The Early Days of AMD

to that episode. So today we're going to learn about a m D, where it came from, and its role in the tech industry in general, because it's a pretty interesting story and I knew bits and pieces of it, but like any subject for tech Stuff, as I dive into the research, I learned way more than I had ever learned before, and I end up going down lots

of rabbit holes. And of course, to understand the history of a m D, WILL need to talk about a couple of other companies first, as they would lay the ground for a m D and a couple of other really big organizations in the microchip industry. And at the center of this prehistory for a m D was William Shockley. Now I've talked about Shockley a few times in past

episodes of tech Stuff. He was, without a doubt a brilliant engineer, though in other ways he was also a deeply flawed human being who harbored some truly horrible traits. And you might think that such a comment isn't really germane to the discussion of technology, But as it turns out, Shockley's personality plays just as important a role in the

emergence of a m D as his experimental work did. Now, Shockley worked at Bell Labs, specifically in their solid state physics research department, and he would play an important part in the development of the transistor, which was an alternative to the vacuum tube technolology that had come before and had proved to be a limitation on technology in general and electronics in particular. So, hey, what the heck do vacuum tubes do? And what do transistors do? Was the

big deal with them? Well, vacuum tubes are also known as thermionic tubes, and they look a lot like light bulbs. They are glass with a filament inside of them, and they work on the principle that if you add energy to a metal, as in, if you heat up that metal, the energy will cause the metal to eject electrons. And you probably remember this from science classes. Electrons inhabit an energy level orbit around the nucleus of an atom, and if you pour energy into that atom, it will boost

the electron to higher energy level orbits. And if you pour in enough energy, you'll cause the electron to leave its atom entirely. John Ambrose Fleming would build the first vacuum tube like device, which he called an oscilla. Should valve way back in four His device had two electrodes, and that would be our good old buddies, the cathode and the anode. The cathode is the negatively charged electrode and is thus the source for electrons flowing through a system.

The anode is the positively charged electrode, and it accepts electrons. Though we also need to remember that current direction is traditionally thought of as the direction of the flow of positive charge, So in other words, the flow of current is actually in the opposite direction of the electron movement. But that's really been Franklin's fault. Will just skip it

for now. Fleming demonstrated that by heating the cathode, which is at one end of an enclosed glass tube inside of which Fleming had induced a vacuum, So there was a vacuum inside the tube. When you heat it up the cathode, it would give off electrons and those electrons could flow across the gap inside the tube between the cathode and the anode. They could leap through that vacuum.

This type of component is called a diode because it only allows electrons to travel in one direction, and it's useful if you want to create a more complex device that works based on where electrons can and cannot go. Now, another smarty Pants named lead to Forest would add a third electrode to this system, and it's called a control grid. And the control grid can serve as both a control switch for how many electrons can travel from cathode to

anode as well as an amplifier. And you can influence the current flowing through the vacuum tube by controlling the amount of voltage going into the control grid itself. So with a tiny change of voltage to the control grid, you can have a larger change manifest itself through the overall circuit. This was a triode and other variants would follow, making complex electronics possible. But while they were useful. Vacuum tubes are also really large, and they give up a

lot of heat too. By the nineteen forties, physicists were looking at an interesting category of materials that might serve as a substitute for these large, bulky hot vacuum tubes, and that material category is called semiconductor material and semiconductors are why we have the electronics of today in the form that they are in. That being said, vacuum tubes still are useful. They are still used in electronics today, particularly in things like music amps. But that's a discussion

for a different episode. So what about a semiconductor, because that is the bread and butter of companies like a M D well. A conductor, at least for the purposes of this episode, is a material that allows for the passage of electrons through that material. It conducts electricity. The opposite type of material is called an insulator. That's material that resists the flow of electrons through it. Semiconductors have

conductivity that lies between a conductor and an insulator. In addition, the conductivity of a metal decreases as the metal's temperature increases, or another way to put it is that a metal's resistance to electricity increases as the temperature also increases. Semiconductors are actually the opposite. Their conductivity increases as temperature increases. There's one part of this picture that I need to

really cover, and it's called doping. Now. Doping is when you take an otherwise pure substance and you add small amounts of some other material to it, so it becomes impure because you no longer have one type of atom in that substance. With semiconductors, we typically talk about silicon, although that would actually come a little later in the development of the transistor. But pure silicon is an insulator.

So if you have a bunch of silicon atoms, they form silicon crystals, and the crystals all bind perfectly with each other. They have perfect covalent bonds between the atoms, so means there's no free electrons available to move around. So if you hit this stuff with a free electron, that free electron can't shake anything else loose. It's all kind of locked in, so it insulates electricity. But by adding small amounts of other materials like arsenic, you can

add in some free electrons. See arsenic has some extra electron or an extra electron compared to silicon. So if arsenic is binding with silicon atoms, then you end up with this extra electron that's not bound to anything, and it will allow for some level of conductivity, all dependent upon how much arsenic per silicon you have in that mix.

This would be called in type doping because you're adding electrons to the actual material and electrons have a negative charge, thus in type doping, or you could dope the silicon with something else like boron or gallium, which would mean the crystal would actually have a few free spaces for electrons or holes. So instead of having these perfect covalent bonds that are tight all the way across the entire material, you would have these little holes that could accept an

incoming electron. This is called P type doping. By putting in type and P type silicon together, you can create a diode, and by adding a third layer so that you either have N P N or a P N P sandwich of doped silicon, which honestly sounds pretty gross, you would get a transistor. And that's what Bell Labs was trying to create back in the late nineteen forties. Now, Shockley's research group was able to develop a transistor that could perform the same tasks in electronics as a vacuum tube.

The first ones were very large and bulky and more like a proof of concept, but it quickly became apparent that this was going to be the future of electronics, and it was what would pave the way for many tourization, ultimately leading to a new era in electronics. And I should also point out that there were other team members besides Shockley who were working on this, like John Bardeen

and Walter Brittaine and Gerald Pearson. They all made equally important contributions to make the transistor possible, but Shockley was frequently sourced as the the head or the the prime contributor, which is not entirely fair. Now, Shockley left Bell Labs and he went on to found his own company called Shockley Semiconductor Laboratory, and he hired many brilliant people to work for him. But his personality and his leadership style

was so confrontational it was demoralizing. He was described as being autocratic and paranoid, and he had a reputation for insulting his employees, building them way up early on and then gradually undercutting them as the relationship would continue in as you might imagine, this led to a pretty unhappy work environment. On a side note, Shockley would later espouse

some truly terrible racist beliefs. And I feel it's important to note this because I don't believe in giving a free pass to someone simply because they made truly monumental contributions to the advancement of technology. We can't deny those contributions. They were absolutely important and they transformed our world. At the same time, we shouldn't ignore the negative aspects of

someone's contributions either. We should take in the full picture. Okay, So Shockley was in the running for world's Worst boss, and it all came to a head in nineteen fifty seven, a little more than a year after Shockley had created the company in the first place, eight employees, all engineers with PhDs, confronted a board member of Shockley Semiconductor named Arnold Beckman. And I'll have to do a full episode

on Beckman at some point. He's another fascinating person. But they voiced their concerns to him, and Beckman heard them out, and he tried to kind of work out a compromise, but it was really too little too late. So the eight decided to leave the company, and Shockley would dub them the Traitorous Eight, very dramatic, and they included Julius Blank, Victor Greenwich, Jean Harney, Gene Kleiner, Jay Last, Sheldon Roberts,

and a certain Gordon Moore and Robert Noyce. These eight individuals approached a company called the fair Child Camera and Instrument Corporation, and that company was actually looking to diversify into the burgeoning semiconductor business at the time. Robert Noyce and Gordon Moore were sort of leaders of this charge, and after coming to terms with fair Child, including each of the engineers sinking five dollars of their own money into the uh the whole endeavor as an initial investment,

they created a new division called fair Child Semiconductor. Now I've covered fair Child in episodes, but granted those episodes aired way back in two thousand thirteen. The company did a lot of really big things in technology, including bringing the integrated circuit to market. Though I should mention that the engineers over at Texas Instruments had also independently created an integrated circuit. Fair Child was just really fast at getting that to consumers um and by consumers, I really

mean other businesses. This is sort of a business to business kind of enterprise. But fair Child was also known for giving birth to other companies, and we sometimes call these other companies the fair Children. In nineteen sixty eight, after working at fair Child for about a decade, Robert Noyce and Gordon Moore decided they were going to leave fair Child and they were going to start their own company,

and they called it Intel. Intel will pop in and out of our story of a m D as it was not just a m d's chief rival and still is, but so has a strangely collaborative relationship with a m D. So there's both competition and collaboration between the two companies. I'll explain more later. Now, in the wake of noise and more departing fair Child, fair Child Semiconductor reached out to a physicist over at Motorola named see Lester Hogan, yet another person I'll have to do a full episode

on in the future, and fair Child offered Hogan. I will modestly call it a pretty darn sweet deal. That's underselling how crazy good this deal was for Hogan. But this is not an episode about fair Child, so they wanted Hogan to come over to fair Child to manage the semiconductor team. So Hogan brought seven Motorola executives with

him and they were collectively known as Hogan's heroes. So we've got the Traitorous Eight and we have Hogan's Heroes, and this makes the early days of Silicon Valley sound

like some sort of Tarantino movie. Hogan the Way had previously worked under Shockley over at Bell Labs, so he had that in common with the trader Is Eight, though he didn't join Shockley's semiconductor company when Shockley left Bell Labs, and Hogan's team had a very conservative management style, something that clashed with another fair Child Semiconductor employee, a guy named Walter Jeremiah Sanders the Third or just Jerry Sanders. And Jerry Sanders will play a very important role in

our story. I'll explain more in just a second, but first let's take a quick break. Jerry Sanders grew up in the nineteen forties. He was raised by his grandparents

after his parents essentially abandoned him. He grew up on the South Side of Chicago, fairly rough part of Chicago at the time, and according to an article in sf Gate and a few other sources, when he was eighteen years old, he rushed to help a friend of his who was being attacked by a gang, and he himself was also beaten up, and he was being up so badly that he went into a coma for a few days,

and a priest actually administered last rites. But he recovered from that and he was able to succeed despite his tough past. He enrolled in the University of Illinois and graduated with a degree in engineering, and he got hired by Fairchild to be a sales engineer and also a marketing manager, and he became known for being particularly successful in that regard. But then Nois and more left and Hogan and his heroes swooped in and they changed things

and they effectively pushed Sanders aside. They essentially they called it a promotion, but it really was a d motion. He went from a director of marketing to being sort of a vice president of marketing, and it was really seen as as more of a let's get this guy out of the way. Sanders was thirty three years old at the time. Now Sanders and seven other fair Child employees would end up leaving fair Child Semiconductor to go

and found a new organization. Now, according to most accounts, if you go and you start searching for history of a m D on the net, you're gonna find a very similar story told over and over again. And the story goes that the Jerry Sanders effectively led this charge

and he got the team together to form this new company. So, according to the history of semi conductor engineering, Jack Gifford, who had become the head of computer marketing at Fairchild in early nineteen nine, saw the writing on the wall when Hogan's heroes swept into the company. He had already been considering the possibility of starting his own analog circuit company,

but he was young. He was just twenty eight years old at the time, and when he decided to take that leap, he found he couldn't get any financial backing for his business. His buddy, Bruce Waterfall, told him that the problem was the financiers thought Gifford was too young

and inexperienced, and therefore he posed an investment risk. So Waterfall reportedly told Gifford that he needed to find someone older and more experienced whom the bankers would find more reassuring, and Jack then thought of Jerry Sanders, who, according to the book, had just left Fairchild himself after being pushed

aside by Hogan. Sanders was apparently considering a new career, going into the recording business in Hollywood, and initially he wasn't interested in Gifford's pitch to start a new analog circuit company. Sanders response was effectively, I'll do it on two conditions. One, I have to be the president of the company, and too, it's not going to be an analog circuit company, but a digital circuit company. Gifford found himself without any real leverage, and he agreed. Getting the

money also proved to be a little tricky. One of the investment groups they approached was called the Capital Group, and there was a guy there named Jim Martin who was working there, and that might have already spelled doom for the new company before things could even get started, because it turned out Jim Martin had previously worked for fair Child, but he had been fired. In fact, he

had been fired by a certain Jerry Sanders. This has me imagining a scene in which Sanders, looking for investment capital for Gifford's company idea walks into the office of a guy he had once fired at his old company in the past. But Jim Martin was also good friends with Jack Gifford, and so he worked with his colleagues at the Capitol Group to provide an initial investment in the new company. And this new company's name would be Advanced micro Devices or a m D, and it incorporated

on May one, nine nine. So from Chockley's semiconductor lab, we can trace a path not just a fair Child, but also Intel and a m D. And while a m D would become known as a competitor with Intel, things would start off a little bit differently. A m D was originally in Santa Clara, California, but quickly moved to Sunny Vale just a few months after the founders formed the company. Their new DIGS had fifteen thousand square feet of space and was valued at half a million

dollars at the time. While the engineers at Intel we're working on creating new microchips, a m d s first order of business was taking products from Fairchild and then redesigning them, essentially optimizing them and tweaking them. This would be something that a m D would get really good at not necessarily building its own products from the ground up,

but taking other products and then optimizing them. These were mostly in the form of integrated circuits, and while a m D started in the business of building logic chips, they weren't yet creating CPUs themselves. Uh, the CPU is the primary logic chip in a computer. Now. To be fair, a m d s founding was right around the time when the concept of a CPU on a single chip was just starting to coalesce, because this was still the

very early days of computers. Earlier, the logic center of a computer consisted of several different logic chips, all wired together, and each logic chip itself was an integrated circuit that would fit into the larger circuit of the central processing unit,

which would, as I mentioned, consists of several chips. But many people, independently or depending upon whom you believe, not so independently, proposed that with the right architecture, you could build all the necessary logic components onto a single chip in an integrated circuit and create what was effectively a computer on a chip. Now, I said, depending upon whom you believe, because there are disputes regarding who first came up with the notion of a computer on a chip.

There's some arguments about who it was that first proposed this, and there's the past ability that people responsible for building what would become the first true single chip CPU may have learned about the possibility from another person who had already proposed it and had worked for them in a previous company. But the unfolding all of that would require

an episode all by itself. The episode about how the CPU on a chip came to be would be a pretty dramatic story that I don't have time to tell today. So single chip CPUs were not yet realized when a m D first started, and it makes sense that they began with basic logic chips what we would consider components of an overall central processing unit today. They were chips

like an arithmetic logic unit and a control unit. So these are all elements that are now integrated into the CPUs we have today, but in the old days, they were all discrete components that you would have to, you know, put together in your circuit. Their first really successful component came out a year after the founding of the company, so in nineteen seventy, and it was called the a M to five O one logic counter. It was the industry's first binary slash hexadecimal up down counter. So what

the heck does that mean? I could just say that and move on, but I feel like without describing what binary hexadecimal counters do, it's meaningless. Right, I could have said any gobbledygook and it would have been just as fine. So binary, of course, refers to the two state basic unit of logic in computers, and we represent binary as being either a zero or a one. So you can think of it like a light switch, right, it's either off or it's on. It can't be both, it can't.

It has to be one or the other. And if you use strings of binary digits series of zeros and ones, you can represent all sorts of stuff, from other numbers to letters, to pictures of cats and so on. But it takes a lot of binary numbers to represent the stuff. And as you work with larger digital systems, you start to discover that working with binary becomes unwieldy. It's very hard to read or write blocks of binary code, and it's super hard to do so without introducing errors in

the process. So one way to deal with This is to group sets of four bits together. A bit is that basic unit of information, a zero or a one, So you can group these four bits together into another type of numbering system called hexadecimal numbers. Now hexa decimal numbers is a base sixteen numbering system. We use a

base ten numbering system. We go from zero to nine, and when we get past nine, you have ten, which is again you start back at zero, and then you have a one in the tens column for that number. But you go all the way back up to nineteen, and then you start over again, and now you have a two in that ten's column. Well, hexadecimal is base sixteen, and that presents a challenge, right because if you're talking about base sixteen, you would normally start with zero. Then

you'd work up to fifteen. But how could you tell a ten apart from the two digits of one and zero that are side by side. Right, If you can have a zero and a one in your numbering system, and you can have a ten in your numbering system and it's base sixteen, you can't tell the difference between

a ten and a one zero. So anything ten or higher ten to fifteen would be confusing, and so for that reason, the digits ten, eleven, twelve, fourteen, and fifteen are in hexadecimal, represented by letters A, B, C, D, E, and F. So hexadecimal digits include zero through nine and A through f to represent binary or decimal numbers. Now,

remember hexadecimal numbers represent groups of four bits. A zero in hexadecimal represents a binary string of four zeros in a row, and F and hexadecimal represents the four bit string of one one one one, Because that actually represents the decimal number of fifteen um you have. You essentially say one plus two plus four plus eight. That's how

those different digit spots represent numbers. So to convert binary into hexadecimal, you would first take your big block of binary code and you divide it into four bit strings. Then you would convert each four bit string into the hexadecimal digit that represents that four bit string, and you have a slightly simpler way of representing all the information. So a m d S first successful logic counter could do this task and it became an important early component

in mini computer systems of the early nineteen seventies. At the time of the A M to five zero one release. A m D had fifty three employees. The company had established a wafer fabrication lab that could make two inch silicon wafers and then a m D would then use that as a platform for integrated circuits, and the company was able to build circuits with elements on the seven micrometer scale, or if you're old school, the seven micron scale.

A micrometer is one millionth of a meter, and today microprocessors are built on the nanometer scale that's one billionth of a meter, But in nineteen seventy the micrometer scale was pretty darn impressive. In an a m D engineer named Sven Simonson led a group that designed another successful A m D product, and it was a chip that handled multiplication. It was called the A M two five oh five and it was at the time the industry's

fastest multiplier chip. So the company was making a name for itself building out these components that were outperforming other manufacturers that we're working in the same industry. One was also the year that a m D began to produce random access memory or RAM chips, and anyone who has gone shopping for computers has seen stuff about RAM, and I think most people realize it has something to do with computer performance, but they might not know what it

actually is all about. Well, first you know it's it is memory, and memories purpose is to keep a record of information, and there are different types of memory. This is true for people and it's true for computers. So in computers, you have read only memory or ROM, you have random access memory or RAM, and then you have auxiliary memory, which we usually refer to as storage. So auxiliary memory is where information lives when you've saved it. It's in a way sort of analogous to our long

term memory as human beings. But retrieving information from auxiliary memory takes a little bit of time. A computer has to go through the directory, find the information, retrieve it, and pull it up into the current moment. So if a computer had to refer to its auxiliary memory every time you want to run any sort of process related to that data, it would feel like it was really taking forever. Random access memory is more sort of like

our our short term memory. It's used to temporarily store information for the purposes of working with that info and making it faster. So rather than having to pull up the data from storage every time the computer can store it temporarily in RAM. So the more RAM you have, the more information you can hold in this working memory. And that's why people tend to talk about having more

RAM with your computer makes your computer faster. What's really doing is it's cutting down on how frequently your machine needs to consult it's auxiliary memory. So if it can load more data into RAM, then it doesn't need to pop back into the library as frequently. RAM is often referred to as volatile memory, and it will only hold information as long as the computer is powered on. Upon

losing power, traditional RAM relinquishes all that information. Read only memory, by the way, or ROM has pre programmed information that's hard coded onto the memory itself and generally is used to hold stuff like basic sets of instructions that the computer has to follow in order to boot up and get the system ready for use. All right, so that's

RAM in a nutshell. I'll have to do a full episode about later on to talk about the nitty gritty stuff, and when we come back, I'll talk more about the early days of a m D. But first let's take another quick break. In two, a m D made the move to become a publicly traded company and held a night po Chairs of a m D were valued at fifteen dollars each. The following year, it would open its

first overseas manufacturing facility in Malaysia. So the company was expanding early on, and the company continued to manufacture components and grow, And that's pretty much what the company did for its first few years, building logic chips, growing the company. And there's not really much to say about those years apart from the fact that Sanders established himself as a

bit of a flamboyant leader. He had already been seen as similar in uh fair Child, and I read in an Ours Technica article that Francis fran Barton, who was the chief financial officer at a m D in the late nineties, described Sanders as being part Indiana Jones, part

don Keyxote. So that's pretty darn flamboyant. Now, Uh. I do want to say that if I were to cover every single thing they ever put out, this would sound like a technical manual and all the names are a M and then a bunch of numbers that would become unmanageable right away. So I'm going to be skipping around a little bit. So by the spring of nineteen seventy four, five years after the company had started, it had grown

to just under fift employees. A m D also reinvested in its manufacturing facilities, which is a necessary and critical part of the semiconductor business. Gordon Moore, you know, that guy who used to work at Fairchild and then became a co founder of Intel, had famously made an observation back in nineteen sixty five that market demands would create the incentive for semiconductor companies to cram about twice as many components onto a square inch silicon chip every two

years or so. To meet that demand, companies like Intel and A m D had to frequently overhaul not just the design of the chips, but the manufacturing process itself to create ever smaller transistors and pathways in order to

stay true to that observation. Today we call that observation Moore's law, though these days we tend to think of it in terms of computing power, and how computing power tends to double and strength every eighteen to twenty four months, as if it were magically doing that on its own.

The Gordon Moore's point was that there was going to be a continuing demand from the marketplace for ever smaller components on microchips, which in turn also means that the microchips are able to do a lot more stuff because you've crammed more components onto it than the previous generations microchips, and that as long as that market demand is there, then it would create the incentive to continue investing in that.

So his was a market driven vision. We tend to think of it as some sort of innovation law, but that means that's sort of like getting it backwards anyway. Even in those days, a m D and Intel were competing.

While Intel had started to develop computers on a chip in the early nineteen seventies, releasing the Intel four zero zero four micro processor back in nineteen seventy one, it was also still the business of building logic chips, and a m D sales for certain products we're starting to catch up to and in some cases exceed Intel sales, So things were looking good for a m D. Jerry Sanders initiated a special program called Run for the Sun

in nineteen seventy five. This was a sales target for a m D. The sales target was to make ninety three million dollars in sales that year, So why ninety three million dollars, Well, that would be one dollar for every mile between the Earth and the Sun, and Sanders again had a certain flair for the dramatic. By the way, A m D would very nearly make that goal. They came up less than a million dollars short of that figure.

Really impressive considering where they were. But I guess that means they didn't burn up on the surface of the Sun, so that's good. Also, in nine, A m D did something pretty clever. Engineers took a very close look at a photograph of the die that Intel was using to build the company's A D eight eight bit microprocessor. So the microprocessor was called the eight and it was an eight bit microprocessor. Am D looks at this their engineers.

They start to set out to reverse engineer Intel's work and make their own version of Intel's chip, So a m D S version would ultimately be called the a M nine D e D. Now you might imagine that Intel was pretty head up about the fact that A m D had managed to figure out their secret sauce and then reverse engineer it to create their own variation

of Intel's chip. But what actually unfolded was one of the more unusual business deals in tech history, in Intel and A m D entered into a cross licensing agreement between the two companies. This initial agreement had to do with microcode, which is the code on top of a CPU that allows it to interact with the computer's other systems,

and it gets pretty complicated both logically and legally. But an interesting thing to point out is that A m D and Intel, without while they were still being fiercely competitive against each other and even engaging in lengthy and acrimonious legal battles over their history, would continue to renew patent licensing agreements every decade because they were set to expire after ten years. While in nineteen seventy seven, A m D and the German company Siemens entered and do

a joint venture to develop micro computers. Those are the type of computers we think of as desktop personal computers in other words, Now together the companies formed a third entity called Advanced micro Computers, and they established it both in the United States and in Germany. The main focus was to create computers that had a dialog Z eight thousand micro processor as the CPU. A m D was actually a second source for those type of chips. Now, that means that a m D had acquired a license

from Zilog to produce Xilogs chips using Zalogs designs. So it's it's as if you, let's say that you make a soft drink like Coca Cola, and then Pepsi comes up to you and says, hey, we can't meet our demand. We have way more demand for our product than we can personally manufacture, so we are willing to strike a deal with you where you can make our stuff and sell it because the demand is there, and you'll pay us a little licensing fee so that we get some

money out of this. But that way we meet our customer demands and you make money and I make money. And that's sort of the idea that a m D had with Zilog, where they were allowed to produce Zilog chips in return for this licensing fee. Now, before long it became clear that Siemens and a m D had very different visions of where advanced micro computers should go.

And in nineteen seventy nine, just two years after entering into the joint venture, a m D would buy out Siemens stake in that company, and A m D would continue to operate advanced micro computers for a short while, but would choose to shut it down in N one because of another big opportunity. Then, opportunity came straight from

their old nemesis, Intel. In the nineteen seventies, Intel had developed the eight six microprocessor and by extension, what has become known as the X eight six instruction set architecture. This was a significant advancement over Intel's eight bit processor, the D eight that was the one that A m D had managed to reverse engineer and effectively clone a couple of years earlier. The eight six microprocessor was a top candidate when another tech giant, IBM, was looking at

microprocessors that might power its upcoming IBM PC. But Big Blue had a concern. IBM was worried that the demand for the IBM PC would quickly exceed Intel's manufacturing capacity, and that would result in shortages and delays in the supply chain, which in turn would make IBMS customers unhappy. Plus, if something should happen to Intel, then IBM would be up the creek as far as its computers were concerned.

They'd have no supplier for their microprocessors. So IBM essentially told Intel, hey, we can make a deal, and it's going to be a big one, you know, make you lots of money, but you have to figure out how to license your technology to another manufacturer so that we can get the number of chips we need to meet our demand. Intel, not wanting to lose this valuable contract, agreed to IBMS terms and then turned to a m D. Thus Intel an a m D entered into an agreement.

Intel would supply a m D with the proprietary information about how the eight six and by extension, the x A D six instruction set architecture worked, and a m D would start producing some of Intel's chips, and the

two competitors joined forces to meet IBMS expectations. Now, in another episode, I talked about how IBM's decision to rely heavily on off the shelf components would lead to its eventual departure from the personal computer market because other companies would replicate IBM computers by getting hold of those same

basic components and putting them together themselves. And part of that had to do with Intel and a m D not having to sign any sort of exclusive deal with IBM, So not only did these companies make a killing off IBM, they also benefited from all the IBM compatible manufacturers that grew out of that era, and both Intel and A

M D were raking in the cash. Intel would continue to supply A M D with database tapes for the design of the six, the six, and the A D two eighty six, which gave A D the ability to make clones of those chips, plus the variants like the eight and thee Those were variations on the x A D six architecture. Now A and D did not put all its eggs in the Intel second source basket. It was also developing chips for RISK computers. RISK or r

I s C stands for reduced instruction set computer. It relies on a processor design that follows a simplified set of instructions, and it's an alternative to complex instruction set computing or c I s C. The idea of r I s C computers is that they do fewer things, but the things they do they can do much more quickly and efficiently. The power PC microprocessor architecture, which was a joint venture between Apple, IBM, and Motorola, relied on risk chips. The A M D line was known as

the A M twenty nine thousand series. So in nine three, A M D ruffled the feathers over at Intel a little bit. The company produced it's a M two eighty six licensed clone of Intel's eight two eighties six, And typically we just refer to these as two eighty six microchips or two eighty six computers. It's really saying that the computer, which was an IBM compatible computer had inside of it an A D two eighty six microchip or a M D S version. So the fact that A

M D was making this chip totally fine. That was completely covered under this licensing agreement between the two companies. That was not the problem. A M D was doing exactly what it was supposed to be doing. It was taking Intel's chips, and it was making them and making them available to these UH manufacturers that are making the actual computers. The two chips were identical from an architectural perspective, but a M D S version had a higher clock speed.

Intel's clock speed for the two eighty six topped out at twelve point five Mega hurts, and a M D S could go up as high as twenty mega hurts. So that raises the question what are clock speeds? So let me answer that very quickly. With a processor, the clock tells us essentially how many internal operations the microprocessor can perform each second, and we describe this in cycles per second, and a HURTS is one cycle per second.

So Intel's two eight six microprocessor could complete twelve and a half million cycles every second, but a m D S could do twenty million cycles every second. So a m D S chip was able to process information at a faster rate than Intel's which led the industry to say that a m D was effectively producing better Intel chips than Intel could, and as you can imagine, that didn't go over so well at Intel. Intel did not want to see companies going with their competitors version of

their own chip instead of them. But things were going great at a m D. The company was named one of the Fortune five hundred companies in nineteen eighty five. Tony Holbrook would become the president of the company in nineteen eighty six, and Jerry Sanders would turn into the CEO, not turn into he was just that was his role. Didn't have a magic fairy come down and Grant him the wish of becoming CEO those some days. I think

that's how business works. Also, in nineteen eighty six, Intel terminated their contract with a m D, which was problematic as the second source deal between the two companies that that had started back in nineteen eighty two, and it was supposed to last ten years. But if I'm doing my math correctly, two plus ten does not equal nineteen

eight six. So I guess Intel wasn't too happy with getting a reputation for making the second best Intel microchip in the industry, and the company was gearing up with its three eight six update. So Intel said no dice to a m D, and a m D sued, alleging that Intel had breached the contract. But these legal battles take time. This particular legal battle would take almost ten years, and in the meantime am D had to figure out what else it had to do. That what else would

end up being up two pronged attack. One would be to develop their own CPUs, actually designing their own microchip architecture from the ground up based on the x A D six instruction set. The other prong of this two pronged attack was to look at what Intel was doing and reverse engineer it again and maybe even do it better than Intel could again. So in our next episode we'll talk about this two pronged attack and what a m D did and how it continued to evolve and

what's going on with the company today. But this is time to wrap up this particular episode, So thank you so much for the suggestion. Greatly appreciate it. If any of you have any suggestions for future episodes of tech Stuff episodes, you can write me the addresses tech Stuff at how stuff works dot com. You can drop on

by the website that's tech stuff podcast dot com. You're gonna find an archive of all of our shows there, plus links to the social uh presence of tech Stuff and to our online store, where every purchase you make goes to help the show. We greatly appreciate it, and I will talk to you again about a m D really soon. Tex Stuff is a production of I Heart

Radio's How Stuff Works. For more podcasts from my Heart Radio, visit the i Heart Radio app, Apple Podcasts, or wherever you listen to your favorite shows.