Post AxVXoS18MffMJoFG1w by tomjennings@tldr.nettime.org
(DIR) More posts by tomjennings@tldr.nettime.org
(DIR) Post #AxTseZM3wFJ6tvFAuG by tomjennings@tldr.nettime.org
2025-08-24T02:35:43Z
0 likes, 0 repeats
@nina_kali_nina The history of computing is really the history of fast memory. Its pretty damn fascinating. Acoustic delay lines, cathode ray tubes, drums of spinning capacitors, all sorts of crazy stuff until magnetic core. ....
(DIR) Post #AxV36yEkDNVksYxiW8 by tomjennings@tldr.nettime.org
2025-08-24T16:07:36Z
0 likes, 0 repeats
@nina_kali_nina Wow nice! Me and a friend talked about making a Williams tube, now that we're out of the 1940s analog scopes are essentially flawless, and rigging up an opamp to an Arduino etc to prototype one would be not a lot more than a weekend project. Lol maybe. With that knowledge it could then be reduced to minimal electronics and maybe more bits. 256 would be an easy 8 x 8. I wanted to make a drum. I spent a year designing a tube computer, serial, for a drum, 75 tubes, little one address instruction set, simulator, and assembler. Slow...... For subminiature tubes, which are or were all over ebay , cheap. Or dumb old transistors. Or twice as many if I switched to long tailed pairs. Too much work though.
(DIR) Post #AxV3BfUlK6EOCVh7Nw by tomjennings@tldr.nettime.org
2025-08-24T16:08:29Z
0 likes, 0 repeats
@nina_kali_nina (Coat the *inside* of the drum....)
(DIR) Post #AxV75MFeMEy2b2s8lk by scruss@xoxo.zone
2025-08-24T16:52:08Z
0 likes, 0 repeats
@tomjennings @nina_kali_nina plenty of crazy stuff after magnetic core: how about sequential serial memory using shift registers, as in the Datapoint 2200?
(DIR) Post #AxVN5yaRQUmIl3ujLc by phloggen@expressional.social
2025-08-24T19:51:29Z
0 likes, 0 repeats
@tomjennings @nina_kali_nina Delay lines recently got crazier:Sandy Toksvig claimed in her Turing lecture that Turing thought gin would work as well as mercury.
(DIR) Post #AxVNUDmoZAjQqgGKno by phloggen@expressional.social
2025-08-24T19:55:55Z
0 likes, 0 repeats
@tomjennings @nina_kali_nina Back then the limiting factor was impurities in the phosphor coating, I wouldn't be surprised if you find 256 bits slightly tricky, and 1024 very, very hard...
(DIR) Post #AxVXoS18MffMJoFG1w by tomjennings@tldr.nettime.org
2025-08-24T21:51:37Z
0 likes, 0 repeats
@nina_kali_nina Yes, metal foil over the crt face. The whole thing is insane. Ever read about them? Its absurd! Lol FC was apparently a wild inventor. When you squirt an e-beam to phosphor, that negative charge creates a positive charge on the face. But thats the writing event, not the memory. Memory is the phosphor glow, more correctly the pile of electrons still hanging around for T in the millisecond region.Turns out this spot is like a crater, elections bounce out (long known as secondary emission) and it makes like a puddle....But how do you read that? Well you can't. But! when you write to an area that's blank you get that initial current flow you got from the initial write. So reading memory is done by doing a second write, off to one side, but close to the defined spots X Y screen location. Remember that the spot is really a puddle. So if the read beam (off center second write) generates current on the foil, there was no previous write there! A 0! And so you've just added some undesired noise to that 0 spot, but it's mostly still a 0... If your test write ("read") generates no current, then the cell contained a 1.Williams tubes needed constant refreshing. So refresh logic used scrambled addresses and had to interleave with legit read and write. And Williams tubes have a serious access problem I think called "read around"; if you read the same location(s) too often memory corrupts. Its up to the programmer to not do that! WTs are fkn awful. But they were fast! I think a Los Alamos machine used them (MANIAC?) Once you get your head around the weird physics of beams and phosphors and energy wells they inherent problems become comically awful. But I think a 256 bit tube with Arduino and op amp would be fascinating. The poor bastards had only tubes. What was done in ye olden dayes was a parallel crt displayed what the memory tube was displaying, exactly wires parallel, as diagnostic device. I bet it would be really neat to watch.
(DIR) Post #AxVaF0JHggHg74WYCm by tomjennings@tldr.nettime.org
2025-08-24T22:18:45Z
0 likes, 0 repeats
@nina_kali_nina hysteresis makes mag core practical. WTs are terrible! I read that a big problem with them was traced to the factory, which had been a mattress factory in it's past and quality control was lax and there were dust fibers getting into phosphor mixtures making dead spots etc. Brit ww2 something.What's amazing to me is that folk like FC Williams did all this stuff half in his head, just knowing the physics. They couldn't probe the innards of tubes while working, and nothing was fast enough to see those side effects in operation.correspondingly, today, we DO have all that, plus 1000X more. I really think we could pull off an 8 x 8 WT with minimal stuff today, and since we wouldn't be running code from it read-around etc wouldn't matter.
(DIR) Post #AxVc4ZScrEOZr02vnk by tomjennings@tldr.nettime.org
2025-08-24T22:39:22Z
0 likes, 0 repeats
@phloggen Oh for sure. But I bet a 1970's Tektronix surplus 'scope instrument would be pretty good! For a lash-up at least.@nina_kali_nina
(DIR) Post #AxVcmmPY4NaWlCgg3E by tomjennings@tldr.nettime.org
2025-08-24T22:47:21Z
0 likes, 0 repeats
@nina_kali_nina Yes, time-based scopes we not common. Most were analog repetitive waveform tools, so everything AC coupled and slow as shit, drifty as hell. I'm not that familiar with the history of o'scopes, I'm sure someone here knows them well. Certainly, pre-ww2 they were 1 to 10 MHz AC coupled slobby things, and expensive for that. (WIkipedia https://en.wikipedia.org/wiki/History_of_the_oscilloscope sez triggered sweep started 1945ish, mentions Tek 511, and brief ref to 'something seen in germany')So JvN would be referring to the old shitty Dumont types common then. Hell, everything with tubes drifted, and substantially. I mean like discernable percentages up to 10. Radio receivers, you'd leave them on, let them literally get up to temperature, to minimuze drift, in the 10's of KHz in the HF bands, and that's for good receivers.You'll see the acronym "stalo" in ww2 era stuff (always liked the word). "Stable Local Oscillator" as compared to all the other kinds of oscillators.To be electronic is to drift! lol The old days sucked.@phloggen
(DIR) Post #AxVljHqs8BqQhWsR8q by tomjennings@tldr.nettime.org
2025-08-25T00:27:29Z
0 likes, 0 repeats
@nina_kali_nina @scruss There's a 4000 series 1024 bit serial shift register that just screams to be made into serial machine.. lol
(DIR) Post #AxVloyVowpkgSSk6q0 by Raffzahn@mastodon.bayern
2025-08-24T17:52:48Z
0 likes, 0 repeats
@scruss @tomjennings @nina_kali_nina @tomjennings @nina_kali_nina Fast? Not really. Most of the time it was about getting capacity up and price down. All the weired early stuff, like delay lines, were about (some) storage at an affordable price. Same for serial memories and so on.In fact, even most early semiconductor memory was all about price, not speed or capacity. It's a constant issue that emerging technologies are usually slower than older ones. The very reason why core memory was still kind in the mid 1970s when smaller and slower machines were already using semiconductor storage.
(DIR) Post #AxVlozeihESs0LujLc by tomjennings@tldr.nettime.org
2025-08-25T00:28:34Z
0 likes, 0 repeats
@RaffzahnRight! "Fast", but only as compared to slow, like tape! @scruss @nina_kali_nina
(DIR) Post #AxVpvJHB3qhcPUAZnc by Raffzahn@mastodon.bayern
2025-08-25T01:14:32Z
0 likes, 0 repeats
@tomjennings @scruss @nina_kali_nina Not really, as delay lines, drum and even core predates tapes. Tapes are about mass storage with fast sequential access - punch cards on steroids :)
(DIR) Post #AxWJ3f4kn9dj0Xh26C by phloggen@expressional.social
2025-08-25T06:41:00Z
0 likes, 0 repeats
@tomjennings @nina_kali_nina Before oscilloscopes there were oscillographs.BSTJ has you covered:https://archive.org/search?query=oscillograph+bstj