A PDP-11 On A Chip

If you entered the world of professional computing sometime in the 1960s or 1970s there is a high probability that you would have found yourself working on a minicomputer. These were a class of computer smaller than the colossal mainframes of the day, with a price tag that put them within the range of medium-sized companies and institutions rather than large corporations or government-funded entities. Physically they were not small machines, but compared to the mainframes they did not require a special building to house them, or a high-power electrical supply.

A PDP-11 at The National Museum Of Computing, Bletchley, UK.
A PDP-11 at The National Museum Of Computing, Bletchley, UK.

One of the most prominent among the suppliers of minicomputers was Digital Equipment Corporation, otherwise known as DEC. Their PDP line of machines dominated the market, and can be found in the ancestry of many of the things we take for granted today. The first UNIX development in 1969 for instance was performed on a DEC PDP-7.

DEC’s flagship product line of the 1970s was the 16-bit PDP-11 series, launched in 1970 and continuing in production until sometime in the late 1990s. Huge numbers of these machines were sold, and it is likely that nearly all adults reading this have at some time or other encountered one at work even if we are unaware that the supermarket till receipt, invoice, or doctor’s appointment slip in our hand was processed on it.

During that over-20-year lifespan of course DEC did not retain the 74 logic based architecture of the earliest model. Successive PDP-11 generations featured ever greater integration of their processor, culminating by the 1980s in the J-11, a CMOS microprocessor implementation of a PDP-11/70. This took the form of two integrated circuits mounted on a large 60-pin DIP ceramic wafer. It was one of these devices that came the way of [bhilpert], and instead of retaining it as a curio he decided to see if he could make it work.

The PDP-11 processors had a useful feature: a debugging console built into their hardware. This means that it should be a relatively simple task to bring up a PDP-11 processor like the J-11 without providing the rest of the PDP-11 to support it, and it was this task that he set about performing. Providing a 6402 UART at the address expected of the console with a bit of 74 glue logic, a bit more 74 for an address latch, and a couple of  6264 8K by 8 RAM chips gave him a very simple but functional PDP-11 on a breadboard. He found it would run with a clock speed as high as 11MHz, but baulked at a 14MHz crystal. He suggests that the breadboard layout may be responsible for this. Hand-keying a couple of test programs, he was able to demonstrate it working.

We’ve seen a lot of the PDP-11 on these pages over the years. Of note are a restoration of a PDP-11/04, this faithful reproduction of a PDP-11 panel emulated with the help of a Raspberry Pi, and an entire PDP-11 emulated on an AVR microcontroller. We have indeed come a long way.

Thanks [BigEd] for the tip.

47 thoughts on “A PDP-11 On A Chip

    1. The T-11 was a beautiful piece of work. Although limited in the amount of memory, it was very flexible and could be configured for various memory maps with an appropriately strapped shift register at power up. I bought one way back in the mists of time with the idea of making a palm-sized, wire wrapped PDP-11. Sadly, I rather consistently wire wrapped my circuits mirror-imaged and was terrified of blowing it up…

  1. Heh from link “He got the IC count down to 9, while increasing the memory to 128KW, ” yup, that sounds about right for the era, enough to heat a decent sized house :-D

    ˙ʇı ןıods ʎɥʍ ʇnq ‘spɹoʍ-oןıʞ ɹoɟ puɐʇs ʇɥbıɯ ‘ʎɐʞo

    1. ˙sǝıʇıɔɐdɐɔ ʎɹoɯǝɯ buıuoıʇuǝɯ uǝɥʍ sǝʇʎq puɐ spɹoʍ ɥʇoq ǝsn ʎǝɥʇ ‘sǝʎ puɐ ‘0761 ‘ʞooqpuɐɥ 11-dpd ǝǝɹʇpɐǝp ǝɥʇ ɥbnoɹɥʇ pǝddıןɟ ʇsnظ

    2. I have few doubts that an 11/70, the machine the J-11 was supposed to emulate, would indeed draw about that much power. I used one that lived in its own room with a raised floor. The AC kept the room something like 60 degrees F. If it got much warmer than that, an alarm would sound and the system would shutdown. The great thing about core memory was once the problem was fixed, you could fire it up and continue from where it shut down…

      128KW/256KB isn’t much by today’s standards, but that was adequate for an 11/70. We had 192KB of core running IAS(a RSX/11D+ derivative) supporting more than a dozen users. I suspect a smallish FPGA or a couple of ATMEGAs could emulate enough peripherals to get one of those OSes to boot and run…

      Which is not to take anything away from Sytse with his vhdl implementation. That would be the most practical way to go. But having actual DEC hardware, even a small amount…

    1. The LSI-11 was treated as an early candidate for a home computer. It was covered in Byte early on, and the Southern California Computer Society had t least one group buy. If you were coming from a different direction, the LSI-11 was familiar, yet “cheap” compared to minicomputers. This was probably the foundation of Heath choosing it for one of their Computers in 1977.

      There was also the Intersil 6100, basically an IC version of the PDP-8. It never saw much use in hobby circles, but it was familiar to some, and 12bits when everything was still mostly 8bit. It was also an earlyish CMOS processor, when the RCA 1802 was about the only other one. Someone I knew used it in a controller for record presses, the CMOS having advantages and the PDP-8 instruction set meant he could find someone used to the set to do programming. But it didn’t include a front panel in the IC, I remember someone grumbling about that ince much of te existing software expected ä front panel.

      Michael

  2. I’ve never owned a PDP-11, although I learned Pascal on one in high school in ’85. I did used to collect VAXen, though – a VAX 4000/200 with two racks full of HVD SCSI drives, a VAXStation 4000/60, various MicroVAX III variations (including one with a QDSS 3-board graphics kit), and a 16-node cluster of VAXStation 3100. Ah, those were the days. Never needed the furnace to keep the basement warm. And all of it ran NetBSD! Just don’t ask how long it took to compile the world.

  3. Take a look at USSR processors http://www.cpu-world.com/CPUs/USSR-1801/ They are real single-chip PDP-11 versions. Unlike most of Soviet IC’s they were not a copy of some Eastern IC, but original design. Moreover, it was very popular, there was a lot of devices with that chips, from desktops and industrial controllers to home computer BK-0010 (something like PDP-11 version of Sinclair ZX Spectrum) and PDA MK-85 that look like Casio FX-700P, but was PDP-11 inside.

  4. I remember that chip. I designed a PDP11 coprocessor board that ran in another computer. I can say that the chip had one of the strangest interfaces I’ve ever worked with. The fun part was that DEC told me that the company I worked for and I were only one of five companies/engineers outsite DEC that was allowed to design the chip into a product not controlled by DEC. Little more information. The chip was called the J11 aka Jaws11. Here is a picture of the J11 https://en.m.wikipedia.org/wiki/PDP-11#/media/File%3AKL_DEC_J11.jpg

  5. In the early ’80s there was also the General Instruments CP1600/1610 (used in the Mattel Intellivision games console, a would-be Atari 2600 killer), which was based on the PDP-11. My first job after leaving school was reverse engineering Intellivision’s system ROM. My main memory of it is how slow it was, even back then, running at less than 1MHz.

    1. Motorola 68000 had a lot of similarities to the PDP-11 (paird A & D registers, auto-increment/decrement, etc).
      I talked to a Moto sales rep who said that some of the designers had liked these features of the 11 so much , they included them.

      Remember, at the time the 68000 was designed, DEC was donating or selling at a discount, a LOT of 11’s to colleges and universities under the assumption that if you did your assembly language course on an 11, you would be predisposed to buy them when you got into industry…

      1. We learned bout the 6809 and the 68000 about the same time (I can’t remember how close together, but decades later it was about the same time), both pats up from the 6800. But a big contrast, the 6809 built on the 6800, but the 68000 didn’t build on the 6809, and didn’t resemble the 6800 that much. You very much see a different philosophy, though I didn’t know much about the PDP11 at the time so didn’t see the similarity. But the cleaner architecture of the 68000 was clear. I like the 6809, but it’s very different from the 68000.

        Michael

        1. I did a bit of 6800/7802 programming, and the first thing you notice, is that the X index register had only auto increment and auto-decrement modes, when what would have made it really useful would have been some arithmetic operations on it. This was fixed in the 6809.

          68k was a whole different design team IIRC. We used 7 of them (4 serial data processors, one data stream allocation processor and two TCP/IP processors) on the Data General ITC/128 terminal server board…which made our Motorola sales rep very happy indeed :-)

    2. May I ask which company you worked for where you did the Intellivision system ROM RE? I’d love to know more about your job and the company. Was it one of the published 3rd party software developers for the system? Did you RE the ROM with the goal of using the EXEC for games, or just to be able to boot a game developed on the metal?

  6. Lackluster article. Makes it seem like the J-11 was the only integrated PDP-11, but in reality the J-11 was actually the LESS popular of DEC’s integrated PDP-11 chips. The T-11, for example, found use in several arcade games, including the venerable Paperboy.

  7. Fun Atari PDP-11 facts

    The original Atari 2600 dev system was a cross assembler running on a PDP-11
    Atari used the T11 one chip PDP-11 in their System 2 hardware – Super Sprint, 720, Paperboy and APB. Their system 1 hardware used a Moto 68K so its possible they chose the T11 because of its similarity.

    I’m sure countless other arcade games were written on PDP mainframes but I’ll leave that for the retrogaming historians.

  8. I’ve been looking for PDP-11 and HP 9000 parts, if anyone has leads on where to get such things (Other than eBay, of course). Recently inherited a good sized dairy farm from my grandparents, sold most of the land to a neighboring dairy and kept the 10,000 sqft Milking shed to turn into an electronics lab (The building is already made to be easily sterilized with very strict environmental controls, to comply with FDA regulations). With such a building, I’ve been looking into restoring parts for the old mainframes (Or building fresh, 100% compatible replacements). I have the facility and enough money to invest in such an undertaking. I just need to finish clearing some space (The milking and processing machines are taking up a significant chunk of spaces right now).

    There are a ridiculous number of PDP-11s out there being used in some of the most critical roles possible, and they’re always going to need replacement parts.

    1. Of course. Any Turing-complete machine can emulate any other. All it takes is an interface to external memory, lots and lots of it. The Arduino just needs to run an interpreter sketch.

Leave a Reply to SytseCancel reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.