Advent of Computing - Episode 7 - Attack of the PC Clones

Episode Date: June 30, 2019

Today, I want to share with you the story of the first PC clones and how they cemented the rise of the x86 chipset.   Most of this story takes place between 1981 and 1984, but I think it's fair to say that these 3 years are some of the most influential for the PC's rise to domination. So lets start the story with a discussion of the IBM PC, how it was special, and then examine how reverse engineering it lead to the current x86 monoculture we see today.

In June of 2005, Steve Jobs announced that Apple would stop using their custom PowerPC architecture and instead start using Intel's x86 chipset in all future Macintoshes. At the time, this shocked a lot of diehard Apple users, since this change would essentially make every new Mac just another personal computer. would essentially make every new Mac just another personal computer. And in the process, this would also destroy the last bastion of resistance against Intel's monopoly in the home market. The next year, 2006 came, and this was the first year that Intel Macs went on sale. Since then, almost no non-Intel based systems have shipped from any large-scale manufacturer. These days, all consumer laptops and desktops and even most enterprise-grade hardware, with a few notable exceptions, are running Intel-compatible chips under the hood. But this transition wasn't anything new, and it didn't start in 2005.

Far from it. In fact, the shift first started sometime in the 1980s. So how did Intel's x86 chipset take over the world, and how is that connected to some of the early efforts to clone IBM hardware? Welcome back to Advent of Computing. This is Episode 7, Attack of the PC Clones. I'm your host, Sean Haas. Today, I want to share with you the story of the first PC clones and how they cemented the rise of the x86 chipset. Just to give some quick terminology right off the top, x86 is the shorthand to refer

to a family of Intel chips, specifically those modeled after Intel's seminal 8086 processor. There are other companies that make x86 compatible CPUs, you may be familiar with AMD for instance, but at the release of the original PC, Intel was the only show in town. Most of the story is going to take place between 1981 and 1984, but I think it's fair to say that these 3 years are some of the most influential and important for the PC's rise to dominance. So let's start the story off with a discussion of the IBM PC, what it was, how it was special, and then examine how outside companies reverse engineering it led to the current Intel monoculture that we see today.

Sometime in early 1980, a team of IBM engineers in Boca Raton, Florida started an ambitious new project. Up until that point, IBM had primarily been a mainframe company. They had tried to break into the home market in the 70s, but the machines never really caught on at home. The Boca Raton group, called Project Chess, was trying to change that. Many inside of IBM had doubts that any machine they could produce would be able to capture a large enough percentage of the home market to make the endeavor worthwhile. But after only a year in development, Project Chess would prove all doubts wrong. 1981 rolled around and the secretive project

was released as the IBM 5150, better known to history as the IBM PC. Almost overnight, the PC became a total hit, finding its way into both the home and business markets. In fact, it sold so well that sales quickly outpaced supply, leading to a shortage in units that IBM could ship. A lot of this machine's success was due to its competitive price and very good feature set for the time. The base model retailed for just over $3,000, which was on par with other home computers in that era. But the PC had something special under the hood that made it stand out. One of the key design features of the IBM PC was that it used, almost exclusively, off-the-shelf

parts. Only a few components were proprietary to IBM. Now, this had a few large implications. First off, using generic components was really the only way that the PC was able to be developed so quickly in just under a year. But that also meant that IBM was not able to patent the machine, at least not outright. Mainly just because there wasn't enough trade secretness to patent the machine, at least not outright, mainly just because there wasn't enough trade secretness to patent in the system. In addition, it was easy to repair or replace broken components in a PC, and this also meant

that third parties could freely design peripherals and upgrades to the machine. The PC was centered around the Intel 8088, one of the first generation of x86 central processing units. This was an early 16-bit microprocessor targeted at the smaller computer market, in a time where most home computers were only 8-bit. For the time, this was a really capable chip, running at around 5 MHz and able to address up to 1 MB of RAM at any one time The rest of the system came down to disk drives and a couple peripheral cards

The stock option PC came with two five and a quarter inch floppy disk drives for storage This was before our hard drives were really common or at a reasonable price, so this was as good as you could get. The video output was handled by either a monochrome card, able to display up to 80 columns of text, or a color graphics CGA card. The system came with 16 kilobytes of RAM, but could be expanded through new RAM chips socketed directly into the motherboard, or larger RAM expander cards. Just as a quick aside, the PC also came bundled with the very wonderful IBM Model F Buckling Spring Keyboard.

According to many enthusiasts, myself included, this is among the best feeling and most reliable keyboard ever made. In fact, most of the research and scripting for this episode was done via an IBM Model F. The final puzzle piece was the PC's expandability. The motherboard came with five internal expansion slots. For the time, that alone isn't remarkable. A lot of computers could also be expanded internally. But it's how IBM handled the internal expansion that's important, and I'll explain more about that in a little bit here. Now, I'm rattling off the machine's specs to get to an important point. The PC wasn't really a powerhouse, even in the early 80s. It was a very capable machine, but it wasn't a bolt from the blue in terms of new exciting functionality.

So what made the PC such a smash hit? Well, it turns out that breaking IBM's normal proprietary nature gave the PC a huge edge in the market. But this went beyond just using off-the-shelf parts. Despite the original secrecy surrounding the development of the PC, IBM would end up releasing the documentation covering all the inner workings of the machine, known as the IBM PC Technical Reference Manual. This information allowed third-party developers to make new hardware expansion and software for the machine

without having to work directly with IBM. As a result, the PC was widely supported even from its initial release. The personal computer would end up becoming the basis for our modern systems, but there were a few stumbling blocks in its way still. It turns out that despite the PC project breaking from IBM's tradition, there was still one proprietary component of the computer. So where was the secret IBM proprietary-ness and copyrights? Well, the first one of two was the logo on the badge of the case, was the logo on the badge of the case, which really isn't that important to the full story.

It turns out that the copyright that matters wasn't even in the hardware, but instead, a small piece of code nestled deep inside the PC, the BIOS, otherwise known as the Basic Input Output System. The system BIOS is a very small program stored on a ROM chip on the computer's motherboard. The purpose of this program is to provide a way for other software to manage the computer's hardware. This ranges from being able to boot up the machine, running boot time diagnostics, and providing an interface for any other software

running on top of the computer to access the underlying hardware. Basically, you can think of the BIOS as the glue that holds the hardware together. Without this chip, a PC can't boot. But beyond that, a lot of PC software also require the BIOS to function. So, on a PC, you can directly access the hardware, but that's somewhat complicated. It's harder to program. The BIOS, however, provides a simple-to-use interface for programmers, making it easier

and faster for them to develop new programs. And if we have any recurring theme in this show, it's that programmers are lazy and fickle, so a lot of them just defaulted to using the BIOS. are lazy and fickle, so a lot of them just defaulted to using the BIOS. To strengthen that, IBM also strongly recommended that developers use the BIOS exclusively instead of attempting direct hardware access. As with most conveniences, however, using the BIOS came at a cost. Turns out that like most conveniences, the BIOS runs a little bit slower than direct hardware access. So some programmers did disregard IBM's specification and didn't use the BIOS at all. But most did stay with the BIOS train.

The net result of all of this is the PC's hardware is dependent on the BIOS to operate, but also almost all software has to have access to IBM's BIOS to run. If you remember back, IBM had spilled all the beans in terms of the technical docs for the PC, and this included the source code listing for the PC BIOS. But the complication here is that IBM held the copyright to that code. This meant that even though the IBM PC Technical Reference Manual laid out a whole blueprint with all the required hardware and software to make a PC, no competitor could totally copy the machine. And it would be impossible to create a PC without some form of IBM-compatible BIOS. But if you just copy it directly, you're violating a copyright.

And that's grounds for IBM to sue your company out of existence. So that was the state of things directly after the August 81 release of the PC. Sadly for Big Blue, things wouldn't stay that way for long. By December of that year, there was already talk of companies working to develop their own versions of the PC. After seeing IBM's unprecedented success and their inability to meet demand, a lot of computer companies wanted to get in on the PC bandwagon with a system of their own. And the key to that strategy was to make a fully compatible PC clone, since that would make it easy to slide right into the new booming market and profit off IBM's short-term supply issues.

But to create a fully functioning PC, the proprietary BIOS would have to somehow be dealt with. One option would be to license the BIOS from IBM, but to my knowledge, no company ever went this route. My guess would be that simply IBM didn't want to lose their monopoly on the market. Another way to circumvent the BIOS would be to create a competing standard. Just write enough code that you could bootstrap the machine and then offer some different hardware interface. The problem with that method is that a lot of software did rely on the BIOS environment.

If a company broke from the established standards, then they'd lose most software compatibility and have to try to carve out a niche in the market for themselves. The final option was to create a legally distinct clone of the PC BIOS. was to create a legally distinct clone of the PC BIOS. The key here is to rewrite the BIOS code in such a way that it's 100% compatible with IBM's version, but also would be impossible for IBM to sue you over. A lot of companies skirted the line between a total reimplementation of the BIOS and custom machines. This resulted in quite a few PC lookalikes with small tweaks and limited compatibility. This included machines like the Tandy 2000 or the Deck Rainbow.

While these systems were capable, they had issues in the market partly due to this lack of compatibility. However, with all the competing PC-like machines, it was only a matter of time until someone would crack the market wide open with a 100% compatible clone. Sadly, it turns out that finding which computer was the definitive first PC clone is a somewhat difficult task. There were dozens of systems put out to the market in the first few years after the PC that could fit the bill, but in most cases, the product or the company failed. This means that there's not really that much of a paper trail out there about the development of a lot of these computers from failed companies, except for some articles on old PC magazines and mentions from trade shows. articles on old PC magazines, and mentions from trade shows. On top of that, a lot of these computers boasted totally clean BIOS with 100% compatibility,

but either had compatibility issues or were violating IBM's copyright. Suffice to say, I'm not going to be coming in with the big truth about the definitive first IBM PC clone, but the information is somewhere out there. However, I would like to discuss one machine that did make a splash in the market and continued to stay relevant. That machine was first announced in November of 1982, the Compaq Portable. This new and exciting Compaq machine was nearly identical to an IBM PC. It came stocked with two 5.25-inch floppy drives, an Intel 8088, and 128 kilobytes of RAM. Off the bat, it has more RAM, but that's just about it.

And most importantly, all software that worked on the PC worked on the Compaq Portable. So how did Compaq solve the BIOS issues for themselves? They used a technique called clean room design. This is a way to reverse engineer a program without violating the original copyrights. For Compaq, it worked something like this. One of their engineers read through the source code listing that IBM had published for their BIOS. They then wrote up a specification for the BIOS, basically a document that outlined everything that the BIOS did and how it interfaced with the hardware and other programs without

mentioning how it accomplished that on the source code level. Then, a separate team that had never seen IBM's code so they were quote-unquote clean, wrote their own BIOS from scratch using the specification as a guide. Once done, Compaq had their own work-alike of the PC BIOS that was legally distinct and safe from IBM. There was, however, one key difference between the PC and the Compact Portable, and, well, it's right there in the name. The Portable was a portable computer. Specifically, it's what was known as a Luggable. That shouldn't sound very familiar because it's not a form factor that really exists

anymore. A Luggable computer is the same size and weight as a desktop computer, but it has an integrated screen and keyboard. You can quite literally lug one of these bad boys around from point A to B. In the case of the compact model, it had an integrated 9-inch monochrome display and weighed somewhere in the neighborhood of 28 pounds. Development took about a year, but once done,

the machine proved to be a massive hit. While there were other PC clones on the market around the same time, Compaq was the first to make a PC clone that really stuck. In the first year on the market, Compaq shipped over 50,000 portables. But more importantly, IBM never sued. This success would lead to many more PC compatible systems flooding the market, including from Compaq. Each company brought their own clean room BIOS to the table. While this did increase the popularity of the PC platform, there was still a key piece of the PC platform, there was still a key piece of the puzzle missing. It turns out that Compaq

and other manufacturers weren't thinking big picture enough. If you could reverse engineer IBM's BIOS, then why not make that your product? In 1983, Phoenix Technology would do just that. The business plan was simple. Phoenix would create their own cleanroom BIOS and then sell licenses to use their code to other PC clone makers. By making an IBM compatible BIOS available, this meant that computer manufacturers didn't have to go through the difficult process of creating their own BIOS version. So how did Phoenix create their BIOS? It turns out that they used a method very similar to Compaq, but with a few more safety measures, at least legally speaking. To start

with, Phoenix had a team of programmers read and analyze IBM's code and then produce a specification for recreating it. So far, this sounds exactly the same as what Compaq did. But the next step was different. Instead of having another whole team of programmers, only one programmer was hired. And beyond that, the programmer was chosen explicitly because they had no experience with the PC platform or the x86 CPU family. Instead, he was most familiar with the TI TMS9900 chipset. Basically, for the 80s, this was the furthest you could get away from Intel's chips. This ensured that the new BIOS code would be as clean as possible, even from accidental code contamination.

The finishing touch was the paper trail. The team that wrote the spec for the BIOS and the programmer only communicated via written memos, which were all kept as documentation of the process. This way, Phoenix could easily and very explicitly show what information their programmer had access to. So if IBM decided to sue, it would be easy to prove that no copyrighted information from Big Blue's code had made it into Phoenix's chip. To shore up this protection from IBM, Phoenix even took out an insurance policy against future possible lawsuits.

Licenses for the new BIOS started selling almost immediately. Some of their first clients included giants like AT&T, HP, and Tandy. Manufacturers loved this arrangement, since it let them very easily and cheaply create new PC compatibles while shifting any legal weight onto Phoenix. This made the barrier to entry into the PC clone market almost non-existent. If you could build a computer, you could now just as easily and legally build a PC clone, so there's really no reason not to get in on the burgeoning compatible scene. As IBM released new machines with new upgraded BIOS chips, Phoenix kept up the same clean room procedure. When the new PC-AT was announced by IBM, Phoenix took only six months to release their updated BIOS.

In an unsurprising plot twist, Phoenix wouldn't be alone in the new licensing market for long. In 1986, American Megatrends released their own clean BIOS called AmiBIOS. This new BIOS had a twist. While Phoenix licensed the compiled and complete BIOS chips, AmiBIOS was licensed as source code. This meant that manufacturers could further customize and bespoke their own BIOS without having to create it from scratch. spoke their own BIOS without having to create it from scratch. Essentially, there was now no downside to using an off the shelf BIOS alongside your off the shelf hardware. By 1994, 70% of the PC clone market was using AMI BIOS.

This included giants like Dell. In a lot of ways, the clone had become the cloned. And really, this helped the PC become the platform of choice in the 80s and beyond into the present day. Access to generic but compatible BIOS made it easy and cheap to make generic but compatible computers. By the 90s, IBM lost all dominance and control in the home market as generic PCs took the world by storm. With the gates now open, the PC market became a race to the bottom price-wise, and that made the platform all the more attractive to consumers. For a cheap

entry price, you could get a home computer for the first time, and it would be compatible with every other home computer everyone you knew owned. Alright, I think it's just about time to wrap up for this week. I want to end off with a few closing thoughts. Prior to the PC being cloned, the computer market was fractured into an uncountable number of incompatible and usually very expensive systems. There had been attempts at similar marketing strategies through cloning the Apple II, but Apple quickly crushed those with lawsuits. Between IBM's published PC designs and generic BIOS from Phoenix and other similar manufacturers, a go-to design for new computers started to form.

And the coming year, cheaper and cheaper generic PCs hit the market, bringing more and more computers into the home. And that takes us to where we are today. Almost by accident, IBM was able to create an insanely popular computer that was insanely easy to reproduce. As the barrier to enter the PC market lowered, the PC architecture became the dominant force and basically a default choice for new manufacturers. And for better or for worse, the x86 architecture became THE architecture for computers into the modern day. Thanks for listening to Advent of Computing. I'll be back in two weeks time with a new episode. This time, I'm thinking of telling the story of how the BBC created a new generation of programmers.

Until then, if you like the show, go ahead and share it with a friend. You can also rate and review on iTunes. If you have any comments or suggestions for an episode, go ahead and shoot me a tweet. I'm at Advent of Comp on Twitter. And as always, have a great rest of your day.