Federico Fagin joins Intel to turn Ted Hoff's vision into silicon reality. Less than a year later, he and his team created the 4004 microprocessor, which was introduced in November 1971. The first serious problem for Fajin was the development of a methodology for using the new silicon shutter technology. This allowed us to create complex schemes in a different way. Since no one had done anything like this before, we had to start from scratch. And Federico Fagin succeeded - he independently developed the chip in just 9 months of 1970. This is a fantastically short time compared to the Intel processors of the next generations. For example, it took 100 man-years to create a 32-bit chip.
Design
The silicon shutter design created by Federico Fagin made the first microprocessor a reality in 1971. It was unusual since integration of such complexity had never been achieved before. Fagin was able to develop an Intel 4004 processor only thanks to its innovations in MOS technology for the production of integrated circuits (ICs). The bootstrap and hidden contact became the ideas that formed the basis of the design methodology that he first applied, saved the Hoff architecture and implemented it in 1970. Without it, it would not be feasible, because the result would be too slow and costly to have practical application . Thus, the invention did not consist in developing a model of a simple CPU, which was not lacking at that time, but in creating and implementing a technology that for the first time allowed placing all the functional blocks of the processor on one chip.
Ted Hoff hoped that his proposed architecture and set of commands could be placed on the same chip. However, he could not assess the feasibility of the project or implement it, since he was not an MOS developer. It was Federico Fagin who invented the design and layout of 2300 transistors of arbitrary logic on a chip measuring only 3x4 mm, inexpensive, with 5 times the speed of work and twice as high density of the elements than the existing MOS technology at that time.
Fajin's methodology was a breakthrough and was used in all of the company's early microprocessors. The Italian engineer led the project to a successful conclusion and played a role in promoting the new Intel processor, demonstrating to company management that the chip can be used not only in calculators.
Specifications
The Intel 4004 processor specifications are as follows:
- Crystal area: 12 mm 2 .
- Maximum clock frequency: 740 kHz.
- Cycle Time: 10.8 ฮผs (8 cycles / instruction cycle).
- Command execution time - 1 or 2 command cycles (10.8 or 21.6 ฮผs), 46300โ92600 commands per second.
- The addition of two 8-digit numbers (32 bits each) takes 850 ฮผs, i.e. 79 cycles of instructions, about 10 cycles per decimal digit.
- Separate storage of programs and data. Unlike designs based on Harvard architecture using separate buses, the 4004 has one multiplexed 4-bit bus for transmitting 12-bit addresses, 8-bit instructions, and 4-bit data words.
- Direct addressing of 51220 bits (640 bytes) of RAM, organized in the form of 1280 4-bit "characters", of which 1024 represent data and 256 represent status.
- Direct addressing 32768 bits of ROM (4096 bytes).
- A set of 46 teams (of which 41 are 8 bits wide and 5 to 16 bits wide).
- 16 registers of 4 bits.
- An internal stack of routines 3 levels deep.
Busicom order
A computer with a stored program, which was used as a calculator in the 1950s and 1960s, was one of the best achievements of the post-war era and was familiar to all engineers working in the semiconductor industry.
In 1969, Japanese calculator maker Busicom turned to Intel to translate their logical design into a series of calculators in silicon. Their approach repeated the implementation of the world's first desktop programmable calculator Olivetti's Programma 101, presented at the World Fair in New York in 1965 and went on sale that same year. Programma 101 had a CPU (central processing unit) and ROM (read-only memory) with sequential read and write, which were made of discrete components. Busicom proposed a similar architecture, which provided for the implementation of the processor on three MOS chips, ROM and register on two more, with two other input-output chips.
MCS-4 Architecture
Head of Applied Research Ted Hoff acknowledged that Busicomโs design complexity was the use of serial memory, and since Intel was developing its first dynamic RAM (random access memory), he saw that the design could be greatly simplified using a traditional and more universal computer architecture based on on RAM. With the help of Stan Mazor and through interaction with Busicom engineers, including Masatoshi Shima, Hoff formulated the MCS-4 architecture, reducing the design from 7 to 4 chips. Hoff believed that the CPU could be implemented in a single 4-bit microprocessor, but neither he nor Mazor were MOS developers, and Intel did not have specialists who could create complex ICs with an arbitrary logical structure. Therefore, Hoff's proposal stood idle for about 6 months, until in April 1970, Federico Fagin, who led this project, was hired.
Innovative technology
Federico Fagin joined Intel from Fairchild Semiconductor, where in 1968 he developed the MOS technology with silicon gates and based on it the world's first commercial integrated circuit 3708. The technology was adopted by Intel and subsequently throughout the world semiconductor industry, and during 40 years was the basic structure used in almost all microcircuits. Gordon Moore, co-founder of Intel, acknowledged that this step was the main component of her early success. MOS technology with silicon gates ensured the competitive characteristics of the Intel 4004: the chip was about 5 times faster, had 100 times less junction leakage and contained 2 times more logic transistors of any type compared to a chip of the same size made with aluminum gates, and dissipated equal power. This allowed the creation of the first commercially successful dynamic RAM, image sensors with a CCD matrix (charge-coupled devices), non-volatile memory devices and microprocessors. For the first time in history, the processor contained all the components of a general-purpose computer.
Create a new design and layout
Ted Hoff was not a designer of MOS circuits. His role was to create architecture and further support products. After defining a set of teams, the project was transferred to a team of MOS developers led by Federico Fagin. The work was carried out very quickly, and in about 9 months 3 main chips were created. The last of them in January 1971 appeared Intel 4004 microprocessor.
According to Stan Mazor, Fagin's merit was that he carried out engineering design, and Hoff - in creating an original concept and architecture. Mazor himself was a kind of intermediary who helped as he could and did what he could.
Federico Fagin has developed a methodology for designing silicon-gate MOS structures used in arbitrary logic circuits. This was necessary because the new technology required a different design, and especially the layout.
According to Fadzhin, he decided that instead of separately designing the logic, and then the circuit, they should be done together on one sheet. In this case, it is necessary to take into account the location as much as possible, so that the conductors and transistors are located as close as possible to the final layout. Obviously, for this it was necessary to carry out general planning of the chip beforehand in order to know the location of the various blocks. It was then that he refined the methodology for creating this type of circuit.
Designing a new Intel processor and managing the MCS-4 project from the beginning of development to production could only be mastered by someone who was able to innovate in process technology, prototyping of microcircuits, circuitry, logic design and computer architecture. Fagin gained such skills and knowledge through his education and work experience before joining Intel. After graduating from a technical university in Vicenza (Italy), he participated in the development and creation of a small transistor experimental computer with magnetic core memory at Olivetti in Borgolombardo (Italy) at the age of 19. He then graduated with honors from the University of Padua and began developing MOS technology by creating 2 commercial microcircuits when he worked at SGS-Fairchild (now ST Micro). In 1968, he was sent to Fairchild Semiconductor R&D in Palo Alto, California, where he created silicon-coated MOS technology and others.
Real innovation
The conceptualization of the first processor, which became Hoff's main contribution to the 4004 project, took place in other companies. The same conclusion was reached by several groups independently of each other. Therefore, the main thing in the invention of the microprocessor was the creation of an economically viable product. Only one person in the world knew how to take the next step and translate architecture into working design. It was Federico Fagin. Without it, the first microprocessor would never have been built. Back at Fairchild, he invented the technology that formed the basis for future devices. After starting work at Intel, he corrected the errors of the missing Hoff, and then made the first Intel 4004 chip, after which he led the development of 8008 and was the chief architect of 8080.
At that time, engineers knew how to create small computers, make logical CPU designs, and create programs. The idea of โโa microprocessor, that is, of placing a universal computer on one chip, was also in the air. Some architectures have already been implemented on several MOS chips. However, no one knew how to install 2300 arbitrary logic transistors - the minimum amount needed for a simple processor - into a chip that was small enough to be cheap, work at high speeds, and power dissipation sufficient to fit in existing enclosures.
Thus, the real innovation in the microprocessor was its layout on a single silicon chip, since everything else had been done before. And Fagin succeeded without any significant help from Ted Hoff and Stan Mazor.
The only one who helped him was Busicom engineer Masatoshi Shima. He came to Intel to check on order progress a couple of days after Fagin was hired. He realized that no progress had been made in the previous 6 months. Given the delay in the project and the absence of any Intel engineer who could help, Sime was allowed to stay for 6 months to expedite the work. However, he knew little about integrated circuits and, although he was very useful, Fajin made all the creative decisions. The head of the latter, Leslie Wadash, was so preoccupied with the design of 1103 (the first 1024-bit dynamic RAM, considered the future of Intel), that he could not provide technical control over the MCS-4 project. After success with 4004, Fagin led the implementation of 8008, as well as conceived and determined the architectures of the most successful of all the first processors - 4040 and 8080.
Developer Doubts
According to Stan Mazor, he and Ted Hoff believed that the Intel 4004 was too aggressive. They were not sure that it could be done, so they started with another chip called 4005. It was a joint project with MIL, which was Intel's partner in Canada. They identified a much simpler architecture than 4004. A Canadian company had to develop a chip, and Intel - to provide memory. It turned out that she could not make 4005.
Hoff and Mazor were unsure of the possibility of implementing 4004 in 1994, which is why a few months after Fajin joined Intel, they created a simpler 4005 architecture and handed it over to the Canadian company MIL for development. But the MIL engineers failed to make a microprocessor. It became clear that even creating a simple chip was far from routine work. In addition, Hoff and Mazor doubted that the 4004 might be useful for applications other than calculators, cash registers, etc. They thought that only 1201, and later 8008, would have a sufficiently universal architecture to be used in various applications. After the completion of the 4004 project, Fagin demonstrated that the microprocessor can be used in various control systems and urged management to bring the Intel 4004 to the market.
Failures with 8008
Another example that proves how necessary Fajin's methodology was is Intel 8008, whose architecture was originally developed by Computer Terminal Corporation (CTC). Work on the chip, originally named 1201, began before Fagin joined Intel, but the project assigned to the developer of arbitrary logic processors, who moved from General Instrument, did not go far because at that time there was no methodology and libraries circuits. Work on 8008 was suspended and resumed only in the year Intel 4004.
Microprocessor TI
Another example is the first single-chip CPU, which became the second source for the 8008 ordered by CTC from Texas Instruments. Announced in the press in mid-1971, just a few months after the successful creation of 4004, this processor never worked and never sold. It was created using a MOS technology with a metal shutter by a company that had many years of experience developing ICs with complex arbitrary logic. Compared with the Intel-8008 processor, the size of the TI chip was twice as large, providing the same functions. Speed โโand power dissipation have never been made public.
Exemplar
After the 4004 project was completed, other engineers, both inside and outside Intel, were able to study the methods used by Fajin by examining the design under a microscope. The same style was used in all other early Intel and Zilog microprocessors.
Finally
4004 was the first ever processor to be made using silicon shutter technology. It was the most advanced integrated circuit of the time. Its creation required not only extraordinary creative abilities and skills from the designer, but also a deep knowledge of the new technology that only its developer could have. In addition, for the successful completion of the project, which was required to be completed in 10 months due to previous outstanding obligations to the client, great courage, motivation, management skills and steady hard work were needed.