For the vintage computer enthusiasts: vid of booting a PDP-11 from tape drive

[culturejam]

It blows my mind how complicated the PDP-11 was, and it was considered a "mini" computer in its hey-day. That one part with all the wire connections is crazy (why didn't they use insulated wire!?!?!). Dude replaced some old monster electros with some MKT gi-normous box caps.

What really took me aback was how long it took to call up a directory listing (around the 13:45 mark). 

[reddesert]

I actually used a PDP-11 once upon a time in the mid-90s, but it was an LSI-11, which is the more modern/compact variant that used LSI integrated circuits. Yeppers, the original PDP-11 used no LSI chips, meaning the whole thing was made out of ICs probably about as complex as a CMOS 4xxx chip, discrete resistors, and so on. Of course, by the mid-90s the LSI-11 was a legacy item; they were easy to interface to hardware control, so had been used a lot for lab equipment.

Somewhere I have one or two "Flip Chip" boards and a magnetic core memory board, from a DEC-10, I think, that I was given as a souvenir. You can see a whole lot of these Flip-Chips with their plastic handles at 2:00 in the video. https://en.wikipedia.org/wiki/Flip_Chip_(PDP_module) The number of components on a single flip chip is impressively small; less than a medium complexity pedal.

The Flip Chip boards are plugged into a wire-wrapped backplane, which appears at 3:30.  It looks nuts but was a sensible fabrication method at the time. Wire-wrapping is now nearly extinct, but I remember seeing a few hobbyists build PCs out of it when I was a kid. It's actually incredibly robust. The wire is very thin and has a thin insulation coating, like magnet/pickup wire.

I think wire-wrapping likely made sense when you had a lot of pins that needed to be connected directly from one IC or board connector to another; it's easier to route than a single sided PCB because you can cross wires. (That is, in digital circuits you often see a complex set of connections between a small number of ICs or boards with a lot of pins, while in analog circuits there are more discrete components and the traces reflect the topology of the circuit.) Also, wire-wrap was used for prototyping and hobby builds because BITD PCB fab was complex and expensive.

[stecykmi]

that's a pretty crazy machine. how do you actually write programs for it? manual input with individual bit entry register by register? i guess there is some machine language code that you have to follow?

[cooder]

Wicked piece of gear. You probably could build a cool stereo tape delay out of it... somehow.. 

Sorry to offend all vintage computer enthusiasts.

[thesmokingman]

the pdp-11 ran unix among other operating systems ... lots of C, FORTRAN, and COBOL ... took ages for that piece of hardware to go away compared to today's standards of lifespan

[culturejam]

I always wanted to check one out in person, but could never find one that was running.

I summer interned at a place that had this abandoned office space on the property (they just let it go wild after the place was bought out by a competitor), and I had to go in there one day to hunt for something the boss wanted. I almost shat my pants because in the back of this place (amid the dust and encroaching wild plants and spiders and whatnot) was a CRAY system of some sort. I was like WTF!! Who the hell just leaves a supercomputer rig just laying around. Craziest shit I ever saw...in an abandoned building owned by a company worked for, that is. (hell of qualifier, I know).

[alanp]

It's mostly industry enthusiasts preserving them now, but I suspect that in forty years time, "real" museums are going to go, "oh shit, did anyone save those early dinosaurs of computing?"

[culturejam]

It's mostly industry enthusiasts preserving them now, but I suspect that in forty years time, "real" museums are going to go, "oh shit, did anyone save those early dinosaurs of computing?"

It's a pretty involved process to just boot the OS. And upkeep can't be all that easy with all the moving parts. I would be stunned if there are working units in 20-ish years.

[alanp]

Depends if they want a static display, or a working one.

[juansolo]

It's mostly industry enthusiasts preserving them now, but I suspect that in forty years time, "real" museums are going to go, "oh shit, did anyone save those early dinosaurs of computing?"

It's a pretty involved process to just boot the OS. And upkeep can't be all that easy with all the moving parts. I would be stunned if there are working units in 20-ish years.

Parts are going to be the biggest problem by far. Most enterprise stuff really well engineered but very bespoke. I have a friend who for one reason or another has got into restoring old Quantel kit. He's had to reverse engineer how the unlock IC's were programmed so he can build new machines out of what parts he has and generate keys for them. Amongst other fun and games, and they aren't that old (30 odd years).

I suppose then it's going to be media, if it's new enough to use magnetic media, that's degrading at such a rate now it's untrue. I've had to abandon it on all my kit. Older stuff that uses paper tape and punch cards with ironically still probably be ok. For a bit of trivia, my brother used to work for British Telecom up until about 10 years ago or so, and not long before he left, the directory inquiries machine was still running punch cards, because it was so damn reliable!

But there are people out there hoarding this kit already for museums. I suppose most will be static displays, certainly supercomputers as they're just going to be too complex to power, cool and maintain. I still want an IBM SP frame though... Used to work on them and they just look so freaking cool.

[alanp]

If they're getting supercomputers from three-letter-agencies, then my guess is that someone is going to go through with a pair of gardening shears and deliberately cut the wiring harness in several places, before they let them go.

[culturejam]

If they're getting supercomputers from three-letter-agencies, then my guess is that someone is going to go through with a pair of gardening shears and deliberately cut the wiring harness in several places, before they let them go.

I think you are overestimating government employees. 

[thesmokingman]

when it comes to disposal security, there's a physical hard drive wiper(electromagnet) and that's about it ... put the hard drive on there, flip a switch, no more hard drive.
with regards to legacy equipment, even ten years ago I was having trouble with upkeep on the silicon graphics indigo 2, ibm rs/6000, and sun sparcstation I had. I let them go to the computer lab in the sky because the cost and availability of operating systems, software, hardware, and space were too much to keep up with from a hobbyist standpoint when you could get the same kind of cheap thrills playing around with BeOS, Sun Solaris, or BSD on regular hardware. I even threw out my empty pdp-11 rack when it came time to move because I just wasn't going to fill it with rack-mount hardware any time soon. Now that I think about it, there's a grand missed opportunity there. BeOS would be the perfect OS for the raspberry pi.

[culturejam]

even ten years ago I was having trouble with upkeep on the silicon graphics indigo 2, ibm rs/6000, and sun sparcstation I had.
...
BeOS would be the perfect OS for the raspberry pi.

My university used to have computer hardware "clearance" sales for old stuff that was retired. Back around 2000, I went to one of these events and picked up a couple SunSpark 5 "pizza box" units and a Sun IPX "lunchbox". It was fun messing around with them, but when I moved to NYC, I had to ditch them.

Haven't heard anyone mention BeOS in a long time. It was a pretty neat OS, as I recall.

[reddesert]

A PDP-11 system fits in a 19" rack so is in the capability of a hobbyist or small museum to operate. I agree that mechanical systems are likely the hardest things to keep operating. IIRC, there are hobbyist communities passing around advice for keeping tape drives alive, etc, but tape itself is quite fragile. Availability of software is an issue. I think DEC was pretty good about making the software available, but there are other old computers where either the software, source code, or the media are hard to obtain or can't be licensed.

A typical PDP-11 would have been programmed in a high-level language. The users would have sat at a printing terminal (DECwriter, descendent of a teletype, used accordion-fold "computer paper") or a CRT terminal like the standard VT100. I think they were usually booted off mag tape or disk though apparently, one could boot them from paper tape. Because the OS was good for real-time control (more so than many timesharing OS), they were often used for hardware control; my advisor programmed one in Forth to control the instrument we used.

I doubt the TLAs are passing supercomputers along to surplus in an accessible way; anyway, I think the storage space and power requirements would put a contemporary supercomputer, even civilian ones, out of reach or interest for a hobbyist. For some time, the academic supercomputers I know about have been massively parallel, and I don't know if the individual nodes are particularly interesting from a history or hobbyist point of view.