VCF East: PR1ME And AT&T Unix Boxes


At the Vintage Computer Festival last weekend, there was a wonderful representation of small 8 and 16-bit home computers from the 80s, an awful lot of PDP and VAX-based minicomputers, and even some very big iron in the form of a UNIVAC and a Cray. You might think this is a good representation of computing history, but there was actually a huge gap in the historical reality. Namely, workstations and minicomputers that weren’t made by DEC.

[Ian Primus] was one of the very few people to recognize this shortcoming and brought his PRIME minicomputer. This was a huge, “two half racks, side by side” computer running PRIMOS, an operating system written in FORTRAN. Of course this made it extremely popular with engineering teams, but that doesn’t mean [Ian] can’t have fun with it. He had two terminals set up, one running Dungeon (i.e. Zork pre-Infocom) and a text-based lunar lander game.

Because the VCF East is held in New Jersey, it’s probably no surprise a few vintage AT&T Unix boxes showed up. [Anthony Stramaglia] brought in a few very cool vintage Unix workstations, dating from the early to mid 80s. In the video, he shows off two AT&T boxes. The first is a UNIX PC, containing a 68010 clocked at a blistering 10 MHz. Next up is the UNIX PC’s bigger brother, the 3B2 400. This is the workstation found on just about every desk at Bell Labs in the 80s, meaning this is the same computer [Ken Thompson] and [Dennis Ritchie] used for their work on UNIX.


44 thoughts on “VCF East: PR1ME And AT&T Unix Boxes

  1. Oddly enough he and I discussed them. I mentioned that at the last one, the 3B2 and its terminal were there. However they chose to not cooperate with their people. He pointed out that the systems were not his then. That is they were not exhibited by him.

    Oddly enough I did look at the Prime systems that were there, I found myself entranced by them.

    And of course by the Vaxes and the PDP-11s.

  2. I couldn’t make it to VCF, but if I did, I would have brought along several reels of 20+ year old 9 track tapes, with the hope that someone there would be kind enough to dump the tapes to a another machine/Flash Drive.

    Did the 3B2 have Ethernet or what???

    Were any of the exhibiters capable or willing to read 9 track tapes – or 80 column punch cards???

      1. Thanks! I’ll try the next one.

        Any idea as to how the dump could be made to current day media?

        USB or CD/DVD??? (not on old hardware), Ethernet?

        Was there a Punch Card reader?

        1. As a matter of fact, yes. David Gesswein had a PDP-8 with a mark sense card reader, which he had connected to a USB interface. It would read the cards and dump the output to a laptop and onto a USB flash drive. Not sure in what format, but he was reading batches of cards for some people for exactly the purpose you mention. He can probably do the same for punch cards.

          1. EP, Thanks.

            I must make it to the next VCF in New Jersey.

            I wonder if my 20+ year old IBM mainframe mag tapes are still readable. The IBM tapes use 800 bpi for file labels and 6250 bpi for data.

    1. The standard 3B2 didn’t include Ethernet – you did everything with serial or sometimes parallel ports. Eventually there was an Ethernet card made for it for networking. But mostly you used dumb ASCII terminals over RS232 or networked with modems, and the 3B2’s I/O bus wasn’t really fast enough to keep an Ethernet very busy.

      The 3B2/300s were half the height of the 3B2/400 shown there. The 400 had some double-height cards inside, and could have more I/O cards on the side. I had a scavenged /300 on my desk for quite a while in the late 90s; if it had had an Ethernet card I might have done something useful with it, but since it didn’t, I mostly used it as a monitor stand. (The RS232 ports were built with RJ45 connectors, back when every designer felt that they needed a different mapping of RS232 signals to RJ45 pins, so you needed custom connectors and cables, so I would have also had to spend a lot of time soldering cable adapters.)

  3. What memories! Chemistry Department. They drug a PDP-11 into the lab and had me hook it up to the Finnegan GC/MS4000. A REAL computer! Dad got to touch them and told me, and now I had one. It was weeks of work to code the two into one. The linoleum floor was enough for naps they told me I took. Grad students donated food and quietly organized a feeding schedule. Marathon work…. but I was touching a REAL computer for once! 1982. The 6502 user group I’d organized thought I’d died. The wife had lost weight by the time I came home. All was good! Thank you for the memory. Thank you very very much.

  4. I used to work for the Pr1me R&D section in Canberra, great times until they closed it down and moved comms R&D back to the US. Of course not long after the entire company was gone.

    1. I used a Pr1me 20 years ago in college (it was pretty old even back then). My college had a system with 20 or so terminals (no where near as modern looking as the one in the video!). As part of a C course we had to write C programs that compiled & run on a Vax, 286 PC and the Pr1me.

      The green-screen terminals had a ‘soft scroll’ option which was pretty cool!

      I remember my CS teacher saying the Pr1me never crashed (it did grind when 20 students were compiling C code). The Vax crashed occasionally. DOS 286s crashed all the time :-)

      1. I still maintain 7 Pr1me Computers – I have parts to build several of them in my “wizards lab” – Recently – I put my TEST BED Pr1me into he DMZ – so link-lookers could find it.

        Over he weekend – I got like 25 hacking attempts. Most of the people attempting to login thought they were talking to a CISCO router. In PRIMOS – you can set it to log to he system console – unsuccessful login attempts – the OS tracks the source IP address the Login Name Attempted – as well as the failed password guess.

        Lots of “ADMIN/PASSWORD” tries and few root and cisco attempts as well!


  5. As part of my Elec. Eng. Degree, I had to do work experience in my summer holidays. I got a job in a civil engineering company to convert their library of TI-59 design programs to run on a brand new Fortune 32:16 68000 Unix workstation. The only language they had purchased was Fortan 77.

    So, over the summer, I learnt to read TI-59 programs off small paper strips and learn all about Fortran at the same time. The machine was really nice to work with and even had mono graphics so I could show appropriate diagrams.

    Fun times now long gone.

  6. One thing that is rare to see but was very common in the midwest was Cromemco mini computers running Cromix, a AT&T Unix clone.

    I had one of these as my second computer, my father had bought it from an auction and nobody knew what it was, I spent 3 months trying to crack the root password to get into it at age 12.

    1. @fartface: “… I spent 3 months trying to crack the root password to get into it at age 12.”

      When I was given my SGI Indigo, I had the same problem. Turns out, the box is actually quite secure. And it uses an old 50-pin SCSI HDD, so reading the drive on a modern Linux box wasn’t going to be easy. However, they had left the “demo” user active. No password required for login, and this was old UNIX, so /etc/passwd was world-readable. I copied the (encrypted) password string to my Linux system and ran “John-the-ripper” on it. Took less than 10 minutes to crack, and I was able to log in.

      Four months later, the head of IT stopped by and handed me an envelope he found when he was cleaning out his desk…with the root password inside. (He was the first person I asked)

      1. On our Honeywell Bull at college, another 68K Unix-runner, our always-faithful backdoor was to copy /etc/passwd to a file of our own. Edit the file. Then delete the original /etc/passwd and copy ours in place. Finally chown and chmod it to the SU, and bob’s yer uncle.

        Obviously there were several holes left open here. Anyone care to list them? I’m not quite sure what the modern solutions are, but I wonder why we could delete passwd without having write access to the file.

        All permissions were set properly, and all ownerships, on passwd itself at least. I suspect making changes to /etc’s own entry would’ve made it more secure. I presume it was in this state from the factory, unless students gone by had deliberately left a hole in.

  7. I have an SGI Indigo sitting on my desk. It runs. Whenever I get the urge to go back and see old-school UNIX, I can boot it up and go back to the eighties.

    I did not realise PrimeOS was written in FORTRAN. Wikipedia says it’s true. And the explanation they give, is that the original authors were Real Programmers. And, as we all know, a Real Programmer can write anything in FORTRAN.

    Gardner Hendrie, the designer of the first 16-bit minicomputer (Honeywell DDP-16) was my boss when I first started working at Data General in the late 70s.

    1. “I have an SGI Indigo sitting on my desk. It runs. Whenever I get the urge to go back and see old-school UNIX, I can boot it up and go back to the eighties.”

      I had a SGI too in the ’80s that did not have an OS installed. When I mentioned this on a SGI BBS forum, a SGI employee said I could bring him the HD and he would load up the current software. Those were the days!

    2. I was in a California startup that used Primes as their mainframe. It had Primos, but on top of it we ran Software Tools
      (Kernighan). I remember that the documentation said that if you did something nasty, Primos would terminate and leave you staring face to face with Primos (dread). Software Tools was unix like having pipes and a lot of the unix commands.

      1. That’s exactly right. I have Primos Source Code as one of the “MFD’s” on my box.
        SPL, PLP, and PMA… didn’t see Fortran – at all – that was probably rewritten due to some of the awkwardness – of FORTRAN. However there are a BUNCH of FORTRAN subroutines. And lots of primitives like TNOUA. (If you remember THAT one – you’re ANCIENT… like me!)

        Honestly – my Prime actually gets started up a few times a month… for support for the few remaining boxes out there… that I still maintain. Ahhhhh SPL and PLP!!! yep… you’re a primate! :)

  8. “You might think this is a good representation of computing history, but there was actually a huge gap in the historical reality. Namely, workstations and minicomputers that weren’t made by DEC.”

    You must have missed my HP minicomputer exhibit. I had a HP2109E mini, 7970E tape drive, 9895 dual 8″ floppy unit, 2645A and 2647A terminals, running RTE-6/VM. Vintage 1983.

  9. The Prime computer’s microcode was on floppy disk? Awesome. I imagine that could lead to some really hair-pulling debug issues if a byte or two of that floppy disk got corrupted in just the right way…

    I used to have a Prime computer keychain given to me by my dad, whose company used Prime computers for control systems for automated wire-wrap board machines.

    1. Some PDP-11s had a microcode patch SRAM, so any fixes or user enhancements to the microcode ROMs could be dynamically loaded at boot.

      Unlike the PR1ME, you toggled in the bootloader on the front panel on every power up. Usually this was enough to cue up a peripheral like a tape drive or paper tape reader to start reading a more complex bootloader into memory and execute the results.

    2. Take a look at the ICL/Three Rivers PERQ, it had a completely programable microcode and a CPU built out of discreet logic. It was based on the Xerox Alto machine and was one of the first GUI workstations, but sold poorly. They switched over to the 68020 by the third version but the first two were pretty much unique for the time (hardware accelerated, portrait orientation video at 1024×1280 for example).

  10. ” This is the workstation found on just about every desk at Bell Labs in the 80s, meaning this is the same computer [Ken Thompson] and [Dennis Ritchie] used for their work on UNIX.”
    Actually Unix is from the early 1970s and Thompson and Ritchie wrote it on a PDP-11. They may have used the 3B2 but most of their work on Unix was probably long over by then.

      1. Yup. And once Pike and Locanthi built the BLIT terminals, the folks in Murray Hill all got them instead of using dumb terminals. The terminal on top of that 3B2/400 was an AT&T-built mostly-VT100-clone (newer ones had color, not just green or amber!), but the most common terminals we had in the 80s were HP 2621 (usually with the printer) or sometimes DEC VT100/VT102, but there were lots of other brands of ASCII terminal around. The Ann Arbor Ambassador was the really cool terminal because you could get 60 lines of text on it (in really-small-font mode.) There were also some Tektronix vector graphics terminals around, but they were a lot more specialized.

        The 3B2 was a late-80s machine; before that we had the 3B5 (roughly 3′ cube), made with the same WE32000 microprocessors, and before that the 3B20 (a VAX-sized bitslice machine.) But Ritchie, Thompson, and gang were more likely to be using VAXen or mainframes (with Unix on an Amdahl 370-clone.)

  11. Kind of silly to emphasize a lack of one make or model type. Perhaps you forgot about the two HP exhibits (one mini focused, one workstation focued)? I say people should decide for themselves – see for photos of ALL exhibits and for a complete list of exhibits. Bare in mind that last show we had a lot of IBM exhibits but none this time. There is an ebb and flow as to what is covered each year, there have been 9 VCF East’s – the depth and variety of exhibits cannot be disputed when one takes a larger view.

  12. As a kid with a modem growing up in Dallas in the ’90s, my lifeline to the Internet was the 3b2/500 that ran. That was my first run-in with a UNIX shell, real email (via elm), real USENET (via tin), and getting stuck like a DOS noob in vi.

  13. I worked with Prime computers for about 20 years in Denver. Used the PRIMOS O/S… but our applications were developed on a relational database called Prime Information.
    It was a derivation of the PICK O/S.
    The first Prime we had… used 16 toggle switches to set on the front to start the boot process. Our first 300MB disk drive was about the size of a filing cabinet and cost $10,000.

    1. Cray Y-MP EL. There’s one pic of it above. Don’t know if they run it (or have software for it). They were running the PDP 11/40. That’s the building’s heating system.

      1. Is that what was doing it? I walked past it and noticed it…… Anyway yes their market is famous all over the community. And that was the session after the storm that made a big mess all over NJ. The years earlier was amazing. (The market, not just the whole show.)

  14. I used a Prime with Fortran and C for about 2 years at work. At the same time, my dept bought a 3b2-400 which I then used to teach C, Fortran, and Pascal as well as OS principles. When it was surplussed, I bought the 3b at auction and still have it. This one had 48 RS 232 ports, 8 parallel ports, and 10base 2 Ethernet.

  15. Too bad I grew up too late to have anything to do with these “Real Computers”…
    The closest I have ever come to anything like this would be one time on a school trip to a place that handled grain for export and they had some sort of big-ass IBM Mainframe off in a room somewhere (probably running some sort of special software to keep track of all the grain)

  16. Primos was initially written in FORTRAN, but later all new code was written in PL1 subsets (PLP and SPL compilers). Near the end, I think some routines were also written in Modula2.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s