He forgot to mention that there are still at least two computers still using core memory - the Voyager probes! Still going strong (ish) after 40 years! Now that's reliability!
Well, not to ruin the joke... but core memory was actually cool for that reason specifically. You could turn it off and on again, and lose no information. In fact, you can turn it off for hundreds of years, and not lose any information! It's one of the reasons people still bought core even after CMOS memory was available, and even when it was cheaper. No load time! Just turn it on and plug away. I would love to get a set of core memory for my PDP-11/03. The irony is that with windows crashes, core memory would be a disaster! When you hit the reset line on the bus, it'll jump right back into the crash unless you completely reload it from paper tape! You're not doing that in space...
I remember core store lectures (because we might come across old mainframes during our careers as electronics engineers) at college in 1983. Advantages (or why the millitary had a liking for them even duringthe 1970's and 1980's) was that (A) totally EMP resistant, (B) core memory will hold the "bit" stored without power for just over 4 centuries and (C) not as fragile or sensitive as the (then) new static and dynamic RAM devices. I designed and built a 4 bit computer using TTL static RAM, TTL ROM, 4 x TTL shift registers, TTL numerical LED drivers and a 74181 ALU for a weekend homework project. I still have it and it still works.
Hasn't the US Millitary have some ancient super computer buried inside a mountain from back in the 1980's?? That was using core stor memory and there was a documentary on BBC TV back in the 1980's about it. Kind of rapidly superceded by events and technology. Core Stores did have a longer existance in the computing world than DTL logic devices. I know I'm growing old when I can remember old obstelete technology and how it functioned.
There is no way to subscribe to this discussion so I'm forced to leave a message I can get back to (UA-cam might decide to scramble comment order, who knows..). Sorry, keep talking.
P-3? I worked P-3C out of Diego Garcia AMID W/C 610 the VQ-3 EC-130Q TACAMO. I'm trying to remember the manual programmer we used. Collins or Rockwell 7404-1a or 7401-a or something like that 32 toggles in 8 groups of 4 and 32 lights above them with a send/encode button. I was rather good at programming in binary.
I worked with a core memory based computer. The core had to be in a temperature-controlled chamber to work right. I once had to replace the heating elements. This was in the late 1970's, and that computer was well into obsolescence at that time. Oddly, it had a 24-bit word length. As I remember, the memory was only 2048 K, and programs had to be partially run and swapped back and forth from core to the memory drum unit. The drum unit was the size of a washing machine and held 256 K words. It had to be pressurized with helium. The computer controlled a steel rolling mill and was made by General Electric. And yes, the programs had to be compact and efficient! I used an assembly language called PAL for most of it, although it had a FORTRAN compiler too.
When I was in school I had two huge dumb terminals that used core memory and hall effect switch keyboards. The fans in the things sounded like a hurricane. That was early '80's. I took the planar mem boards out and hung them from the ceiling. When people asked what it was, I told them memory-belia.
I've seen people talk about this type of memory for years, but you are the first to actually explain how it was read and worked in fairly simple terms. Awesome video!
Dear Mr. Ganssle, this was a very good vid for me, brought back memories, and explained much. I built my first analog resistive "computer" in 1956, I was so excited. For me there was no storage. Thanks again, pierre from New Mexico
I was a maintenance man on the NORAD SAGE Computer system (IBM ANFS/Q7). The largest (physical) computer ever built. When they finally shut down the SAGE systems in the 1980s I managed to snag a piece of the ferrite core memory, which I have displayed in my hime office to this day. As an aside, Hollywood bought some of the components of the old SAGE systems for use in TV and movies. I've seen the maintenance consols in shows like the Six Million Dollar Man, Battle Star Galactica and even one episode of Columbo. In the movie Austin Powers, there is a console that housed the long range radar input channels that can bee seen in the background in Dr Evil's hideout.
Ali Mirjamali That's for sure. My computer is only 7 years old and I get ragged about it. People ask me "So when are you going to get a modern computer?"
***** Probably when I wake up one morning and see that it's raining hundred dollar bills. I probably have enough spare computer parts in my garage to build one myself. It might look like a Frankenstein computer but as long as I can install Chrome on it I'll be in business. = )
And I thought I made out like a bandit when I managed to cadge a 8kb upgrade chip for my TRS-80 Model 100 out of the manufacturer as a "sample". Cost me nothing, and SAVED me $300 in 1985 money. I managed to write my own ISAM storage and retrieval method for the TI 99/4a. There WAS a disk drive unit available but it was far too expensive, so I stored data for a record-keeping program I wrote for my Dad. I saved the data as indexed-sequential records on the cassette tape the device used for program storage. It was supposed to be "impossible" to use the tape as a data-storage device on that machine, especially since I didn't have the "Expanded" BASIC cartridge, and had to make do with the more limited on-board BASIC. Years later, I worked for a company that sold it's own OEM IBM XT clones. I spent hours putting DIP RAM chips on the motherboards. Got to be very good at it, too. I even managed to cadge a few NEC V20 chips from NEC, so I could upgrade my own XT at home. The V20 was a little speedier than a stock Intel 8088 chip, and I could also run CP/M on it. I had a large backlog of CP/M programs from when I used to write some of early forms of shareware in Z80 Assembler.
Those were the good old days. I still remember my TRS 80. My first 'puter. I used to store my written code on cassette from my TRS80 as I wrote it. I try explaining to my kids that back then we did not have hard disk drives and I don't think they grasp the concept that 64k of RAM and a cassette on a computer was enough to do a lot of work. Thanks for the memories. Take care buddy.
Hey, 64K on a home machine was going big! The Atari 400/800 were only offered with 48K max with factory 16K cards. Then someone (Atari's founder's new company) began selling "Ramcram" cards for them, 32K on a single card. A friend hacked one into 2 selectable 16K drives to speed up loading his games!
Crazy days, but it gives you a warm fuzzy feeling too. I'm only 35, but I went to a 4 year technical course in high school before my information systems degree, and they had a massive wealth of historic artifacts and technical data that really gave me an appreciation for technology and it's roots. The teachers were phenominal, and even fired up a large tesla coil, and taught us the truth of Nikola Tesla.
For extra entertainment you should see how we used to do read-only memory for things like bootstrap loaders - the very first thing a computer runs when you turn it on, which today is lodged in a small piece of flash memory and is called the BIOS. Let's say you needed a 256-byte bootstrap loader. The manufacturer would send you a blank edge card with a grid of 256 x 8 pairs of solder pads. To program a 1 into a location you had to solder a jumper wire, or a 1N34 germanium diode, into its solder pad pair. Field-programmable read-only memory ICs were invented by the Air Force in 1956, but they were REALLY expensive, much worse than diode cards. We used diode cards with 1N914 silicon diodes before we moved wholesale to PROM ICs. (And you know you're old when you know the part number for a germanium diode off the top of your head.)
I had one of these kits as a kid in 1966 which started me off in electronics. It had AC126 and AF116 transistors and an OA79 diode. The transistors were expensive for me to replace if I broke a lead or burned one out. Ah the memories.... www.hansotten.com/uploads/philips/jlee20c.jpg
The 2732 EPROM was my first memory. I used the 74LS138 as address decoder ... and knew many other TTL chips number. On the more crazy part of my self introduction to microprocessor, I studied in detail the hardware of the video game miss pac man. The steps were: - creating the electronic schematic by tracing each connection between the about 80 TTL chips * 14 pins : about 1120 pins tested with ohm meter - read the EPROM content manually - write the byte value on note books, one byte per line - identify the Z80 opcode and optional data using the appendix of a programming manual which listed each instructions over 250 pages and a table sorted by numerical value in the appendix I made my first EPROM reader using a weller solder iron plastic box on which I installed 12 switch for selecting the address and 8 red led to display the 8 bit value stored in the EPROM. I was flipping the address switch in binary and was writing address and data in the notebook. A second plastic box was used to create a debugging emulator. It had one switch connected to the clock input of a Z80 processor and one 24 pin socket for a 2732 EPROM. It also had 16 led to show the address selected by the Z80 and 8 led to display the data line. Pressing the clock switch was triggering the execution of each instruction step by step. The Z80 was fully static, so the clock could be stopped for minutes or hours and still continue to the next operation. Eventually, a 555 oscillator was added with another switch to skip some repetitive loops then resume clocking manually.
I am a tech nerd and I love seeing and hearing about old tech. It amazing how complex memory used to be. We have come so far in such a short amount of time!
Thank you for sharing. It's great to see the pioneers alive and kicking. Before the turn of the millennium I didn't know the first thing about a computer. The world is indebted to the great minds of the leaders in the industry and your video is tribute to their legacy.
I used to program an IBM 360/30 that had, I think, 64KB of core memory, and progressed through a few mdels to a System 370/145 running VM (two concurrent operating systems.... OS and Dos27 or so). Our software had to be very efficient back then, which is why most of it was done in Assembler, although later on we used COBOL and RPG, and sometimes FORTRAN, then PL/1. Apart form I/O macros, Assembler was one machine instruction for each instruction coded.... nothing was wasted! Slow to code but it ran very fast for the speed of the CPU. Nowadays I have 64Gig SD cards for under $50, while back then a Gigabyte was science fiction... Progress indeed! I still have my old programmer reference cards and flow-charting templates somewhere!... :-)
sounds lovely. today the developers get more and more lazier because theres more processing power available they just are too lazy to optimize. i'd love to see the kind of interest and effort everyone had to program as efficiently as possible back then. the best example i can think of is NASA, oh god, their code has to be flawless and as perfect as possible
most of the moon missions were running with error code 1102? (computer overloaded) they couldnt find someone who knew what it meant or why it was happening, of course all this occured during the flight so no tests ? You would hope they got better. Nasa is far from perfect they launched discovery knowing it had a 5% chance of failing (ie 1 in 12 will kill everyone on board in a giant ball of white death) but regan wanted something to talk about during state of the nation, so they launched with frozen o rings :( turned out he got to talk about brave school teachers touching the face of god .. ugh
Andrew P It's one thing to say the Challenger disaster was from a pervasive cavalier attitude that allowed operation outside the normal flight parameters under pressure from the Reagan administration to make shortcuts to save money. It's another, sheepish, cowardly thing to say it was murder so he could recite a poem from 1941. A lot of shameful things happened under the Reagan administration, but that's some real tin-foil-hat shit right there. Grow up.
Andrew, two Space Shuttles were lost during the program. Challenger was the one that blew up with the frozen O-rings. Columbia broke up over Texas. Discovery survived the program and is now in the Smithsonian.
exactly, many of the spaceships failures are mechanical. of course there are some software bugs, but the software developers at nasa are famous for making the most reliable software jmowreader
My uncle worked for Sperry Univac some time in the past. He died some time back and I inherited an old airstream from him. I Have recently been cleaning it up and found two ferrite core memory cards, plus a NOS Sperry univac badge. I got a shadow box and mounted them with the badge, it looks great on black felt and something I'm very proud of. My Uncle Bob was a cool guy and I really like having my own memory. Although mine is 256 bits per card. It still keeps me in awe of computers. Since I'm part of a makerspace we have it proudly hung on the wall to show others and inform them off how computers used to be as well.
Love your videos, just watched tons of them in a row.. Oddly enough your voice makes me feel really at ease in a way! Listening to your videos in the background while studying :)!
robuk1981 I wonder how a simple digital Casio watch would stack up against early 50s computers. What would the modern analogue be, a digital timed washing machine?
the biggest computer I've ever seen was about the size of a shed, it's still used to control the railroads, the next time you pass some railroads and see a small house made of steel near the tracks, there's a huge computer in there and the walls are stacked and stacked with batteries and live electrical wires. It's a pain in the ass to set up those computers and make sure every little feature works
Well, even the early computers had way more input/output bandwidth than a $5 watch. The problem with early non-transistorized computers were that they weren't very reliable. But anyway, back then it was about computers handling large amounts of information through batch processing. That was where their power was at, not so much at complex calculations. In fact, analog computers were better for a lot of things up until the mid 60s.
I visited Kennedy Space Center back in 2011 for the final shuttle launch STS-135. Back then we still had iPhone 3GS and iPhone 4 phones. The replica display of Mission Control for the Apollo moon missions said that modern smart phones (ca. 2011) had more power than the entire Mission Control center that put humans on the moon.
Cool stuff, I was born in 79. my dad was into computers and I can remember playing many many games on his C64, Which became mine:) wasteland, hunt for red October, gunship, stealth fighter, summer games, winter games, comando, rambo, raid on bungling bay...omg wayyyy to many to list. if you had a C64 you know how it was.. I was instantly hooked
While in the Navy, I worked on a computer that was literally an IBM 360 in a compact military package. It had 32K Word (16 bit) core memory. It was used in an attack aircraft. That computer was amazing because you could power it down and when you fired it up again, it would pick up exactly where it was when it was shutdown. No loading OS, no feathering it’s nest, just ran what it was doing when shutdown. It was also used in SkyLab. And 3 of them were used in the Instrumentation Unit on the Saturn V rocket. Yeah. I’m that old.
I ran across this old digital computer patent 3190554 that use compressed air instead of electricity. It even had it's memory stored in compressed air. ( I think) Did you ever run across such a computer if so what was it used for? Think one could be built today using 3D printing using some Nano Tech ? If Babbage had gone this route do you think that his Difference Engine could have been built by the folks who made pipe organs?
Surprised the hell out of me and I don't know squat about computers. I didn't read the patent to close but I think it had air memory. Let me if try to build one using 3D printing.
+ufoengines Player pianos used air blown through the holes in the scroll of paper to operate. I kind of doubt the people who made pipe organs made the Difference Engine. Large buildings use(d) compressed air to run their climate control systems. Electropneumatic controls were and still are quite popular, and they are still used in lots of cars to open and close valves & dampers on their climate control system. They run off the engine vacuum. Air can be under or over atmospheric pressure, so why not?>"
+ufoengines I'm imaging a seriously steampunk computer, sort of a cross between a computer and a pipe organ. (Calling Captain Nemo?) In all seriousness a pneumatic computer has some interesting possibilities, it would be virtually immune to radiation of any kind and could be powered by a tire pump.
I have this idea that if Babbage had run across this concept that he could have had the pipe organ guys build his his Difference Engine for him and Lady Ada would have invented COBOL. Of course if Da Vinci had invented the phonograph.instead of Edison would have had five hundred years of recorded music. Dig that!
great video, beings back memories. I became a programmer in 1963 on a Univac SS 80 with 3,200 words of memory. This memory was stored on a cylindrical magnetic drum that was spinning at about 18,000 RPM. magnetic read write heads read the memory contents as the drum rotated. Unlike discs these heads were static, they did not move, YOU had to wait for the required memory locations to pass beneath the R/W Heads to access the contents. 20% of the drum has 4 sets of heads at 90 degrees to each other so that you only had to wait for a quarter of a rotation to access the memory. When writing programs you had to keep track of where the heads were in relation to the memory locations you needed to access, if you did not do this the program ran very slowly. Each WORD of memory contained 10 4 bit characters, these bits had the binary values 1 2 4 5, yes 5, it was known as BIQUINARY. Programming these machines was a nightmare BUT GREAT FUN.
Hi, Jack! I think I used to visit that same store on Commonwealth Ave in Boston back in 1976! I bought my copy of BYTE magazine issue #1 there! They used to have a bunch of surplus PDP-8's and 11's and parts, since DEC was close by. I was fascinated by the place, and went every weekend I could get free from Ft Devens (Army). I bought my first computer in 1979, a Mattel (can't remember the name) that had 4k Static RAM, and had BASIC built in. I bought an 8k SRAM and made my own module to plug into iit, and programmed the game of LIFE on it. FUN! Back in 1974, in high school, we did an extension class at U of Montana where the main computer was a DECSystem10 with 56K (36-bit) words of core mem. We were allocated 100K disk space while logged in, so I bought a 10" magtape and stored all my programs on it - Fortran, Algol and Macro-10. It was a wonderful time!
Oh indeed we will. Memristor-based storage will (likely) change everything. Solid state non-volatile storage will be as fast as DRAM, which will remove the need to use volatile memory anything except on the CPU die itself. And even that might go away, since memristors can be used to build computing logic as well as memory. I'm not sure if there are speed reasons not to, but we could have just a giant array of memristors which dynamically change from providing computation to memory. Many people are excited about using memristors to build artificial neurons and artificial brains, but I'm far more interested in how they might radically alter the entire landscape of computers. There is every reason they might be more successful and ubiquitous than the transistor. (The transistor isn't even a fundamental element of circuits, which memristors are. Resistors, capacitors, inductors, and memristors are the fundamentals, we just didn't know how to build memristors until recently.)
Hi Jack and thanks for posting the information on magnetic cores. I was in USAF during the Vietnam War and in SAC. We had a system called the 465L that used magnetic cores. The system was quite a bit like the modern internet in that it allowed SAC to communicate with all of its bases through digital communications. It used modified IBM selectric typewriters as both keyboard input and real time printers. The system we worked with was my first exposure to magnetic cores, though ours was much smaller than the one you showed in the video. Ours was only 90 words of memory and served as a buffer for the data as it was sent out. We also had a device used for encryption called a KW26. This device was extremely old and used tubes and what were called bi mag cores. The bi mag cores were used to store information as it was encrypted and decrypted. Each core was a single toroid that was mounted with others on a board. the cores themselves were about an inch square and about half an inch high. Lots of fun way back then. Thanks for the memories.
Why did I watch this? My eyes just crossed and I drooled a little bit. My ears are ringing and I can feel my brain shrinking!!!! You puter geeks are the cutting edge, no doubt about it.......
Wish I'd saved some of those on the systems I worked on. The magnetic drum memories were quite interesting, too. Ahhh the good old days. Thanks for the trip down memory lane, literally... My first time seeing your channel.
Tim Barton You misunderstood me. I'm not a fucking idiot lol I know there were computers. I meant that I wonder if the technology back then would interest me enough to become a programmer as the technology today did. To much extent I feel that there is little room for new things these days. Most of what I can tell almost all products today are just better versions of what was already out there, and that sucks. And the chances of me starting a pc company in my garage and being a billion dollar business is about 0% possible. However I do think there are new things that will show up, but not at a small business level. One thing I would love to see in my lifetime is quantum computers personally. I don't know much about quantum stuff, but if I am correct it would offer much faster non computational instructions compared to normal ram and cpu's.
I started out in electronics in the late 1960s. My interest in electronics arose purely out of curiosity. I had a really cheap electric guitar and amplifier. I lost interest in learning to play the guitar, but I wondered how the amplifier worked, so I took it apart. I was hooked. One of the first projects I built was a 4-bit binary counter using flatpack J-K flip-flops. I was pretty proud of it, so I entered it in the science fair in sixth grade. Nothing. I kept on plugging anyway. Eventually got an EE degree, worked in minicomputers, hard disk media, telecommunications, and aerospace before going for my MSCS. Now I build web-based applications managing big data.
In DRAM, not only does the charge on the capacitors have to be refreshed every few milliseconds, but it also, like core, has to be rewritten every time it is read. That happens automatically inside the chip, so we normally don't think about it. Ferroelectric RAM (FRAM), a nonvolatile memory technology, also has this hidden rewrite on read, which is why it has limited read endurance, unlike EEPROMs and Flash, which generally only have a limit on write endurance.
Thanks for posting this. The first memory I ever worked on was a core like the one you show. Cards full of hand stitched ferrite beads exactly as you describe them. I can still remember being taught about the hysteresis loop created by the write wires causing teh ferrite to hold a magnetic '1' or '0'. Strange, but actually understanding computer memory and logic from these sort of hard wired principals has enabled me to adapt to many other aspects of electronics throughout my working life. Strange to think that just maybe I belong in a museum alongside some of the exhibits now.
I did not realize just how far computers have come so quickly. that's crazy. to think at one point memory was handled with a fluid bed of mercury and some springs, that just blows my mind. now we've got laptops with 1tb+ for $500
I still have the core memory I salvaged in the early 1970s that I intended to use for a DIY computer while I was in college. I think I got it free as a donation from Xerox after making a few calls. Ended up first building one from an 8008 chip that I had trouble getting working, and ultimately, my first working computer was an IMSAI 8080 kit, which I then utilized for years-- still have that and it still works. But you forgot to mention drum memory. I also had a vacuum tube computer for a while, the LGP-30 that had 4K words of 31-bit drum memory-- bit 32 was a "spacer" bit on the drum and couldn't be written. And even the CPU's accumulator was stored on the drum, they saved multiple copies there in order to increase the access time. The drum also had a clock track which produced the master clock for the system-- it was bit-serial so it would clock data on and off of the drum as it ran.
this is the most fasinating video ive watched all day my grandfather always tells be about when he would work on the computers with hole cards and it just amazes me what we have evolved into technology wise
Many years ago I worked for British Rail. As we had very little money we had two old second-hand IBM 370-155 machines on which we ran the reservations system (one machine as backup). One day a drive transistor on one plane of the core memory failed. Our IBM engineer loved these machines, I think they reminded him of his youth! It turned out that IBM no longer stocked the part, at least not in the UK. So, he looked up the transistor's characteristics, found an equivalent in the Tandy (UK version of Radio Shack) catalogue, popped out to the local store, came back and soldered it in. I don't know if his expenses claim was accepted though...
Hey good post! Thank you for a walk down memory lane (pardon the pun), my first job back in the day was with Dataram Corp. plant (Trini Data owned by Mr. Michael Armoogam) here in Trinidad. Later on, I moved to United Telecontrol Electronics Inc. (Qualitek Electronics Plant, again owned by Mr. Michael Armoogam and Dr. Winston Akong) where we manufactured those Ferrite Core Memories. I went in as a trainee test technician and quickly moved up the ladder to Senior Test Technician / Assist. Plant Manager. I was very fortunate to be mentored by such engineering greats a Dr. Winston Akong and Dr. Art Ericksen. So yes, Trinidad and Tobago does have some history in the manufacture of "State of the Art" electronics. I believe Dataram Corp. was headquartered in NJ then. BTW our biggest memory back then was a folded, 128K stack and the "holy grail" of oscilloscopes was the Tektronix 500 series (my lab had two 545A's) and Nixie Tubes were the common digital displays, before moving on to the HP 7-segment LED displays (the segments were dots, don't recall if it was indeed a dot matrix device ....I doubt it).
I'd imagine that the material would change to something less temporal, if you go on Ebay for one of these you see alot of the connections severed. My idea, Crystal capacitors!
I was in the US NAVY 1963-1973. Yes, we used core memory in our SOSUS systems. As a computer technician, I can assure you they were a bugger to tweak for proper operation.
I loved the fact that you could halt a computer with core memory and start it up later at the exact point it left off (if you made a note of the program counter register). The old PDP 11/70s were great in some ways!
I wish people would not use the SI value for K (1000), but instead use the one that's always been used in the industry of 1024. It seems those hard drive manufacturers in the early 2000's looking to make thier disks look bigger turned the tide. Even though all other parts of computer architecture still uses 2^2 ie, 1024 systems. In fact, even filesystems used on those disks still use 2^n.
my god that is so interesting (the bit about old memory with the springs and mercury) its my opinion that its valuable for us to remember how things were originally done if we ever need to start back over.
Great video. If I remember correctly the first computer I worked with, a Univac 418 which had a (literal) core of 16 kbit. It was manually booted loading binary words into the main registers to call up tiny program that read a paper tape. The paper tape contained a program that loaded the operating system. The “large” memory was a pair of Fastran magnetic drum memories that were half the size of an automobile and could hold a couple of Mbytes. We definitely have made advances - as I type on my IPad Pro with 128 GB.
Came across this by accident. Had to watch. My Dad worked on the first computer at Rice University. They had the second computer for NASA in the late 50's. Was amazing to hear what that computer did, how it was built and then see what we have today. We had boxes of IBM punch cards that became Christmas wreaths. Might still have a few of those cards somewhere, going to have to look next time I go to Moms..
As a professional programmer that was taught assembly in my computer science degree the lack of memory forced better programming practices than are the norm today. Nowadays languages with garbage collectors mean most programmers that never learned assembly or even C know nothing about memory management and consequently code as if memory is infinite. For a single user system that might be just about viable but if you need to scale all the old memory limitation problems are still there. Great video.
Random fun fact: Those units were handcrafted, and usually by textile worker women. Here in Hungary from where i write this comment, such "ferrite-ring" memories were manufactured in the '50s. Countrywoman who otherwise made sophisticated lace overlays and ornaments at home were gladly hired for this job, for they had the patience and cold hand to produce such matrices of small pearls. This way about 1bit/square mm of data density could be achieved, which was a huge leap after the Williams-Kilburn tubes. The manufacturing process was never automated here, for by the time the technology had become available for that, the advent of semiconductors made the whole stuff obsolete.
I was the last person at Iowa State University (~1969) to use the core memory out of its old Cyclone Computer. The memory was in a console that was about three feet wide and four feet high and about 12 feet long. It had four stacks of 4 K, 40 bit words. When I used the memory it was used as a buffer memory for an A/D - D/A converter. I remember loading punched paper tape and magnetic tapes. I was told that the Cyclone Computer had gone through mercury memory and the CRT memory years back.
Not sure why you're thankful it's gone as from what I read core memory was pretty reliable comparable to semiconductor memory. The Apollo AGC and 4PI had it and where so reliable they were used in flyby wire systems and some old mainframes that had it were it were still in use until the early 1990s. Now mercury delay lines and Williams tubes on the other hand were known for being fickle and buggy to the point of frustration.
jakejakeboom Of course not. But it was as a huge leap over previous memory technologies. One big disadvantage of core it was not only bulky but was very labor intensive to manufacture. When the 1103 1K DRAM came out in 1971 it was a significant an improvement in cost and packaging and quickly began to replace core memory in many applications.
Only one I see wasting comment space is you. The others are at least having a civilized discussion about the "shit old tech" that eventually evolved into the modern shit that you value so much. Pull the stick out of your ass please.
my first exposure to those memory cores was with the Control Data 160A mainframe. That was already a dinosaur, but great for learning the basics of computers back in the 70s even. Later, even into the late 90's, the same basic memories were at the heart of Thomson CSF surveillance radar systems, used in the SDGS (synthetic data generation system) for creation and display of vector based characters on Secondary Radar displays. Really miss those times now... The technology at 2:20 was another that was used on the same radar. The device was called a Scan Converter Tube, for converting standard radar PPI circular sweep, into a TV styled raster scan. In basic description, there was a TV tube, being faced and bonded to a TV Camera tube, which actually read the charge left by the source tube upon the interfacing screen at the junction of the tubes, but there were no visible 'dots' in that case
In 1960 working as an assembler at IBM San Jose, I happened to see another assembler slip during assembly of a core memory for the IBM 360 and shove the screwdriver through the entire assemble of 12 planes of cores. At the time the assembly was worth about $2,500 1960 dollars and he literally cried when it happened. I worked doing wire wrap panels for the 360. Occasionally I would be loaned to other groups working on RAMAC, Stretch 7090 and other systems. Started at $86 a week. Good start for high school grad and I worked in electronics for 45 years as technician, engineer and manager before retiring.
Rope memory? Related to core, it could only read. Think of it as early ROM. The flight control programs for Apollo were stored on woven ropes. The system used AC in the interrogation and sense leads, which prevented having to re-write the bits on read. The ropes were literally hand-laced to program the bit-patterns.
I remember looking at a computerized phototypesetting machine around 1975 of so. It was state of the art and I was really impressed. The big problem was that it used something called "bubble memory", and the instant the machine was turned off (or lost power), the memory would vanish. Around 5 years later, a computerized typesetter had what I think was a small ROM memory (something like 650 bytes of memory) that it used in starting up, and then it relied solely on one-sided 4.5" floppy disks for operation. It was a lot of trouble to operate with memory like that, and there was a 5 megabyte hard drive available, but the hard drive cost US$10,000, so I never got it.
Reminds me of coming across a cabinet of core memory in the basement of the HPER building at IU Bloomington, which housed the Wrubel Computing Center in one wing - must have been in the summer of 1980 or 1981. Taking a FORTRAN programming class, and we were still using punch cards. There were all sorts of fascinating machines down there from my point of view - HUGE punch card duplicators and printers, for instance. Big line printers, and a CDC6600 (which we learned programming on). Only five 300 baud terminals. A lot of this stuff would be gone in the next 3 years. The memory was in a cabinet, and this had toroids that were maybe ten times the size of yours. Big, visible wires running through the thing. It was on it's way out to the recyclers, I'm sure - it was in the middle of the hallway in the non-computing side of the building. Wish I'd at least had taken some photos.
Very interesting. I remember core memory but at that time we were moving to solid state TTL memory. I actually worked on a device that had RTL logic! A very nice trip down memory lane. Thanks.
I remember back in 1991, when I had a computer with an 80286 processor, running at a blazing 16MHz, and it had a 40 Meg hard drive. It came with 1 Meg of memory and I wanted to add an extra Meg. One Meg of memory cost $100 at the time. I had a friend who was doing computer repair and he sold me two 512K DIMMS for $50. I thought that I had gotten the deal of the century. Now, we can purchase 8 Gigabyte thumb drives for about $6.00 In fact, a lot of vendors at trade shows hand out 8, 16 and 32Gig flash drives for free. Also, in the late 90’s, I traded a used 1.2 Gigabyte hard drive for a brand new, good quality, 8 man camping tent and we both thought that each of us a got a great deal. It’s amazing when we think about how far we’ve come.
Fascinating video. My big story I tell my younger friends is that I recall the day (circa 1983) I upgraded my Atari 800 home computer (before there were true IBM-clones, aka, PCs) from 48K to 64K. That's right - I added an additional 16K or 16,000 bytes of RAM to this unit with this huge 11/4" x 10" x 5" slot-shaped plastic card. It was heavy, bulky, and the chips were not exposed to the eye. I often wondered how stupid this was as there was zero ventilation for these circuits should they run hot.
I used to work on core memories when I first started work in 1980. Testing used massive valve and relay based machines but to change a core took skilled operators using binocular microscopes and ultra thin needles.
+Andy Reid I remember working, at Boeing, with one of the techs who manufactured core memory at Burroughs in Downingtown, PA. He introduced me to the saying that "a good company will keep a tech until they begin to drool. I now know why!
Thanks. I have only seen pictures of core memory so to have it explained is great. I also just saw your interview with Dave (That crazy Aussie bloke!). Really interesting. :)
Amazing. I recently began messing around with tandy color computer 2 with 64k memory. My brother got one in 1984 & I was always blown away by it and it's programming guide even though I was only 6 and far too young to know what the hell I was looking at. I guess I had inner geek then because even though I was clueless about it I knew I liked it.
back in early 1980s , stationed at mtn home afb our repair shop ran test equipment using burroughs d-84 computers with core memory, i believe these computers were from early 1960s, 2 large ampex tape drives is where the test programs were located, very stable and ran the shop testing just fine. quite big though, great to be reminded of them, thanks j
Mr. Ganssle: Overall, a very interesting video; but my attention was suddenly riveted when you put up pictures of: ≈ 2:25 to 2:47 - the Williams tube When I was old enough to comprehend some things about computers, my father described what it was like to work on one of the very first of them. He was back and forth between MIT (where he was doing his doctorate) and Princeton (where J. von Neumann's electronic baby was), and I didn't catch which location this was, but he talked about having to deal with computer memory that was on an o-scope screen. And here's the crazy part - when some of the memory spots on the screen would become faulty, programs for that machine had to be written to specifically *avoid* those memory addresses! Anyway, thanks for connecting me with images (as well as the name!) of what my dad must have been dealing with back in the early 50's! PS. I still have his copy of the owner's manual for the IBM 704, which, some years after his doctorate, was the machine in use at his work.
Really like your videos. I love microcontrollers and computers, because they bring so much history to the human race. Can't wait until retirement tire to have time to play with all this stuff.
This video reminds me how I once said the internet would never quite replace TV. This at a time where streaming was typically .RM files, so not even all that long ago, just long enough ago that most of those web pages are gone.
It's crazy to think that we still use physical structures to form computer memory, but we managed to get the density millions of times higher over the last few decades.
The magnetic cores depend on a couple of phenomena for operation. The most important is hysteresis, a resistance to changing magnetic field until a threshold is reached. In the case of a typical core memory, you need about 300 milliamps through the core (ferrite bead, core, donut) to switch its magnetic field. Because of transformer action, this abrupt switch of magnetic field will induce a tiny current in a "sense wire". There's typically only one sense wire woven diagonally back and forth through the entire plane of cores. Each core has three wires: X wires, Y wires and a sense wire. So what you do is shoot 200 milliamps down ONE of the wires in the X direction, 200 milliamps in ONE of the Y direction wires. All of the ferrite beads along these wires ignores the 200 milliamp current BUT where these wires come together and cross inside the hole of a ferrite bead, now you have 400 milliamps and that's enough to switch the magnetism unless of course it was already magnetized in that direction, in which case nothing happens. So you repeat this process for each of the intersections and watch for a pulse on the sense wire. If there was a pulse, you just erased a "1" (or a zero; you can also use negative logic). Before you continue you'd better put those "1" back in because the act of reading the memory sets everything to zero (or 1). The speed of these things was measured in microseconds rather than nanoseconds. They are non-volatile; I've done some tricky things with core memory used in aircraft, solved a malfunctioning boot loader by using a good computer to load the program, then physically remove the core memory and install it in the aircraft and away they go with a functioning program already in memory. They were VERY expensive; it is a great souvenir.
Dr An Wang, who held the patent on core memory, and sold the patent to IBM always claimed they paid another inventor to challenge the patent during his negotiations for the sale. Although the other inventor had a weak case, it created doubt and lowered the sale price quite a bit. Dr Wang continued to invent things, and built Wang Computers into a world power. The company disappeared in the 90's after his death - along with a bunch of other computer companies like DEC. I was a computer tech in the 70's/80's and worked on some older Wang Computers and programmable calculators that used core memory. IIRC, these were the Wang 600 and 700 series, and they were pretty simplistic, but very programmable. Storage was on paper cards and cassettes. Engineers could make them do some pretty amazing things by switching cards in and out. Ah, the good old days.
>getting arrested while hitchhiking to buy computer memory
It doesn't get much more cyberpunk than that, I gotta say.
I'd give a nut to live in those days.. for a couple days.
Ahh the days of making VCR backups of your computer overnight.
lol right
perfect plot for a video game
in the year 2157 memory is outlawed. all random access info has to be stored on hard disk
He forgot to mention that there are still at least two computers still using core memory - the Voyager probes! Still going strong (ish) after 40 years! Now that's reliability!
seriously? this is amazing!
Aslong as they're not using Windows CE...
It's hard turning a launched space probe off and on again!
Send a reboot command and pray.
Well, not to ruin the joke... but core memory was actually cool for that reason specifically. You could turn it off and on again, and lose no information. In fact, you can turn it off for hundreds of years, and not lose any information! It's one of the reasons people still bought core even after CMOS memory was available, and even when it was cheaper. No load time! Just turn it on and plug away. I would love to get a set of core memory for my PDP-11/03.
The irony is that with windows crashes, core memory would be a disaster! When you hit the reset line on the bus, it'll jump right back into the crash unless you completely reload it from paper tape! You're not doing that in space...
Meanwhile in Kubernetes land...
I remember core store lectures (because we might come across old mainframes during our careers as electronics engineers) at college in 1983. Advantages (or why the millitary had a liking for them even duringthe 1970's and 1980's) was that (A) totally EMP resistant, (B) core memory will hold the "bit" stored without power for just over 4 centuries and (C) not as fragile or sensitive as the (then) new static and dynamic RAM devices. I designed and built a 4 bit computer using TTL static RAM, TTL ROM, 4 x TTL shift registers, TTL numerical LED drivers and a 74181 ALU for a weekend homework project. I still have it and it still works.
I was flying navy patrol planes in the late 80's into early 90's that had 64k iron core memory with 32 bit parallel word processing.
Hasn't the US Millitary have some ancient super computer buried inside a mountain from back in the 1980's?? That was using core stor memory and there was a documentary on BBC TV back in the 1980's about it. Kind of rapidly superceded by events and technology. Core Stores did have a longer existance in the computing world than DTL logic devices. I know I'm growing old when I can remember old obstelete technology and how it functioned.
There is no way to subscribe to this discussion so I'm forced to leave a message I can get back to (UA-cam might decide to scramble comment order, who knows..). Sorry, keep talking.
P-3? I worked P-3C out of Diego Garcia AMID W/C 610 the VQ-3 EC-130Q TACAMO. I'm trying to remember the manual programmer we used. Collins or Rockwell 7404-1a or 7401-a or something like that 32 toggles in 8 groups of 4 and 32 lights above them with a send/encode button. I was rather good at programming in binary.
***** Real bootstrapping. :)
I worked with a core memory based computer. The core had to be in a temperature-controlled chamber to work right. I once had to replace the heating elements. This was in the late 1970's, and that computer was well into obsolescence at that time. Oddly, it had a 24-bit word length. As I remember, the memory was only 2048 K, and programs had to be partially run and swapped back and forth from core to the memory drum unit. The drum unit was the size of a washing machine and held 256 K words. It had to be pressurized with helium. The computer controlled a steel rolling mill and was made by General Electric. And yes, the programs had to be compact and efficient! I used an assembly language called PAL for most of it, although it had a FORTRAN compiler too.
When I was in school I had two huge dumb terminals that used core memory and hall effect switch keyboards. The fans in the things sounded like a hurricane. That was early '80's. I took the planar mem boards out and hung them from the ceiling. When people asked what it was, I told them memory-belia.
Should have taken the hall effect keyboards too. Worth decent money to collectors these days.
rafflesmaos I know. I wish I would have kept them. They were bulletproof.
Thanks for sharing your memory.
Badumpum.
i heard that music in my head
I downloaded a drumming app for my phone just so I could make a dumdumtsss sound when I made a pun. Used it twice so far. Totally worth it.
A lot of labor went into making that 50k bit module. Pretty neat. Thanks for sharing.
I've seen people talk about this type of memory for years, but you are the first to actually explain how it was read and worked in fairly simple terms. Awesome video!
Thanks for the walk down Memory Lane....
This is pure gold in knowledge, I need to watch it again to fully appreciate it. Thank you!
How much did you pay for your core memory, back in the day, and did you ever implement it?
Dear Mr. Ganssle, this was a very good vid for me, brought back memories, and explained much. I built my first analog resistive "computer" in 1956, I was so excited. For me there was no storage.
Thanks again, pierre from New Mexico
I was a maintenance man on the NORAD SAGE Computer system (IBM ANFS/Q7). The largest (physical) computer ever built. When they finally shut down the SAGE systems in the 1980s I managed to snag a piece of the ferrite core memory, which I have displayed in my hime office to this day. As an aside, Hollywood bought some of the components of the old SAGE systems for use in TV and movies. I've seen the maintenance consols in shows like the Six Million Dollar Man, Battle Star Galactica and even one episode of Columbo. In the movie Austin Powers, there is a console that housed the long range radar input channels that can bee seen in the background in Dr Evil's hideout.
Great history lesson. Makes you wonder how archaic today's technologies are going to seem when we look back 40 years from now! (Greetings from NJ).
Or maybe 4 years from now.
Ali Mirjamali That's for sure. My computer is only 7 years old and I get ragged about it. People ask me "So when are you going to get a modern computer?"
ldchappell1 So, when are you going to get a modern computer?
***** Probably when I wake up one morning and see that it's raining hundred dollar bills. I probably have enough spare computer parts in my garage to build one myself. It might look like a Frankenstein computer but as long as I can install Chrome on it I'll be in business. = )
And I thought I made out like a bandit when I managed to cadge a 8kb upgrade chip for my TRS-80 Model 100 out of the manufacturer as a "sample". Cost me nothing, and SAVED me $300 in 1985 money.
I managed to write my own ISAM storage and retrieval method for the TI 99/4a. There WAS a disk drive unit available but it was far too expensive, so I stored data for a record-keeping program I wrote for my Dad. I saved the data as indexed-sequential records on the cassette tape the device used for program storage. It was supposed to be "impossible" to use the tape as a data-storage device on that machine, especially since I didn't have the "Expanded" BASIC cartridge, and had to make do with the more limited on-board BASIC.
Years later, I worked for a company that sold it's own OEM IBM XT clones. I spent hours putting DIP RAM chips on the motherboards. Got to be very good at it, too. I even managed to cadge a few NEC V20 chips from NEC, so I could upgrade my own XT at home. The V20 was a little speedier than a stock Intel 8088 chip, and I could also run CP/M on it. I had a large backlog of CP/M programs from when I used to write some of early forms of shareware in Z80 Assembler.
I burnt plastic army men and insects with a chipped magnifying glass.
It's also fun to use a propane torch for that, pretend they got napalmed
Derek Jeter mnj
Those were the good old days.
I still remember my TRS 80. My first 'puter.
I used to store my written code on cassette from my TRS80 as I wrote it.
I try explaining to my kids that back then we did not have hard disk drives and I don't think they grasp the concept that 64k of RAM and a cassette on a computer was enough to do a lot of work.
Thanks for the memories.
Take care buddy.
Hey, 64K on a home machine was going big! The Atari 400/800 were only offered with 48K max with factory 16K cards. Then someone (Atari's founder's new company) began selling "Ramcram" cards for them, 32K on a single card. A friend hacked one into 2 selectable 16K drives to speed up loading his games!
Crazy days, but it gives you a warm fuzzy feeling too. I'm only 35, but I went to a 4 year technical course in high school before my information systems degree, and they had a massive wealth of historic artifacts and technical data that really gave me an appreciation for technology and it's roots. The teachers were phenominal, and even fired up a large tesla coil, and taught us the truth of Nikola Tesla.
For extra entertainment you should see how we used to do read-only memory for things like bootstrap loaders - the very first thing a computer runs when you turn it on, which today is lodged in a small piece of flash memory and is called the BIOS. Let's say you needed a 256-byte bootstrap loader. The manufacturer would send you a blank edge card with a grid of 256 x 8 pairs of solder pads. To program a 1 into a location you had to solder a jumper wire, or a 1N34 germanium diode, into its solder pad pair. Field-programmable read-only memory ICs were invented by the Air Force in 1956, but they were REALLY expensive, much worse than diode cards. We used diode cards with 1N914 silicon diodes before we moved wholesale to PROM ICs. (And you know you're old when you know the part number for a germanium diode off the top of your head.)
I had one of these kits as a kid in 1966 which started me off in electronics. It had AC126 and AF116 transistors and an OA79 diode. The transistors were expensive for me to replace if I broke a lead or burned one out. Ah the memories....
www.hansotten.com/uploads/philips/jlee20c.jpg
The 2732 EPROM was my first memory. I used the 74LS138 as address decoder ... and knew many other TTL chips number.
On the more crazy part of my self introduction to microprocessor, I studied in detail the hardware of the video game miss pac man. The steps were:
- creating the electronic schematic by tracing each connection between the about 80 TTL chips * 14 pins : about 1120 pins tested with ohm meter
- read the EPROM content manually
- write the byte value on note books, one byte per line
- identify the Z80 opcode and optional data using the appendix of a programming manual which listed each instructions over 250 pages and a table sorted by numerical value in the appendix
I made my first EPROM reader using a weller solder iron plastic box on which I installed 12 switch for selecting the address and 8 red led to display the 8 bit value stored in the EPROM. I was flipping the address switch in binary and was writing address and data in the notebook.
A second plastic box was used to create a debugging emulator. It had one switch connected to the clock input of a Z80 processor and one 24 pin socket for a 2732 EPROM. It also had 16 led to show the address selected by the Z80 and 8 led to display the data line.
Pressing the clock switch was triggering the execution of each instruction step by step. The Z80 was fully static, so the clock could be stopped for minutes or hours and still continue to the next operation. Eventually, a 555 oscillator was added with another switch to skip some repetitive loops then resume clocking manually.
I am a tech nerd and I love seeing and hearing about old tech. It amazing how complex memory used to be. We have come so far in such a short amount of time!
Thank you for sharing. It's great to see the pioneers alive and kicking. Before the turn of the millennium I didn't know the first thing about a computer. The world is indebted to the great minds of the leaders in the industry and your video is tribute to their legacy.
I used to program an IBM 360/30 that had, I think, 64KB of core memory, and progressed through a few mdels to a System 370/145 running VM (two concurrent operating systems.... OS and Dos27 or so).
Our software had to be very efficient back then, which is why most of it was done in Assembler, although later on we used COBOL and RPG, and sometimes FORTRAN, then PL/1.
Apart form I/O macros, Assembler was one machine instruction for each instruction coded.... nothing was wasted! Slow to code but it ran very fast for the speed of the CPU.
Nowadays I have 64Gig SD cards for under $50, while back then a Gigabyte was science fiction... Progress indeed!
I still have my old programmer reference cards and flow-charting templates somewhere!... :-)
sounds lovely.
today the developers get more and more lazier because theres more processing power available they just are too lazy to optimize. i'd love to see the kind of interest and effort everyone had to program as efficiently as possible back then. the best example i can think of is NASA, oh god, their code has to be flawless and as perfect as possible
most of the moon missions were running with error code 1102? (computer overloaded) they couldnt find someone who knew what it meant or why it was happening, of course all this occured during the flight so no tests ? You would hope they got better. Nasa is far from perfect they launched discovery knowing it had a 5% chance of failing (ie 1 in 12 will kill everyone on board in a giant ball of white death) but regan wanted something to talk about during state of the nation, so they launched with frozen o rings :( turned out he got to talk about brave school teachers touching the face of god .. ugh
Andrew P It's one thing to say the Challenger disaster was from a pervasive cavalier attitude that allowed operation outside the normal flight parameters under pressure from the Reagan administration to make shortcuts to save money. It's another, sheepish, cowardly thing to say it was murder so he could recite a poem from 1941. A lot of shameful things happened under the Reagan administration, but that's some real tin-foil-hat shit right there. Grow up.
Andrew, two Space Shuttles were lost during the program. Challenger was the one that blew up with the frozen O-rings. Columbia broke up over Texas. Discovery survived the program and is now in the Smithsonian.
exactly, many of the spaceships failures are mechanical. of course there are some software bugs, but the software developers at nasa are famous for making the most reliable software jmowreader
My uncle worked for Sperry Univac some time in the past. He died some time back and I inherited an old airstream from him. I Have recently been cleaning it up and found two ferrite core memory cards, plus a NOS Sperry univac badge. I got a shadow box and mounted them with the badge, it looks great on black felt and something I'm very proud of. My Uncle Bob was a cool guy and I really like having my own memory. Although mine is 256 bits per card. It still keeps me in awe of computers. Since I'm part of a makerspace we have it proudly hung on the wall to show others and inform them off how computers used to be as well.
Great to see Jack is still at it writing and making great videos about tech.
Love your videos, just watched tons of them in a row.. Oddly enough your voice makes me feel really at ease in a way! Listening to your videos in the background while studying :)!
Pretty scary to think just how much computing power you have in your phone these days when you watch vids like this.
robuk1981 I wonder how a simple digital Casio watch would stack up against early 50s computers. What would the modern analogue be, a digital timed washing machine?
the biggest computer I've ever seen was about the size of a shed, it's still used to control the railroads, the next time you pass some railroads and see a small house made of steel near the tracks, there's a huge computer in there and the walls are stacked and stacked with batteries and live electrical wires. It's a pain in the ass to set up those computers and make sure every little feature works
Well, even the early computers had way more input/output bandwidth than a $5 watch. The problem with early non-transistorized computers were that they weren't very reliable. But anyway, back then it was about computers handling large amounts of information through batch processing. That was where their power was at, not so much at complex calculations. In fact, analog computers were better for a lot of things up until the mid 60s.
I visited Kennedy Space Center back in 2011 for the final shuttle launch STS-135. Back then we still had iPhone 3GS and iPhone 4 phones. The replica display of Mission Control for the Apollo moon missions said that modern smart phones (ca. 2011) had more power than the entire Mission Control center that put humans on the moon.
Cool stuff, I was born in 79. my dad was into computers and I can remember playing many many games on his C64, Which became mine:) wasteland, hunt for red October, gunship, stealth fighter, summer games, winter games, comando, rambo, raid on bungling bay...omg wayyyy to many to list. if you had a C64 you know how it was.. I was instantly hooked
While in the Navy, I worked on a computer that was literally an IBM 360 in a compact military package. It had 32K Word (16 bit) core memory. It was used in an attack aircraft. That computer was amazing because you could power it down and when you fired it up again, it would pick up exactly where it was when it was shutdown. No loading OS, no feathering it’s nest, just ran what it was doing when shutdown.
It was also used in SkyLab. And 3 of them were used in the Instrumentation Unit on the Saturn V rocket. Yeah. I’m that old.
Thanks Jack, your post was as fascinating as it was nostalgic.
What a pleasant manor as well as informative, thank you.
What manner of manor might that be? Elizabethan? Georgian? Or a different era altogether?
You certainly know how to mind your Manors.
wow!yousureshowedhim!bethethinkstwicebeforenotdoingnotgoodgramarsagin,uhheh,sowstoooped.
Ha, fair play! To offer a small excuse, I do work in Architecture, so perhaps my auto-correct was merely used to my normal wordage :-)
I ran across this old digital computer patent 3190554 that use compressed air instead of electricity. It even had it's memory stored in compressed air. ( I think) Did you ever run across such a computer if so what was it used for? Think one could be built today using 3D printing using some Nano Tech ? If Babbage had gone this route do you think that his Difference Engine could have been built by the folks who made pipe organs?
+ufoengines That is an incredible patent!
Surprised the hell out of me and I don't know squat about computers. I didn't read the patent to close but I think it had air memory. Let me if try to build one using 3D printing.
+ufoengines Player pianos used air blown through the holes in the scroll of paper to operate. I kind of doubt the people who made pipe organs made the Difference Engine. Large buildings use(d) compressed air to run their climate control systems. Electropneumatic controls were and still are quite popular, and they are still used in lots of cars to open and close valves & dampers on their climate control system. They run off the engine vacuum. Air can be under or over atmospheric pressure, so why not?>"
+ufoengines I'm imaging a seriously steampunk computer, sort of a cross between a computer and a pipe organ. (Calling Captain Nemo?)
In all seriousness a pneumatic computer has some interesting possibilities, it would be virtually immune to radiation of any kind and could be powered by a tire pump.
I have this idea that if Babbage had run across this concept that he could have had the pipe organ guys build his his Difference Engine for him and Lady Ada would have invented COBOL. Of course if Da Vinci had invented the phonograph.instead of Edison would have had five hundred years of recorded music. Dig that!
great video, beings back memories.
I became a programmer in 1963 on a Univac SS 80 with 3,200 words of memory. This memory was stored on a cylindrical magnetic drum that was spinning at about 18,000 RPM. magnetic read write heads read the memory contents as the drum rotated. Unlike discs these heads were static, they did not move, YOU had to wait for the required memory locations to pass beneath the R/W Heads to access the contents.
20% of the drum has 4 sets of heads at 90 degrees to each other so that you only had to wait for a quarter of a rotation to access the memory.
When writing programs you had to keep track of where the heads were in relation to the memory locations you needed to access, if you did not do this the program ran very slowly.
Each WORD of memory contained 10 4 bit characters, these bits had the binary values 1 2 4 5, yes 5, it was known as BIQUINARY.
Programming these machines was a nightmare BUT GREAT FUN.
Didn't want this video to end... it's so fascinating, the evolution.
wow very formal information. thank you for the look back. its amazing how far we came.
Hi, Jack! I think I used to visit that same store on Commonwealth Ave in Boston back in 1976! I bought my copy of BYTE magazine issue #1 there! They used to have a bunch of surplus PDP-8's and 11's and parts, since DEC was close by. I was fascinated by the place, and went every weekend I could get free from Ft Devens (Army).
I bought my first computer in 1979, a Mattel (can't remember the name) that had 4k Static RAM, and had BASIC built in. I bought an 8k SRAM and made my own module to plug into iit, and programmed the game of LIFE on it. FUN!
Back in 1974, in high school, we did an extension class at U of Montana where the main computer was a DECSystem10 with 56K (36-bit) words of core mem. We were allocated 100K disk space while logged in, so I bought a 10" magtape and stored all my programs on it - Fortran, Algol and Macro-10. It was a wonderful time!
As usual, an excellent video, shame it was too short... OUT OF MEMORY!
you blew my mind. jesus, I had no idea how complex of a challenge it has been making these tiny arrays!
I wonder if we'll look back on these days as "bad old days" someday.
Oh indeed we will. Memristor-based storage will (likely) change everything. Solid state non-volatile storage will be as fast as DRAM, which will remove the need to use volatile memory anything except on the CPU die itself. And even that might go away, since memristors can be used to build computing logic as well as memory. I'm not sure if there are speed reasons not to, but we could have just a giant array of memristors which dynamically change from providing computation to memory. Many people are excited about using memristors to build artificial neurons and artificial brains, but I'm far more interested in how they might radically alter the entire landscape of computers. There is every reason they might be more successful and ubiquitous than the transistor. (The transistor isn't even a fundamental element of circuits, which memristors are. Resistors, capacitors, inductors, and memristors are the fundamentals, we just didn't know how to build memristors until recently.)
TFW the modern world still runs on tape drives that are as fast as solid state while not being volatile.
10 years ago were 'bad old days', just look at your cellphone..
I sure as hell hope so.
Nice EVE portrait ;)
Hi Jack and thanks for posting the information on magnetic cores. I was in USAF during the Vietnam War and in SAC. We had a system called the 465L that used magnetic cores. The system was quite a bit like the modern internet in that it allowed SAC to communicate with all of its bases through digital communications. It used modified IBM selectric typewriters as both keyboard input and real time printers. The system we worked with was my first exposure to magnetic cores, though ours was much smaller than the one you showed in the video. Ours was only 90 words of memory and served as a buffer for the data as it was sent out. We also had a device used for encryption called a KW26. This device was extremely old and used tubes and what were called bi mag cores. The bi mag cores were used to store information as it was encrypted and decrypted. Each core was a single toroid that was mounted with others on a board. the cores themselves were about an inch square and about half an inch high. Lots of fun way back then. Thanks for the memories.
Why did I watch this? My eyes just crossed and I drooled a little bit. My ears are ringing and I can feel my brain shrinking!!!! You puter geeks are the cutting edge, no doubt about it.......
Wish I'd saved some of those on the systems I worked on. The magnetic drum memories were quite interesting, too. Ahhh the good old days. Thanks for the trip down memory lane, literally... My first time seeing your channel.
Great video. I did not know about early memory. It's amazing how much it has changed.
I almost wish I was born in the 70's/80's. I just wonder if I still would have became a computer nerd or not.
gopro_2027 lol. Commodore 64, Apple computers, DAC, Tandy, Atari the list goes on old yeah and that pesky IBM PC an .......learn your history.
Tim Barton You misunderstood me. I'm not a fucking idiot lol I know there were computers. I meant that I wonder if the technology back then would interest me enough to become a programmer as the technology today did. To much extent I feel that there is little room for new things these days. Most of what I can tell almost all products today are just better versions of what was already out there, and that sucks. And the chances of me starting a pc company in my garage and being a billion dollar business is about 0% possible. However I do think there are new things that will show up, but not at a small business level. One thing I would love to see in my lifetime is quantum computers personally. I don't know much about quantum stuff, but if I am correct it would offer much faster non computational instructions compared to normal ram and cpu's.
the first (analog computer)- the abacus.
Not really analog. Just mechanical instead of electronic.
I started out in electronics in the late 1960s. My interest in electronics arose purely out of curiosity. I had a really cheap electric guitar and amplifier. I lost interest in learning to play the guitar, but I wondered how the amplifier worked, so I took it apart. I was hooked. One of the first projects I built was a 4-bit binary counter using flatpack J-K flip-flops. I was pretty proud of it, so I entered it in the science fair in sixth grade. Nothing. I kept on plugging anyway. Eventually got an EE degree, worked in minicomputers, hard disk media, telecommunications, and aerospace before going for my MSCS. Now I build web-based applications managing big data.
In DRAM, not only does the charge on the capacitors have to be refreshed every few milliseconds, but it also, like core, has to be rewritten every time it is read. That happens automatically inside the chip, so we normally don't think about it.
Ferroelectric RAM (FRAM), a nonvolatile memory technology, also has this hidden rewrite on read, which is why it has limited read endurance, unlike EEPROMs and Flash, which generally only have a limit on write endurance.
You are a talented video producer.
Szr7zMix5kcV6F3 nice icon m8
Manuel Hernandez thanks for motivating me to change it. how do you like the new one? tehehehehee
Szr7zMix5kcV6F3 it's cool I actually had an anarchist icon as well
Manuel Hernandez got tired of it. too many keyboard warriors "not getting it"
+Szr7zMix5kcV6F3 XD same that's why I changed it mate
Thanks for posting this. The first memory I ever worked on was a core like the one you show. Cards full of hand stitched ferrite beads exactly as you describe them. I can still remember being taught about the hysteresis loop created by the write wires causing teh ferrite to hold a magnetic '1' or '0'.
Strange, but actually understanding computer memory and logic from these sort of hard wired principals has enabled me to adapt to many other aspects of electronics throughout my working life.
Strange to think that just maybe I belong in a museum alongside some of the exhibits now.
I did not realize just how far computers have come so quickly. that's crazy. to think at one point memory was handled with a fluid bed of mercury and some springs, that just blows my mind. now we've got laptops with 1tb+ for $500
What a nice stroll down memory lane.
I still have the core memory I salvaged in the early 1970s that I intended to use for a DIY computer while I was in college. I think I got it free as a donation from Xerox after making a few calls. Ended up first building one from an 8008 chip that I had trouble getting working, and ultimately, my first working computer was an IMSAI 8080 kit, which I then utilized for years-- still have that and it still works. But you forgot to mention drum memory. I also had a vacuum tube computer for a while, the LGP-30 that had 4K words of 31-bit drum memory-- bit 32 was a "spacer" bit on the drum and couldn't be written. And even the CPU's accumulator was stored on the drum, they saved multiple copies there in order to increase the access time. The drum also had a clock track which produced the master clock for the system-- it was bit-serial so it would clock data on and off of the drum as it ran.
Thanks for publishing... I enjoyed it very very much...
this is the most fasinating video ive watched all day my grandfather always tells be about when he would work on the computers with hole cards and it just amazes me what we have evolved into technology wise
I remember when a memory dump was called a core dump.
Many years ago I worked for British Rail. As we had very little money we had two old second-hand IBM 370-155 machines on which we ran the reservations system (one machine as backup). One day a drive transistor on one plane of the core memory failed. Our IBM engineer loved these machines, I think they reminded him of his youth! It turned out that IBM no longer stocked the part, at least not in the UK. So, he looked up the transistor's characteristics, found an equivalent in the Tandy (UK version of Radio Shack) catalogue, popped out to the local store, came back and soldered it in. I don't know if his expenses claim was accepted though...
OH JACK YOU ALWAYS KNOW HOW TO PUT A SMILE ON THIS OLD MANS FACE
Jerkamie he did to mine.. and I'm barely a 90s kid!
ZED O_o Let me guess, you were born in 1999? XD
I know how to put a smile on an old mans face.
OBAMA ISNT PRESIDENT ANYMORE! :D
Dragon Butt WANT a bet.
Thank you for this. Really awesome stuff! Love technology!
Hey good post!
Thank you for a walk down memory lane (pardon the pun), my first job back in the day was with Dataram Corp. plant (Trini Data owned by Mr. Michael Armoogam) here in Trinidad. Later on, I moved to United Telecontrol Electronics Inc. (Qualitek Electronics Plant, again owned by Mr. Michael Armoogam and Dr. Winston Akong) where we manufactured those Ferrite Core Memories. I went in as a trainee test technician and quickly moved up the ladder to Senior Test Technician / Assist. Plant Manager. I was very fortunate to be mentored by such engineering greats a Dr. Winston Akong and Dr. Art Ericksen. So yes, Trinidad and Tobago does have some history in the manufacture of "State of the Art" electronics. I believe Dataram Corp. was headquartered in NJ then. BTW our biggest memory back then was a folded, 128K stack and the "holy grail" of oscilloscopes was the Tektronix 500 series (my lab had two 545A's) and Nixie Tubes were the common digital displays, before moving on to the HP 7-segment LED displays (the segments were dots, don't recall if it was indeed a dot matrix device ....I doubt it).
www.dataram.com/blog/?p=163
Fun stories and history, very entertaining!
No doubt that memory will never look as cool as these older devices, I wish I could see what people a thousand years in the future will think
sam goad
they will maybe using a new form of core memory because it can be as fast as an ssd and you can rewrite it endless
I'd imagine that the material would change to something less temporal, if you go on Ebay for one of these you see alot of the connections severed. My idea, Crystal capacitors!
Thanks! It was very informative, and gives perspective on where technology lies today.
I was in the US NAVY 1963-1973. Yes, we used core memory in our SOSUS systems. As a computer technician, I can assure you they were a bugger to tweak for proper operation.
I loved the fact that you could halt a computer with core memory and start it up later at the exact point it left off (if you made a note of the program counter register). The old PDP 11/70s were great in some ways!
I wish people would not use the SI value for K (1000), but instead use the one that's always been used in the industry of 1024. It seems those hard drive manufacturers in the early 2000's looking to make thier disks look bigger turned the tide. Even though all other parts of computer architecture still uses 2^2 ie, 1024 systems. In fact, even filesystems used on those disks still use 2^n.
my god that is so interesting (the bit about old memory with the springs and mercury) its my opinion that its valuable for us to remember how things were originally done if we ever need to start back over.
Great video. If I remember correctly the first computer I worked with, a Univac 418 which had a (literal) core of 16 kbit. It was manually booted loading binary words into the main registers to call up tiny program that read a paper tape. The paper tape contained a program that loaded the operating system. The “large” memory was a pair of Fastran magnetic drum memories that were half the size of an automobile and could hold a couple of Mbytes. We definitely have made advances - as I type on my IPad Pro with 128 GB.
Came across this by accident. Had to watch. My Dad worked on the first computer at Rice University. They had the second computer for NASA in the late 50's. Was amazing to hear what that computer did, how it was built and then see what we have today. We had boxes of IBM punch cards that became Christmas wreaths. Might still have a few of those cards somewhere, going to have to look next time I go to Moms..
Can it run Crysis?
Cantrio X Yea but sadly only on poor quality.
Simmons :(
Simmons is a liar
Its just a joke.
I know lol
As a professional programmer that was taught assembly in my computer science degree the lack of memory forced better programming practices than are the norm today. Nowadays languages with garbage collectors mean most programmers that never learned assembly or even C know nothing about memory management and consequently code as if memory is infinite.
For a single user system that might be just about viable but if you need to scale all the old memory limitation problems are still there.
Great video.
Random fun fact: Those units were handcrafted, and usually by textile worker women. Here in Hungary from where i write this comment, such "ferrite-ring" memories were manufactured in the '50s. Countrywoman who otherwise made sophisticated lace overlays and ornaments at home were gladly hired for this job, for they had the patience and cold hand to produce such matrices of small pearls. This way about 1bit/square mm of data density could be achieved, which was a huge leap after the Williams-Kilburn tubes.
The manufacturing process was never automated here, for by the time the technology had become available for that, the advent of semiconductors made the whole stuff obsolete.
Very educational and good video, thanks for taking the time to make it.
I was the last person at Iowa State University (~1969) to use the core memory out of its old Cyclone Computer. The memory was in a console that was about three feet wide and four feet high and about 12 feet long. It had four stacks of 4 K, 40 bit words. When I used the memory it was used as a buffer memory for an A/D - D/A converter. I remember loading punched paper tape and magnetic tapes. I was told that the Cyclone Computer had gone through mercury memory and the CRT memory years back.
Not sure why you're thankful it's gone as from what I read core memory was pretty reliable comparable to semiconductor memory.
The Apollo AGC and 4PI had it and where so reliable they were used in flyby wire systems and some old mainframes that had it were it were still in use until the early 1990s.
Now mercury delay lines and Williams tubes on the other hand were known for being fickle and buggy to the point of frustration.
you think that having a million of these (6 GB) would be more reliable than 6 gigs of DRAM?
jakejakeboom
Of course not.
But it was as a huge leap over previous memory technologies.
One big disadvantage of core it was not only bulky but was very labor intensive to manufacture.
When the 1103 1K DRAM came out in 1971 it was a significant an improvement in cost and packaging and quickly began to replace core memory in many applications.
comparing shit old tech to other shit old tech, people are sitting on gigabytes of ddr3/4, and none of them give a shit so why waste comment space?
Only one I see wasting comment space is you. The others are at least having a civilized discussion about the "shit old tech" that eventually evolved into the modern shit that you value so much. Pull the stick out of your ass please.
my first exposure to those memory cores was with the Control Data 160A mainframe. That was already a dinosaur, but great for learning the basics of computers back in the 70s even. Later, even into the late 90's, the same basic memories were at the heart of Thomson CSF surveillance radar systems, used in the SDGS (synthetic data generation system) for creation and display of vector based characters on Secondary Radar displays. Really miss those times now... The technology at 2:20 was another that was used on the same radar. The device was called a Scan Converter Tube, for converting standard radar PPI circular sweep, into a TV styled raster scan. In basic description, there was a TV tube, being faced and bonded to a TV Camera tube, which actually read the charge left by the source tube upon the interfacing screen at the junction of the tubes, but there were no visible 'dots' in that case
In 1960 working as an assembler at IBM San Jose, I happened to see another assembler slip during assembly of a core memory for the IBM 360 and shove the screwdriver through the entire assemble of 12 planes of cores. At the time the assembly was worth about $2,500 1960 dollars and he literally cried when it happened. I worked doing wire wrap panels for the 360. Occasionally I would be loaned to other groups working on RAMAC, Stretch 7090 and other systems. Started at $86 a week. Good start for high school grad and I worked in electronics for 45 years as technician, engineer and manager before retiring.
Rope memory?
Related to core, it could only read.
Think of it as early ROM.
The flight control programs for Apollo were stored on woven ropes.
The system used AC in the interrogation and sense leads, which prevented having to re-write the bits on read.
The ropes were literally hand-laced to program the bit-patterns.
I remember looking at a computerized phototypesetting machine around 1975 of so. It was state of the art and I was really impressed. The big problem was that it used something called "bubble memory", and the instant the machine was turned off (or lost power), the memory would vanish. Around 5 years later, a computerized typesetter had what I think was a small ROM memory (something like 650 bytes of memory) that it used in starting up, and then it relied solely on one-sided 4.5" floppy disks for operation. It was a lot of trouble to operate with memory like that, and there was a 5 megabyte hard drive available, but the hard drive cost US$10,000, so I never got it.
Reminds me of coming across a cabinet of core memory in the basement of the HPER building at IU Bloomington, which housed the Wrubel Computing Center in one wing - must have been in the summer of 1980 or 1981. Taking a FORTRAN programming class, and we were still using punch cards. There were all sorts of fascinating machines down there from my point of view - HUGE punch card duplicators and printers, for instance. Big line printers, and a CDC6600 (which we learned programming on). Only five 300 baud terminals. A lot of this stuff would be gone in the next 3 years. The memory was in a cabinet, and this had toroids that were maybe ten times the size of yours. Big, visible wires running through the thing. It was on it's way out to the recyclers, I'm sure - it was in the middle of the hallway in the non-computing side of the building. Wish I'd at least had taken some photos.
Very interesting. I remember core memory but at that time we were moving to solid state TTL memory. I actually worked on a device that had RTL logic! A very nice trip down memory lane. Thanks.
+1 for making me discover delay line memory, Analog computing is fascinating in its complexity!
I thought I was into old computer stuff (born in 82) but I'd never heard of core memory. Nice explanation.
I use this video to sleep every night. I can't go to sleep without it.
I remember back in 1991, when I had a computer with an 80286 processor, running at a blazing 16MHz, and it had a 40 Meg hard drive. It came with 1 Meg of memory and I wanted to add an extra Meg. One Meg of memory cost $100 at the time. I had a friend who was doing computer repair and he sold me two 512K DIMMS for $50. I thought that I had gotten the deal of the century. Now, we can purchase 8 Gigabyte thumb drives for about $6.00 In fact, a lot of vendors at trade shows hand out 8, 16 and 32Gig flash drives for free. Also, in the late 90’s, I traded a used 1.2 Gigabyte hard drive for a brand new, good quality, 8 man camping tent and we both thought that each of us a got a great deal. It’s amazing when we think about how far we’ve come.
I love the somebody did a show about this topic. very interesting.
Fascinating video. My big story I tell my younger friends is that I recall the day (circa 1983) I upgraded my Atari 800 home computer (before there were true IBM-clones, aka, PCs) from 48K to 64K. That's right - I added an additional 16K or 16,000 bytes of RAM to this unit with this huge 11/4" x 10" x 5" slot-shaped plastic card. It was heavy, bulky, and the chips were not exposed to the eye. I often wondered how stupid this was as there was zero ventilation for these circuits should they run hot.
I used to work on core memories when I first started work in 1980. Testing used massive valve and relay based machines but to change a core took skilled operators using binocular microscopes and ultra thin needles.
+Andy Reid I remember working, at Boeing, with one of the techs who manufactured core memory at Burroughs in Downingtown, PA. He introduced me to the saying that "a good company will keep a tech until they begin to drool. I now know why!
Thanks. I have only seen pictures of core memory so to have it explained is great. I also just saw your interview with Dave (That crazy Aussie bloke!). Really interesting. :)
Wow! Thanks, somehow I missed this amazing history!
Amazing. I recently began messing around with tandy color computer 2 with 64k memory. My brother got one in 1984 & I was always blown away by it and it's programming guide even though I was only 6 and far too young to know what the hell I was looking at. I guess I had inner geek then because even though I was clueless about it I knew I liked it.
back in early 1980s , stationed at mtn home afb our repair shop ran test equipment using burroughs d-84 computers with core memory, i believe these computers were from early 1960s, 2 large ampex tape drives is where the test programs were located, very stable and ran the shop testing just fine. quite big though, great to be reminded of them, thanks j
Mr. Ganssle:
Overall, a very interesting video; but my attention was suddenly riveted when you put up pictures of:
≈ 2:25 to 2:47 - the Williams tube
When I was old enough to comprehend some things about computers, my father described what it was like to work on one of the very first of them. He was back and forth between MIT (where he was doing his doctorate) and Princeton (where J. von Neumann's electronic baby was), and I didn't catch which location this was, but he talked about having to deal with computer memory that was on an o-scope screen.
And here's the crazy part - when some of the memory spots on the screen would become faulty, programs for that machine had to be written to specifically *avoid* those memory addresses!
Anyway, thanks for connecting me with images (as well as the name!) of what my dad must have been dealing with back in the early 50's!
PS. I still have his copy of the owner's manual for the IBM 704, which, some years after his doctorate, was the machine in use at his work.
I remember seeing core memory on display decades ago, and assuming it was some sort of old memory, but only just now did I learn what it was called!
Really like your videos. I love microcontrollers and computers, because they bring so much history to the human race. Can't wait until retirement tire to have time to play with all this stuff.
I worked on that exact model of core memory (the cube). It's out of a Digital Equipment PDP-1
This video reminds me how I once said the internet would never quite replace TV. This at a time where streaming was typically .RM files, so not even all that long ago, just long enough ago that most of those web pages are gone.
Watching this, I can smell the room where I learned basic computing. We used computers called Tiki 100. The keys had a very distinct feel and sound.
It's crazy to think that we still use physical structures to form computer memory, but we managed to get the density millions of times higher over the last few decades.
The magnetic cores depend on a couple of phenomena for operation. The most important is hysteresis, a resistance to changing magnetic field until a threshold is reached. In the case of a typical core memory, you need about 300 milliamps through the core (ferrite bead, core, donut) to switch its magnetic field. Because of transformer action, this abrupt switch of magnetic field will induce a tiny current in a "sense wire". There's typically only one sense wire woven diagonally back and forth through the entire plane of cores. Each core has three wires: X wires, Y wires and a sense wire. So what you do is shoot 200 milliamps down ONE of the wires in the X direction, 200 milliamps in ONE of the Y direction wires. All of the ferrite beads along these wires ignores the 200 milliamp current BUT where these wires come together and cross inside the hole of a ferrite bead, now you have 400 milliamps and that's enough to switch the magnetism unless of course it was already magnetized in that direction, in which case nothing happens. So you repeat this process for each of the intersections and watch for a pulse on the sense wire. If there was a pulse, you just erased a "1" (or a zero; you can also use negative logic). Before you continue you'd better put those "1" back in because the act of reading the memory sets everything to zero (or 1). The speed of these things was measured in microseconds rather than nanoseconds. They are non-volatile; I've done some tricky things with core memory used in aircraft, solved a malfunctioning boot loader by using a good computer to load the program, then physically remove the core memory and install it in the aircraft and away they go with a functioning program already in memory. They were VERY expensive; it is a great souvenir.
Dr An Wang, who held the patent on core memory, and sold the patent to IBM always claimed they paid another inventor to challenge the patent during his negotiations for the sale. Although the other inventor had a weak case, it created doubt and lowered the sale price quite a bit.
Dr Wang continued to invent things, and built Wang Computers into a world power. The company disappeared in the 90's after his death - along with a bunch of other computer companies like DEC.
I was a computer tech in the 70's/80's and worked on some older Wang Computers and programmable calculators that used core memory. IIRC, these were the Wang 600 and 700 series, and they were pretty simplistic, but very programmable. Storage was on paper cards and cassettes. Engineers could make them do some pretty amazing things by switching cards in and out. Ah, the good old days.
It's truly amazing that I'm watching this video on my hand held phone only 62 years after that MIT built computer that took up an entire room.
0:33 lol at first I thought he was doing a "memory joke"
It seems like he just forgets what hes doing and starts a new topic
Very interesting, thank you for sharing and posting. I liked and subscribed
Thanks for the memories (reminder)!