He invented the world's first microprocessor

The rising sun has just reached the foothills of Silicon Valley, but Marcian E. (Ted) Hoff Jr. is already scrambling for electronic parts, digging through piles of dusty circuit boards. This is Foothill College's monthly flea market and he rarely misses it.

Ted Hoff is part of the legend of the electronics industry. While working as a research manager at Intel Corporation in Mountain View, he realized that silicon technology had advanced to the point where, with careful design, a complete central processing unit could fit on a single chip. Working with Stanley Mazor and Federico Faggin, he created the Intel 4004, the first commercial microprocessor.

But for Hoff, microprocessors were just a blip in a long fascination with electronics. His passion for the field took him from used electronics stores in New York City to elite university labs, through the ferocious early days of the microprocessor revolution and the turmoil of the video game industry, and ultimately to the job he has today: a high-tech private investigator ( high-tech private eye).

Hoff discovered early in his childhood that the best way to feel less like a child—and more empowered—was to understand how things worked. He began his exploration in chemistry. By the age of 12, he was working in the electronics industry, using parts ordered from Allied radio catalogs, shortwave radio kits, and leftover relays and motors salvaged from the trash of his father's employer, the General Railroad Signaling Company in Rochester. thing. Then in high school he built an oscilloscope using mostly second-hand components, a feat he parlayed into a job as a technician for the General Railroad Signaling Company.

Hoff returned to that job during his undergraduate studies at Rensselaer Polytechnic Institute in Troy, New York. A few summers ago, Hoff went into the General Railroad Laboratory and discovered that two of the researchers' best oscilloscopes were broken. He would tinker with state-of-the-art Tektronix 545s and then move on to more interesting things, like inventing audio railroad train tracking circuits and lightning protection devices, receiving two patents before he was a teenager.

The best thing about the job, Hoff recalls, was that it gave him access to components that were beyond the budget of most engineering students in the 1950s—such as transistors, even the newly introduced power transistors. He wrote an undergraduate thesis on transistors used as switches, for which he received a cash prize that he soon used to build his own Heathkit oscilloscope.

Hoff liked Rensselaer's engineering programs but disliked the narrow focus of the college itself. He wanted to broaden his horizons intellectually and geographically (he had never been more than a few miles west of Niagara Falls), so he chose graduate school at Stanford University in California. While pursuing a Ph.D. There he studied adaptive systems (known today as neural networks) and, together with his thesis advisor Bernard Widrow, was awarded two patents.

His Intel colleague Stanley Mazor, now a training manager at Synopsys Inc. in Mountain View, Calif., recalled meeting Hoff in his Stanford lab. "He had a toy train moving back and forth under computer control, balancing a broomstick," Stanley Mazor said. "I think he was an eccentric inventor, a mad scientist."

After receiving his degree, Hoff stayed on as a postdoctoral researcher at Stanford University for six years, continuing his research on neural networks. Initially, his team trained the network by using a device whose resistance changes with the magnitude and direction of the current. It consists of a pencil lead and a copper wire placed in a solution of copper sulfate and sulfuric acid, which they call a memristor.

The group soon acquired an IBM 1620 computer, and Hoff had his first experience with programming—and with playing against the system. He had to deal with officials at the campus computer center, who believed that all computers should be in one place, managed by experts who handled boxes of punched cards delivered by researchers. The idea that researchers should program computer systems interactively is anathema to them.

"One result of all the work we've done on microprocessors that has always pleased me is that we've taken computers away from those people," Hoff told IEEE Spectrum.

By 1968, student hostility to the Vietnam War was growing, and life for campus researchers like Hoff, who relied on government funding, might look uncomfortable. Hoff was already considering the possibility of industrial work when he received a call from Robert Noyce, who told him that he was starting a new company, Intel Corporation, and heard that Hoff might be interested in a job interest. He asked Hoff where the semiconductor integrated circuit business would find its next area of ​​growth. "Storage," Hoff replied.

That was the answer Noyce had in mind (Intel started out as a memory maker) when he hired Hoff as a technician that year, Intel's 12th employee. Hoff worked on memory technology and soon received a patent for a cell used in MOS random access integrated circuit memory. After becoming an applied research manager, he had his first professional customer contact.

"Engineers tend to have a very cavalier attitude toward marketing, but I've found that if you keep your eyes and ears open in the field, you learn a lot," Hoff said. "It's very helpful to try to understand the problems people are trying to solve. Back to People who are in the lab but don’t have that connection are at a disadvantage.”

One group of customers Hoff contacted was from Tokyo-based Busicom Corp. Busicom hired Intel to develop a custom chip for the low-cost calculator and sent three engineers to Santa Clara to work on the chip design. Hoff was assigned to look after them, get them pencils and paper, tell them where the restaurant was—nothing technical.

But the technical part of Hoff's brain didn't have an off switch, and he quickly concluded that the engineers were going in the wrong direction. Twelve chips, each with more than 3,000 transistors and 36 leads, were used to handle different elements of the calculator's logic and control, and he estimated that the packaging cost alone would exceed the calculator's target retail price. Hoff was struck by the complexity of this tiny calculator compared to the simplicity of the PDP-8 minicomputer he was currently using on another project, and he concluded that a simple computer that could handle the functions of a calculator could Uses about 1900 transistors. He believes that, given Intel's advanced MOS process, all of this can be put on a single chip.

Busicom engineers had no interest in abandoning their design in favor of Hoff's unproven proposal. But with Noyce's support, Hoff began working on the project. He was soon joined by Mazor, then an Intel research engineer, and the two followed Hoff's ideas and developed a simple instruction set that could be implemented with about 2,000 transistors. They showed that a set of instructions could handle decimal addition, scan the keyboard, maintain the display, and perform other functions assigned to separate chips in the Busicom design.

In October 1969, Hoff, Mazor and three Japanese engineers met with visiting Busicom management from Japan and described their different approaches. Busicom managers chose Hoff's approach in part because they knew the chip could have applications beyond calculators, Hoff said. The project was given the internal designation "4004".

Federico Faggin, now president and CEO of Synaptics Inc. in San Jose, Calif., was commissioned to design the chip and within nine months came up with a working prototype chip of a 4-bit, 2300-transistor "microprogrammed computer." ” Busicom received the first batch of equipment in February 1971.

Faggin recalled that when he began implementing the microprocessor, Hoff seemed to lose interest in the project and had little interaction with him. Hoff was already working on his next project, a preliminary design of an 8-bit microprogrammed computer for Computer Terminals Corp. in San Antonio, Texas, which was designated the 8008. Hoff always "had to work very cutting edge," Faggin told Spectrum. "I could see his nerves, always being at the forefront of what was going on."

Mazor recalled that in the early days of Intel, Hoff had many ideas for projects, many of which, while not commercially successful, proved to be prescient: a camera that could work like a digital camera and capture images in memory. RAM chips, video games and moving spaceships, devices for programming erasable programmable ROM, and computer-aided design tools for logic simulation.

Intel's marketing department estimates that total [microprocessor] sales may be only 2,000 chips per year.

At the same time, the microprocessor revolution was accelerating, albeit slowly. Hoff joined Faggin as a microprocessor evangelist, trying to convince people that general-purpose microcontrollers made sense. Hoff said his toughest sales were to Intel's marketing department.

“They were pretty hostile to the idea,” he recalls, for several reasons. First, they figured that all the chips Intel could make would be sold to one company within a few years, so there was no point in marketing them to other companies. Second, they told Hoff, "We have diode salesmen selling memory like crazy. Do you want them to sell computers? You're crazy." In the end, they estimated that total annual sales might only be 2,000 chips.

But word got out. In May 1971, the product was mentioned in an article in Datamation magazine, and the following November, Intel produced the first advertisement for the 4004 CPU and placed it in Electronic News. By 1972, stories about the wonders of the microprocessor began to appear regularly in the media, and Intel's competitors followed suit with their own microprocessor products.

One step Hoff did not take at that time was to apply for a patent, although he had successfully patented several inventions. (Later, together with Mazor and Faggin, he applied for and received a patent for a "memory system for multi-chip digital computers.")