Skip to main content


The CPU is far from being the most sophisticated component of a computer.

At least if we're talking about #permacomputing, or, rather, scavenging and collapse computing. Okay, maybe in open source hardware, too.

Designs of new hobbyist computer architectures are seemingly revolving around inventing a CPU and/or mapping the peripherals on the system bus.

And you could find many simple CPUs based on FPGAs, logic chips, transistors, valves and even relays.

What you usually don't find is custom RAM. Before Intel introduced cheap solid-state RAM in 1969, there were at least six contemporary competing types of RAM used in computers, and at least as many were already considered obsolete.

What you don't find is peripherals. There are rare cool appliances, like punch tape readers. But have you seen a custom hard drive? A printer?

All these are "easy" in terms of relative complexity for industry. But they are simultaneously very hard for a hobbyist/DIYer/tech collapsnik.

Change my mind, show me the good stuff~

reshared this

in reply to Nina Kalinina

there's also lots of great stuff in the retro computing community, but that's usually not prepper focused, so there's no anxiety about the validity of stuff like using an arduino to glue an sd card reader to a beloved old computer. I don't think she's posted about it online, but one of my friends was showing me core memory she made as part of a restoration project.
in reply to aeva

also keyboards are very popular DIY projects these days
in reply to Nina Kalinina

sometimes no. I made one out of wires and cardboard, but it wasn't very pleasant to use unsurprisingly.
This entry was edited (5 months ago)
in reply to aeva

that was a just for fun thing though. my perma computing phase ended when I tried to make my own batteries 🤷‍♀️
in reply to Nina Kalinina

I did a bunch of research, and decided that a gravity cell hit the right sweet spot of usefulness vs being easy to construct mostly from stuff you could easily loot that most people would overlook. The experiment failed because I underestimated the metallurgical requirement, and I decided that figuring out how to safely cast a zinc anode with the right characteristics in my Chicago apartment was probably not in the cards.
in reply to aeva

@aeva I see! Well, you probably shouldn't do it INSIDE of your flat. It was a common* game for pre-00s children in Russia to cast lead on a campfire, and Zinc has a comparable melting point. If you're wondering where children would get lead, the answer was discarded car batteries and such. If you're wondering if that was safe, no, not at all.

Common = I have heard of this from at least three different regions, including a major city

@aeva
in reply to Nina Kalinina

@aeva it was also sometimes done in Britain from 1950s to 1970s in gardens and sheds/workshops (away from the main house), using the same materials
@aeva
in reply to aeva

My conclusion from the experiment is that that it's probably less work to just not let civilization collapse so completely and totally that we have to rebuild all of the good parts from scratch.
in reply to aeva

actually I think the thing that really did it in was I was info dumping to a former room mate about it all and he just deadass said "what the hell would we need webpages for in the apocalypse" and tbh I think he raised a rather compelling point about how useless I'd be in such a world
in reply to aeva

@aeva a counter point: if you ever tried printing with moveable type, especially without a linotype machine, you would very quickly realise that even the simplest of computers could speed things up a lot.

Even some mechanical linotype machines accept punched tape as an input for the type - word processing like it's 1880

@aeva
in reply to Nina Kalinina

that's still putting a lot of carts before a few very important horses
in reply to aeva

@aeva well, we're in a nice time and place to try to solve difficult problems and write the solution down.

I am curious to learn more about things you consider horses!

@aeva
in reply to aeva

@aeva ah, yes 😀 It's funny to think how people consider "homestead" a solution to this.

FWIW, I think a real collapse-like situation wouldn't mean immediate famine or dying out from cholera, anyways. I've read about the collapse of Bronze age, as well as the downfall of Romans. In those cases, humans didn't rapidly lose access to shelter and sanitation, and they had some food to go around. Even the excess mortality wasn't insane - certainly not the levels in some European countries during the WW2. All of those were very sucky times, that's for sure.

@aeva
in reply to Nina Kalinina

look, we'll all go and live in boats and sustain ourselves off patreon, as our ancestors once did
in reply to aeva

anyway, a common thread for these kinds of projects being hard is supply chains. Take the gravity cell thing. That was something that was popular once because you could just buy all of the stuff you needed from the local general store. Or like, photography was invented simultaneously by a bunch of different people working independently of one another worldwide, within a few years one another.
in reply to aeva

@aeva oh, things that could be bought from your local store ❤ I remember reading a century-old magazine that explained the DIY radio with "all you will need is just a piece of tin foil from a tea box". And they meant tin tin foil, not aluminium "tin foil".
@aeva
in reply to aeva

my personal theory these days is that prepper fantasies are something that comes from a complete lack of trust towards your community and neighbors (which incidentally cooperation in your community and neighbors is how humans survive difficult circumstances) and it's soothing to think you can just build your way out of not being in control of everyone.

reshared this

in reply to aeva

@aeva Supposing decline is long and slow with chaos sprinkled here and there we can look at being poor in a rich country or a displaced refugee elsewhere. Entertainment, communication, mapping. People who think reference materials are useless change their mind as soon as they have a weird rash or have to walk for transportation and can't afford a detour.
@aeva
in reply to mirth

@mirth it is important to remember that those are not "profitable" and are basically an expensive hobby or a waste of resources - depending how you look at it. But then, a band aid is a waste of money until you scraped your knee, too.
in reply to Nina Kalinina

Even in extremely dreary conditions like muddy trench in a war zone or a tent camp amidst the rubble of a city under siege people will go to some trouble to charge their phone, sideload some music or a movie onto their phone, and search for places they can get fresh water or other scarce resources like fruit. Supporting morale is necessary and doubly so if there are kids around.
in reply to Nina Kalinina

@aeva I made one with Alps (well, Matias) switches. I always thought those were the better retro style key switch. Loved the old Dell AT101 keyboards.
@aeva
in reply to aeva

@aeva 3d printer as storage device, it prints little boats onto a big array of load cells to set bits and runs the head into them sideways and knocks them off to clear bits
@aeva
in reply to emily, emitter of spooky noise

@emily see the first place I would have went with that idea would have been printing the world's crappiest sounding records
This entry was edited (5 months ago)
in reply to aeva

@aeva I do think it would be hilarious to boot a computer from a vinyl record

it's like a cassette, but the initialization sounds *warmer*

@aeva
in reply to Jari Komppa 🇫🇮

@sol_hsa a tape in is a mandatory part of IBM PC! Booting DOS from this input is less weird to me than from ROM
in reply to Nina Kalinina

I never saw a single PC with a tape drive. I've heard of them, though. also never saw a PC with Rom basic.
in reply to Jari Komppa 🇫🇮

@sol_hsa Colorado drives used to be ubiquitous - I have one in my 286. Our home gaming/server PC has an LTO tape drive, too - it's very nice for storing all the downloads, for the low cost of £10-15 per 3TB used archival-friendly tape.
But I'm talking about something completely different here! On IBM PC and PC Jr, one of 8255s was used for a cassette interface - a 5-pin DIN that can be connected to a domestic casette player
in reply to Nina Kalinina

@Nina Kalinina Just googled (duckduckwent?) what a Colorado drive was, and all the results were about road trips.

I'm not familiar with this term. Could you possibly enlighten me?

in reply to Jeff Shaffer CBET, ret

@CivilityFan I have a Sharp pocket calculator running Basic on a 4-bit CPU with a cassette port, as well. It's a really convenient interface
in reply to Nina Kalinina

hard, but also prob often not worth it. F e why build my own printer, if I can just buy a used older one, or just forego printing?
in reply to Florian Idelberger

@fl0_id you can also buy a CPU, or a computer, or decide that you'll be alright without computers, as billions of people did. The area of permacomputing is outside of interests of a capitalist society. The market of DIY computers is likely similar to the market of DIY clothes: very niche, with some big players trying to capitalise on entry-level kits, and peer to peer sales and services. And I like it that way.
in reply to Nina Kalinina

In my view, most electronics won't last. The aesthetic is fun, but if we're being realistic, I feel we need to go back to treating computing as a human activity, rather than one influenced by machines.

It's easy, if not desirable, to build things that look like foundations in isolation. But when you strip out something as basic as electrical or mechanical power, design gets.. complicated.

Programming languages, computing stacks, should be able to be run on humans.

This entry was edited (5 months ago)
in reply to (wryl)

@wryl as they used to, in the past!

I think, with the knowledge that we have today, computers can be made using 18th century technologies. And there is meaning in having computers like that: engineering, science, even art.

@wryl
in reply to (wryl)

@wryl
A few years ago I made a "card game" that could be used to convert an 8 year old child into a touring complete computer running the brainfuck programming language.

The small stacks are memory cells, the big cards are the program. The figure is the pointer. The player just has to follow the instruments on the cards.

in reply to Hackbroetchen

@hackbroetchen that reminds me of Cardiac computer, but simpler! Do you have a link for the "print and play" version?

Hm, your post gave me a really cool idea, thanks!

in reply to Nina Kalinina

no PDF sorry. It would be in german anyway. And a lot of work to cut out the memory stacks....
Also it was a bit unfinished. It lacks a loop depth register.
Maybe I should finish that project.
in reply to Hackbroetchen

@hackbroetchen gotcha! Thank you for sharing anyway.

As for cutting out, there are simple paper cutter knives that are perfect for this task. I've had a great success with cutting a few hundred small cards with a machine like on the picture, with low effort.

in reply to Hackbroetchen

@hackbroetchen @wryl
This reminds me of the game of Alligator Eggs, which seemingly allows an 8-year-old to learn about lambda calculus: worrydream.com/AlligatorEggs/

We might be able to find similar ideas among other esoteric languages.

in reply to smlckz

@smlckz rule 110 is the easiest to implement but also computationally tedious to the point of being useless. So balance must be found
in reply to Nina Kalinina

I think the realization you're approaching here is that collapse computing, as currently practiced, is largely cyberpunk LARPing.

You will have a lot of trouble manufacturing any sort of nontrivial IC, including but not limited to a CPU, in the apocalypse. If your design includes any such component and it's not readily salvageable from the devices normal people are surrounded by, you've already failed.

An FPGA is decidedly nontrivial and also needs an existing, working, relatively modern computer to program it, so that's worse.

Writing an OS for simple Z80 computers sounds like a fun project, but branding it "CollapseOS" is a bit silly because... where are you expecting to find the components to build a simple Z80 computer? If you're salvaging from Game Boys and TI calculators, why not just write an OS for those? The average person in that scenario will have a lot less trouble actually installing and using the thing in practice.

You know what's easy to find in the apocalypse? Existing working computers. Refurbishing them and building methods of powering them is *significantly* more doable without factories. But building an entirely new computer sounds a lot cooler to most people.

in reply to emily, emitter of spooky noise

@emily yep, I think I agree with you, generally. There is still value in CollapseOS and such - it's an attempt at simplification the stack and bootstrapping, and we need more of those.

I am thinking about this topic in the context of how personal computing stopped being accessible thing; we have lots of computers around, but we often cannot reprogram these, and it's unlikely we could repair or expand them, either. I am also remembering the beginning of 2022 and the effects of sanctions on Russia - the electronics shops quickly ran out of everything, and the marketplaces only had used Pentium 3 and 4, as well as Chinese 8 and 16 bit console clones. These times didn't last, because globalisation and parallel imports, but it was a reminder to me that supply chain hiccups or even total collapse are a possibility.

in reply to Nina Kalinina

Certainly, there are different levels and genres of collapse. All of them share the common theme of "we have some existing working devices and everything else must be built from scratch or from parts salvaged from those devices".

For existing devices, Linux is probably a more effective collapse computing project by accident than any dedicated software project will ever be on purpose, because of the sheer amount of stuff it can already run on and the number of copies of it floating around.

For building new devices, ICs are probably the wrong level to build up from. You can assume no new ones can be acquired in most realistic scenarios (either they aren't being produced at all, or supply chain issues mean you can't get them). If you have devices to salvage usable ICs from, they probably still mostly work, and pulling them apart is much higher risk than just using them as is.

On that note, I was under the impression CollapseOS targeted only homebrew machines, but I see now they also have ports for an assortment of real retrocomputers. That's great! The TI-84+ is the only one I'd realistically expect to be able to find outside a retrocomputing enthusiast's home, but it's a hell of a lot more than I expected, and so I partly retract my mocking of them.

But more than that, I want to see more "I made transistors in my garage" projects. *That* is the primary thing we'd need to actually bootstrap modern technology. Software is near trivial in comparison.

in reply to emily, emitter of spooky noise

@emily you will be surprised by the number of PIC and ATmega controllers in things, there are billions of those, and they are powerful enough to emulate a CP/M machine
in reply to emily, emitter of spooky noise

@emily yep! There are two approaches I've tried for scavenged SOICs.
One is to leave them on the board they're from. It works best on boards with lots of testpads, and gives you a tiny breakout board for the chip.
Another is to glue the chip upside down to something, and solder the wires to each pin directly - with enough flux, the surface tension force makes it relatively easy to avoid shorts. It is a bit tedious for chips with over 20 pins, but it's doable as long as you have reasonably thin wires (like wires from transformers)
in reply to Nina Kalinina

@emily These thoughts reminds me of my writing exploring "what's the least software & hardware required to build an inclusive web browser?"

I think it helped me better wrap my mind around around existing operating systems, & really emphasizes how vital memory is!

in reply to Nina Kalinina

For exhaustive detail: adrian.geek.nz/from-scratch/#b…
(I'm currently preparing a page about dev tools for programming it)

In short: The primary (until visual output's required) component would be a hardware pushdown automaton. Which traverses a labelled graph to parse its input data.

Then I added an "Output Unit" to concatenate & demux the output into a new order, a rudimentary processor for occasional/trivial math, & a couple other things.

Anything specific you're curious about?

Nina Kalinina reshared this.

in reply to emily, emitter of spooky noise

@emily

Also, in a collapse scenario, you're going to be way too busy with basic survival to bother with hobbies. Nor will there be any electricity available.

@nina_kali_nina

in reply to argv minus one

@argv_minus_one that depends on how you define collapse. When political and economic system crashes, and the streets are ruled by violent gangs, one of your best bets is to stay at home and tinker with electronics, if you have the skill.
in reply to Nina Kalinina

I've built a printer from scratch, but it's an electromechanical contraption not directly related to computery things and not exactly a convenient form-factor.
(though watch this space, I still have loads of ink and I've been Having Bad Ideas for future projects after watching too many teleprinter videos... 😳)

oldbytes.space/@zxguesser/1125…

in reply to Nina Kalinina

There’s hobbyist core memory.

The kind of person who makes a custom CPU design out of discrete transistors or valves or relays also often makes a small amount of RAM with flip-flops or individual transistors. It’s definitely hard to scale up.

Open mechanical stuff is indeed really rare, with the notable exceptions of rapid prototypers and that electron microscope from a while ago.

in reply to Zimmie

Turns out there have been more electron microscope designs released than i had thought. U Muenster’s SXM, the ChemHacker effort in ~2010, the recent NanoMi.

An open printer or plotter would be nice. And certainly seems like those would be easier to make than open electron microscopes or CNC machines.

in reply to Zimmie

@bob_zim making a decent printer is likely more complicated than making a CNC. Laser and inkjet require serious know-how, and even dot matrix is harder than it looks. Daisywheel is relatively simple, but it's more of a mechanical project than an electronic one. And mechanical projects aren't that common nowadays
in reply to Luka Rubinjoni

@rubinjoni I know of at least two cases of custom hardware plugging in to existing printing heads/chassis. Reusing inkjet heads is relatively simple, after all - probably simpler than reverse-engineering old-ish printers to make firmware for those. Newer printers probably use some kind of ARM, and in that case should be easier to reverse-engineer.
in reply to Nina Kalinina

I would expect most types of CNC machine to be more difficult than most types of printer, simply because print heads don’t fight back against the positioning system when they act on the feedstock. The only mechanical resistance they deal with under normal circumstances is on an axis the print head doesn’t move along. Pigment delivery and fixing definitely isn’t easy, but I doubt it’s as difficult as repeatable, extremely strong positioning of tools.

I wonder what would be the best pigment delivery system for such a printer. I’m partial to waxjets (e.g, Xerox Phaser) largely because of the low material waste, but some inkjets come close (particularly if the printer has refillable tanks and you get the ink in glass bottles).

in reply to Zimmie

@bob_zim I disagree about complexity because I've worked on building hobbyist CNCs - they are relatively straightforward and surprisingly precise, especially if you don't care that much about the speed. I've been thinking about printers for a long time, and I couldn't invent anything better than a daisywheel or a low resolution riso. I suppose xerography, spark and thermal printing are relatively simple, but the first one might shock you, and the other two require special paper which is hard to DIY and that sort of defeats the purpose.
in reply to Nina Kalinina

Maybe for wood or plastics. It’s really hard to build a two-axis router to carve channels into metal, though. Swiss lathes and 3+ axis mills are difficult, too.

I’m not familiar with any DIY paper process which yields a uniform enough thickness and composition for most printer technologies. Definitely not good enough for traditional laser. Composition would be a problem for any jet-based. Struck prints like dot matrix or daisy wheel would probably be it.

in reply to Zimmie

@bob_zim DIY paper that works with laser is definitely doable. But making it thermo sensitive or cover it with thin metal film is hard 🙁

Soft metal CNCs are okay to make:)

in reply to Zimmie

@Zimmie @Nina Kalinina I'm working on designing a CPU from "scratch" as a hobby project. Being able to build my own RAM would be amazing. I have no idea how to do it for anything more than a few bytes.
in reply to Jonathan Lamothe

@me I'm currently working on two prototypes, both relatively simple and, if working, should be able to hold at least a kilobyte
in reply to Nina Kalinina

@Nina Kalinina I'll try to keep an eye out for that. If you post about it (and you can be bothered/ remember to) please @ me.
Unknown parent

Nina Kalinina
@argv_minus_one again, that depends on how you define collapse. I could offer you counter-arguments, but I'm not sure if you're interested to listen at all 😀
in reply to Nina Kalinina

I wonder if anyone's tried replicating the memory used by two of the computers in the Voyager spacecraft. They each have three computers, two with some kind of wire memory and one with a more modern kind.

All the problems the Voyagers have had is with the more modern memory. The wire memory is still working perfectly afaik.

in reply to Kirtai

@kirtai it is an interesting variation of core memory, somewhat similar to twistor but operating on a different principle, nice.
in reply to Nina Kalinina

It's surprising how many cool ideas have been almost lost from early computing.
in reply to Nina Kalinina

What I am interested in is exploring the GPU in the fpga/open hardware space. Not a lot of projects have attempted this, possibly because it's extremely hard.
in reply to mcc

@mcc Indeed, I couldn't find a single one with even a rudimentary OpenGL support. You probably know these, but I'll share for those interested who don't. There are a couple simple 2D accelerators, like this one: opencores.org/projects/siga
There are a bunch of gpgpu projects, but that's not fun. And then there's FuryGPU but it's Vaporware until it's actually released: furygpu.com/about
@mcc
in reply to Nina Kalinina

Yeah! There's all these projects to make open-hardware laptops but without a GPU that supports standard APIs there's gonna be a hole in the open stack.

A thought I keep having is a good place to start might be to accelerate *just* what is useful for 2D compositing, and pass the rest to Mesa. Then you could get improved desktop performance, and most daily tasks don't involve 3D rendering anyway.

in reply to mcc

@mcc a good thing is that the CPUs are fast enough to run simple GUIs adequately fast even on top of a simple frame buffer. But it is very important to have an accelerator, as you say, for anything else
@mcc
in reply to Nina Kalinina

I think I heard on the fpga mnt reform the scroll performance was kinda bad
This entry was edited (5 months ago)
in reply to mcc

@mcc Vortex¹ ² apparently supports Vulkan 1.0 via a modified SwiftShader—is that at all interesting? (I wonder if it’d be possible to also get OpenGL ES support by layering ANGLE.)

__
¹ github.com/vortexgpgpu
² B. Tine, V. Saxena, S. Srivatsan, J.R. Simpson, F. Alzammar, L. Cooper, and H. Kim, “Skybox: Open-Source Graphic Rendering on Programmable RISC-V GPUs,” in proc. ASPLOS 2023, ACM, Mar. 2023. Online: dl.acm.org/doi/10.1145/3582016…

@mcc
in reply to Hovav Shacham

@hovav I mean, it's a start! I think it's not what nina's looking for but I'd be curious at least to test it.

I'd heard there were projects like this but I didn't know they'd already reached a Vulkan compliance level.

in reply to mcc

@hovav I wonder how I'd go about porting this to this FPGA I've got if I wanted to run a test and see what it looks like. There's test instructions here github.com/vortexgpgpu/vortex/…

But it's pretty much "run our build scripts!" and don't explain what they do or what they target. Like would I need to integrate this with a litex setup, or…?

in reply to mcc

@mcc they seem to target a wide range of platforms: github.com/vortexgpgpu/vortex/…
@mcc
in reply to Nina Kalinina

! Quartus project already set up, this is unusually courteous for HDL stuff
in reply to Nina Kalinina

I don't know ... strictly in terms of STEM learning tools for bright kids, I'll take AdaFruit, Arduino, etc.any day over the GENIAC I had when I was a bright kid!

en.wikipedia.org/wiki/Geniac

in reply to Nina Kalinina

Sorry to jump in but I don't know that end-user hardware or even software preservation are nearly so important as algorithm management and manufacturing equipment specs.
The long-term problem wouldn't be finding hardware to read an old SSD -- but how to make such devices at all. If we look at a modern factory, how many of those tools could be remade 100 years after a collapse? Have we documented them? Or are they all trade secrets?
in reply to Michael T Babcock

@mikebabcock oh, but that is EXACTLY what I'm talking about! Lots of "obsolete" technology is just junk because it's impossible to operate, repair or replicate - and it will include crucial components of our current infrastructure that we neglect for the reasons like "it's obvious, widespread and cheap". Anyone who tried fixing 60 years old hardware, even using the documentation preserved, knows that it is an enormous task. I'm scared we will be forced to reinvent the wheel, even without total production line collapse, in case of "too big to fail" companies going bankrupt and taking their proprietary technologies to the grave.
in reply to Nina Kalinina

I think learning to desolder and solder SMD components could be the top skill for a scavenger technologist. The vast majority of discarded tech is going to be SMD....

That's why I'm working on DIY co-processing using 8 bit CPUs from a begone era 😁😬

in reply to Sasha

@sashabilton don't be afraid of SMD. It's very doable by hand soldering except for ball grid arrays. Even for point to point assembly, the limiting factor is going to be just the width of your wire. And making SMD-ready PCBs at home is relatively simple if, for example, you have a decent analogue photo lab.
in reply to Nina Kalinina

I've ordered some cheap SMD practice kits thanks to your encouragement. I look forward to adding tiny LEDs to *everything* 😄. Thanks!
in reply to Sasha

@sashabilton if you can afford it, I recommend getting "helping hands" and a cheap magnifying glass on a stand. Having a sharp soldering iron needle, good tweezers and decent flux won't hurt, either. There's a few YouTube videos that can help you to start, if things won't work out naturally
in reply to Nina Kalinina

@Nina Kalinina I didn't know the term "ball grind array" but had encountered it in the past and based on context was pretty sure I knew what you were talking about.

Looked it up to be sure. Felt like one of my riskier searches.

This website uses cookies. If you continue browsing this website, you agree to the usage of cookies.