Advent of Computing - Episode 166 - Beyond the PDP-11

Episode Date: October 5, 2025

My trilogy on the PDP-11 concludes with a look at the far flung places this computer can take us. In this episode we look at some issues with claims of the PDP-11's linage, smuggling, Hungarian-made microcode, and much more. Along the way we answer the question: if the PDP-11 was such a good design then where was it during the home computing boom of the 80s?  

I'm now two months into reading and learning about everything PDP 11. I had this informal goal of learning more about deck hardware this year, and, well, I think I've had enough at this point. I think I can mark that goal down as completed. One thing that continues to surprise me is that the PDP 11 story expands so far out. What starts as a simple tale of engineering at deck explodes in scope. actually pretty quickly. We hit a point where the story has very little to do with deck itself, almost like how the IBMPC loses its first three letters after a few short years.

As a software dude, my mind is a little warped when it comes to this kind of stuff. Computers are inhuman things, and instructing computers requires us to kind of bend our minds to think inhuman ways. One of those warps is that I tend to think of computer. as settings for events. The PDP 11 quickly moves into being a setting. Entire sagas of the Unix Wars happen on PDP 11s. Micro processors were designed using software on PDP 11s.

So were many operating systems for home computers. The machine was a cornerstone. Due to that, entire generations of programmers were exposed to the computer. Now, I know I say that, a lot, and it may sound casual, but it's not. Let me try to make this a little more clear. In programming, exposure is a massive factor, especially for people with certain learning styles. Let's say you're working a job and you need to learn a new language. It's actually pretty rare for people to, say, go take a college course for that new language. Some people may hit the books and learn that

way, but that can only go so far and it only works for certain learning styles. What I usually end up doing is diving in headfirst. I read code examples, I look at existing software, and I write piles and piles of bad software. I expose myself to that language or that new tool or whatever I'm trying to learn. One of my professional specialties is PHP. It's an awful language that people shouldn't use in the 21st century, and I'll leave my analysis at that. It's one of my specialties because years ago, maybe over a decade at this point, one of my bosses sat me down and asked if I knew PHP. I said no, but that I would be willing to learn. I was then dumped onto a server that could run PHP and given a project.

Exposure leads to experience, and experience leads to a certain way of thinking. I have some coworkers these days who've never used PHP, and I can tell easily. I've used language so much, been exposed to it so deeply, that I'll slip into using PHP-specific lingo when talking about other programs. As a result, I sometimes have to stop and translate. I'll find myself taking PHP shortcuts in other languages, then get a little confused. some things don't actually execute. If I'm particularly tired, I'll start all my variable names with dollar signs. I'll explode when I should split. When you've been that exposed to a system,

be it a computer or a language, it comes out in these kinds of deep ways. How would you design a new computer if you were taught to program on a PDP 11? How would that shape your view of computing. Now, imagine a whole generation thinking that way. Where could that leap? Welcome back to advent of computing. I'm your host, Sean Haths, and this is episode 166, beyond the PDP 11. Today, we're finishing up my trilogy on the PDP 11. So far, we've seen how the computer was created and how it adapted to the advent of the microprocessor. Now, we'll be leaving deck behind, almost completely, and looking at more strange expressions of the PDP 11's design.

Central to today's topic will be two concepts, influence and theft. We already spent a lot of time last episode talking about the PDP11's influence. Chips in the latter part of the 70s took many design cues from the PDPP. So the most famous were the Intel 8086 and Motorola's 68,000. For those microprocessors, the influence was undeniable. The 8086 is very close design-wise to the PDP 11, and the 68,000 was explicitly modeled after that older machine. What about when influence is less clear?

What about when that influence is called into question? It turns out that the PDP 11 and the IAS machine, both face a similar issue, that being claims to lineage. The PDP 11 introduced so many foundational concepts to modern processor design that lines get blurred. Today, we'll be looking at one example of those blurred lines and how prominence can warp history. On the flip side, we have actual clones.

There were a number of one-to-one clones of the PDP 11. And I'm not talking about like an unlicensed copy, of the LSI 11 microcode. Instead, I mean totally unofficial and unauthorized clone computers. The best known of these is the 1801 series of CPUs, which were developed in the USSR and used in, get this, home computers. This may sound like a bit of a grab bag today, but there is a deeper purpose at work. I want to answer a simple question. If the PDP11 was such a well-designed computer, then why did the microcomputer revolution pass it by? We already have a partial answer to that question.

We saw last time that by the end of the 70s, the PDP 11 was a little bit outdated. That made it vulnerable to newer designs. But I think there's still a little meat on them there bones that we can pick at, so to speak. Besides, this gives us an excuse to look at the strange periphery of computing. So, let's begin. We start off today by placing some blame. There's actually an inciting incident for this entire series. When I was at VCF West this year, I ran into a listener who was trying to pitch me on an idea for an episode.

I'm so sorry. I forget your name, but you know who you are. We had a lovely conversation in the coffee shop of our hotel. He was trying to convince me to, cover the computer expansion for the Intellivision game console. Now, he did convince me of how funny the device's history was. I will probably fold that into an episode eventually. But he also kind of nerd-sniped me.

When we were talking, he said something along the lines of, well, you know, the Intellivision used a CPU based off the PDP-11, right? The answer to that question was, no. I had never heard that one before. Again, this is why I love going to VCF so much, and this is why I love computer history so much. There's always something new. But this did surprise me, and thus, here we are, two months into a wild deck-themed ride. So, let's cut to the facts.

The Intellivision is released in 1979. It's powered by a General Instrument's CP-1610 processor. First off, not a PDP 11 on a chip. But we're not nearly done here. You do know what podcast this is, right? This is where the rambling path begins. The CP 1610 is a revision of the CP 600, which was released in 1975. It's actually one of the first 16-bit microprocessors, which is pretty neat.

It's also somewhat controversial. or at least there's controversy around its influences. Was the 1600 similar to the PDP 11? Well, to start with, let's have a quick refresher on what we mean by PDP11 like. We're looking for lots of general purpose registers, memory mapped I.O. devices, complex instructions, and a few instructions with specific names like Move. So how does that line up with the CP-1600? Well, the General Instrument chip does have a lot of registers.

It actually has an identical loadout to the PDP 11. We get eight general-purpose registers, with two of those pulling double-duty as a stack pointer and an instruction pointer. The registers are even named R-0 through R7, which is exactly the same naming convention used by the PDP-11. So that lines up perfectly, right? At least, that's what it looks like if you read a spec sheet. There's a problem here, though.

When you look deeper into the documentation, you find that these registers are not actually general purpose. The big tip-off is the instruction set. The PDP11's instructions are a master class in simplicity. Each instruction reads like this. operation, source, destination. The source and destination arguments set a few modes to let everything work out nicely. Basically, is this source a register, a location in memory, or something more complicated?

It's very clean, very simple, and it's very orthogonal, meaning that you can use any source and any destination with any operation. Just nice. This makes it possible to tell the PDP 11 to move data from one. and register to another, or from a register to any memory location. It's a really slick setup. The CP-1600, on the other hand, has something pretty different going on. Its instructions read as Operation, Mode, Destination.

You'll notice there's no source there. That is a critical clue. The mode part of each instruction selects the source kinda. I say kinda because there are a bunch of caveats and limitations. Basically, you can use some registers as source and you can use some registers for operations

like referencing memory addresses, so pointers. If all this is over your head, just look at it this way. The CP-600 physically has general purpose registers but I can't talk about all of them in a general purpose way. There are limitations on how it can express that kind of information. Let me give a concrete example. Let's look at the lovely register R5. You know, old R5, it's one of my favorites.

On the PDP 11, it's just 16 bits of storage. On the CP-1600, it's also 16 bits of storage, but it can only be used in one specific mode. Further, it has a trick. When you reference R5 after the operation, it automatically increments itself. Now, in isolation, that just sounds weird, right? Why would you do this?

Well, it's actually mimicking a feature from the PDP 11. There is a specific way to reference a register called auto increment on the PDP 11. That means read that register, then add one to it when you're done. That's really handy for programs that deal with lists of data. You basically can have a pointer that goes from one address to the next to the next to the next. It makes loops easier to write. So the CP-1600 is implementing this cool auto-increment feature

just in a much less general way than on the PDP11. That's a clear influence, but not a copy of the design. The R4 register, another one of my favorites, does the same auto-increment trick. You can use R1, R2, and R3 to do indirect addressing, basically pointers with no auto-increment. Then R0 has no special role. It should be pretty clear here that the CP-1600 is inspired by the PDP-11, but its features don't really implement every key part of the 11's features. What made the 11's instruction set so cool is that it was generalized and orthogonal.

You can do anything using any source and any destination. You can even move from memory to memory, or add data from memory to memory through indirect pointers and auto-increment one of those involved registers. It's powerful, it's flexible, and it's generalized. The CP-1600 has the sheen of the PDP-11, but not exactly the designs to back it up. That said, there are a lot of similarities. Input and output are the next broad category.

The PDP-11 was famous for having no dedicated I.O instructions, but instead, only using memory-mapped I.O. All external devices just appear as if they were part of memory. The CP-1600 follows that pattern to a T, but that's honestly a small point, and not a hugely specific one either. Many other machines had started moving towards memory-mapped I.O. at this point. The CP-1600 also has complex instructions. Let's take its add as an example. You can work up an instruction that takes a value from an address in memory as specified by the value

of R5, adds that value to R0, and then increments the value of R5. That is multiple steps all expressed by one instruction, which is a pretty clear complex instruction. That lines up nicely with the PDP 11, you know, the classic complex instruction set computer. But again, things diverge pretty heavily. Let's consider those instructions again. specifically let's think about one move i've kind of been using that as a quick yardstick for the pdp11 well the cp sixteen hundred doesn't have move at least not really it has two equivalents

move in and move out mv i and mvo those are used to move a value from memory into a register or a value from a register out to memory. Put another way, it's load and store. That kind of makes this look like a load store architecture computer, something that the 11 is not. We can also look at something more basic, like how the 1600 breaks up data. For this to make sense, I need to explain another peculiarity of the PDP 11.

That machine uses a 16-bit word. That's why it's called a 16-bit machine, after all. Each memory location in each register stores 16 bits of data. However, you can break that word down into two 8-bit bytes. That was handy for dealing with string data. On the PDP-11, that's done using specific instructions. There are sets for dealing with upper and lower bits of a word. The CP-16-100 is also a 16-bit processor,

but it can't break down words. Each location in memory and each register is a solid 16-bit chunk, and you just have to live with it. Again, there's the outside veneer of similarity, but the details and flexibility are lacking. So, is the CP-16-100 actually based off the PDP-11? That's like, what, the $1 question, right? If you do any searching around, you'll find a single stack exchange thread about just this question.

This thread gets referenced and linked everywhere I see discussion of the CP-1600. Now, to be fair, this isn't too many places, but the discussions are out there. This thread has a huge post that comes to the conclusion that, no, the CP-1600 is not based off the PDP-11. Now, I like a good nerd fight, but I usually don't get this nitpicky over, you know, a single post on a stack exchange thread. It's just that there's such a small amount of information out there on the chip, and this is so heavily referenced, that I kind of have to address it. Now, I'm going to say that I'm not very convinced by this reputation. The main point of the argument is that the CP-1600 is actually based off the data general NOVA, a machine developed by a group of deck employees who left after the end of the PDPX project.

You remember, we talked about that in episode one of the trilogy. Now, herein lies my issue with this argument. First of all, you can't find anything else online that connects the CP-1600 and the Nova. If you go searching, you literally just are directed to this post. But we can set that aside for now. The argument goes that the CP-16-100 is actually an expansion of the Nova architecture. If that's the case, then that should explain the weird drift between the CP-1600 and the PDP-11, right? So, does it?

Not exactly. at least not, not to my satisfaction. Instruction encoding is one easy way to spot the problem here. The Nova is a pretty old-school kind of machine, all things considered. It has multiple different instruction formats, including special formats to specify input-output instructions. The CP-1600, like the PDP-11, has a single instruction format. That may sound nitpicky, but it's actually a huge deal.

That dictates how code is structured and how the processor processes that code. It speaks to a deep structural similarity between the 11 and the CP-1600's design. Another issue with the argument, if you run into it, is that the author makes a critical mistake, a classic blunder. First is that they have this implicit claim that the NOVA was based off the PDPX. They claim that the CP-1600 is based off the NOVA, which was based off the PDPX, which was related to the PDP-8. Hence, the CP-1600 is actually a PDP-8 in a trench coat. Now, we know that that doesn't hold water. If you've listened to the first part of this trilogy, you know why.

The details of the PDPX were unknown to the public until 2007, when a pile of memos were released. It turns out the PDPX was actually a lot closer to Dex's original design for the PDP10 mainframe than anything else. Neither the PDP 11 nor the Nova took a whole lot of cues from the X's design. The NOVA in particular was very different. And you'll notice that along that lineage, there's no mention of the number 8. So if the CP-1600 isn't exactly a PDP-11, and it's not a NOVA, then what is it? Well, I think it's kind of just its own chip. If we go back to the mid-70s, we can find reporting from when the CP-1600 was released.

It got a nice article in bite, we get manuals and documentation, and later on we even get chapters in books on 16-bit microcomputers. What do those all have to say? In a word, oh, not much. This was, to me, the most illuminating fact. Contemporary documents don't say anything like the CP-1600 is a PDP-compatible or a PDP-like, or it's an expansion of the Data General Nova. It's presented as a new microprocessor. The closest we get are a few articles that compare certain features of the CP-1600 to the PDP-11. The detail here is kind of telling, so let me pull an exact example.

In Osborne's book, 16-bit microprocessors, it's explained that the CP-1600 allows for similar programming tricks as the PDP-11. Specifically, those tricks center around the stack pointer and program counter registers. On these computers, each of those registers is exposed like any other. So you can say, add to the stack pointer, or use the program counter to fetch the current instruction. This kind of flexibility can be used for things like reentrant code. Basically, since you can read and manipulate these registers, you can set up code for execution, save down where you were, run that new program, and then jump back to where you started. That's useful for, say, multitasking. So there are some clear similarities, but again,

the CP-1600 isn't just a PDP 11. That said, I think I'm fine saying it was heavily inspired by the 11. And this is where we get into the weird gray zone. We don't know a lot about the processor's history. Most of what I can find comes from Bolstrov and Sutterman's book, Intellivision. The chip was designed at Honeywell by an engineer named Jeff Stein. It was manufactured by general instruments, and that's about it. The sign is an enigma. The authors of Intellivision got their information by talking directly with Stein. As far as I can tell, we don't have a paper trail on the chip's design. at all. Then why are we talking about the 1600 when it may or may not have had anything to do with

the PDP 11? There are a few reasons and a few lessons we can pull from this. The CP-1600 isn't the only microprocessor with a thin paper trail. As I found, and as I continue to find out, there are actually many computers, microprocessors, programming languages, basically anything technical, that don't have publicly known histories. You basically only have access to documents that say what something is and how it worked. I actually think this is just the phase we are at in the study of computer history. Folk have been really good about fetching and really ferreting out and digitizing manuals. But things like memos and internal documents are harder to come by.

We're starting to see those kinds of papers make their way online, and get publicly distributed, but not at the same level of preservation as documentation. At least, not yet. I think that's going to be the next wave of what starts to come out. In the absence of that paper trail, all we can really do is speculate. That's what I often do here on the show, in fact. But you have to be careful. I try to strike a balance, and I apologize if I fall on the wrong side sometimes.

The CP-1600 looks so much like the PDP-11 that there has to be some connection, but we don't have evidence of what that connection is. So when we say the CP-16-100 was inspired by the PDP-11, we should be very careful to qualify that it kind of looks that way, but we don't have a smoking gun. The final important point is that by the time the CP-16-100 is released, the PDP 11 is just in the culture.

It was a wildly popular machine. Deck literally gave bulk discounts. Because folk were buying PDP 11s in bulk. Many, many people would experience the computer. Entire generations of programmers and engineers would cut their teeth on PDP 11s. So there's a certain amount of background level influence that machine had.

That can also make it hard to find a smoking gun. How do you notate that you designed a computer a certain way because you learn to program on the PDP 11 and that's just sort of how you think now and you can't get out of that mindset? That's not really a line item that you put in a spec sheet. I'm pretty upfront about the fact that most of the history I cover is American history.

I do, when I can, try to break out of that. One particular region that I need to spend more time investigating is the Soviet Union. This episode gives us a fun chance to do just that, because, believe it or not, many PDP-11 compatible computers showed up in the old USSR. Let's start off with a simple question. How did Soviet countries get access to computers back in the day. The actual factor here is COCOM, the coordinated committee for multilateral export controls.

This was a large-scale agreement between Western Bloc countries to prevent the export of certain goods to the USSR. The primary purpose was to keep military technology from crossing the curtain, and that included computers at certain points. COCOM was established in 1949, so right at the beginning of the digital era. So, how did Soviet countries get their computing done? There were three pathways. First and most obvious was to just make their own computers.

There is an early period where new machines are popping up all over the world, and there isn't any kind of industrial support for computing. This would slow down as computing became a market and the U.S. started to do, dominate that market. That happens probably around the end of the vacuum tube era. It's definitely the case when we reach the 1970s. The Soviet Union's first computer, MESM or Messum, was built in 51. That's just two years after COCOM goes into effect. So, point one for homespun hardware. But that's just one avenue. Another route was smuggling. This is best known to have happened during the home computing boom. If you want around an embargo, you can just, well, get around it.

This could be done on a small scale by smuggling in systems to sell to end users. Or it could be done on a large scale. There were a number of IBM systems that were smuggled into the USSR. There were even a few famous incidents of the Soviets attempting to smuggle deck Vax computers into the Eastern Block via South Africa. These smuggled machines could be strictly for personal use, or they could be for something else. This gets us to option three, reverse engineering or cloning. Some of these smuggled systems were ripped apart and recreated.

This wasn't just as easy as getting a parts list and then going out to the store, though. This was actually a huge undertaking. Soviet engineers would have to find ways to source or manufacture completely. components from their side of the Iron Curtain. The results were a series of compatible computers without any hang-up, like licensing. But why reverse engineer and why clone? That's a trickier question. By the time we reached the 1970s, the USSR is undoubtedly falling behind when it comes to computing.

Part of that has to do with the fact that the USSR is somewhat disconnected from international collaboration. While the U.S. would dominate the field, that wouldn't be possible without international breakthrough. Part of this also has to do with funding. The capitalist structures in the West provided pressures that made companies go really hard on developing cheaper, faster, and better computers. In the USSR, pressures were simply different. There's also something to be said about the Soviet focus on cybernetics versus more Western-style computer science. That though is kind of a topic in and of itself for another day. For all those reasons, plus, I'm sure many more that I'm not aware of,

it became advantageous to take advantage of Western designs instead of focusing on new indigenous designs, at least in part. This doesn't mean that all computers developed in the USSR were copies of Western block computers. Instead, that some machines borrowed very, very, very heavily from those designs. To get into more detail, let's travel to Hungary and look at the strange story of TPA, the stored program analyzer.

I've been told that the acronym lines up if you say it in Hungarian. TPA was a line of computers manufactured in Hungary from 1968 to 1992. Most of that series was compatible with some kind of deck hardware. Here I'm working out of translations of a few Hungarian articles about the TPA series. So, again, not primary, and there may be something lost in translation, but it should be neat. These machines were developed by KFKI, the Central Physics Research Institute. According to sources, there were forces against the creation of new computers in Hungary in the 1960s. That didn't sit well with KFKI.

They wanted access to small lab computers, but other member states in the Soviet Union just weren't making those kinds of machines. So they decided they wouldn't make a new computer. Instead, they'd make a stored program analyzer, you know, just some lab hardware that analyzes data, and it uses a stored program to do so. But I assure you, this is in no way a computer. Even making a small computer is a massive undertaking. Luckily, there was a design floating around.

In 1969, Deck published a text called the Small Computer Handbook. It's a full description of the PDP8 series of computers. The intent was to get new users familiar with the platform, its theory of operation, and how to program the thing. Used another way, it's a blueprint for how you would design a computer. KFKI would use this book to create a PDP11 compatible computer. The result, the TPA 1001, was unveiled in 1968, and it was capable of running PDP8 software on purely Hungarian-built circuits.

Before we move on, I just want to strike a very important parallel here. Right now we're talking about computing at a certain point where industrial manufacturing, and markets matter. But the story is very, very similar to tales told in the earlier era of computing. Take the IIS machine or the designs for EDVAC. Those designs were proliferated internationally by papers describing how those platforms work, describing the architectures of those machines. They were very similar to Dex's small computer handbook.

The key difference was those papers were written with the intent of other researchers recreating machines or making similar computers. The small computer handbook was marketing material. I think it's important to point that out because once we hit this period of industrial manufacturing, if someone makes a copy of a computer architecture, I think it's easier for people to be upset, to view that as intellectual property. theft, when historically, prior to this period, the same thing was going on and it was viewed as international scientific collaboration. So back to TPA, it turns out that making this PDP8 compatible was a completely winning strategy. The PDP8 was purpose built as a handy lab computer, and the 1001 fared about the same. It became a workhorse in a number of labs in Hungary. This is, to my ears, pretty unique for the period.

We've dealt with clones before on the show. Take the entire PC-compatible industry as an example. But I don't think I ran to anything exactly like the TPA 1001. It's a re-implementation based off external documents. That's, well, I guess it's similar to the Clean Room reverse engineering that was done, to clone the PC BIOS. The differences here are the scale and the isolation. Creating an entire computer is a lot different than creating a program.

And isolation, well, the Iron Curtain had its holes, but it didn't force a level of restriction that KFKI didn't have control over. But that's just the PDP8. What about the main event? What about the PDP 11? Therein lies some mystery. In 1976, KFKI released the TPA 1140. It was a PDP11 compatible machine through and through.

It could run any software that could run on a real PDP11. It even had a unibus, sort of. TPA machines called it the UBus, and it had a different physical connector, but it was unibus compatible. So if a PDP 11 card made its way into Hungary, it could be hooked. into a TPA machine if you could work up a physical adapter. But how did KFKI go about making these clones? Well, answers on the English net are lacking.

However, there is a book to fall back to. It's called, in English, the TPA story, from punch cards to informatics. And it explains, well, the TPA story. I've been able to translate and read a few key passages from the text. The relevant story goes like this. After the success of the 1001, KFKI started work on a number of successors. Those were all PDP8 compatible until around 1970. During that period, KFKI researchers started to consider the leap to 16-bit.

That resulted in a fully original machine called TPA-70. A number of units were built and used internally, but it was never mass-produced. This gets us into the middle of the 70s. I'm going to quote my computer translation of the TPA story. Just as a note here, MSZKI was the group within KFKI that spun off to focus on computers in this period. Quote, around 1975, the leading computer expert at MSZKI were faced with a big question as to which direction to move forward in. There were two options. Either MSZKI would continue the line it had started with its own TPA-70

and try to create a wide-ranging software background for its own development machines, document them, and create applications, or it would follow the deck machines of the 16-bit line, end quote. Do they continue work on a new design, or do they make another compatible machine? The determining factor would actually come down to a simple reality, human power. KFKI's computer project was small and scrappy. They could do hardware, but that's only half of the equation. That was a lesson that the PDP8 and now the TPA-70 it taught them.

For a computer to really sing, it needs software. And for a computer to see adoption, it needs to come with software. KFKI simply didn't have enough people to support continued software development for a new platform. Deck, however, did. So in a strange way, Deccas, the fabled deck user group, had a truly international impact. It was decided to ditch the new original design and make another deck compatible, specifically the PDP 11. That's how we get up to the TPA 1140. and this machine was microprogrammed.

And this is where we also get back to some uncertainties. From my reading of the machine translated Hungarian, it sounds like the TPA 1140 was a pretty direct copy of the PDP 1140. It's possible that KFKI had access to a physical machine to clone, or that they had access to publicly available documentation, or maybe a combination of the two. So the TPA machine was just a PDP-11. Further compatibles, however, were not. That's the interesting thing about the TPA series.

As the 1970s drug-on, KFKI and others would adapt PDP-11 software to suit their needs. And they would also adapt that hardware. By the 1980s, they were making new machines with new hardware designs. that were only software-compatible with the PDP-11. That, dear listener, is just Dex business model around the PDP-11. Make whatever hardware you want, juice it up however you want, just make sure it can run the same software. So this gives us a pretty neat situation, right?

If you look at KFKI's early PDP-11 clones, you see just that, clones. Then we get this divergence point in the early 80s where those clones turn into their own machines that are compatible with deck software. How wild is that? Over in Hungary, we have a factory turning out new models of PDP 11s. At that point, it's not really a clone.

It's more like, well, a totally new kind of PDP 11. That's just wild to me. And I wish there was a full translation of the TVP 11. TPA story. Now, KFKI isn't the only place turning out new 11s. There are actually quite a few instances of unlicensed PDP11s. One other somewhat related case is that of Sistime computers based out of the UK. At least Sistime is one of the more known names. They were at DECDEC OEM, meaning they bought components from DEC and then manufactured and sold their own computers. it's been alleged that they also violated co-com to sell deck hardware to eastern block countries.

That could explain how the USSR had so many PDP-11 clones. They had their hands on a bunch of hardware. But not so fast. There is a bit of a hitch, which is explained in Kevin Calhills' trade wars, the high-technology scandal of the 1980s. This outlines a few cases of strange federal meddling in 10. techno trade in that decade. And luckily, it explains the entire Sistime debacle. While Sistime was an OEM, they had to deal with DEC's U.K. arm. In order for Sys Time to be profitable, they ended up in the situation where they had to sell to markets the deck did not.

Basically, to sell in the UK, they would have had to undercut DEC's rates, but there wasn't a way for them to do that and keep a profit margin. But this in turn led to some possible license issues. How do you deal with licensing if an OEM in England is buying all their parts to make a PDP 11 from the U.S., building that system in the UK and then selling it in India? Well, apparently, there are legal ways to do it, which is how Sistime stayed afloat.

But DEC, they didn't like that. They didn't have a good way to legally stop. systime, so they went about things from another angle. Deck hired private investigators to snoop on Sistime. Supposedly, those PIs discovered that Sistime was selling their PDP 11s to the USSR. The trick was that they dressed them up as jukeboxes and shipped them through Switzerland, who was not party to co-com. That sounds like a cool premise for a spy movie, right?

But, it turns out we have no hard evidence of this, besides reports written by Dex's hired hands. Sis-Times offices were raided, but no documents were found to point to deals with the USSR. It's possible documents were hidden, removed, or destroyed. It's also possible that Dex PIs were wrong. We kind of just don't know. So maybe this explains how Deck clones showed up in the USSR. They had computers smuggled in, ripped them up. up and set to work. But there's an easier answer. Kocom had ebbs and flows. There were periods

when U.S. companies could ship computers to the eastern block by just filling out the right paperwork. Trade Wars points out that in the late 70s, DEC exported a significant amount of hardware to Soviet nations. While there is isolation, it's not exactly absolute. As we've seen, this can lead to some interesting outcomes that diverge from what we got in the Western block. Let me kick this last part off by restating the initial question, yet again. If the PDP 11 was such a cool machine, why didn't it end up on every desk in the world? The short answer was that machines that use the 8086 or Motorola 68K could do all the same things as the PDP 11 and much, much more. By the time computers were fighting for the desktop space, the PDP 11 was getting pretty old.

I want to complicate that part of the story a little more. It can be easy to slip into this view that the most capable computer always wins. We know that's false. Take the IBM PC as the prime example. It's capable, but far from a powerhouse. Its design has all sorts of shortfalls and has led us to something of a dead. end. However, forces a line to make that the winning platform of the 1980s. In 1982, the Commodore 64 comes to market. That's a year after the PC. The C-64 is an 8-bit machine. It uses a pretty

outdated chip set. It's slow. It's simple. It's not all that capable compared to the rest of the market. But it's cheap. Over its lifespan, which would lead into the 1990, it sold tens of millions of units. So while some people were unboxing Macintoshes, someone a few doors over could have been doing the same with an 8-bit micro. It's conceivable that there's a world in which the PDP 11 made the jump to home microcomputers and actually stuck that landing. Something like a cheap little thing to plop on your desk.

If you could make it cheap enough, then it could fight out the lower parts of the market, market. I mean, people in the 80s were still buying 8-bit machines. The PDP 11 was far more sophisticated and powerful than a Commodore 64. I mean, one of them can run Unix, after all. It turns out that such a world existed, namely, again, it was the Soviet Union. There were actually quite a few different models of PDP 11-based home computers that circulated in the USSR in the 1980s. The best known is, at least in the West, the Electronica BK, which was released in, get this, 1985. That's a year after the Macintosh. That's the same year as the Intel 386. But those were all decadent Western computers. How could they even compare

to Soviet engineering? The Electronica BK is fascinating for a few reasons. It's built like an 8-bit micro. By that, I mean it's a wedge. The entire computer is fit into a chunky wedge-shaped case with a keyboard on top. The ports are all around the back. It plugs into a TV for output. Discs and all other peripherals are optional and external. We do know that exteriors can be misleading.

Most of these little wedge machines were 8-bit, but there were some notable. outliers. In the West, the only one I can think of is the Ti-994A, which has many computer-derived silicon inside. Its architecture is actually better suited to time-sharing than a cheap home computer. That's also true for the Electronica BK. This computer is a micro through and through. It's not using an old-school board-based LSI 11. It has a single-chip-11 inside. By the end of the 70s, deck had some pretty slick single-chip options. They managed to shrink the LSI 11 down to just one package. But the Electronica BK was bootleg all the way.

It didn't have a single genuine deck part inside. Instead, it was rocking an 1801 series CPU. This is another place where, sadly, the language barrier poses a big problem for me. Here's what I've pieced together. 1801 processors were microcoded. The core, the actual thing that executed the microcode, was designed and built by Soviet researchers. I've seen it claimed that 1801's used an Electronica NC core,

but I don't think that's the right name. There were a number of computers called Electronica NC, so I think there's some issue with translation going on there. The point is, like TPA, we have another case of homespun hardware using microcode to gain access to imported software. If you go looking around, you'll see claims that the KGB stole deck software to freely distribute it around the USSR. But I don't know if that's really a primary factor here. Remember that there were periods where sanctions were weakened.

We can add two other facts here. first that there were always holes in the iron curtain and second that piracy is a constant factor when it comes to software especially older software I think it's little wonder that someone would want a way to run Western software Western software would have been if not easy at least relatively easy to get a hold of the 1801 was only one of these PDP 11 compatible chips

There were a handful of other clones, each with varying degrees of performance and originality. Some would show up in micros like the Electronica BK, others would be used as the heart and brains of larger machines. Now, you may be asking at this point, does this mean that in the USSR, Unix on the desktop was commonplace? Or maybe you aren't asking that. Maybe I'm just odd. among the Linux community there's this joke that 2025 will be the year of Linux on the desktop or 2024 or 26 maybe

the year changes but we'll we'll catch up to it eventually so when I first read about these PDP 11 compatible micros my first thought was well does this mean that Unix was on Soviet desktops back in the 80s

the answer is mixed So, technically, an Electronica BK can run Unix because it's PDP11 compatible. There is a somewhat recent port of Unix that runs on the platform. There were also versions of Unix kicking around the Soviet Union that ran on other PDP11-like machines. It's possible that someone rigged up a micro to run Unix. I just can't find reference to it in English. The other piece here comes down to availability. Computers were expensive back in the day.

That was doubly true in Eastern Bloc countries. It also wasn't always possible for a private person to buy a computer for personal use. Remember, the quote-unquote market in communist countries operated on completely different principles than markets under capitalism. So it wasn't exactly common to see a home computer. So sadly, no, I don't think. we get to look at a spike in Unix users in the USSR in the 80s. But let's get back to that big question.

If the PDP 11 was so good, then why didn't it wind up as a winner in the home computer boom? If you look at things from a different angle, then I think we can actually say that yes, the PDP 11 did end up on many desktops in the 80s. Allow me to explain my reasoning. For this to make sense, we need to go back to what made the PDP. P.11 unique and why it will be cool to have in the home. I think we've beat all the architectural details to death at this point, but I can add a few more. The unibus would have been especially compelling for a home user. You could buy a PDP 11 in 1981 and keep expanding

and upgrading it over the years. Need a hard drive and finally have the scratch? Just get a new unibus card and throw it into the old 11. A desktop PDP 11 would also give you access to a sophisticated machine at home. That may sound a little trite, but look at it this way. Many early home computers were, frankly, closer to toys than serious machines. With a desktop PDP 11, you have a very serious computer. You can write sophisticated code without jumping through the hoops that more primitive platforms place in front of you.

As we've also seen, software is a huge draw. If you have a PDP-11, you can run PDP-11 software. You get access to slick operating systems. You get access to programming languages, and you get access to Deccas' catalog. That was so enticing that even the Soviets were interested. Of course, that would be really, really nice to have in the home. Last but not least, you get Unix. That opens up an entire world of sophistication and power.

plus access to even more software. Again, we have this over and over during these episodes. Buying the PDP11 platform is more than just getting a new computer. It's entering a whole ecosystem. Is this starting to sound familiar yet? We actually did see something very similar to this in the 1980s. It's the IBM PC. Did I manage to trick you?

The architecture of the IBM PC is markedly similar to the PDP-11. It's not just that the X-86 family of processors is very PDP-11-like. It's also that the PC memory maps most of its I.O. devices. It supports direct memory access. The PC also has an expansion bus that, while less fancy, is used in a very similar way to the unibus. When you plug in an expansion card, it takes up a chunk of it. of memory, and it can connect to interrupt lines. That is some classic Unibus stuff.

As for software access, well, things are a little bit different. The big draw with Deck was, in part, first-party software. Now, there is a bit of an argument here. Deccas wasn't technically owned by Deck, but it operated so closely with the company that I don't really have qualms with calling their catalog first-party. The IBM PC launched with some software, but not a whole lot. The software catalog for the PC grew in a very distributed manner. In the first half of the 80s, the PC gets cloned and spread out as a major platform.

That forms a feedback loop. As more people have more PCs, more developers want to write software for the PC. As there's more software for the PC, more people want to buy PCs. What do you do with a PC? well, you run software. By the middle of the 80s, there's a staggering amount of code written for the PC and 100% compatible. It starts to make sense to buy a PC because there's so much software for the PC. That could be actual commercial software, could be shareware, could be freeware, or could be the pirated stuff. Expansion is particularly interesting to consider here. Remember that

quirk about how the unibus was used, Deck said the unibus made the PDP 11 immune to obsolescence. That meant that Deck could always crank out an updated PDP11 by making some new unibus cards and mixing up parts in a new chassis. But that didn't mean that an end user could keep upgrading their PDP 11 for 30 years. With the PC's bus, the ISA bus, things end up differently. The ISA bus allowed for expansion of the PC, but there are some limits that I shouldn't know. First is that, with some exception, processors aren't put on the ISA bus.

You have a motherboard that holds the processor, and the ISA bus is exposed on the motherboard. There were some expansion cars that had processor upgrades on board, but that wasn't the order of the day. It's also worth noting that memory didn't necessarily hook in over the ISA bus. There was a separate set of sockets on the motherboard for memory. But there were also some cards that added more memory. Many things with the PC are messy. Anyway, expansion. The ISA bus turned into an end-user thing.

Manufacturers sold cards to users. Users would then crack the case and add the new card. By doing so, they could make their PC last longer, give it new features, and new life. It was possible for a PC to have a certain level of immunity to obsolescence, but on the end-user's side instead of on the manufacturer's side. You could have the same PC chassis for years and years and years, far after the processor would have been considered obsolete. It's pretty well documented where the ISA bus concept came from.

It wasn't borrowed from the PDP 11. It was developed at IBM for use with an earlier personal computer. But still, it bears a striking resemblance to the unibus. What I'm getting at is the simple fact that we did see PDP11s on the desktop just in a different form. The technical features that made the PDP 11 an exciting machine became entrenched in the fabric of computing. As a result, many other machines began to follow suit.

Some were clones and compatibles as in the Eastern Block. Others were simply inspired designs or very, very similar outlines. That's what we saw with the PC. And hey, the PC can run Unix after all. That's it. I'm done. I think I've said everything I want to say about the PDP 11. We've seen its origin, how it changed with the times, and how its design proliferated into strange places.

It's over, and I am free. With this trilogy over, we can move on to the next event, Spook Month 2025. Didn't think I'd forget, did you? As you listen to this episode, I'm already deep in the books pulling out weird, wacky, and frustrating stories from Computing's past. I'll see you in two weeks' time with the next episode. Thanks for listening to Avon of Computing. You can find links to everything.

over at avonofcomputing.com. And as always, have a great rest of your day.