Slashdot Mirror
EDSAC Computer To Be Rebuilt
nk497 writes with this bit from PCPro: "The first working stored-program computer is set to be rebuilt at Bletchley Park, home to the UK's National Museum of Computing. The Electronic Delay Storage Automatic Calculator ran its first programme in 1949, and was two metres high. Its 3,000 vacuum tubes took up four metres of floor space, and it could perform 650 instructions per second. All data input was via paper tape. The EDSAC used mercury-filled tubes for memory, but in the interests of safety, the replica will use an alternative non-toxic substance. Rebuilding it will take four years, and the public can visit to watch the work as it happens."
Or Linux for that matter...
The war will have been lost to Jerry by then!
Should have been enough for anybody. I bet it could calculate my tax return in the time it takes me to log in to gnome.
http://michaelsmith.id.au
In fact Alan Turing himself pointed out that a mixture of alcohol and water would do the job as well as mercury (he wanted to use gin.) Perhaps "Mercury delay line" just sounded more techie to the Civil Service.
From scarped cliff or quarried stone she cries "A thousand types are gone, I care for nothing, no not one."
The first stored program computer was the Manchester Baby
http://en.wikipedia.org/wiki/Manchester_Small-Scale_Experimental_Machine
Actually the code that is run on the minimalistic instruction set reminds more to BrainFuck then to anything resembling a OS.
I was at Bletchley Park a couple of months ago and by chance the National Museum of Computing was open that day. They've got some interesting displays of old computers, and their goal is to get them all running again. They cover everything between EDSAC and modern computers. Their oldest computer is a Harwell WITCH from 1951 (a decimal computer), this is being restored at the moment. Other fun stuff includes a collection of calculators, and a BBC micro with a working BBC Domesday Project laserdisc installation.
It's a separate museum on the Bletchley Park grounds, and its opening times are a bit limited (esp. in winter), so check before you go.
... the rebuilding of the rebuilding of the Mark 1: http://www.computer50.org/mark1/index.html
:)
There's a simulator here if you want to do some old-school coding http://www.davidsharp.com/baby/
(The whole abortion was replaced with a small PCB containing mercury-wetted relays, which contain only tiny amounts of mercury - the electrician who built the panel didn't know the difference when he resd the spec.)
From scarped cliff or quarried stone she cries "A thousand types are gone, I care for nothing, no not one."
I'm not sure that's a genuine stored-program computer. From what I recall, it was a very clever adding machine. Either way, it was a prototype rather than a fully complete, reliable machine that was used for research or commerce.
So let me get this straight ...
A FedEx driver screws up a delivery. A manager/supervisor at the local FedEx office takes it upon himself to actually be helpful and sympathetic to the situation. He uses his free time to deliver the package, and probably has to jump through spinning hoops to handle the details of the delivery in his computer system (a fact you kindly point out yourself). Likely he will also have to answer to his superiors about the way he bypassed established procedures in order to help you out.
And you are giving him a hard time about it by complaining about his actions to the corporate complaints office???
What am I missing here???
- Jesper
My security clearance is so high I have to kill myself if I remember I have it...
At a guess, finding enough of the required types of valves (aka "tubes" in other languages) is a time consuming activity
politicians are like babies' nappies: they should both be changed regularly and for the same reasons
I love the fact that there is a common desire to preserve our historic technological achievements.
Working reproductions of dying / dead machines are a great learning tool -- We are all truly standing on the shoulders of giants today.
I feel that efforts such as rebuilding the EDSAC are in the same vein as those that would create emulators for our out of production computers and video game systems as a cheap way to preserve the past.
What good is the EDSAC or an Emulator without a sampling of the programs the systems used to run? Surely different people would attribute different degrees of importance to different programs -- Thankfully digital storage is abundant and cheap enough that we are capable of preserving entire catalogs of programs.
Notice however, that the more relevant, beneficial and useful a replica or emulator is, the more illegal it is to produce due to patents and copyrights.
I fear that if the current copyright laws could be enforced absolutely, we stand to loose important parts of our history and culture for no other reason but greed. Given the long terms of copyright, it's a safe assumption that much of our digital heritage could decay and be lost before it's legal to reproduce it -- Even under good conditions CDs, Magnetic and Solid State Drives will all fail before 70 years after the author's life has elapsed.
I'm very wary of DRM and the DMCA -- Today we can recreate past works to better understand the significance of the shoulders on which we stand; Tomorrow we may find ourselves searching for footing that has long since crumbled away.
In fact Alan Turing himself pointed out that a mixture of alcohol and water would do the job as well as mercury (he wanted to use gin.) .
Turing failed to include a dash of Angostura . . . with enough alcohol, the computer can shoot shit out, but everyone is too trashed to give a damn.
Schroedinger's Brexit: The UK is both in and out of the EU at the same time!
None of the early Zuse machines were stored program computers - they had a relay memory for data and got their instructions from punched tape. The table in the Wikipedia page about the Z3 seems about right:
http://en.wikipedia.org/wiki/Z3_(computer)
The Manchester Baby was the first stored program machine, quickly followed by the modified ENIAC (the original used patch panels and cables) and then the EDSAC. Since the Baby was created to explore ideas for the EDSAC rather than as a usable machine on its own, I guess if you squint enough the article is right in an Obi-wan kind of way :-)
The method of storage in the Baby - a static charge used to represent 1 or 0 - proved to be the most effective form of storage for RAM (as static and dynamic CMOS) and is becoming more and more of a competitor for hard drives. Though CRT memory was short lived, in the long run Williams proved to be right. The Baby was prescient.
From scarped cliff or quarried stone she cries "A thousand types are gone, I care for nothing, no not one."
"The Electronic Delay Storage Automatic Calculator ran its first programme in 1949, and was two metres high"
Which also happens to be the height of a killer robot. Coincidence? I don't think so.
they have an app for that.
Barely.
Maybe. In theory you might get a scaled down bare bones Linux to run, but even so you would be hard pressed to run any programs with it. Vaccuum tubes were replaced by transistors in the '50s and '60s, which in turn were largely replaced by ICs in the '60s and '70s. A single vaccuum tube preforms the same functionn as a single transistor. The Z80 CPU chip, which came out in 1976, had 8,000 transistors, more than twice the tubes of this entire computer.
The Z80 processes around 40k instructions per second, compared to EDSAC's 650 IPS. That's sixty times as fast as the EDSAC. Imagine how long it would take just to boot!
As I noted in Growing Up with Computers, UNIVAC, a more powerful computer with 5200 vaccuum tubes that first shipped the year I was born, was less powerful than a Hallmark Greeting card.
Free Martian Whores!
We'd have the mind-numbing processing power to get my garage door open, and one less US state.
I swear to God...I swear to God! That is NOT how you treat your human!
It's great to see that EDSAC will be rebuilt! I wonder if Maurice Wilkes, the project leader, was told before he passed away just this last November? He was probably the last of the "first generation" computer pioneers to pass away. Several slashdot stories of his passing were submitted, but I don't think it ever made the main page. At least he can get his props here now.
Can no one look up and confirm well-known facts? Heck, this stuff is still within living memory. The article claims that EDSAC was the "first working stored-program computer" and that is just wrong.
The Manchester Small-Scale Experimental Machine [often known as "Baby"] was the first stored-program computer, not EDSAC. Baby was operational on June 1948; EDSAC didn't run anything until May 1949. Please don't play semantics with the word "working"; Baby worked, and in any case, all of these early computers were wimpy if you measure by storage or speed. EDSAC is important in computer history - don't take anything away from THAT - but let's get the facts right.
- David A. Wheeler (see my Secure Programming HOWTO)
So.... hamsters?
Anybody want a peanut?
I'm guessing the decision was not based on safety but practicality.
A sealed tube of mercury is not a significant safety hazard in a one off application like this machine. There are an awful lot of wall thermostats out there with sealed glass bulbs of mercury in them.
The EU has regulations on RoHs (Reduction of Hazardous substances) that apply to electronics. It's likely it just would have been too much paperwork hassle to get an exception.
I'll bet for similar reasons the solder they use in connecting it will be lead free.
Besides, if they used gin, like Turing wanted, it'd definitely be extra geek points.
... since LEO, the first commercial business computer, was based on the EDSAC design. Amazingly LEO computers were still in use in 1981. Check out the LEO Computers Society.
Andrew Yeomans
Edsac was not the first stored program digital computer.
Konrad Zuse's Z3 was running in 1941... turing complete, vacuum tubes, and all.
I dunno, but in 1949 they were already ... half a dozen at least that would fit into this characteristic.
The oldest coming into my mind: http://en.wikipedia.org/wiki/Z1_(computer) while mechanical and being a bit unreliable due to being handmade from scrap metal... still fits into the description.
Mark I and others also should be noted, in 1949 it was definitely not the first...
The development of computers that have all of the architectural features we consider standard took about 15 years and there were several steps in the process with each one having some sort of bragging rights. And deciding when the process was "done" and we had a fully modern architecture is something of a matter of judgment.
Back in the 1980s I researched exactly this question for a CS course project, and I examined the architectural details of every early computer to MANIAC and IAS or so. EDSAC was the computer I identified as being the first to have not only stored programs BUT ALSO "programs as data" - one that could rewrite their own instructions and thus (for example) load programs dynamically in the course of computation. Without this feature the concept of an "operating system" is essentially impossible. The EDSAC was my pick for the best claimant to the "first modern architecture" computer.
Second class citizen of the New Gilded Age
Can in run Linux???
Very...slowly...
Stupid safety theater. People all over the world sit RIGHT UNDERNEATH mercury-filled tubes. They're called fluorescent lighting.
Do these tubes explode spontaneously or something? Maybe they should give everyone eye protection and breathing apparatuses.
Gamingmuseum.com: Give your 3D accelerator a rest.
You'd need to reduce Linux's total footprint to 1024 instructions+data with no swapping, no hard disk, no networking, and all I/O through punched tape, but within those limitations it should run just fine.
Who would win this election: Andrew Weiner vs Andrew Weiner's weiner.
The Z80 processes around 40k instructions per second, compared to EDSAC's 650 IPS. That's sixty times as fast as the EDSAC.
That's pretty unfair on the Z80, too. To get anywhere near that figure, you'd have to take your basic 1MHz Z80, and have it continually execute the longest possible instructions, which were rarely used and took up 23 T-states (clock cycles) - that would give you 43,478ips.
In practice, most Z80 instructions executed in the range 4 to 13 clock cycles, and just about every Z80 I ever met back in the day was the 4MHz part, so you're talking between 300k and a million instructions per second. So more like a thousand times faster than poor old EDSAC.
(Nowadays of course, the modern Z80 clones/cores run the basic instructions in just 1 clock cycle.)
As mentioned elsewhere, the Baby was the testbed for EDSAC -- you could think of it as "EDSAC Lite". but it WAS a computer in its own right, so let's keep shaming the submitter and editors ;)
> Maybe [snip]
Or, no.
Thing is, early computers were immensely different than the computers we have today. Addressing modes weren't fully thought out, instruction sets were esoteric and more suited for hand assembly, and even just getting information to/from memory wasn't quite what you'd expect. Both delay-line and drum memory were delay based, you had to have extremely tight timings to get the word you wanted.
Linux leverages many modern conveniences and paradigms. Without heavy modification, it cannot run on anything older than a M68000/i80386 processor with appropriate support hardware, and the older you go, the more you have to gut, change and cripple.
Also of interest is the 1949 CSIR Mark 1 (CSIRAC), which is held at the Museum of Victoria in Melbourne (unfortunately no longer on display). Because of its historical value, there is no intention to restore it to working order.
I'd love to visit Bletchley Park one day though if I'm ever on that side of the world.
Perhaps it is unfair; I just did a quick google of Z80 for a comparison.
Free Martian Whores!