Forgot Valentine's Day? Never mind, today marks 75 years of the Electronic Numerical Integrator and Computer Happy ENIAC Day! Join us for a celebration of the 75th anniversary of the launch of the Electronic Numerical Integrator and Computer. Yes, we know that things actually kicked off in 1945 rather than 15 February 1946, but today's the day when, 75 years ago, the machine was launched at the University of Pennsylvania. In 2011, …
Just as with the supersonic aircraft, British scientists and engineers tempted over the pond to "share" information, Turing and others found that the US had basically nothing and the transfer was all one way. Meanwhile, the British government sat on everything that had a military angle, largely suffocating it.
A lot of great stuff in this world has been invented, step by step, in several countries at about the same time.
Also the Americans tend to forget the Germans rather often.
"Konrad Zuse (German: 22 June 1910 – 18 December 1995) was a German civil engineer, pioneering computer scientist, inventor and businessman. His greatest achievement was the world's first programmable computer; the functional program-controlled Turing-complete Z3 became operational in May 1941. Thanks to this machine and its predecessors, Zuse has often been regarded as the inventor of the modern computer. ......"
https://en.wikipedia.org/wiki/Konrad_Zuse
Also about the Z1:
"The “Z1” was the first freely programmable computer in the world which used Boolean logic and binary floating-point numbers, however it was unreliable in operation.[3][4] It was completed in 1938 and financed completely from private funds.".
http://www.enseignement.polytechnique.fr/informatique/INF431/X09-2010-2011/AmphiTHC/SynchronizationPrimitives.pdf has a summary of computer history covering these (Ch 2.1). According to that the Atanasoff computer was never completed so it seems the Colossus was the first actual implementation of an electronic computer while Zuse's electromechanical Z3 was the first actual implementation of a programmable computer.
I don't want to get into whose country is bigger than who's on this (so please don't hate me).
I think the machine you are referring to was the first digital computer - and it wasn't programmable. It was the ABC, or Atanasoff-Berry Computer.
Colossus is usually acknowledged as being the first programmable electronic computer.
If anything, we Brits stifled things a bit by keeping it under wraps, so no one knew about it until others had been invented elsewhere.
Konrad Zuse's electro-mechanical machine predates it. But the fall-back position is always Charles Babbage (though latterly, Ada Lovelace).
And before Charles Babbage there was:
"Wilhelm Schickard designed and constructed the first working mechanical calculator in 1623.[9] In 1673, Gottfried Leibniz demonstrated a digital mechanical calculator, called the Stepped Reckoner.[10] Leibniz may be considered the first computer scientist and information theorist, for, among other reasons, documenting the binary number system. In 1820, Thomas de Colmar launched the mechanical calculator industry[note 1] when he invented his simplified arithmometer, the first calculating machine strong enough and reliable enough to be used daily in an office environment.".
https://en.wikipedia.org/wiki/Computer_science#History
Absolutely. And before that there was the Antikythera mechanism.
And before that, various stone structures built to compute astronomical movements/seasons/whatever.
Mankind is the key. Not artificial borders and coloured rags - the frantic waving of which does tend to stifle things a little (as well as try to rewrite history sometimes).
The Iowa State machine was programmable by hand, not by stored program. As were Colossus and ENIAC and the early Zuse machines. The first stored-program machine was the Manchester prototype in June 1948. Zuse had the idea, but wasn't there first due to wartime and post-war conditions in Germany.
But it's a fairly pointless argument. It was Zeitgeist.
As for the British failure, read John Hendry's "Innovating for Failure". Published in 1990, but still very much to the point.
Rather than endless arguments about whose ancestors were best, here are some useful facts.
Atanasoff–Berry machine (the Iowa State machine). Designed 1937-1938, tested 1942. Electronic, separate memory and computation, but not programmable, not Turing-complete.
Colossus. Designed and built 1943-1945. Electronic, programmable but not stored program (program was on plugboard), not Turing-complete. It has more recently been shown that if you took all of the colossi (or enough of them, and I'm not sure how many 'enough' is but it's 10ish, and that's how many there were) and wired them together then you could have made a Turing-complete machine. But if I take enough transistors and wire them together I can do that too: it's a very interesting thing but it doesn't make Colossus Turing-complete.
Zuse Z3. Built in 1941. Electromechanical (relays, not valves), stored program (program was on tape). Turing-complete in a slightly mad way: there are no conditional branches, but it could speculatively execute every branch.
Zuse Z4. Completed in 1945. Electromechanical (relays), stored program, had the beginnings of a high-level language (symbolic assembler, perhaps is better name). Turing-complete, initially in the same mad way as the Z3 but later had conditional branches.
ENIAC. Completed in 1945. Electronic, programmable but not stored program (programs on plugboards and switches). Turing-complete.
SSEM (the Manchester Baby). Ran its first program on 21st June 1948. Electronic, stored program, Turing-complete. Designed as proof-of-concept for Manchester Mark I.
Manchester Mark I. Operational in April 1949. Electronic, stored program, Turing-complete. Commercialised as the Ferranti Mark I, from 1951.
EDVAC. Operational in 1949. Electronic, stored-program, Turing-complete.
Pilot ACE. Ran its first program 10th May 1950. Electronic, stored-program, Turing-complete. Not the first at anything but it was based on Turing's design.
There may be some here I've missed. Which you want to count as 'the first computer' depends I think, on which set of ancestors you want to worship.
One important thing to realise though, is that after Turing's 'On computable numbers' paper in 1936, anyone who cared knew what a stored-program, Turing-complete machine was and it was clear that such a thing could be built: they weren't feeling their way towards something that they didn't yet know what it was.
The reason people didn't build such things earlier – the reason Colossus was neither a stored-program machine nor Turing-complete for instance – was not that they didn't know what that meant, but that they had a rather specific problem they wanted to solve, and they built a machine to do that: reading the program from tape and storing it in some writeable memory would make the machine slower, more expensive, and less reliable than reading it from a plugboard, so they didn't do that. Similarly the people involved in building it very definitely knew what Turing-complete meant, but they didn't need it to be that.
Much the same is true with ENIAC: if your problem is to compute artillery firing tables or simulate some of the processes in nuclear weapons, you have very specific programs you want to run many times, so why waste money, performance and reliability when you could spend the time to manually enter these programs on plugboards.
It really wasn't until people wanted to either do open-ended research or, probably more importantly, sell machines to people who wanted to write their own programs that they started needing stored programs and Turing-completeness.
But Zuse does need more credit than he gets.
IBM's tabulating machines (which were, in fact, computers) existed long before even Colossus or Turing's Bombe.
I'll be nice and not mention the Polish Bomba, much less Hollerith's tabulating machines that were used in the 1890 US Census.
That's not to take anything away from the British and German contributions to computing, far from it, I'm just trying to open a few rather blinkered eyes to the wider world ...
"If I have seen further it is by standing on the shoulders of Giants." —Isaac Newton, 1675
While tabulating machines were very definitely extremely important, they weren't computers in the sense that we now mean that term. The four criteria I was using were really: electronic (does not matter theoretically but makes a huge practical difference)? programmable? stored program? Turing-complete? Tabulating machines fail all but one (programmable, and even then they may fail that if you're fussy).
Neither were the bombas (I think the right plural is 'bomby', which is just a fantastic word which I shall use from now on whenever I can). These really fail all of them except possibly programmable. But that's not being rude about what Rejewski, Różycki and Zygalski did: they both showed that Enigma could be broken and built a machine which could break it. That machine wasn't a computer, but it didn't need to be, and in fact building a general-purpose computer would have been a good way of failing to solve the problem they needed to solve. The Turing-Welchman bombe (which was derived from but not the same as the bomba) also wasn't a computer, despite its designers pretty certainly knowing what a computer was, and again it wasn't a computer because being a computer would have been a way of not solving the problem they needed to solve.
I'm sorry if I gave the impression of being pro-British or pro-German: that wasn't the intention. I was just trying to make a list of early machines which either were or nearly were universal computing machines: tabulating machines aren't in that list, neither are bomby, or bombes. As far as I can see ENIAC was clearly the first Turing-complete electronic machine to run: it failed the stored-program thing until ?1948? but that doesn't matter as much as the 'being a universal machine' thing. The Zuse machines were universal (but this wasn't known until 1990-something I think) but they were not electronic and were universal only in a mad way which would have meant exponential slowdown on almost everything. If you have the age of the universe to play with they're useful: not if you don't.
And again: Rejewski, Różycki and Zygalski didn't build a computer. They did however solve a very hard problem and shorten the war, and they should be remembered for that much more than they are.
"Then the earliest computer is thousands of years ago"
That would be accurate, yes.
"and this is all a waste of time."
You said it, not me.
"I'll stick with narrowly defining the ruleset so it puts my country first, because that makes me feel all warm and fuzzy."
FTFY
@tfb
You get more or less everything wrong regarding Zuse, for no good reason at all as there is a lot to read about him.
"Zuse's work went largely unnoticed in the United Kingdom and the United States. Possibly his first documented influence on a US company was IBM's option on his patents in 1946.".
PS. As I am nor British or German I can speak for or against depending on just facts and I don't have to rub my penis in nationalistic fervour all the time.
So, what exactly did I get wrong about him? If you are objecting because I didn't include the Z1 and Z2: neither were Turing-complete, and I was not including non-Turing-complete non-electronic things, which is why they weren't there (if I was going to include them I'd end up including hundreds and hundreds of things back to antiquity, including the silly suggestion of someone that slide rules should count). The Z3 was Turing-complete which is why, despite being a non-electronic machine, it's there. The ABC is there because despite not being Turing complete, it was electronic. And so on.
And if you are accusing me of being nationalistic about this you are clearly really badly misreading what I wrote.
"And if you are accusing me of being nationalistic about this you are clearly really badly misreading what I wrote."
Then re-write it so it doesn't come across that way. Moving the goalposts after the kick is made isn't in any rule book I've ever read. A computer is a device or group of devices used for calculating.
The UK government didn't stifle everything. Turing moved to Manchester University and programmed computers there (teaching himself base 32 arithmetic in order to read the displays).
See, for example, Andrew Hodges' excellent biography of Turing, the Enigma of Intelligence. (Avoid the movie 'the imitation game', there are several howlers in it, although the acting is excellent.)
And it wasn't just Turing: a lot of people came out of Bletchley Park who had experience of what was possible: not really that universal machines were possible, which everyone knew (and universal machines weren't built during the was at Bletchley Park anyway), but that large-scale electronic machines could be reliable enough to be practical and could also be used to solve really hard problems.
So after the war there must have been a great influx of people into universities & elsewhere who couldn't talk about what they'd done, but who knew that certain things were practical to do. So people like Donald Michie come to Edinburgh (via Oxford, interrupted by BP during the war) and help start what became the AI department there, and the same thing happened in lots of other places I'm sure. And obviously not just in the UK: the same thing must have happened in the US at least.
Anyone watching this carefully (and perhaps people were) could have inferred that something interesting involving computation happened during the war I think.
Of course, one consequence of a bunch of people realising that quite hard problems were tractable is that many of those same people then believed that really hard problems were tractable, leading to the first AI hype cycle.
150 KW is not huge, considering that ENIAC was a Generation Zero valve machine. I don't know what its clock speed was.
By comparision, the mid-range ICL 1903 mainframe I was using in 1969 ate 40KW with all cabinets powered up: these were the CPU (including 32Kwords magnetic core memory with logic based on 1st gen chips), a disc controller, 2 x 60MB disks, 6 x tape decks, 1350 lpm printer, card reader and paper tape reader plus an ASR33 teletype operator's console. IOW, the 1903 mainframe still consumed 25% of the power needed by ENIAC despite using solid state logic running at a whole 3.5 uS clock speed (28000 cycles/sec).
Maybe a better comparison would be with the HP28s calculator I got in 1990. This, still working perfectly, runs off 3 N size alkalines and, even when used much more regularly than when I got it, has never run for less than a year on a set of batteries.
Reminds me of waiting in line in a restaurant in London a the counter. American gentleman in front of me to assistant: (points) "What is cam-em-ber-t?"
Assistant, with barely concealed contempt: "It's a soft, French, Cheese" (pause) "sir."
The American was quite happy to try this novelty, though I don't know what he thought of it.
Whilst I don't doubt your account, I fear you must have encountered that rarest of species, the Uninformed Parochial American. Sadly, they do exist, albeit in ever decreasing numbers (stop sniggering at the back there, you cynics).
Here in my own quiet corner of the Pacific Northwest we're perfectly au fait with such furrin fancy-pants concepts as Camembert. Heck, even our regular Safeway grocery stores sell Ekte Gjetost/Brunost and California's excellent Humboldt Fog, two true Cheeses of the Gods (or at least, Cheeses of Nazareth, h/t Robert Rankin for that one).
Still can't find mulled wine anywhere, though, but at least Glüwein is readily available.
You don't buy mulled wine, you heathen. You make it.
And you don't make it, either, if you can help it ... Instead, one strives to make wine that is drinkable without polluting it.
As a side-note, you can add Cowgirl Creamery to your cheese collection.
https://cowgirlcreamery.com/
As an aside, try to listen to the Christmas edition of 'Cabin Pressure' by John Finnemore. It includes the rather unfortunate fate of a bottle of Petreus (which includes a microwave oven and tic-tacs). Wine lovers of a nervous disposition should avoid this episode.
A long time ago our engineering team was visited by an American colleague who was distressed to see what he assumed were the carcasses of large dogs hanging in the local butcher, sheep presumably being uncommon in his flyover state of origin.
Folks in the so-called "flyover states" are usually considered more rural than the states of the folks doing the flying over, which is the point of the attempted insult. As such, your colleague was probably not from such a state. Us rural folks know what food looks like before it gets packaged by Safeway.
Sheep are common in pretty much every state .... however, we don't really eat a lot of sheep (or goats) here in the US when compared to other protein sources. So it wouldn't surprise me if even one of our more rural denizens went a lifetime without seeing a freshly harvested sheep hanging. But I'd be flabbergasted if he thought the carcass was that of a dog. He certainly wouldn't be distressed about it.
A 'Chinese' 'take-away' cafe in Leeds, got shut down several times for having dogmeat in the larder. As in the carcasses of dogs hung up ready to cook for the discerning (mainly student) population.
I only ever got prawn dishes from him, so reckon I did not consume any German Shepherd.
Here in my own quiet corner of the Pacific Northwest we're perfectly au fait with such furrin fancy-pants concepts as Camembert.
Drive East for a bit...over the mountains...where the Real Americans (or, so they would claim) live.
"Do the Merkins know what real cheese is though?"
Yes. Do you know what a merkin is?
"They think cheese come out of a squeezy bottle!:
No. So-called "American" so-called "cheese" is no more American than it is cheese ... The inventor of the narsty artificial plastic cheese substitute was a dude named James Lewis Kraft, who was Canadian.
For that matter, I've never seen a "cheese spread" or "cheese food" in a squeeze bottle. Cheez Whiz, when I last saw some (years ago), came in a glass jar; Easy Cheese in a metal can (because it's pressurized). A squeeze bottle would be barbaric.
"For that matter, I've never seen a "cheese spread" or "cheese food" in a squeeze bottle."
You don't remember "Squeeze Cheeze" from the mid-'70s? Came in a fat, sausage shaped plastic tube with a red X-shaped nozzle that you had to open with a couple pokes of a knife through a film seal. It also had a cousin, "squeez-a-snak" a little later (late '70s?) which was often flavo(u)red with things like bacon or chives, with the same X-nozzle but clear plastic instead of red.
The pumpable orange goo that goes onto everything from nachos to hotdogs to fries at sporting events was invented by a guy named Frank Liberto in 1976. He was looking for a way to get customers in and out of his concession stand as fast as possible, and liquid cheese helped. Howard Cosell made it famous in a Monday Night Football broadcast in 1978 after he found the name "nachos" amusing.
Certainly "Nacho Cheese" was available in 1 gallon (#10) cans for the restaurant/ballpark trade by the late 1970s, and I seem to vaguely remember it in squeeze bottles at the local supermarket around the same time.
There were other, similar, products in the same time frame. I seem to remember one that was advertised as "astronaut food", sold in single-serving tubes that you squeezed into your mouth.
I'm mildly disturbed that I remember this.
When your debugging session required a technician and line-by-line computation verification, you were heavily incited to think about your code twice - if not more.
18,000 vacuum tubes (or 19K, who's counting ?). Can you imagine a modern datacenter based on that ? You'd have twenty guys replacing the broken ones day in and day out. The vacuum tube industry would be more important than Exxon.
The industry has come a very long way in a very short time. Always makes me wonder what will come by in the next ten years.
ENIAC had guys pushing shopping carts with spare vacuum tubes in them around the site. Dead tubes would be replaced while the machine was running. Calculations would be run three times; if you got the same answer twice, that was probably the correct answer.
And science fiction stories about giant computers with lots of vacuum tubes proliferated; Asimov’s MULTIVAC stories are probably the most famous, ending in the religion-destroying “The Last Question”. Clarke had two religion-destroying stories, but only “The Nine Billion Names of God” used computers, and he didn’t specify how the computers were built. (Note that the ‘AC’ in MULTIVAC stood for ‘Analog Computer’, digital computers were considered second-rate.)
I have all of Gerry Anderson’s 1960s puppet shows around somewhere; I still remember the astonishing speed of the ‘latest French supercomputer’, from the first episode of _Joe 90_, which could retrieve one byte in 5 nanoseconds. And it used lasers, not transistors or vacuum tubes… And the way that BIG RAT depended on magnetic tape. Lots and lots and lots of magnetic tape.
And then there's the Bat-Computer from the 1966 Batman television show, which featured a voice-recognition digital assistant (Batman would just address it as "Bat-Computer" and then ask a question aloud), but responded with a punchcard that Batman would have to read, if memory serves.
In the Stratemeyer Tom Swift Jr. books, Tom invents his "Little Idiot" handheld computer, sort of a programmable calculator. A handheld calculator wasn't a visionary idea in 1954, but it was forward-thinking and probably the most plausible of any of the things Tom is credited with inventing in the series.
This is complicated. 'two foot wide panels' is, I think, probably correct in my home-counties English – it's an 'eight metre long shark' not an *?'eight metres long shark'. But 'two foot wide panels' is also horribly ambiguous: are these this pairs of, foot-wide panels, or single panels which are two feet wide? 'two feet wide panels' neatly resolves the ambiguity at the cost of causing the tops of grammar pedants' heads to come off. You can avoid the whole grammar-pedant-head-detachment thing by saying 'panels which were two feet wide' but this is both two more words and clumsy.
I'll take language which is clear over language which is 'correct' any time.
And of course 'correct' is dialectical: in my dialect 'two feet wide panel' is not right (but 'two feet wide' not *'two foot wide'), but it could easily be correct in either the dialect of the person who wrote the article, or whatever dialect it's being subedited into (if it is being subedited). And those dialects are just as good as mine. We (collectively) get to choose the rules of the languages we speak: they are not handed down by God, or posh people.
(And for instance, the moment I write 'panels which are two feet wide' I realise that there are dialects where what would be correct is '... two foot wide': 'how far away is that? oh, about two mile down the road' or 'how long are those timbers? about six foot' are both fine in some dialects.)
Show me a style guide that advocates one, and I'll show you another style guide that advocates another. And there's probably a third style guide that advocates yet another way,
Me, I avoid the cranial-rectal inversion and attempt to make myself understood as best I can. Most people in RealLife seem to appreciate that.
After having owned a few tube oscilloscopes, I'm in awe that that engineers could keep this machine in a working state. The components from back then weren't very reliable and their parameters drifted like crazy. Several parts could conspire to create a fault that can't be found by testing or replacing any single component on its own.
Yes, I think that's right (although I'd forgotten). Another trick, certainly used for Colossus and I bet for others as well, was to ramp up the voltage on the heaters slowly, so you never get a big thermal shock. You need a really big rheostat, I suspect (I wonder if the rebuild does that or if they use some fancy switch-mode-PSU magicness to tweak the heater voltages up from zero: I bet they do the latter, it's what I'd do).
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
