subsequently by the entire world semiconductor industry where for 40 years it provided the basic MOS structure used in nearly all ICs. Gordon Moore, Intel’s co-founder, acknowledges that a major component of Intel’s early success was due to their adoption of the SGT.
The SGT was about five times faster, had about 100 times less junction leakage, and could integrate twice as many random logic transistors for the same-size chip made with aluminum gate, while dissipating equal power. It was also the technology that allowed the production of the first commercially successful dynamic RAMs, CCDs (charged coupled devices) image sensors, non-volatile memories, and the microprocessor, providing for the first time in history all the components necessary for making general-purpose computers entirely made with solid state devices.
Creation of New Silicon-Gate-Based Design and Novel Layout
"I am not a MOS designer.” “My role was primarily in doing the architecture and then later doing support for the products." “After the architecture was done, the instruction set defined, it was turned over to the MOS design team, they carried it in there.”
Ted Hoff, head of Intel Application Research, in a 1988 video interview – IEEE Larson collection interview.
"Federico worked at just a furious pace,I mean, he really, really went all out and in the space of about nine months, designed the three major chips."
"I believe the last one [chip] that Federico did was the 4004, and I think he had first silicon on that sometime in January of 1971."
Ted Hoff in a 1995 Silicon Genesis Interview by Rob Walker
"Because I was not able to do both a circuit design and also layout. What I was able to do was to implement logic and generate the test pattern."
"Even if the specs and logic came out, without Federico Faggin we would not have been able to make the chip."
"Federico did a good circuit design, introducing many, many new techniques, such as bootstrapping."
Masatoshi Shima in a 1994 interview with William Aspray
"So Faggin really gets the major credit for making the thing happen by actually doing the engineering design and Ted really gets the credit for doing the original concept and architecture. And I was sort of the middleman helping out around the project as I could and making contributions as I could."
Stan Mazor, in a June 9, 2000 Silicon Genesis interview with Rob Walker
At Intel, Federico Faggin set about immediately to create the silicon gate design methodology (SGDM) necessary to design complex random logic circuits with the SGT. This methodology was needed because the SGT was a new technology requiring a different way to do the design, and especially the layout.
"So I decided that instead of doing a separate logic design, and then a circuit design, to do logic and circuit design together in a single sheet. But also with the notion of the layout so that to the extent possible, you try to put the lines and the transistor placement in this sheet as close as you can guess to what it will be [on the final layout]. Obviously, you had to do the overall planning of the chip ahead of time, so that you know roughly where the various blocks are.. It was during this period that I refined the methodology of doing these type of designs."
Federico Faggin in 2007 oral history panel interview with David House (CHM)
The design of the first microprocessor and the design leadership of the MCS-4 project, from the beginning of its development to production, could only be done by someone capable of innovating in process technology, chip layout, circuit design, logic design and computer architecture. Faggin had acquired such skills and knowledge through his education and job experiences previous to joining Intel. After graduating from the technical high school in Vicenza, Italy, Faggin co-designed and built a small transistorized experimental computer, using magnetic core memory, at Olivetti in Borgolombardo, (MI), Italy, at the age of 19. He then graduated in Physics, summa cum laude, from the University of Padua, Italy; and then he developed the MOS process technology and designed two commercial MOS ICs while working at SGS-Fairchild (now ST Micro), in Agrate Brianza, Italy. In 1968 he was sent to Fairchild Semiconductor R&D in Palo Alto (CA) where he developed the SGT and other technologies.
Real Innovation
"…the conceptualization of the microprocessor, which was Hoff’s principal contribution to the 4004 project, was independently conceived in other companies" (pg. 216)
"...a number of different groups independently came to Hoff’s concept of a computer on a chip. It is clear that what was important in the ‘invention’ of the microprocessor was not the conceptualization but rather the implementation of an economically sound product" (pg. 246)
William Aspray: "The Social Construction of the Microprocessor A Japanese and American Story," in Facets: New Perspectives on the History of Semiconductors, ed. Andrew Goldstein & William Aspray (New Brunswick: IEEE Center for the History of Electrical Engineering, 1997)
"In fact, probably only one person in the world did know how to do the next step ["translate the architecture into a working chip design"]. That was Federico Faggin.” (page 9) “Without Faggin, the first Intel microprocessors would never have been built. It was Faggin, while still at Fairchild, who invented the MOS technology from which the Intel devices would be built. Then, newly arrived at Intel, he patched the mistakes of an absent Hoff, then made the first 4000 chip set a reality. After that, he led the development of the 8008, and most important of all, was the chief architect of the 8080.” (page 19)
Michael S. Malone, "The Microprocessor – A Biography," Springer-Verlag, 1995.
At that time, engineers knew how to architect small computers, do the logic design of a small central processing unit (CPU), and create software programs. The idea of a microprocessor, i.e. of integrating into one chip the CPU of a general-purpose computer, was also in the air. In fact, some CPU architectures had been realized in several MOS chips. However, no one had yet been able to figure out how to fit 2300 transistors of random logic -- the minimum number for a simple CPU -- into a chip small enough to be manufacturable at low cost, with the necessary high speed to be useful, and with the low power dissipation to fit into existing packages.
Thus the real innovation in the microprocessor was its implementation into a single silicon chip since everything else had been done before. And this was done by Faggin without any meaningful help from Ted Hoff and Stan Mazor, who were not chip designers.
The only help Faggin had was from Masatoshi Shima, the Busicom engineer who came to Intel to check on the progress a couple of days after Faggin was hired, only to find that no progress had been made for the previous six months. Given the delay in the project and the lack of any Intel engineer that could help Faggin, Shima was allowed to stay for six months and assist Faggin to expedite the project. However Shima knew very little about integrated circuits and, although he was very helpful, all creative decisions were done by Faggin. Faggin’s boss, Leslie Vadasz, was then so preoccupied with the design of the 1103, the first 1024-bit dynamic RAM, considered the future of Intel, that he could not provide technical supervision to the MCS-4 project. After completing the 4004, Faggin directed the implementation of the 8008 and he also conceived and defined the architectures of the 4040 and of the 8080, the most successful of all early microprocessors.
Doubting Architects
“Ted and I thought it [the 4004] was a little too aggressive and we weren’t so sure it could be done, so we started with another chip, called the 4005, and it was a joint project with MIL, which was an affiliate of Intel in Canada. And so we defined the architecture to be much, much simpler than the 4004 and the idea was the Canadian company MIL would actually design the chip and we’d provide the memories……it turned out they could never make the 4005 so it went away.”
Stan Mazor, in a June 9, 2000 Silicon Genesis interview with Rob Walker
"More than ten years ago, when I returned to Intel, I asked Ted Hoff why he left the project. He said, 'Oh, I was busy with some other things.' That was true. Also he said, 'I thought we could not make it [the 4004].'"
Masatoshi Shima in a 1994 interview with William Aspray
Hoff and Mazor were actually not sure that the 4004 could be done; that’s why they came up, a few months after Faggin joined Intel, with the architecture for the 4005 CPU, a simpler chip than the 4004, that was given to MIL in Canada for its development . But the MIL engineers could not make the chip, demonstrating that even a simpler chip than the 4004 was far from a routine undertaking. Hoff and Mazor also doubted that the 4004 could be useful for applications other than calculators, cash registers and the like. They thought that only the 1201 (architected by Computer Terminal Corporation, and later called 8008) had a sufficiently general-purpose architecture to be useful for a variety of applications. Faggin, upon completion the 4004 project, demonstrated that the 4004 could have many applications in control systems, and urged management to introduce the 4004 to the general market.
Microprocessors Needed the SGT
Another example to show how necessary was Faggin’s methodology to designing Intel’s early microprocessors is the 8008, whose architecture was originally developed by Computer Terminal Corporation (CTC). The chip definition of the 8008 (originally called 1201) was already under way when Faggin joined Intel but the project, assigned to Hal Feeney, a random logic chip designer recently hired from General Instrument, did not go very far because Intel at that time had no methodology and no library of circuits for random logic design. The 8008 was then suspended and was resumed in January 1971, under Faggin's direction, after he had successfully designed the 4004.
A further example is Texas Instruments’ first CPU-on-a-chip, which happened to be a second-source for the 8008 commissioned by CTC to TI. Announced in the press in mid-1971, only a few months after the 4004 completion, this chip never functioned and it was never commercialized. It was made using MOS metal gate technology by a company that had many years of experience in complex random logic design. Yet the chip size was twice the size of the 8008 to perform exactly the same function. The operating speed and the power dissipation were never revealed.
The 4004 Established the Design Style for Early Microprocessors
After the 4004 was completed, other engineers, within and without Intel, could learn the techniques used by Faggin by examining the design under a microscope. Faggin’s design style was used for all the other early Intel and Zilog microprocessors.
Summary
“The silicon design is the essence of the first microprocessor”
Federico Faggin, "The crucial role of silicon design in the invention of the microprocessor"
The 4004 microprocessor was history’s first random logic integrated circuit made with the Silicon Gate Technology (SGT). The 4004 was also the most advanced MOS integrated circuit ever done up to that time. The making of the microprocessor demanded not only extraordinary creativity and skills from a MOS designer, but also the intimate knowledge of the new SGT that only its developer could have had. In addition, bringing to successful completion a project which was beyond-state-of-the art and needed to be done in 10 months because of previous unmet commitments with the customer required great courage, motivation, management skills, and sustained hard work.
|
|
