The Intel Story Part Two

semiconductor industry itself. Two of the co founders of Intel, Gordon Moore and uh and Rob Noyce. They had come from a a an organization that didn't treat them very well, and so they, along with six other people, left that organization. They became known as the Traitorous Eight and founded the fair Child Semiconductor business under the umbrella of a larger company that did camera and instrumentation work. Worked there for about eleven years before they left that company to co

found Intel. And in our last episode, I ended right around the time that they had introduced their first micro processor, the four zero zero four. Now we're gonna pick up from that point forward. We're also going to backtrack a little bit just to explain some other details, and we're gonna explore what Intel was all about since its founding up until present day. Now, the four zero zero four started off as one of a set of four chips. It's like a package of four chips that were designed

specifically for another company. That company was the Nipon Calculating Machine Corporation. Now, the original designation for those four chips was the MCS DASH four. Along with that four oh four or four zero zero four, I should say chip, were three others. Right. You had a chip that was a read only memory chip or ROM chip that was responsible for storing the custom applications for calculating machines. So essentially the programs of these calculating machines were stored and

read only memory. That meant that you could not write over that information. It was always going to be there. It was hard coded onto the chip itself, and that way it made it efficient and reliable and you didn't have to worry about accidentally erasing it. There also was

a random access memory chip or RAM chip. Random access memory allows you to store data in a temporary storage space, so that way the processor can call upon that data quickly without having to search a deeper storage UH solution. The fourth chip was a registered chip, and that was handling the input output port, so taking in the information from the input devices and being able to translate that

into whatever the commands were for the CPU to process. Now, it was the CPU that really ended up changing everything. So Intel designed the CPU as sort of a general purpose programmable processor. Now it wasn't a true general purpose processor yet. That wouldn't come until a little bit later, but it showed the promise of this strategy. It wasn't locked into a single practical application. So up until the early nineteen seventies, hardware for computing machines was heavily customized

for each specific machine. In other words, the chips you would find in one type of computer, we're completely incompatible with every other type of computer. To design the processor for these machines, you essentially had to go back to the big inning and rebuild the wheel every single time you wanted to do it. Uh. There wasn't really a concept of a general purpose processor the way the four zero zero four was, and the concept of plug in

play would be decades further into the future. Intel's move to create this programmable processor allowed for an explosion in various uses. So the four zero zero four would provide the foundation for Intel's future chips, which then would become an instrumental component in countless types of technology, not just computers or calculators. It virtually guaranteed Intel's success in the market.

Beyond that, the processor was able to take advantage of the advances and manturization that had followed the invention of the transistor. Now you'll remember from part one of this series that Gordon Moore the one of the co founders of Intel. He was the guy who came up with what we now call Moore's law. All he had made an observation that observation would become Moore's laud But the observation said that the balance of technology, economics and manufacturing

processes would mean that for the foreseeable future. And remember he made this production back in nineteen sixty five, the number of discrete elements on microprocessors, on semiconductor chips would double every two years or so. So if you were to get a semiconductor chip in nineteen sixty seven, it would have twice the number of transistors that you would have found on a on a semiconductor in nineteen sixty five. The ones in nineteen sixty nine would have twice as

many as the ones from nineteen sixty seven, and so on. Now, by the time Intel was ready to produce the four zero zero four chip, they could make a single CPU microprocessor that was as powerful as the famous Eniac computer. Now, the Eniac computer was one of the first electronic computers ever in the history of mankind. Uh. It was constructed in the forties, came online more or less in and at the time it was one of it was really

the most powerful electronic computer in existence. It was one of the first ones, so obviously it didn't have a whole lot of competition, but it took up an entire room. It had a lot of very large components that generated a ton of heat. It used vacuum tubes instead of transistors. But for this supercomputer of the early days, it would

take up an entire room. Well until's first microprocessor chip had the equivalent amount of power stored on a chip of semiconductor material that was the size of your fingernail. So you went from a device that took up the entire room in a building, and it was a big room. It wasn't like a little office to something that could fit on your fingernail with the equivalent amount of processing power.

This was a transformational moment in computer history, and one of these days I'm doing I'll do a full episode about the Eniac computer and also how it came to be and the people who worked on it. It was a fascinating device all on its own. In the decade that it was in operation from around nineteen to nineteen fifty five, researchers estimate that it ran more calculations during that ten year period than all the calculations that had

been performed by humans leading up to that moment. So in ten years, it did more calculations than all of humans had done throughout history leading up to the creation of the Eniac. That is incredible. But then it ended up dying after it was struck by lightning. So it has a tragic end to that story, which also means it would make a great subject for a podcast in the future. I think that maybe thour got a little miffed that we humans that we're getting really confident, and

so he can kind of nipped in the bud. But that's another story. Let's get back to Intel Now, if you want to know what the clock speed was of the four zero zero four, it's initial speed was one d eight killer hurts. Now, clock speeds, in case you aren't aware, they're measured in hurts, and it's a measure of how many clock cycles a CPU can perform in a second. So a hundred eight killer hurts clock speed means that the processor can perform one eight thousand clock

cycles every second. In an ideal world, every single individual instruction sent to a processor takes up one clock cycle. So you could translate this as saying this processor was capable of completing one hundred eight thousand inst ductions every second. That's not exactly true because not all instructions take a single clock cycle, and efficiency makes a big difference, but it gives you a general idea of the capabilities of

this microprocessor. Now, really, when you get down to it, everything that a processor does takes up a certain number of clock cycles, and it depends on what the processor needs to do, but it tells you there's a physical limit to the number of tasks a processor can complete within a given amount of time. So a second in this case, if you are throwing stuff at the processor, that takes fewer clock cycles than what it is capable

of doing. Things should run pretty smoothly, right. If the number of instructions you're sending to the processor is less than the processors clock speed, it should be a pretty smooth experience for you on the user side of the computer. But as you start to throw more processor hungry tasks at the chip, you can start to slow down because you might be throwing more directions per second than it can handle, and then you start getting lag and things

get jittery and slow. There are a lot of other factors that affect us as well, including how efficient that processor is. Some processors are more efficient and can do the same task with fewer clock cycles than a similar processor.

So if you get two different processors and they have similar clock speeds, but one of them is designed and optimized so that it's much more efficient, you can do more with that processors than you can with the other one, even if they're both rated at the same clock speed, just because the other one handles tasks more efficiently than it's uh than it's peer. This is true even if

you have two processors that have different clock speeds. If you've got one processor that has a measurably faster clock speed than your first one, but the first one is still more efficient. You can still end up having a better experience using the quote unquote slower microprocessor because it is more efficient in its design. So that's something to keep in mind if you're ever shopping for microprocessors. Just looking for that clock speed is not really an indicator

of how good that microprocessor is. It is an indicator, but it is not the only one you should pay attention to. It's kind of like when you go shopping for digital cameras and you start looking at megapixel numbers. Larger megapixel numbers doesn't necessarily equate with better pictures. There are a lot of other factors that are very important when it comes to color, representation, contrast, all of these

other elements that go into creating digital images. So I just want to bring that out here in this part of the podcast, just in case you are looking at building a computer and you want to find a really good micro processor. Just going for that high number is not a guarantee that you're getting the absolute best for your money. You have to take these other elements into consideration. Uh So, I know it's a bit of a tangent.

But it's important to remember just because otherwise we oversimplify and say a higher clock speed equals a faster, better processor. That's not always the case. Uh. Anyway, back in Killer Hurts was wicked fast. That was a really fast processing time. Today, something like Intel's Core I seven KB Lake processor can run at four and a half giga Hurts. That means it can run four and a half billion clock cycles per second. It can handle four and a half billion

instructions per second. Compare that to the four zero zero four that was a hundred eight thousand, four and a half billion. You start to see the power of Moore's law how that plays out over time, and being able to go into the billions of instructions per second would probably have amazed even Gordon Moore back in the day, because when he was making this observation he didn't necessarily

think this would be sustainable indefinitely. He thought that eventually we would run into some sort of fundamental limit as to what we are able to accomplish, and at that point the actual progression would break down and you would no longer be able to have a processor that's effectively twice as powerful two years into the future because we would have bumped up against some sort of fundamental limit

that would prevent us from doing that. Now, for the four zero zero four, Intel was using two inch wafers, which is interesting because typically semiconductor companies were using twelve inch wafers to design microchips, but Intel was using two inch wafers for this one. And that might sound delicious. You're hearing wafer, you might be thinking cookies. It's not. The wafer is called that because it's a big circular disk, or in the case of the four zero zero four,

they're using small circular discs. But a wafer is a substrate. It's a foundation. It is the ground upon which you build a processor or you build a circuit. It's made of semiconductor material It's frequently silicon these days, it's more often silicon than anything else, but there are other types of semiconductor materials, so I don't want you thinking silicon

is the one and only type. There are others. Silicon wafers are usually twelve inches in diameter, not two inches in diameter, so this was a bit of a departure. But you have to imagine in a really shiny disk that has like a grid like pattern across it that's repeated over and over and over again. Uh, the patterns that you see, that's actually the patterns of where circuits will be laid. So it's not just a design, that's actually the physical architecture of where the circuits are going

to be. They don't just come out that way. They have to be etched that way. When you first create these silicon wafers and you and you get them polished, they are just reflective. They don't have any grid like patterns on them. Now, for the geometrically minded out there, you might wonder, why the heck would you produce round silicon wafers when microprocessors are rectangular. Now, obviously you can build a bunch of microprocessors on top of a wafer

and then you cut them out. So you use your wafer to be the the foundation for all of your microprocessors. You you etch it out using a repeated pattern over and over again. You build out the microprocessors, then you cut them all out, and you get your little dies of of CPU chips or other microchips, doesn't have to

be just ae CPU. Other microchips also followed this pattern, but you're thinking, well, if it's round and ultimately the chips are square, that means there's a lot of waste, right, Like, you start to get where the edges of the the individual microprocessors are coming up against the edge of the disk. Well, square and circle, they're not going to match up perfectly, so you're gonna have some wasted material. Why would you

go with a disc in the first place. Well, in order to answer this question, I'm gonna have to talk about how these chips are made today for the most part. But remember that the process back when Intel made it's four zero zero four chip was similar but not nearly as sophisticated as the way we do it today. Um, the microprocessor manufacturers cut wafers up to make several microprocessors per wafer, but that does create some way. So here's why we have round wafers instead of say, square wafers.

It's because we grow the wafers and it's done in a very interesting way. So imagine that your goal is to create a sheet of silicon that's perfect from a microprocessor. That means you have to have an incredible amount of pure silicon, or as close to pure silicon as you can possibly manage. Any sort of impurity will introduce electrical elements into your substrate that you don't want because that's going to create errors. So you need it to be

as pure as you possibly can make it. You then treat it to allow transistors and other components to transmit electricity across the circuit without having it bleed through or leach away the substrate. Because silicon is a semiconductor and can either conduct or insulate based on properties, it is ideal for this. And we start with sand. Sand has a very high percentage of silicon in it, and again

silicon is our semiconductor material of choice. The sand gets melted down and then you eventually end up with a material called polly or poly silicon, and you use this to um you separate out as much of the impurities as you possibly can to you get to pure silicon. You melt down ingots of this stuff into a purged furnace at a temperature of more than two thousand, five hundred degrees fahrenheit, which is more than one thousand, three

d seventy degrees c intigrade. The furnace must be purged with oargon gas to make certain there's no unintended impurities present. You want to make sure that you've gotten rid of anything that could end up affecting the silicon as you try to create a wafer. Now, at this point there's about one non silicon atom per billion silicon atoms, so that's incredibly pure. You get a billion silicon atoms and then one thing that is not silicon. That's pretty pure.

You then have all of this molten silicon inside a crucible, which is like a giant, uh cylindrical column, and it keeps that molten silicon nice and hot. The crucible starts to spin in a given direction. For the purposes of this discussion, will say it's spinning witter shans, also known as counter or anti clockwise, so it's spinning anti clockwise.

You then insert a seed crystal of silicon. It's about the size and shape of a pencil, and you put this it's lowered down by machine, into the molten sil con and it acts as sort of the nucleaic spot for other silicon crystals to form. And this way you get a a very regular crystalline structure, a mono crystalline structure of silicon, and this seed will turn in the opposite direction of the crucible, So in our example, it

would turn clockwise while the crucible turns counterclockwise. Now, as this happens, you end up creating this column, this cylindrical column of pure silicon or mostly pure silicon. Uh, and you get that monocrystalline structure all the way throughout the entire thing. When you withdraw the cooled silicon from the crucible, it looks like a giant silicon log that tapers at an end because that's where you've been pulling the sea crystal out, where you've been very slowly drawing it upwards.

You you draw it incredibly slowly, like a millimeter and a half per minute, so it's a very very slow process. You're not just whipping the crystal out of the molten silicon. Doing this ends up creating, like I said, a tapered column. And the silicon has great tent sile strength, meaning you can suspend it from a thread like amount of silicon, and even though it weighs between two and four pounds

or between a hundred and two hundreds, it'll hold it. However, is very brittle, so while it has great tent sile strength, if you were to just try and cut it. It would cut very easily. The column or pole of silicon would be about twelve inches or so in diameter, and you would then use that to slice it into wafers. So imagine this log of silicon and you put it through a device that uses very thin wire cutters or

wire blades. I guess I should say they're not cutters, they're they're blades made a very very thin wire that then just zoom through this column and chop it up into these very thin wafers. They're about a millimeter thick, and you can get quite a few out of one column of this silicon material. At that point, you send them to another device to polish them out, because the cutting creates uneven set sections on the surface of the wafers, so you have to polish it to remove as much

of that unevenness as you can. But even that's not good enough, because when you're building microprocessors, you're working on the microns scale or smaller. These days, you're working on the nano scale. At that scale, even the tiniest of bumps or grooves is going to look like an enormous mountain range or incredible valley. You have to buff it all out and get it as level as you possibly can.

So after putting it through the polishing machine, you typically would need to chemically treat the wafers to get them as smooth as they possibly can be. Now, the next step would involve etching the wafers to create the pattern for your circuits. So this is sort of like building the blueprint for the circuits directly onto the substrate itself.

Uh So, obviously, any sort of flaw in either the silicon material or it's physical properties, whether it's the unevenness or maybe there's an impurity that has fallen down and touched the wafer, it's enormous when you get down to that scale, so the tiniest of imperfections can completely ruin a chip. For etching, Intel uses a light sensitive layer called photo resist and coats the surface of the wafer. That's essentially an etch resistant material. Anything that UH encounters

it is going to resist being etched. And they let this layer harden and then they use other little stencil like devices called masks. The masks cover parts of the chip while allowing other parts of the chip, or rather I should say cover parts of the wafer and allow

other parts of the wafer uh to be exposed. You then use ultra violet light, which turns the photo resist material it encounters soluble, and through the process of building a circuit, you have to use lots and lots of different masks, essentially, lots of different stencils, and lots of applications of this ultra violet light. Because circuits are really three dimensional creations. They're not just two dimensional. They're not just width and and height. There's also depth to them.

So you use a sequence of these masks and you expose the wafer to ultraviolet light. Each time the ultra violet light hits through the mask and contacts the photo resistive layer below, it makes it soluble. You then treat the wafer with a chemical that removes all the soluble material. So that's you're left with everything else. You've etched away, all the stuff you do not want. All the stuff you do want remains on the wafer. That is the blueprint for your circuit. At that point, then you would

implant some ions, which are charged particles. They can have either a positive or a negative charge, depending upon the application you need them to be. You would either put in positive ions or negative ions, but you use that in the silicon itself. This is called doping. This is used in semiconductors all the time to specifically dictate how the semiconductor performs under specific circumstances. While you're etching, you're

creating channels for what will become transistors. The transistors themselves must be deposited into the channels, and today Intel uses a method called atomic layer deposition to apply materials to the wafer surface at a level of precision necessary for components on the nanoscale, because at that scale you're talking about something so small you can't even see it with

an optical microscope. You would need a scanning electron microscope or something similar in order to even get a look at it, so obviously you have to have atomic precision with this. Intel also uses electroplating to deposit copper ions onto the transistor, and if you listen to my History of Electricity episodes, you'll hear more about electroplating and how

that works. Now, since the individual components on microprocessors now measureing the nanometers, it's critical that from this point forward, all potential contaminants have to be eliminated from the fabrication area. That means anyone working within the environment has to wear a special suit often called a bunny suit, to eliminate

the possibility of dust, skin, or hair getting into the environment. Also, they tend to have very powerful air conditioning systems which are circulating and filtering the air on a very frequent basis. Uh intel They're setup has air coming in from the ceiling vents and the ceiling allow air to come down, and then vents on the floor pull air away and it does this constantly, so you're constantly circulating and filtering the air to remove any potential pollutants that could ruin

a microprocessor. This is what gives us the term clean room, and the clean rooms in semiconductor facilities tend to be a Class one clean room, meaning there more free of

pollutants than even the most advanced hospitals. Are so extremely clean environments, obviously, all the equipment that's being used has to be absolutely spotless, because again, you introduce a tiny impurity and you ruin a microchip, or worse yet, you might ruin an entire wafer, which means all of the microchips that would have been produced on that wafer are

a loss. You you'd have to throw them out. From the beginning, Intel had to work on ways to design, miniaturize, and imprint circuit layouts onto silicon, and this continues to be an engineering challenge as companies like Intel attempt to keep up with the observations Gordon Moore made decades ago. The individual transistors are acting like switches, so they either complete a circuit and allow electrons to move through, or they break a circuit and they keep electrons from moving through.

And they're turned on and off by these devices called gates. So a gates either open or it's closed, and that tells you you know, essentially that the microprocessors. Ultimately their job is traffic management. They're managing the movement of electrons because they represent single pieces of information, either a zero or a one and off or an on, a no,

or a yes. And collectively, when you get lots of these dull pieces of information, you can describe much more complex concepts than just on or off, and that's where you get into the very basics of computer science. So all of this, these different components have to be etched onto the silicon, and that's because in the later stages of manufacturing, the individual elements of the microprocessor have to

be deposited on the wafer. That can be up to thirty layers of of material put onto a wafer before it becomes a chip, and each incredibly tiny element needs to be in its proper place. And these days, chip manufacturers use variation of lithography to essentially print the components onto the circuit etchings to build the microprocessor layer by layer. It is incredibly precise and difficult to imagine. But we'll talk more about that a little bit later on. First,

let's take a quick break to thank our sponsor. Back to the four zero zero four. It had an impressive number of transistors for the time. It had about two thousand three transistors on this one chip. Uh, the micro processors of today leave this processor behind in the dust.

That doesn't exist in those clean rooms I just talked about. So, for example, the Broadwell E family of Intel processors, which is actually from a couple of generations back well, they launched in two thousand and sixteen and have a trans sister account of about three point four billion. So you went from two thousand, three hundred way back in nineteen seventy one to three point four billion in two thousand, sixteen.

So yeah, Moore's law is no joke. While today's chips have components that can measure just a smidge more than a dozen nanometers in width, the four zero zero four circuit line was ten microns wide, or ten thousand nanometers, So the transistors of nineteen seventy one measured about ten thousand nanometers wide. Today it's more like fourteen if you're getting a top of the line processor, fourteen nanometers instead

of ten thousand. If you're wondering exactly how much that is, because it's still kind of hard to imagine even like ten thousand nanometers, how white is that, well, that's one tenth the width of your typical human hair. Human hair tends to be around a hundred thousand nanometers wide, not long wide, or at least so I'm told I have to take a lot about hair on faith these days. I miss having hair. Well. The four zero zero four

launched and changed the world of electronics and computers. In nineteen seventy two, Intel would expand, opening up an assembly plant in Penang, Malaysia. This was Intel's first international manufacturing facility. Intel also acquired a company called Microma, which was experimenting with a new technology themselves. They were offering up digital watches that had liquid crystal displays. This was pretty new

in the early seventies. So Intel got into the wearables business just a couple of years after it was founded, and people would think, oh, I thought Intel got into the wearables business pretty recently with providing chips that could be found in lots of different types of products out there now. As it turns out, this early attempt to

get into wearables didn't really pan out for Intel. Uh. Intel held onto Microma for about six years, but they found that they had real trouble meeting consumer expectations and figuring out exactly what consumers wanted. Because Intel was not a consumer electronics company. Intel was building products for other businesses. So Intel would create a processor that some other computer

manufacturer would use in its products. Intel wasn't building stuff specifically for the end consumer, and as a result, they weren't really good at running Microma as a business. They held onto it for about six years and then they sold it for a big loss. They lost about fifteen million dollars on it. They sold it for less than what they bought it, so it was a a tough lesson for Intel in those early days. Jumping back to two again, Intel created the first eight bit micro processor,

also known as the eight zero zero eight. Now, this one obviously was more powerful than the four zero zero four, but it still wouldn't really transform the world. It wouldn't be until nineteen seventy four when the company introduced the eight zero eight zero or the eight. This was the

first true general purpose microprocessor. So the four zero zero four and the eight zero zero eight had laid the groundwork for the eight, but it was this a D eight that would become the basis for microprocessors in everything from computers and calculators to traffic lights because it was a true general purpose processor that could be used for

all sorts of different applications. So we often think of Intel as the company that makes the processors in computers, but the truth is they make processors that are in all sorts of different gadgets and devices and products, not just computers and phones. This made into one of the most competitive companies in the space. They offered up a computing solution in a small package for an aggressive price, and as a result, Intel became the global leader in

microchips and held onto that title for a while. The cost of the eight was just three d sixty dollars, which you know they were. Intel was saying, this is a computer on a chip. You get all the power of a computer on a single chip that can be incorporated into lots of different applications, and it's three sixty bucks, which was much cheaper than full computer systems. So Intel

was doing really, really well in these days. It was also dominant in the area of memory chips at the time, so remember Intel had started by making chips that were for memory, not for processing. By the end of nineteen seventy four, Intel held eighty two point nine percent of the d RAM chip market. Now this would change over the next decade because other companies would get involved in

making memory chips and the competition got really fierce. And once you get a ton of different companies all competing with each other making the same stuff, you'll see that product your your market share will start to drop over time, unless you're making crazy deals by undercutting your competition. So by a decade later, Intel's sharing the d RAM market would have dropped all the way down to one point three per cent. But while that memory chip business had

become incredibly competitive. Intel was still dominant in the microprocessor market, so it wasn't as big a deal even though their memory chip business was leaking away over time, or at least was getting uh was being less dominant in the market over time to the point where they had dropped down to one point three within a decade. Because they were the definitive name in microprocessors and more than balanced out, they were able to recapture a lot of that success

with the microprocesss. In nineteen Robert Neiss, who was again while the co founders, became the chairman of the board at Intel. Noise was known as an executive who issued the lavish trappings that many CEOs and chairpersons have indulged in. He remained in leadership positions at Intel until his retirement, which must not have suited him very well because after he retired, he then went on to become the leader

of a semiconductor manufacturing consortium called Sema Tech. And Noise would pass away in nineteen nine at the age of sixty two. His leadership style would remain a very powerful influence at Intel. It became sort of the model of how leaders were expected to behave over at Intel, largely that you were supposed to have kind of a let's get to work sort of attitude and not have too many lavish trappings of the executive's life. Uh, the idea

being that everyone should benefit, not just whomever's at the top. UM. I don't know if that's how Intel's culture is now. I haven't ever been at Intel's corporate headquarters, but I know for a long time his values specifically were upheld as the model for future Intel executives. Now, as you might imagine, Intel's line of processors became more powerful with every generation and stuck pretty close to Gordon Moore's predictions. The processor's power tended to double every two years or so.

In nine, Intel introduced the sixteen bit eight eight six eight six could work on to eight bit instruction sets simultaneously, so it had a parallel processing component in it. However, at the time, most software relied on eight bit processors and couldn't take advantage of this sixteen bit capability, so it's almost like it was future proofed that the software that was out wasn't really able to enjoy the benefit

of the sixteen bit processor at the time. However, there's a generally held belief that's been proven true multiple times that if you build a really super strong processor, it does not take long before someone figures out a way to take advantage of that to build the software that makes the best use or at least takes up a great deal of that new super fast processors abilities. Uh. In fact, you can get to a point where software

bloat outpaces processor performance. So it feels like generation over generation, your processors are getting slower, But that's not really the case. It's just that software is getting more bloated. Sometimes it's a combination of the two. Actually. Well, by this time, Intel was starting to face competition with Motorola, which had introduced its own line of microprocessors in a line known

as the sixty eight thousand. This was the type of microprocessors that would end up being used in Apple products for a very long time. Motorola was the big uh provide leader of micro processors for Apple. For many years, there was a race to see which microprocessor architecture would end up becoming the industry standard, which one is going to be used in personal computers the most, and a big,

juicy contract helped assure that space for Intel. The contract came from International Business Machines Corporation, which is better known as IBM. IBM chose the eight bit processor from Intel, called the eight eight, to be the processor of choice in the official IBM personal computer. Line eight was based off the design of the eighty six, the sixteen bit processor, but the eight was an eight bit processor, so essentially the eight had half of the address bus disabled to

make it an eight bit processor. Eventually, the pairing of Intel chips with the Windows operating system a few years down the line would lead to the nickname Windtell for I b M Compatible machines, and it just showed that Intel was considered a standard part of the IBM compatible universe. You had an Intel processor and you had the Windows operating system. If you didn't have those things, people didn't

really think of it as a PC. They thought of it as something else, uh, which is a little weird, but that just goes to show that, you know, Intel had made some very savvy moves to become indispensable in the personal computing industry, which ended up taking off like gangbusters in the eighties. Now, around this time, Intel also introduced a new innovation in computing called coprocessors. Now, these were chips designed to take on certain tasks that typically

would go to the computer's CPU. But by freeing up the CPU from doing those little tasks, they could run software more efficiently. They could concentrate on the harder computational problems, and these coprocessors would free up the CPUs by taking on those more mundane tasks because they were very specialized microprocessor chips and they were very efficient at handling very specific types of computer problems. This kept the chips efficient and kept their price down, and it also opened up

a new market for Intel to dominate. The company was really growing rapidly at this time. Back when it was founded, Intel had a grand total of twelve employees, but by that number had grown to more than fifteen thousand employees. Intel's founders tried hard to avoid falling into the same traps they had encountered over at fair Child Semiconductor. They were really striving to have open communication between executives and

everybody else. They wanted to make sure that their plans were transparent, that everyone knew which way that the company was going, and that people's concerns were listened to. Because the founders of Intel had come from an environment where they felt like they weren't being listened to, and they didn't want Intel to turn into the same sort of thing. But I've see that becomes a challenge when your workforce numbers in the thousands of employees. So it was not

an easy thing for Intel to try and do. Now, if you remember my podcast about the story of Macintosh, you know I don't like to focus on every single product that a company offers, because it just bogs down the show. With Intel, it's particularly important. I don't do that because the company has has had dozens of different microprocessors in its history and hundreds of other types of products. So we're gonna focus on a few important highlights rather

than a rundown of all the chips. Now, I don't want to be here for a week going over an impressive but exhaustive list of specs. I do want to talk about some of the important early processors, though, because I think it's it shows Intel strategy and also some of the mistakes the company has made over its years. The first one I would like to talk about is the eight one six, which followed a sixteen bit architecture.

Is based off that eight six line from before starting with the until began to incorporate components in the micro processor that would normally be independent on a computer's motherboard. So, in other words, Intel was looking at different parts of the mother board and saying, well, we can lump this in with the CPU. We can increase the response of the CPU, we can make it more efficient, we can decrease latency, and as a result, the entire computing experience

goes faster. This does not necessarily mean the CPU itself is clocking faster. It's just again making that architecture more efficient. So that was Intel strategy. The integration made sense. It made the CPUs more efficient, made the computers much more fast. And later that same year, Intel would release the eight two eighties six, and at this point we would just call it the two eight six. We would also end up calling the next two computers the three six and

the four eight six. They were all from the same kind of family of microprocessors. There were obviously changes and redesigns with each generation, but they were all based off that same architecture. In Intel Creative It's first reduced instruction set computer or RISK microprocessor. These are more specialized processors that concentrate on a relatively narrow band of computer instructions, which means it can go really fast as long as the instructions sent to it are within that narrow band.

It's kind of like a bureaucracy. Bureaucracies can handle forms that are filled out properly very well. That's what a bureaucracy does. But if you have a case that is outside of the normal line, bureaucracies are terrible because they're not signed to process stuff that's outside the norm. If they can process something that's inside the norm, it should

go very fast. Same sort of thing is true with risk microprocessors, except we're talking about computational problems, not licenses and other issues that we would have when we encounter bureaucracies. The first person to theore rise a risk microprocessor was John Coke of IBM Research in nineteen seventy four, and David Patterson, who taught at the University of California at Berkeley,

has the credit of coming up with the term. Intel's I nine sixty risk microprocessors and its descendants would find use in many applications, including military systems. In the three six debuted, and it was the first CPU Intel designed to be fully backwards compatible with previous CPUs. This was a design decision that was carried forward into future microprocessors, and it was also Intel's first thirty two bit x

a D six processor. It could support up to four gigabytes of system memory, which at the time was a truly enormous amount. Nobody really expected to use for gigabytes of memory back in the mid eighties, but the architecture texture was capable of supporting it. Now, that's not to say that everything Intel touch turned into gold at this time. The company also designed a processor called the i A p X thirty two, and as you would guess by the name and number, this did not follow the same

architecture as the X eight six chips did. This was meant to expand its product line and differentiate microprocessors by following a brand new architecture, but the design process did not go smoothly. There were lots of flaws that were introduced into the final processor design, and it was a much more complicated design than the eight six line, and it wasn't particularly efficient. So ultimately Intel would end up shelving the product and saying that this was not a success.

Later in another attempt to break away from the x A D six processor line, Intel would introduce the I eight sixty Risk. While the early Risk microprocessors weren't intended to go into personal computers, this was not true of the I eight sixty. Unfortunately, like the I A p X four thirty two, the microprocessor was flawed. It would stall out when the processor encountered computational problems that were

outside of scope. So again, it's like that bureaucracy. When a form doesn't have the right capability of capturing the information you need, and you go to the bureaucracy, you're gonna get the run around, and it's gonna be painful and slow and laborious. Same thing was true with this processor. It could not handle all computational problems efficiently, and that eventually led to it not being a particularly powerful or

popular UH product in Intel's line. But the four six was a different story that the first x A D six CPU to incorporate an element called the functional processing unit, or FPU, which until that time had been a separate component from the CPU. This decreased latency between the FPU and CPU, which translated again to faster computing speeds. The first of these were running at clock speeds of fifty

mega hurts or fifty million clock cycles per second. Starting a n Intel's ad campaign, Push made the company's slogan Intel inside a common phrase in computer circles, into on a really clever way of convincing computer companies to include Intel chips and to use this phrase Intel inside, both

in the marketing strategy and on the computers themselves. The way Intel convinced companies to do this as they said, Hey, if you include our chips and you include the phrase Intel inside in your marketing, we will pay for half of your marketing costs. So they would shoulder half of the marketing budget of these companies for print and television ads. That's an enormous amount of money and Intel spent millions

of dollars doing this. But as a result, they were able to essentially guarantee that their name would become synonymous with computers. People thought, well, if I'm buying a computer, a PC at any rate, not a Mac, then it's gonna have that Intel and side sticker on there. It's gonna have Intel inside on all the advertising, and Intel became a household name, so it was a very valuable move on Intel's part. In n Intel introduced the Pentium

line of processors. I remember when that happened because it confused me because I was used to two eighty six three eight six for eighty six, and then it went to Pentium. These were still built on the X eight six processor architecture, but they did depart from that eighty six naming system. They decided to go with more trademark

names and less about just numbers. They part of this was probably to appeal to a larger audience of computer consumers, people who probably didn't really care for referring to processors by what seemed to be meaningless numbers to them. Now, as you can imagine, these chips were more efficient and faster than their predecessors, and this helped usher in an era of PC gaming. Among other things, the faster processors allowed game developers to create software that was graphically intense

and fast paced. And while games had always been part of personal computer history ever since personal computers had first come on the scene, they now could rival consoles, which are essentially specific computers designed to run very particular types of software. So a console video game console like the Xbox or back in the day of this the Super

Nintendo Nintendo Entertainment System. It's just a specialized computer. But uh, Intel was able to show that with these Pentium processors, personal computers could run games that would rival even the best consoles. But there was a big problem. Those first Pentium chips had a design flaw in them that would

cause the microprocessor to make it incorrect calculation for certain processes. Now, this incorrect calculation didn't happen frequently, but it did happen, and while Intel had hoped that perhaps the flaw was so minor as to go unnoticed, that did not happen. The reason that was noticed was due to a mathematician named Thomas Nicely. It was a professor Nicely was actually working on a math problem involving twin primes. So a twin prime is a set of two prime numbers that

differ by just two. So three and five are both prime numbers and their twin primes because five miles three equals two, five and seven are twin primes. Seven milus five is two. But as numbers get larger, primes become less frequent, and twin primes become even more rare. The philosopher Euclid created a proof that suggested there's an infinite number of prime numbers, and he believed that the same was going to be true of twin primes, but he

had no way of proving it well. Euclid was thinking that way back in three hundred b c. It wouldn't be until the twentieth century before someone came up with a means of describing this mathematically, and that person was Vigo Brune, a Norwegian mathematician. He said that the sum of the reciprocals of twin primes would converge to a constant sum that later on we would call Bruns constant.

And in case you're rusty on your mathematical terms, a reciprocal of a number is what you get when you take one and divide it by that number in question. So let's say you start with five. The reciprocal of five is one five, because again you're looking at the one divided by the number you had started with, or

point two if you prefer. So. Brun figured out that if you took the reciprocals of twin primes and you added the two reciprocals together, they converged on a number that would end up again being called Bruins constant, and by nineteen seventy six that constant was calculated to be approximately one point nine oh two one six o five four for twin primes up to one hundred billion. But hey, that's not good enough for math. We need to be

more specific than that. So enter Thomas Nicely, who bought a computer in n with a Pentium processor and he wanted to run calculations related to Bruin's constant. But as he did so, he was coming up with errors that should not have been there, and he was eventually able to track it down to a design flaw and Intel's chips that only showed up if you were doing some really big calculations. That's when the flaw would make itself apparent,

but most people would never see it. At first, Intel tried to shirk this issue, but intense pressure caused the company to change its tune. So eventually Intel issued a recall offering a free replacement of the affected print Pentium chips with a modified chip that corrected this error, and that recall would end up costing Intel four hundred seventy

five million dollars. Ouch. Well, we've got a bit more history of Intel to talk about But before I wrap all this up, let's take another quick break to thank our sponsor. In Intel would debut the Pentium two. They actually had a slight missed up between the Pentium and the Pentium two called the Pentium Pro that didn't go over so well, but the Pentium two was a slightly different story, and Intel also introduced two new product lines

of processors, the Cellern and Zion brands. Now, these were based off of Intel's existing architecture, but they had very specific implementations where some of the bells and whistles weren't present on them. This allowed them to be used for very specific implementations, like you could have mid range UH servers for example, or lower priced computers because you had the same sort of processors but with some of the features turned off essentially in these more mid range machines UH.

That was Intel's attempt at serving all different levels of the market, not just the premium customers. So later on, from that point, Intel introduced the Pentium three, which would become the first microprocessor to break the one giga hurts

clock speed. Intel had been in a race with its competitor a m D to create the first commercial processor that could hit a giga hurts clock speed, and at one point Intel even had a model of the Pentium three that had a processing speed of or a clock speed I should say of one point one three giga hurts, but that one was unstable. Uh A. A review found the performance was unstable and so that one got recalled. But still gigga hurts clock speed meant that it could

complete a billion instructions per second under ideal circumstances. By the late nineteen nineties, Intel was starting to get into the business of buying other businesses. It was acquiring other businesses to get into new markets that included wireless communications products, networking components, and controlled chips for specific types of applications, such as the kind you might find in v equal control systems. And if you listen to my shows about

the Macintosh, you know that. In two thousand five, Steve Jobs shocked the tech world when he announced that Mac computers would move away from their traditional CPUs and use Intel CPUs instead. Intel maintained a competitive advantage over others in the space, but this also came at a pretty stiff price. In two thousand nine, the European Union found

guilty Intel guilty of alleged monopolistic actions. In other words, they said that Intel was effectively acting like a monopoly within the European Union, and they hit Intel with a

fine that was one point four five billion dollars. That same year, Intel had to pay its competitor, A m D, a hefty sum of one point to five billion dollars as part of a settlement in which am D had accused Intel of using leverage on computer company is two put Intel products in instead of other processors into those computers.

So essentially am D was saying Intel was being anti competitive and really putting the screws to computer manufacturer saying you've got to use Intel products and not anyone else, and uh, that is not legal, which is why the

lawsuit happened, and eventually the two settled. Now I could talk about how every generation of Intel's chips were faster, included more on board memory, had faster bus speeds, greater clock speeds, lower latency, but we can just sum it up to say the company kept following a strategy that it branded the tick talk strategy. So let me tell you what TikTok it means and and why that has

recently changed. So TikTok is a two part strategy to developing the microprocessors of today or really yesterday at this point, because again we're just now beyond TikTok. The tick part involves looking at the architecture you designed on your previous generation of micro chips. So you look at that architecture and you look at the size of it, and you decide, how can I shrink down these components to the next

smaller size. That is your tick. So you might look at a generation of processors that have forty five nanometer components, and your challenge is to figure out how to shrink down that same design so that the components measure thirty two nanometers not forty five. And you follow essentially the same blueprint as you did before, only you can cram more stuff into that blueprint because you reduced the size of the individual components. You freed up some space by

making everything smaller. However, at these sizes, how those components are arranged matters more and more. So you could just keep shrinking the components down as your research and development allows you to make ever smaller pieces, but that only gets you so far. What you really need to do is figure out how to arrange those new smaller pieces in the ideal configuration to get the most efficient use

out of them. And this is the TALK step. So in TICK you figure out how to shrink stuff down further, and talk you figure out how to rearrange these new smaller components so that they work the best way you possibly can make them. So in our example, we would be looking at those thirty two nanometer components and figuring out the right architecture to maximize their efficiency. And the TikTok.

Generations of a single family of processors are going to have the same size opponents, They're just gonna have different configurations. The TICK is going to be based on the previous generation. The TALK is the new architecture to maximize the efficiency of the new sized components. And then your next TICK is going to be taking those thirty two nanometer components and figuring out how to shrink them down even further, but within the same general architecture as your previous generation.

Tick talk, tick talk. That was pretty much how Intel ran things uh for quite a few generations of processors.

Now Intel follows a new strategy. It calls it the process architecture Optimize pattern others will call this tick talk talk because again, the first one is shrinking components down to a new smaller size, the second means finding a better arch at lecture for those sized components, and the third one is refining that design even further, so you're staying on the same size of components for three generations

in a row. UH. This means that you don't have to spend so much time trying to figure out how are you going to shrink things down even further as

you gradually get closer and closer to a fundamental physical limitation. UH. That being where quantum physics comes in and doesn't play nicely with your designs anymore, and you get things like electron tunneling where electrons seem to leak through transistors, and since transistors are meant to control the flow of electrons, this is what we in the computer biz often call

a bad thing. It introduces the possibility of errors and miscalculations, and you don't want that in your processor, so it ends up extending the amount of time Intel spends on a specific size of dye for their their chips, but it also maximizes the efficiency of that while engineers continue to work on the next breakthrough. One other thing I need to touch on before I conclude is the concept of multi core processors, because Intel story doing that as well.

Intel's work with parallel processing way back in the nineties provided the basis for multi core processors. Multi core processors can handle several computational problems simultaneously, which brings the clock speeds to bear on the problems to solve them in parallel rather than in sequence. So one core might be working on one problem, another core could be working on a separate problem, So dual core processors. You could do two of these, quad core four and so on and

so forth. And there's also threading as well, you can thread different copy stational problems. But ultimately the concept we need to really focus on here is that idea of parallel processing, because for certain types of computational problems, parallel processing is much much faster than doing sequential processing where

you're just going down a list of instructions. Even if you make a really really super fast CPU, if it's tackling a very long list of instructions that could otherwise be divided up, a multi core processor might be more

effective than a very fast single core CPU. I like to use an analogy whenever I talk about this, and long time listeners of tech stuff are probably familiar with us because I've used it before, but I find it's very helpful if you're trying to understand the difference between a super fast CPU and a fast but not as crazy fast multi core processor. So we're going to imagine a math class and you've got essentially sixteen kids in this math class, right, one of those kids is a

math genius. She's a prodigy. She's so smart she can complete any problem in a fraction of the time of any of her other classmates. Her classmates, by the way, are smart. They're they're not They're not slower anything, they're just not geniuses like she is. Now, on any given singular problem, the genius is always going to solve it faster than her her other classmates. They're just never going to be as fast as she is for any one problem.

But imagine the teacher hands out the test, and the test has fifteen problems on it, So there are fifteen problems you have to solve, and they're sixteen kids in the class, including the genius. The teacher gives the genius a test that has all sixteen, all fifteen problems on it. So she has to solve all fifteen of those problems as quickly as she can. But he tells the other fifteen students, you each will tackle one of these problems,

and he assigns them. Student one has problem one, Student two has problem to, and so on all the way down the fifteen problems, and they have their job is to complete their problem, their one problem before the genius can complete all fifteen problems. Well, this is sort of

what multi core processors are capable of doing. They divide up problems into different parts, and collectively they can solve that big problem faster than a really super fast processor could nine times out of ten, ninety nine times out of a hundred, nine times out of a thousand. Those fifteen students are going to finish their individual problems before the math genie can work through all fifteen on her test. And you might say, well, that's not fair to like,

that's not the point. It's an analogy to explain how multi core processors work from a kind of high level approach. It gets a lot more technical than that, obviously, but that's just to explain that for certain types of computational problems, parallel processing is much more effective. Now, there are types of computational problems that cannot be broken into parallel processing, and for those a super fast CPU is still more often than not going to be more effective than a

multi core processor. So it really just depends upon the application. But more and more we're seeing parallel processing problems being the type that computers tackle. Now, how long will Intel and other microprocessor manufacturers be able to keep Moore's law alive? Because people are constantly predicting the end to Moore's law has happened numerous times throughout history. From the eighties on.

People will say, oh, Moore's law is coming to an end because it's not physically possible for us to keep up with it. But so far, engineers have been able to stave that off, partly through innovative architectures that don't necessarily increase the number of components by a factor of two, but rather increase the output of the processor by a factor of two every two years. So it requires some

reinterpretation of what Moore's law means. It may mean that you fudge a bit on the amount of time necessary to get to that factor of two. And it might mean reinterpreting it as processing power versus number of discrete elements. But I think the important thing for us as consumers is the performance, not whether the chip actually has twice as many transistors as the one from two years ago.

So that is still a battle that's going on today. Uh. It does mean that maybe we'll hit that fundamental limit at some point and we will not be able to continue More's law using traditional microprocessor designs, and then we may need another true revolution in computer processing to create a new means of keeping up with that power output. But maybe stepping outside of what a processor is supposed to look like, that's a possibility today. Intel is doing

pretty well as of the recording of this podcast. When I checked it was its shares were trading at around thirty five dollars a share. That puts the market cap value of Intel at about a hundred and sixty five billion dollars. The company may nearly sixty billion in revenue in twenty six, which is not too bad for a company that was found by a couple of traders. Well done, More and Noise. That wraps up our two part episode

on the story of Intel. Obviously, there's a lot more I could talk about from the various sensors and processors Intel has been developing for all sorts of applications to their partnerships with various companies throughout its history, and maybe in future episodes I'll touch on some of those, but I thought that it was really important to just kind of hit the high points to understand where Intel came

from and how it developed over its years. If you guys have any suggestions for future episodes of Text Stuff, whether it's a topic I should cover, someone I should have on to interview, a guest co host who might be able to tackle a specific subject with me, let me know. Send me an email. The address for the show is tech Stuff at how stuff works dot com, or you can drop me a line on Twitter or Facebook. The handle for both of those is text Stuff h s W. Remember you can tune in on Wednesdays and

Friday's to see me stream this podcast live. You can watch me record it live, which includes all the ridiculous mistakes I make and when I have to restart, and sometimes I'm chatting with the chat room and just answering questions or shooting the breeze. Go to twitch dot tv slash tech stuff. To see the schedule there and join me sometime, won't you. And in the meantime, I hope you guys have a great week and I'll talk to

you again really soon. For more on this and thousands of other topics, is it staff works dot com.