Ever heard of VENIX? There were lots of variants of Unix back in the day, and VENIX was one for the DEC Professional 380, which was — sort of — a PDP 11. The 1982 machine normally ran the unfortunately (but perhaps aptly) named P/OS, but you could get VENIX, too. [OldVCR] wanted to put one of these back online and decided the ST-506 hard drive was too risky. A solid-state drive upgrade and doubling the RAM to a whole megabyte was the plan.
It might seem funny to think of a desktop workstation that was essentially a PDP-11 minicomputer, but in the rush to corner the personal computer market, many vendors did the same thing: shrinking their legacy CPUs. DEC had a spotty history with small computers. [Ken Olsen] didn’t think anyone would ever want a personal computer, and the salespeople feared that cheap computers would eat into traditional sales. The Professional 350 was born out of DEC’s efforts to catch up, as [OldVCR] explains. He grabbed this one from a storage unit about to be emptied for scrap.
The post is very long, but you get a lot of history and a great look inside this vintage machine. Of course, the PDP-11 couldn’t actually handle more than 64K without tricks and you’ll learn more about that towards the end of the post, too.
Just as a preview, the story has a happy ending, including a surprising expression of gratitude from the aging computer. DEC didn’t enjoy much success in the small computer arena, eventually being bought by Compaq, which, in turn, was bought by Dell HP. During their heyday, this would have been unthinkable.
The PDP/11 did have some success because it was put on a chip that ended up in several lower-end machines, like the Heathkit H11. Ever wonder how people programmed the PDP computers with switches and lights?
Minor correction. Compaq was bought by HP not Dell.
Which is why x86_64 VMS only supports one type of bare metal server, and it’s an HP. Otherwise, it has to run in a VM. Oh the irony of running VMS under a thinly-veiled {Li|U}nix OS.
HP also bought Tandem.
HP under the dumbass witch bought lots of overpriced companies.
That’s how corporate growth worked in the first .com era.
None were a worse buy then EDS.
Tata consulting had already stolen their business model, and is frankly better at it.
But that’s just because they have a larger supply of incompetent techs available to do the needful.
Autonomy must be in with a shout for that accolade, surely!
Er, not quite – it was HP who eventually bought Compaq.
There were a few Soviet PDP-11 compatible desktop workstations and home computers (ДВК-3, БК-0010)
and Cray.
The chip-based 11’s like the 11-03 had some success in reseach and lab applications. Had the fun of programming one in assembler. HP was also competing in that market.
After a number of years of working with assembly on several 8 bit processors, the PDP11 architecture was like a religious experience for me. My only complaint was that the tools all worked in octal rather than hex.
But if you’re working on a PDP11, octal is the base that makes sense. Since most PDP11 instruction addressing is done by specifying a 3 bit address type, and a 3 bit register specification, if it’s in octal you can easily just read what they are. If they used hex or decimal, it would be harder to just look at them and see where the 3 bits fields values’ were.
Octal made sense for when you looked at core dumps. But using octal for data and addresses was a pain. But mixing octal, decimal and hexadecimal in the same project is asking for problems.
Given DEC’s roots in 12 bit machines, octal was natural. Coming from a 16 and/or 8 bit world, not so much. Being able to depend on a byte having 8 bits and a word having 16 changed a lot of how things were done.
Yup. with the 9825 and 9830…..don’t ask me how I know.
I used Venix on a DEC Pro/350 in the mid to late eighties (I was a teenager). We had a DG Eclipse MV 4000 running DG/UX. The OS was sort of late Alpha or early Beta at that point – this was before the AViiON had even started. Most stuff worked, nroff very definitely didn’t work all the way (the MS macros had something that would cause a core dump, for instance). Serial cables were a thing that existed, so I banged out a couple of shell scripts. One ran on the DG and would “uucp” the source file over to the Venix box. Then it would use “uux” to start a script running on the Pro. That script would nroff the file for you and uucp it back to the DG. Move the completed output file back over from the spool and it was nearly as good as having nroff locally.
I was very proud of myself.
The /350 and /380 had so much potential but it was obvious they had been intentionally handicapped so they wouldn’t eat into PDP sales or even, honestly, VAX sales. Seriously… on a “MIPS per user” basis they would clobber a VAX 11/750 or 780. That assumes a heavily loaded system, of course, but back then if your VAX wasn’t heavily loaded then you were very clearly doing something wrong.
I wish I hadn’t googled this… “Xhomer” is a Pro/350 emulator. There went today’s productivity.
I dream about new system on similar machine, some serenity os , haiku or similar. no gui but fast .
Meybe new minix 4 or fuzix on steride
I was working at a university that was given 6 or 8 of the single board PDP’s and the cartridge tape drive. After making a few plywood housings with power supplies we got to work on doing something useful. Booting from the tape took well over 30 minutes! We never got past he boot and debugging. I think they were throw out when new faculty cleaned out the physics department 20 years later – along with all the Weston cells and Leads & Northrup meters and ballistic galvanometers, old spectrometers, Kerr cells, a big Beckman analog computer, etc! I wish I had known.
Usagi Electric (YouTube channel) has been working on a very similar machine, a PDP/11 board and a power supply mounted on plywood, presumably used with some sort of spectrometer, I wonder if this was a common thing to do…
I disagree with the closing remark that PDP/11 was popular because of the single-chip that was put into lower end machines. The PDP/11 was hugely successful in the era when CPUs were multi-chip.
One of the nice things about it is that the instruction set is nicely regular and complete. And it includes instruction features that were thought to be “cheap” when cpu clocks were still measured in kHz. If you use a loadable up-down-counter as a register, a pulse on the increment or decrement input is “for free”. So you had post-increment and pre-decrement addressing modes. Those stop being “for free” once you get pipelining and other more advanced features. So nowadays those are no longer included in instruction sets. But back then they were great to have.
In my early twenties and in the early 90s I got called into a medical publishing house to fix their PDP/11. They had an entire publishing management system on it that had been written in assembly, in about six weeks, by one guy back in the 80s. They had tried to fix it themselves, including messing with the wire wraps on the boards. None of their serial ports worked as a result so they could not connect to it. So i was staring at a closet full of computer with no way to interface with it.
I went to Radio Shack and got a wire wrap tool. I had no docs except for what was printed on the boards themselves. I proceeded to methodically rewrap the serial line posts until I got a connection on my, get this, Tandy 102. Once a got one of their VTs back online I could see the crash codes but knew not what to do
I managed to get back in touch with the guy who wrote the software. He had me set him up with a connection via a 9600 baud modem and then send an interupt that crashed the system into a “micro code shell,” as I remember him calling it. I sat by the modem on the phone with him for an hour or so while he tooled around in his old code. Ultimately he got it working again. I’ll never forget that experience with that tech as it was fading into history.
That picture reminds me of my first PC — A DEC Rainbow. I’d bought it through the college for 1/2 price which was still expensive and upgraded to 256K of RAM (rather than default 64K). Made good use of it during those years as I could write my papers at home and program in Turbo Pascal for CS class projects. Upload to VAX for formal submission over 300, then 1200 baud modems.
Luxury!
We used to use computers with 4K of ram using a NTSC TV as a display.
You can thank Carly Fiorina for tanking HP. She would never have been hired if Bill and Dave were alive. She is a total failure and failed trying to run for president. A total waste of skin.