back to article UK computing museum starts reboot of 65-year-old EDSAC

The United Kingdom's National Museum of Computing has begun the process of booting up a replica of the Electronic Delay Storage Automatic Calculator – aka EDSAC – a computer commissioned in 1949. As The Reg has previously detailed, EDSAC has a distinguished place in computing history because its design influenced that of LEO, …

  1. frank ly

    Imagine ....

    .... being there when they boot it up and hearing it play its startup tune for the first time in decades. Awesome!

    1. jake Silver badge

      Re: Imagine ....

      More to the point, the smell of the thing.

      Yes, it has a "startup tune", despite having zero audio gear[1] ... but the ozone produced by these things absolutely takes my breath away.

      [1] Ever do the Winchester shuffle with an IBM 3340?

  2. jake Silver badge

    Totally awesome.

    Old kit needs to be rebuilt & remembered.

    1. Uffish

      Re: Totally awesome.

      They say they are trying for an authentic reconstruction without, as far as possible, using 'modern' components. So where do the valves, huge resistors, paper capacitors etc come from. I would like to imagine gently rotting, abandoned warehouses full of WWII spares but I can't.

      1. jake Silver badge

        Re: Totally awesome.

        Well, Uffish, I use & when I'm restoring old kit. The parts are available, if you look.

  3. 's water music

    Can it run...



  4. James Anderson

    LEO -- a valve based computer from 1950s superior to S360 a transitor based computer from the1960s.

    I think not.

    1. Fuzz

      LEO vs s360

      I think the author refers to the LEO line of computers vs the s360 line of computers. You'd be comparing the first s360 to the LEO III

      1. Tom 7

        Re: LEO vs s360

        What always amazes me about Leo is that it seemed to do more for automating Lyons tea rooms than some (most?) ERP setups I've seen running on several billion times the computing power.

        It certainly pisses all over BTs ordering system!

  5. Tom 7

    Brilliant - but its more fun pramming like its 1948!

    C'mon folks get unwithit

  6. Frankee Llonnygog

    Mercury delay lines

    Cathode ray storage tubes

    Old memory tech is mind-boggling

  7. leon clarke


    Have they actually restored it, or is this a replica? I was unaware that any significant parts of it survived.

    All the same, incredible achievement!

    1. Essuu

      Re: Reboot?

      It's a replica -

  8. leon clarke

    Oh, and zdnet has a rather good picture of the new EDSAC's boot ROM (which is a bit of a mechanical phone exchange)

    The boot ROM is one of those details that makes EDSAC so brilliant. Everyone else building computers at the time thought that building an electronic computer was an impressive achievement by itself. The EDSAC team also thought through what else was needed to let other people in the university to do something useful with it - so they invented boot ROMs and subroutines and so on.

    1. sarahemmm

      Oh boy! I remember those selectors!

      I spent several years wiring them for lift controllers in late 70s to early 80s. The had 25 selectors, so there was some clever design to link them for US lifts, which were allowed to go 40 floors...

  9. Primus Secundus Tertius

    Symbolic machine code

    In 1967 I started programming a Ferranti Argus minicomputer. It had a primitive assembler that alllowed addresses such as v12, v13, etc. So not a fully fledged assembler by the standards of the contemporary IBM 1620, another machine I used.

    I later learned that EDSAC programs were written in a similar primitive assembler: clearly an ancesral work.

    Neither EDSAC nor Argus had floating point hardware, so for science and engineering calculations you had to understand scaled fraction arithmetic. Not many people did.

    1. Mike 16

      Re: Symbolic machine code

      > Neither EDSAC nor Argus had floating point hardware, so for science and engineering calculations you had to understand scaled fraction arithmetic. Not many people did.

      Not many people understand floating point to this day, but that doesn't stop them from programming stuff that depends on that understanding. Von Neumann considered F.P. suspect, at least initially, and his similar dismissal of "computing" "random numbers" is oddly apt today, as such things as the gaffed eliptic curves are made known.

  10. Version 1.0 Silver badge

    Bring back the old machines

    There are lessons that can be learned when you program in that environment - lessons that many modern programmers never had to learn, such as how to write "Hello World" and produce an program image that's less than 10Mb ... or an operating system that's under 500Mb (hello Lollypop).

    I came late to the art of programming but I can remember the challenges of writing bootstrap code that fit into a 1702 EPROM - and the joy when I succeeded.

  11. viscount

    Imagine the OS updates they'll have to install after 65 years...

  12. FartingHippo

    "EDSAC ran at 500kHz"

    Are you sure about that? Seems improbably high to me.

    For context, the Z80 ran at 3Mhz in the mid-70s (or thereabouts). That's less than an order of magnitude more after 25 years of unbelievably rapid development. The Z80 chip was smaller than a postage stamp, as opposed to a wheezing hulk taking up a large room.

    500Hz, on the other hand, I could well believe...

    1. the spectacularly refined chap

      Re: "EDSAC ran at 500kHz"

      "EDSAC ran at 500kHz"

      Are you sure about that? Seems improbably high to me.

      Yes, it really was that fast in terms of clock speed at least: the comparison with most later machines isn't really valid, since EDSAC was a serial machine working one bit at a time rather than in entire words. As a result each instruction took several hundred clocks to execute.

    2. Tom 7

      Re: "EDSAC ran at 500kHz"

      Colossus ran at 5.7Mhz - valves were a lot lot quicker than early transistors and much faster than the earliest microcomputer chips. But as you point out micorcomputers had some advantages that made it worth investigating...

  13. EDSACqm

    It's not the original - but it's very accurate and looking good!

    Sorry to disappoint everyone - the article got one key basic fact completely wrong. This is not the original EDSAC - most of that lies at the bottom of the council tip near Cambridge.

    This is a very accurate replica, created over the last 4 years by a team of volunteers from almost nothing but a series of old monochrome photographs of the original machine. Just the detective work to do that was extremely impressive.

    We hope to have it running fully by the end of 2015.

    Meanwhile visit and in particular have a look at the excellent series of videos made by David Allen, who produced the BBC's ground-breaking series of educational TV programmes featuring the BBC Micro back in the 1980s, and which introduced a whole generation to computing in the UK.

    Parts? Well, there ARE big warehouses full of valves - they are not the problem. But value bases (holders) in good condition are much harder to come by. And the trouble with other components is that resistors drift from their nominal value with age, and capacitors can deteriorate to the point where they are either unreliable or downright dangerous! So we use the largest, most authentic-looking, over-rated components we can find! The machine also uses about 1,000 tiny induction coils - our first challenge was to find a special winding machine capable of making these!

    Authenticity is very important to the team. Compromises are sometimes forced upon us, but we try to be as faithful to the original as we can be. Our chief engineer is a stickler for authenticity!

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon