Intel’s forgotten past: Wayne D. Pickette, African American father of the microprocessor

Intel’s forgotten past: Interview with Wayne D. Pickette, African American father of the microprocessor

By Philip S. Stinard

Microprocessors have revolutionized the world of technology, allowing computers to evolve from room-sized devices costing millions of dollars, accessible only to research institutions, to small, powerful, inexpensive devices that fit on one’s lap or in the palm of one’s hand. They are a part of everyday life for millions of people, serving as controllers for devices ranging from cell phones and DVD players to automobiles and airplanes. The microprocessor is arguably one of the greatest technological advances of the Twentieth Century.

The official history of the development of the first microprocessor (“History of the 4004”, IEEE Micro, December 1996, pp. 10-20) leaves the reader with the impression that Federico Faggin, Marcian (Ted) Hoff, Stanley Mazor, and Masatoshi Shima were the sole designers and inventors of the 4004 microprocessor. These men, however, stood on the back of a giant, an unassuming young black inventor by the name of Wayne Douglas Pickette. Wayne’s contributions to the development of the 4004 have largely gone unrecognized by his former employer, Intel. Wayne has maintained collegial relations with former Intel coworkers, and when they speak, they often reminisce about those early days of microcomputer development. Wayne’s coworkers recognize his accomplishments; Intel, however, remains silent.

Wayne Pickette was a self-taught childhood electronics prodigy. Wayne recalls reading 2,000 words per minute with 90% retention by the time he graduated from San Martin (California) Elementary School. His early interests included designing a computerized video system for sharing library materials over TV, which he worked on during his high school years; and robotics. In 1967, at the age of 17, he purchased a PDP-8/S computer in order to perform calculations for his projects. The need for smaller computers for controlling self-contained humanoid robots led to the first design for a microcomputer, a PDP-8 with all of its logical functions placed on a single microchip. Wayne presented his design to Fairchild Semiconductor in February of 1968, but the company rejected his idea as “crazy.”

Wayne worked for IBM during the summer of 1968 performing logic design for IBM's Project Winchester, which led to the creation of the hard drives commonly found in today's personal computers.When Wayne started working for Intel in Spring of 1970, he brought his ideas of computer architecture and miniaturization to the project to develop a business calculator, which then became the first microprocessor, the 4004. The following interview captures Wayne’s reminiscences about his crucial role in the development of the microprocessor. The author verified Wayne’s statements with a trusted Intel employee who wishes his name to be kept private for fear of reprisals.

How did you get the idea of putting a computer on a chip?

I had a science fiction book lying on the table next to my bed that I hadn’t read yet, called “I, Robot.” When I got about two thirds of the way through the book, a few lights started to click on in my head, and I said, “You know something? This is not science fiction. This thing is possible.” And I started making a list of stuff that had to occur before you could make a robot. The list included power, skeletal, brain, eyes, the whole thing. When I got down to the computers, I figured it would take 300 computers inside this robot to make it work in a human-like manner, all of them working together, a minimum of 300 computers.

And you basically had details as to what their functions would be.

Yes, specifications. Not all of them the same, but 300 computers. I figured that if I wanted this robot to be normal, without having a semi-trailer following it full of air-conditioning and generating equipment, I had better figure out a way to get the computer down to thumbnail size from the size that was sitting on my desk, which was 100-some pounds, a monstrosity, a PDP-8/S talking to me through a teletype.

And, after all that got through popping through my head—I already knew logic—I made this original drawing. [Brings out 1968 drawing of a circuit diagram, with neat penciled handwriting on thin, delicate crumpled paper.]

That’s my organization. That was the diagram that I presented to Intel. It’s a PDP-8/S normalized down to 19 integrated circuits, using serial communications between the registers, and parallel communications through the registers, which means that it uses shift registers and latches and stuff, which are small logical components. The core of this thing is a 74181 arithmetic element. And it turns out that the instruction set for this 74181 matches the PDP-8 exactly, so I didn’t need an instruction decoder—it was already built in. The general idea was to shift data into the accumulator, do an operation on it, shift it out and put it back in memory, shift data and program into the appropriate registers, perform an operation, shift the data back out to memory. Basically, just what the PDP-8/S does. The only difference is that it was miniaturized down into a little board about that big. [Holds hands about a foot apart.] And it ran off of 5 volts, and it talked to a teletype though a serial interface.

When did you do this diagram?

Late ’67, early ’68. Well, when I called up Fairchild after I finished this, they told me everything had already been done in computers. They said, “What you want to do is impossible. By the way, you’re crazy.” They hung up on me.

Why did they think this was crazy? Did they give any reasons? Are these components that already existed at the time, and they just had to manufacture them?

No. This was putting something on an integrated circuit, whereas before, the only thing that had been put on an integrated circuit was logic functions. I wanted to put the arithmetic element, instruction logic, timing logic and register logic, taking all of these different IC’s and combining them on one chip. Inside here, there are about 1,840 gates, and the limit at that time was 3,000. So, it was within the realm of creation.

[Conversation shifts to a discussion of the 4004 microprocessor.]

The 4004 wasn’t like this. The 4004 was dictated by [the Japanese company] Busicom, because Busicom wanted a calculator, and Busicom’s Masatoshi Shima wasn’t even about to allow anything not meant for a calculator in his calculator chip. Yung Feng, working for Intel’s Federico Faggin, gave me the first working 4004, 4002 & 4003 chips in late summer 1970. It was a real war to get that thing introduced; one reason was that Busicom went bankrupt. Intel management gave up on the product. They basically said that it was too technical, and no engineer would fool with it. Ted invited me to dinner over at his house, and we talked about the microprocessor. He gave me a copy of a book called “Future Shock,” which depicts society’s resistance to change. He said he hoped that would tone me down a bit—it didn’t.

When we had to vote [on whether to introduce the 4004 at the 1971 Fall Joint Computer Conference in Las Vegas], I was a 21-year-old black kid sitting in the conference room with all these high mucky-mucks, [Intel President] Dr. Robert Noyce, [Vice President] Gordon Moore, [Director of Operations] Andy Grove, [4000 series creator] Federico Faggin, [Intel Memory Systems Manager] Torstein Lund, and Ed Day. Ed Day was the head of marketing at the time. A couple of other managers were there, there were at least nine people at this table. Ted Hoff, myself, Stan Mazor, Hal Feeny. Shima wasn’t there—he wasn’t an Intel person at this time.

Anyway, when we voted whether or not to take it to the show, I was the only aye at the table. I was completely flabbergasted. I stood up, banged my fist on the table, called them all blithering idiots, and stormed out of the conference room.

Why didn’t they want to show the microprocessor in Las Vegas?

They didn’t think it was going to be successful. They thought it was too limited, that it was too complicated. No engineers would even look at it.

I asked for a private meeting with Dr. Noyce. He said, “What is it you want?” I said, “I came to plead with you in the case of the 4004 Demo System. I want to show it in Las Vegas.” And when we talked, he finally said, “All right, you can take it there, but these are the conditions. If everything turns out okay, you don’t have anything to worry about. If it doesn’t turn out okay, YOU and IT can stay in Las Vegas.” I said I would happily take those terms.

It was all at my own expense. They did not give me any money advance. I jumped into my car and headed for Las Vegas, probably about 3 in the afternoon. I made it to Las Vegas sometime around 4 in the morning. We got set up on a table probably not much bigger than a lamp table. Intel Memory Systems had the back room in this suite, and I was out in the front room, on this little table, while Stan Mazor went down and distributed some leaflets on the floor. Then, people started to show up: About 150 people the first day, about 500 people the second day, and over a thousand on the third day. I mean, I had a line going out the door, all day long. The door was open all the time, but the line was going out the door. And I was taking them one at a time. One of the people in line on the third day was Dr. Noyce. And he said, when he got to me, “You’re doing a good job.” He said, “I don’t care what you do between Thursday evening [that day] and Monday morning, but you’d sure as hell better have your butt back in Santa Clara Monday morning,” and he handed me an envelope. I didn’t look in the envelope. I just said, “Thank you,” and stuck it in my jacket pocket, and went on to the next person. Later, I opened the envelope and found that he had given me $2,500.

I learned many years later from Paul Metrovich, another coworker, that they intended to fire me when I came back.

Why?

Because I called it a microprocessor, and I took it there. So I came back with success. Dr. Noyce had already given me $2,500, and they were still intending to fire me. Paul said that the phone was ringing off the hook Monday, so they forgot to fire me. [Laughter.]

I found this paper from the IEEE Micro Journal on the history of the 4004, by Faggin, Hoff, Mazor, and Shima.

Oh, I’ve read it.

Do you feel that they’re rewriting history?

They rewrote history, because very simply, they couldn’t adjust for me, and I know that when I walked into Intel, they had no inclination of building this thing the way it was built. Federico hadn’t even been hired yet. He was hired after me. When Tor Lund came on his first visit to the company, he was told by Ted that the most amazing thing had happened. Ted had always wanted to build a computer, but he never thought that he’d be putting a computer on a strip of silicon! Tor said I was already there at the company the day he arrived in March, 1970, and Ted said this to him. What does that tell you? In fact, Ted was still overwhelmed with all that had occurred as he attempted to convince Tor to join Intel!

They were going to build a set of calculator chips, custom calculator chips for Busicom, and Federico’s time and work estimate was much too expensive to accomplish these four chips. He had brought this information to Ted’s office while I was sitting in there talking to him about this computer on a chip. Ted looked the estimate over for a couple of seconds. After Ted finished looking it over, he looked up and said, “This will not work out. It’s much too expensive.” And I knew it was the calculator project that Federico was working on. So I said, “Ted, we’ve just been talking about a general-purpose computer on a chip. Why can’t we make this computer on a chip and program it to act like a calculator, along with other applications? In fact, it would be cheaper, because we have more applications.” And he said, “That’s a thought I’ll take into consideration. I have to go to a meeting, you wait, sit right there.”

Ted had to go to this speakerphone meeting in Gordon Moore’s office between Busicom and him, with Shima nearby, and they were talking about the calculator project. Shima was mad, because they hadn’t done anything. I mean, Federico was just starting to look at the chip design. He hadn’t done anything, and Shima was expecting all these chips to be done already. Shima was walking on the ceiling. But anyway, Ted hit them with this intriguing idea, and they went for it, and Busicom agreed to pay half the cost of the creation of the 4004. Ted came back to his office with me waiting there; his eyes were wide behind his glasses! He said one sentence, “They went for it!” At that point, it was defined to have 90 percent of the same instructions that the calculator chip would have had. It was a software investment, so what Shima had already created would not be lost. Like I said, my attempts to get him to include an Exclusive OR instruction fell on deaf ears. He wouldn’t have anything to do with it. He knows of my existence, if only for those three or four days that we were defining the 4004 and making sure that everything was copasetic, and getting descriptions over to Federico, who immediately began to work. Of course, Federico said that all he got was scraps of paper in his interview, he said he got three scraps of paper with some block diagrams on them. Now, if they’d been working on it for two years, how come he only got three scraps of paper with some diagrams on them? Come on! Right there in his interview, he puts the lie to the idea that Intel and Busicom had been working on the 4004 for three years. Moreover, even if Intel was formed in late 1968 and Busicom began discussions in 1969, my diagram was completed in early ’68 and the prototype logic hardware was completed while INTEL was still forming. Sorry, guys, I was there first.