I was writing about the early 70s.Pete -------- Original message --------From: Gregg Tracton <[email protected]> Date: 4/29/18 4:05 PM (GMT-05:00) To: Pete Soper <[email protected]> Cc: Tadd Torborg <[email protected]>, Ken Boone <[email protected]>, Trianglerobotics <[email protected]>, [email protected] Subject: Re: [TriEmbed] Core Memory shield for Arduino
I worked at Burroughs in the 1980's. Internally, we did not use algol but a set of more efficient languages. The guy who maintained the compilers sat in the next office (remember offices?!? Pre cubes). Core storage, to a programmer, had words that were 53 bits, 48 for the data and 4 for the type of the number (int, float, EBCDIC - a character encoding, even 4-bit and binary). 1 bit marked the word as read only. (There were other bits that only the hardware could access.) 8 6-bit characters were packed in each word (total 48 bits) which made for some very confusing math to find a specific character. They were really fast machines for the time, way faster than PCs, but did crash at least once weekly. Many crashes required calling a repair guy and waiting for him (they were all male) to travel to our plant, and bring parts. Many fixes actually required them to rewire the boards! Editing code and documents was tedious. You displayed a page of code, made your edits locally, then transmit that back to replace the original page, and get the next page. If your new code exceeded a page, you had to pre plan that out. Moving blocks of code was hard, so I usually wrote my code on paper first, then typed it in, to avoid changing my mind later. -g Sent from mobile... Pls excuse brevity and typo's. On Apr 29, 2018, at 3:08 PM, Pete Soper via TriEmbed <[email protected]> wrote: More than a couple of decades ago I visited a friend going to Georgia Tech. He was most keen to show me the Buroughs B5500 (a mainframe with system software written in Algol at a time when all other manufacturer's systems used assembly language. As we were making our way toward the Burroughs we passed by a huge cabinet several feet long and wide and coming up a bit over waist high. John's eyes sparkled with pride as he said "this our megabyte", meaning a megabyte of core for the huge IBM 360 surrounding us. A few years later I watched a guy use a clever trick to get past the fact that Data General minicomputer A lacked a paper tape reader but that was needed to get the second level bootstrap loader for A's disk drive set up. He pulled the core cards out of A and put them in B that had a reader, loaded the tape, halted and shut it down, then put the core back into A, started it up, and got the disk bootstrapped so the second level loader could be installed on the drive where it could subsequently be brought in by the primary bootstrap loader. Speaking of, the university lab I worked for couldn't bear the few hundred dollars for a ROM-based loader (just a bunch of diodes and trivial 7400 logic) so we had to put (IIRC) 17 instructions into core with the front panel toggle switches. While my hippie friends were blistering their fingers with guitar strings I blistered mine with those switches developing an alternative loader that would work with a digital tape drive we got cheap, but with zero driver code. One particularly nice feature with DG's early operating systems was a "checkpoint/restart" aka hibernation scheme that was reliable. So you could invoke the checkpoint, shut down, and later come back to the same state. It was decades later that I saw Unix systems get that right (Solaris 7 got it right 80% of the time, but the cookie-barfing other times trained me out of relying on it), with PCs getting this solid eventually too. I think it was because the DG peripheral controllers and their drivers were too simple to have state that was hard to save and restore properly. The OS was very much simpler, too, but the very predictable nonvolitilty OC core provided a definite advantage. Pete (next to a creek outside of Boone) -------- Original message --------From: Tadd Torborg via TriEmbed <[email protected]> Date: 4/29/18 1:29 PM (GMT-05:00) To: Ken Boone <[email protected]> Cc: Trianglerobotics <[email protected]>, [email protected] Subject: Re: [TriEmbed] Core Memory shield for Arduino Oh my. This is amazing. I want one. I remember seeing 256bit core-memory cards back in “the day”. A couple of decades ago a buddy of mine showed me that he had several pickle jars filled with tiny ferrite beads. These were surplus COREs! He told me that he always wanted to own a million of something and there it was. He owned several million ferrite beads and you could look at them. Counting not recommended. Tadd / [email protected] NC FM05pv “Packet networking over ham radio": http://tarpn.net/t/packet_radio_networking.html “Raleigh-centric ham radio resources page": http://torborg.com/a On Apr 29, 2018, at 1:07 PM, Ken Boone via TriEmbed <[email protected]> wrote: For the people who remember the computers that used them. Just $39.00. I have several core memories in my collection including a 6X6X4 inch multiple layer one. The early space shuttles and earlier maned missions like the moon also used core memories to keep high energy radiation for flipping memory bits. Keep up the building. Ken https://www.tindie.com/stores/kilpelaj/?ref=offsite_badges&utm_source=sellers_kilpelaj&utm_medium=badges&utm_campaign=badge_medium _______________________________________________ Triangle, NC Embedded Computing mailing list To post message: [email protected] List info: http://mail.triembed.org/mailman/listinfo/triembed_triembed.org TriEmbed web site: http://TriEmbed.org To unsubscribe, click link and send a blank message: mailto:[email protected]?subject=unsubscribe _______________________________________________ Triangle, NC Embedded Computing mailing list To post message: [email protected] List info: http://mail.triembed.org/mailman/listinfo/triembed_triembed.org TriEmbed web site: http://TriEmbed.org To unsubscribe, click link and send a blank message: mailto:[email protected]?subject=unsubscribe
_______________________________________________ Triangle, NC Embedded Computing mailing list To post message: [email protected] List info: http://mail.triembed.org/mailman/listinfo/triembed_triembed.org TriEmbed web site: http://TriEmbed.org To unsubscribe, click link and send a blank message: mailto:[email protected]?subject=unsubscribe
