Talk:ENIAC/Archive 1

Archive 1

Archive 2

Archive 3

First para rephrase

First para is currently confusing: was ENIAC actually the first to fulfill all the qualities in this sentence, or was it first thought to fulfill one set of criteria, and is now known to fulfill a different set? I've had a bash at clarifying, using the article and above discussions. Please correct if I've introduced errors.

ENIAC, short for Electronic Numerical Integrator and Computer, was long thought to have been the first Turing-complete electronic computer. It is now known to be the first large-scale, electronic, digital computer capable of being reprogrammed to solve a full range of computing problems. Earlier computers included the German Z3, designed in 1941 by Konrad Zuse, and the British Colossus computer of 1944, designed by Tommy Flowers. The Z3 was the world's first general purpose, electro-mechanical computer. It was Turing-complete, fully programmable by tape and used relays for all functions, so was not electronic; Colossus was all-electronic and could be reprogrammed by rewiring, but was not fully general purpose. Until 1948, ENIAC also required rewiring to reprogram. Both ENIAC and Colossus used thermionic valves, that is, vacuum tubes. The confusion as to priority arose because the machines were developed independently as part of each country's war effort in WWII. The Z3 was destroyed by Allied bombing of Berlin in 1944, and although selected US teams were introduced to the Colossus, its existence remained classified until the 1970s.

It might be appropriate to move most of the material about the Z3 and Colossus further down, as per the Z3 article. Thoughts please... JackyR 20:03, 12 February 2006 (UTC)

Looks good. -- RTC 21:36, 12 February 2006 (UTC)

I agree that the other stuff should be moved. -- Dominus 22:14, 12 February 2006 (UTC)

The University of Pennsylvania (where it was built and still resides) bills it as the first "large-scale general-purpose electronic digital computer", which I believe is correct, although all the qualifiers are needed there. For example, Colossus was a large-scale electronic digital computer, but was not general-purpose; Z3 was a general-purpose digital computer, but was not electronic; an so forth. -- Dominus

OK, I've done the changes in two steps. The first drops the above § in for the intro, the second rearranges the material to make the intro more about ENIAC, less about other machines. If the result isn't satisfactory, pls feel free to revert to step 1. Thanks for watching over these changes :-) JackyR 01:02, 13 February 2006 (UTC)

Added hidden suggestion

I've added a concealed comment to the sections that mention other computers, because these seem to attract edits that might be more appropriate for those machines' specialist articles. Comment is: "THIS ARTICLE IS ABOUT ENIAC. If you wish to add very detailed material about other computers, please consider adding it to that computer's own article or History of computing hardware. Happy editing!" Hope this doesn't ruffle any feathers - pls remove if it does. Cheers, JackyR 00:18, 17 February 2006 (UTC)

Intro again

I'm not enthusiastic about having a 1976 work cited in the first § to show ENIAC's form of priority. Gnetwerker, can you tell us exactly what the source claims? Is it pre- or post- the revelations about Colossus? Do the authors know anything about the Z3? As stated further down, a lot of histories of computing were factually wrong because of the secrecy surrounding these projects. Unless this book's authors display accurate knowledge of the other machines, I wouldn't use it for backing up priority claims. Blainster has edited the § so that WP records the priorities correctly: does this now make the book citation false? Please to check, everyone... Cheers, JackyR 00:25, 15 March 2006 (UTC)

I grabbed that off the bookshelf because it was handy. Of course, 1976 seems recent to me! I can't type in the whole text, but it references the Z3 (noting it was built from telephone relays), talks about the Harvard Mark I and other "electro-mechanical" machines, then says that the ABC was "the earliest known electronic digital computing device", but notes that it was special-purpose, then mentioned ENIAC and EDVAC, noting (pg. 491) "the concept of the large-scale electronic computer is due to ...Eckert and Mauchly". On pg. 540 it says "It (ENIAC) was the first electronic automatic computer". Finally, Engines of the Mind (Shurkin) devotes an entire chapter to the ABC vs. ENIAC controversy, and finally says (pg. 298) "Who invented the modern computer? Eckert and Mauchly."

This has been discussed at least half to death on the ABC and Computer pages, and I have provided complete references there as well. Shurkin's book is from 1996. The comprehensive sources all say:

1. Harvard Mark I and pre-war Zuse machines were "program-controlled calculators"
2. Mark I and Z3 were "electro-mechanical";
3. ABC was special-purpose, but the first digital electronic computer;
4. ENIAC was the first electronic, general-purpose, reprogrammable digital computer.

The Encyclopedia of Computer Science goes to some trouble to define digital, general-purpose, electronic and automatic in the process of making its definitions. I don't think much has changed since 1976. The ridiculous "first computer" battle (here and elsewhere) is over definitions, not facts. -- Gnetwerker 01:09, 15 March 2006 (UTC)

P.s. -- The aforementioned Encyclopedia of CS has a full (two column) page and a large photo devoted to Zuse, and mentions the Z1, Z2, Z3, Z4, Z11, Z22, Z23, and Z64 (plotter). -- Gnetwerker 01:13, 15 March 2006 (UTC)

Many thanks for that. Yes, I'm in no way impugning ENIAC's credentials - but from what you say I'm now pretty sure the ref is not good: could we use Shurkin's 1996 book (or one of the other"comprehensive sources" you mention) as the intro ref instead? The reason being that the Enc of CS clearly has not heard of Colossus and the statement "the concept of the large-scale electronic computer is due to ...Eckert and Mauchly", is now known to be false (unless it adds "in the US": it's manifestly untrue for Britain). Do we really want to refer readers to a book which contains (through no fault of its own) inaccuracies?

(I didn't pick up on 1976 because of not being recent, but because the first person to break the secrecy about Bletchley Park published in 74, and didn't write much about Colussus anyway. Details of that, and the public realisation that the Manchester and Cambridge teams knew where they were going, came later).

Sorry to be nit-picky, but given, as you say, the ridiculous squabbling about so-called "first computers", I'm trying to make this article bomb-proof. Cheers for your patience! JackyR 16:09, 15 March 2006 (UTC)

I should have just removed the Encyclopedia citation when I took out the sentence it was attached to. I did not write the first sentence in the article— it was already there— I just moved it to the front of the paragraph where it originally was. The present description in the article explains the overlapping nature of the various computing inventions better than a lot of outside references do. Each of the above machines could probably be labeled as "first" with the right qualifiers. I will add another new book to the "further reading" list. Electronic Brains not only discusses these machines, but the first Russian computer, an early Australian computer, the British LEO, and an interesting hydraulic computer built at Cambridge. The last chapter of the book tells the tale: "It's not about being first". --Blainster 22:36, 15 March 2006 (UTC)

Bravo! I shall retire for the evening with the weight lifted from my heart :-) JackyR 23:35, 15 March 2006 (UTC)

Or I will, if Gnetworker is OK about changing the ref. Sorry folks, I have none of these books myself, so can't do that bit. JackyR 23:43, 15 March 2006 (UTC)

(De-Indent) You are correct, Enc CS doesn't mention Collosus (didn't know that was in play), but Engines of the Mind (1996) and Computer: A history of the Information Machine both do. I am not an ENIAC zealot, and the latter (Computer) comments that the 1948 Machester computer "established incontrovertibly the feasibility of the stored-program computer" and talks about it leading to EDSAC (1949). Engines gives a little more coverage to COLOSSUS (spelled there in all caps). It says: "By any standards, COLOSSUS was a working electronic computer, perhaps the world's first. The machine was limited because it was designed for a particular function and was virtually useless for anything else." (pg. 143). This is in the same chapter as the Z3 discussion. I would not in any way object to changing or removing the reference, but the important point is that, with deference to the many other "firsts", the ENIAC is the (for lack of a better term) grand-daddy of modern computing, notwithstanding earlier efforts.

Part of the problem is that we have had an ABC activist over on the ABC page, saying things like "binary is the same as digital" and throwing up lots of smoke from the 1973 Mauchly vs. Atanasoff proxy fight (Sperry v. IBM I think). I think all of this is making the collection of early computer pages inconsistent and of lower quality. Perhaps we need a page First Computers that goes into it all in detail. -- Gnetwerker 00:17, 16 March 2006 (UTC)

Ref changed - I pasted in the Shurkin from ABC (hope OK). Re First computers, actually History of computing hardware ain't bad - and is the natural home for such. But it's not great, either. You're quite right about the inconsistency: a lot of these articles seem to be from different planets - which I guess reflects very accurately the original circumstances of the machines, and their historiographies! What was Wiki born for, if not to make all clear? JackyR 01:00, 16 March 2006 (UTC)

PS I didn't mention Colossus?! RTQ! ;-) JackyR 01:00, 16 March 2006 (UTC)

I meant that Colossus wasn't really in the running as "first computer". Was it? -- Gnetwerker 02:00, 16 March 2006 (UTC)

Isn't that the "historiography" thing again? Colossus wasn't in the running for anything - cos no-one writing the history had heard of it. In technical terms, it seems to be in the same grouping: was reprogrammable but not general purpose, seems to be an earlier use than ENIAC of valves, etc. And I've just checked: someone has made a "first" claim at Tommy Flowers (looks OK). And History of computing hardware#1940s: first electrical digital computers describes it as one of the "first-generation computers", along with the Z3 and ABC.

Btw: there's also a bit that seems to be missing from all the WP accounts: Tommy Flowers built the Colossus against the wishes of his bosses, who didn't believe a worth-while machine could be made using valves, because of their unreliability when switched on and off. Flowers took out personal loans and built the thing in secret (from early 1943), to present as a fait-accompli. I know this from first big documentary on Bletchley but don't have books so can't add it - if someone has the refs, some work is needed here. But this has nowt to do with ENIAC.

Why am I a Colossus cheerleader (other than being British)? Because I'm a Tommy Flowers fan: he spent his own money (never fully recompensed) and personal loans to build a machine that had been rubbished by his superiors; shortened the war by, they say, 2 years; dismantled his creation and burnt the plans with his own hands; swore never to speak of the machine's existence; and spent the remainder of his life as an obscure Post Office technician hearing other people credited for "inventing" computers. During the interview for "Station X" (doc), this 90-year-old man was almost in tears describing how he destroyed his machine. Again, I'm on the wrong page for this, but you can see why I bang the drum when histories ask: Was it the ABC or ENIAC?! JackyR 17:33, 16 March 2006 (UTC)

Image caption vandalsim?

The image caption reads "Two women about to get boned by the ENIAC" - is this some term I am not aware of, or is it somebody trying to be funny?? - User:Dawidl

I thought that, but checking the history shows it was vandalism, I've reverted the page. 155.198.63.111 15:10, 9 May 2006 (UTC)

Comparison?

The idea of the stored-program computer with combined memory for program and data was conceived during the development of the ENIAC, but it was not implemented at that time because World War II priorities required the machine to be completed quickly, and it was realized that 20 storage locations for memory and programs would be much too small.

Currently, this is under Comparison, but it doesn't do much comparing. I'd be more comfortable if either (a) context were added to compare to the other machines (I can't do ths bit), or (b) the sentence went back into Description. Opinions? JackyR 01:08, 17 February 2006 (UTC)

Come on guys, the thing about Konrad Zuse's Z3 not being electronic because it used relays is a joke! The only difference between a vaccuum tube and a relay is that one is mechanical and one is solid state. The Z3 is what we would call the first modern day computer, bar none. The Z3 needs to be featured more prominently and explicitly stated in this article that is was the first. JettaMann 16:45, 10 July 2006 (UTC)

No, this article is about ENIAC. The main material about the Z3 should be under Z3, and the main comparison of the different machines should be under History of computing hardware, as a daughter page if necessary. That article currently needs a little structural work, and a feature-by-feature comparison job of the early machines would be a plus, so if you felt like wading in that would be wonderful. JackyR | Talk 17:49, 10 July 2006 (UTC)

I agree actually. The comparison section seems fine as it does in fact mention the Z3, which is all that is required. JettaMann 17:25, 27 July 2006 (UTC)

History

Do we know if ENIAC still exists? What happened to it after the 1950s? Are its pieces in a museum? Pictureuploader 22:11, 7 March 2006 (UTC)

Several portions still exist. The University of Pennsylvania where it was built has four units operational (the Cycling Unit, the Master Programmer Unit, an accumulator, and one Function Table), the Smithsonian has some of it, the Computer History Museum has one Function Table Panel (on loan from the Smithsonian). -- RTC 07:20, 16 March 2006 (UTC)

Much of the ENIAC still exists; not all of it has been restored for display, and of the pieces that have been, the Smithsonian is not aware of the locations of all of them. An exhaustive worldwide search for ENIAC components would be a worthwhile endeavor, one which I attempted at one point but proved to be overwhelming. Robert K S 14:39, 2 August 2006 (UTC)

Comparison table

Table is great: it should also be at History of computing hardware, where this subject should be expanded on (any volunteers?). But please don't try to simplify ENIAC's particular "first" claim - you end up with a statement that is untrue. (It's not true, for example, that ENIAC "combined for the first time both high-speed and rudimentary programmability", as the Colossus was both of these and was fully functional earlier.) JackyR | Talk 22:36, 30 July 2006 (UTC)

The simplest claim of priority would be "first general-purpose electronic digital computer," a designation agreed upon by Arthur and Alice Burks. Robert K S 15:50, 2 August 2006 (UTC)

Nice to see the table to make comparison easier. As I mentioned on Talk:Computer#Comparable_dates I think the dates need to be made as comparable as possible. In particular, the ABC date of 1937-42 makes it look much earlier than the detailed information suggests it actually was. Making the date column heading more specific as "Shown working" would, I think, put the ABC date at around 1941. Including 1937, when it was only at the concept stage, doesn't look comparable to the other dates - concept dates for the other computers are earlier.

I think it would also be nice to have a column saying what each of these machines was first to do. They are all tremendous achievements and it would be good to say in two or three words what "first" each one achieved. Adrian Robson 10:14, 3 August 2006 (UTC)

Currently we're using the "first" designation used at the Moore School (#ENIAC designed and used to be Turing-complete? above). I'd be inclined to stick with that, as it's very specific. I believe "general purpose" is actually shorthand for "theoretically Turing-complete", rather than "could in practice be used for any purpose" (pls correct me if I've got that wrong): so let's use the technical designation rather than the potentially misleading shorthand.

I'd also not be keen on a "first in what" column, as it gives the impression of single track progress, whereas early computers experimented with a range of solutions, some of which were later abandoned (eg ENIAC's use of decimal rather than binary puts it). And to put ENIAC's "first" in context, one would have to include EDVAC, EDSAC and LEO - really beyond the scope of the ENIAC page but very proper for the History of computing hardware.

Overall, the purpose of these pages on Wikipedia is to give a thorough, neutral-as-possible view of the different machines in the history of computing, and any inter-relations, whereas the purpose of some of the source material may be to champion a particular machine and prove it was somehow the "best". JackyR | Talk 14:08, 3 August 2006 (UTC)

Jacky, thanks for this clarification on your interpretation of "general-purpose." I can see how it may be more ambiguous than the terms currently being used. Robert K S 15:19, 3 August 2006 (UTC)

Honeywell v. Sperry Rand Ruling summary

"...voided the 1964 ENIAC patent as derivative of John Atanasoff's ABC Computer."

Other than the (double!) redundancy at the end of this remark, it is misleading as it is a simplification of the ruling, omitting what were actually numerous reasons cited for the invalidation of the ENIAC patent (among several other Eckert-Mauchly patents). If Blainster does not object, I will reword. Robert K S 14:34, 2 August 2006 (UTC)

The phrase could be improved. I'm OK with "John Atanasoff's computer" or the "Atanasoff Berry Computer" instead of the the "ABC Computer". Just saying "Atanasoff's ABC" leaves the reader in a quandry. But the intro paragragh should be kept very simple. The details are present in the articles on Mauchly and Atanasoff, where they are more appropriate. We should just mention the patent problem and link to details in other articles. --Blainster 18:20, 3 August 2006 (UTC)

I agree. As-is, the wording is not controversy-neutral, which the article should try to be. Ultimately, perhaps, there should be a Honeywell v. Sperry Rand page and any article will just be able to link to it with a minimum of explication and be done with it. Any further duking out over the interpretations of the ruling can be done there. Robert K S 18:59, 3 August 2006 (UTC)

Your recent wording is fine on its own, except that it eliminated the WP internal links to existing discussion of the court case (eliminating the external link from an introductory par. is a good move, though). That prevents a reader from learning more about it. We should either redirect the redlink to an existing article, create the caselaw article, or otherwise link to the WP info about it. --Blainster 22:19, 3 August 2006 (UTC)

Caselaw article created. Please feel free to improve it. I hope additions and corrections will make the article more neutral rather than less neutral. It is not my intention to have created a place for controversy to blossom into vandalism, reverts, airing of unsubstantiated claims, etc. Robert K S 08:51, 5 August 2006 (UTC)

Court case

I believe a section should be added highlighting the court case and the fact that for a significant period of history, it was believed to be the first modern computer. The related section on the Atanasoff–Berry Computer#Controversy page would do for a simple copy and paste job! Would a Controversy section do the trick? 220.239.157.232 05:01, 2 September 2006 (UTC)

An entire article discusses this (linked in the third paragraph of this article). Please see Honeywell v. Sperry Rand. --Blainster 22:25, 6 September 2006 (UTC)

Reference required

"As of 2004, a chip of silicon measuring 0.02 inches (0.5 mm) square holds the same capacity as the ENIAC, which occupied a large room." Can this be supported? Colin99 21:59, 26 March 2006 (UTC)

The "ENIAC-on-a-chip" built by the Moore school in 1997 was 7x5 mm with 175,000 transistors [1]. So the statement can be changed to reflect this unless someone can update it to reflect additional "shrinkage" in the last nine years. --Blainster 19:17, 11 April 2006 (UTC)

"Mauchly had borrowed some ideas from the Atanasoff-Berry Computer." This statement has been challenged in enough places that it demands support. Unless someone can really state, in a simple sentence, one definite "idea" that was borrowed, I suggest we delete this. (hint: "computing with tubes" is not a specific or a stealable idea.)--Zebbie 16:39, 11 April 2006 (UTC)

The ABC proponents will start quoting to you from the court case that found in favor of Atanasoff. There are a lot of problems with that court case, and with the judge's conclusions, but on its face it does support that statement, despite Mauchly's later denials. -- Gnetwerker 18:54, 11 April 2006 (UTC)

Here is the relevant section from the judgement, which was never challenged by Univac/Sperry: "3.1.3: Honewell has proved that the claimed subject matter of the ENIAC patent relied on in support of the counterclaim herein is not patentable over the subject matter derived by Mauchly from Atanasoff. As a representative example, Honeywell has shown that the subject matter of detailed claims 88 and 89 of the ENIAC patent corresponds to the work of Atanasoff which was known to Mauchly before any effort pertinent to the ENIAC machine began." – Honeywell Inc. v. Sperry Rand Corp. et al. October 19, 1973. (emphasis added)

The main idea that Mauchly "borrowed" is the idea of a digital computer. The court showed that Mauchly had no concept of a digital computer prior to seeing the ABC. He had only worked on analog devices up till then, and throughout the years leading up to and through the trial, he was never able to find a witness or provide a document showing that he had broached the concept of digital computing prior to his visit to Iowa in June of 1941. This does not denigrate Mauchly and Eckert's fine work on the ENIAC, which used many concepts distinct from the ABC due to their engineering prowess. --Blainster 19:17, 11 April 2006 (UTC)

Thanks for the answers, but I can't get behind them. I know what the court said, and it is the judge's inability to articulate a definition of computer that has caused much of the controversy. No doubt you are aware of the lengthy debate at [Computer | Talk] about the definition of computer. In brief, if the judge meant "a digital computing device" then the category includes Babbage, Stibitz, and every IBM tabulator. If the judge meant "a machine designed for manipulating data according to a list of instructions known as a program" (Wikipedia's definition) then Atanasoff's machine does not qualify as a computer.

I'm looking for the specific definition of a computer feature that originated in the ABC and is found in the ENIAC. For example, "Mauchly had borrowed the idea of using tubes to perform binary addition" or "Mauchly had borrowed the idea of using a series of digital pulses to represent a number" or "manipulating data according to a list of instructions." Something concrete.

In other words, ignore the judge. With 60 years of hindsight, what important idea was borrowed from the ABC and included in the ENIAC? Is it claim 88? --Zebbie 16:45, 20 April 2006 (UTC)

I think your question was answered. Without meeting Atanasoff, examining the ABC, and discussing Atanasoff and Berry's invention with them, Mauchly wouldn't have invented ENIAC, at least not as early as he did. Mauchly's own words in a letter to Atanasoff after his June 1941 examination of the ABC: "A number of different ideas have come to me recently anent (regarding) computing circuits, some of which are more or less hybrids, combining your methods with other things, and some of which are nothing like your machine." Ken Estabrook 05:27, 1 July 2007 (UTC)

I was always under the impression that the two had met after Atanasoff approached Mauchly. Why would he locate Mauchly after a lecture and ask him to come look at his machine if Mauchly had not already proposed similar ideas? Why would he suppose Mauchly would have insights into this type of computing if it was not something Mauchly was known to be an authority on, or at the very least interested in? It would be strange to travel to visit another professor, listen to his lecture, and then ask his advice on a topic that you have no reason to believe he could help with. —Preceding unsigned comment added by Lusid42 (talk • contribs) 20:11, 23 October 2007 (UTC)

Programmability and von Neumann

The third paragraph in this section is largely wrong. Von Neumann never intended to take credit for himself, and it was well known by Eckert and Mauchly that he was writing a report. They gave him the go ahead. It was a draft and was released prematurely by the Moore school (Goldstine specifically) because of the large amount of doubt in other academics that the project would succeed. Having such an influential man behind it added a lot of credibilty. This section cites no sources, and needs to be edited drastically. Sadly I have no time at the moment.

See Mauchly, "Amending the EDVAC story" for verification of the above. Stuk 20:08, 10 November 2007 (UTC)

Well, I've tried to pare down some of the unsourced "POV"-y material. Feel free to take another stab at it/provide sourcing. Robert K S 01:52, 11 November 2007 (UTC)

"Rewiring to reprogram" vs. Turing-completeness of ENIAC?

Until 1948, ENIAC required rewiring to reprogram, like the Colossus.

How does this "rewiring to reprogram" relate to the Turing-completeness of ENIAC? Unfortunately, I can't find any information about this in the article, except that the comparison table says "yes". - Was this proven by anyone, like the Turing-completeness of the Z3? Without Turing-completeness you can hardly speak of a true general purpose computer. MikeZ 08:41, 13 August 2006 (UTC)

This thought occurred to me too when I read the article, so I was happy to find the question already raised here. However I don't believe it is problematic if understood as follows. A Turing machine U is universal when there exists a convention for encoding any Turing machine T as a word e(T) such that for any Turing machine T, when U is started on a tape containing e(T) followed by w, U halts if and only if T halts when started on a tape containing w, and if U halts it leaves the tape in the same configuration T would have left it in when it halted. (So U's task ostensibly is to decode e(T), run T on w, and if T halts clean up all the tape except for T's output.) Universal Turing machines exist, however most Turing machines are not universal in that sense: their finite state control is typically "prewired" to perform some more particular task. Now if it is possible to rewire the ENIAC with a program for reading a list of tuples describing a Turing machine and then emulating that machine, the ENIAC with that rewiring would be as universal as a universal Turing machine. Thus if we understand the ENIAC as a class of computers, and a particular wiring of the ENIAC as making it a particular computer, it is fair to say that the ENIAC is as universal as a Turing machine. From that perspective one can no more say that all rewirings of the ENIAC are universal than that all Turing machines are universal. The crucial distinction from the Colossus is that while the ENIAC was designed to be rewired by its users as a normal part of its operation, the corresponding rewiring of the Colossus would have to be done "at the factory" with a soldering iron and additional vacuum tubes. If you could claim that the Colossus was programmable in that sense, you could equally well call a sedan a convertible because one can saw off the roof. If the user of the ENIAC can program it to read Turing machine tuples, the ENIAC is as universal as the Turing machine. --Vaughan Pratt (talk) 20:29, 26 March 2008 (UTC)

The Story Interview

The Story recently interviewed Mrs. Bartik, and may provide some additional information or references for this article. Jouster  (whisper) 18:43, 16 January 2008 (UTC) This post restored after vandalism Cuddlyable3 (talk) 19:34, 29 November 2008 (UTC)

ENIAC designed and used to be Turing-complete?

Someone wrote: "ENIAC also differed from earlier calculating devices in that it was designed and used to be Turing-complete — that is, a truly universal computing device". What is the evidence for this claim? ENIAC was meant to solve differential equations etc., and to my knowledge its builders were not interested in Turing-completeness, although maybe they got it as a by-product. Are there any documents supporting this statement? Z3 16:33, 22 Mar 2004 (UTC)

The issue is not documenting the statement, but tuning it to say what it surely intends. The point of the Turing-machine model is that it takes almost nothing to be Turing complete, but that near-nothing is the important distinction between a calculator and a fully programmable device. Yes, they cared about it: they were responsible engineers who knew of Turing's work, and they would not consider a programmable device short of T-C one, unless they already knew exactly what program they were going to use.

Numerical approximation, including diff-equ, was a well-researched subject, and they knew they would do a lot of optimizing of algorithms. T-C is not all that they would have considered bcz efficiency of computation is also crucial, but they absolutely wanted to know they weren't going to get bitten from behind when they tried to tune the thing up, by the one thing they knew how to guarantee: every flow of control is possible, even if not efficient. If this is not clear, it's a sign of the need for an article on something like Foundations of engineering design. --Jerzy(t) 20:14, 2004 Mar 22 (UTC)

Well, the reason why I am bringing this up: Konrad Zuse's earlier Z3 also was Turing-equivalent, but back then Zuse and others didn't care and didn't even know; they just wanted to be able to program their machines for the typical engineering applications of that era. None of these old machines had enough storage to exploit Turing completeness, as suggested by the current version of the ENIAC article. As far as I know, none of them was used to implement compilers translating one language into another (language independence is the central aspect of Turing completeness). I guess one should delete this. Z3 09:31, 23 Mar 2004 (UTC)

The sign at the Eniac museum at the Moore School of Electrical Engineering states that Eniac was the world's first "Electronic Large-Scale, General-Purpose Digital Computer", and I believe that all the qualifiers are necessary to make the statement true. Colossus was an electronic, large-scale digital computer, but not general-purpose. The Z3 was a large-scale general-purpose digital computer, but not electronic. The Atanasoff-Berry device was an electronic digital computer, but was not general-purpose, and perhaps not large-scale. I cannot think offhand of an electronic large-scale general-purpose non-digital computer. -- Dominus 14:44, 4 Jun 2004 (UTC)

ENIAC was general-purpose, but not [Turing complete]. This is because no finite computational device meets that standard. Lawdroid (talk) 10:16, 1 February 2009 (UTC)

The above statement limits the term "Turing complete" in a way that (a) is not in line with the term's typical use and (b) renders the term meaningless as applied to practical devices. When the term "Turing complete" is applied to a computing device, it means that a computer would have all the capabilities of a universal Turing machine given infinite memory store. The Turing completeness article covers this. Robert K S (talk) 08:21, 3 February 2009 (UTC).

Programming

Surely von Neumann's early ENIAC programming was on the atom bomb not the hydrogen bomb? --TedColes (talk) 11:46, 8 July 2009 (UTC)

It was the hydrogen bomb, and von Neumann wasn't involved. The program was run by Frankel and Metropolis. Robert K S (talk) 12:45, 8 July 2009 (UTC)

So are you saying that GroveGuy's recently added paragraph is quite wrong? --TedColes (talk) 15:36, 8 July 2009 (UTC)

Sirs: I was reading a book about the H-bomb and saw references to ENIAC. I looked at the wiki article and other articles about ENIAC and was amazed that there was no mention of the H-bomb. I read Goldstine and came to the conclusion that when the Los Alamos people became aware of ENIAC they kind of took over the project - not in terms of design or construction - they simply wanted to use the machine. Indeed, in Goldstine there is no mention of any use of the computer during construction except to run the H-bomb problem. Even after completion most of the production time was spent on H-bomb calculations. Ballistic calculations just got the leftover time slots. When ENIAC was completed its secret classification was removed. But the H-bomb was still ultra secrect. I believe that is why the literature on ENIAC doesn't mention the H-bomb. So, the sentences I added to the ENIAC article were to explain the relationship of ENIAC to the H-bomb project. GroveGuy (talk) 22:03, 8 July 2009 (UTC)

I moved GroveGuy's addition on the H-bomb calculations from the lede paragraph because the intro is already too long. He objected so I added a brief mention back in. The info in this external link on the article page attributed to the Computer History Museum says that the program run at the public unveiling in Feb 46 was a ballistics trajectory program. But of course how would anyone have known? --Blainster (talk) 05:08, 9 July 2009 (UTC)

Bainster - I am sorry you characterize my words as "complaining". I was trying to use rigorous logic to support having my contribution remain in the lead paragraph. It really doesn't belong in the programming section. As for what was running on the computer on the day it was unveiled, I think they were telling a white lie. This article says they postponed moving ENIAC to Aberdeen (until November 1946) in order to complete the test of the H-bomb calculations. A million punch cards - it boggles the mind. GroveGuy (talk) 08:41, 9 July 2009 (UTC)

Metropolis and Frankel ran the hydrogen bomb problem from December 1945 to February 1946. AFAIK the whole ENIAC construction staff and programming team was involved with the problem inasmuch as the problem represented a shakedown cruise of the ENIAC architecture, though no one outside of Los Alamos was apprised of the calculations' significance or the meaning of the result. With hostilities concluded, the firing tables were less of a priority than the Los Alamos work, so it was only natural that the ENIAC would be put to this use as a shakedown problem. It was assuredly not the first test problem run on the ENIAC hardware (e.g., Lehmer ran one of his number sieves over the previous Thanksgiving weekend) but it was probably the first major problem of military importance. Little is known about the calculations since they remain unreleased, but several sources, including McCartney (page 104), indicate that the results of the problem run on the ENIAC showed that Teller's proposed formulation was flawed and prompted a revised theory. Judge Larson's decision invalidating the ENIAC patent was justified in part on this H-bomb shakedown cruise, calling it a "non-experimental public use" that occurred prior to the patent filing's critical date. Hope this clarifies. Robert K S (talk) 14:07, 9 July 2009 (UTC)

Size and Reliability

The article says ENIAC occupied about 680 sq. feet. However, the dimensions provided would put it at around 2000 sq feet, and the specifications in a cited article indicate that it is 1800 sq feet. Could someone resolve this significant discrepancy? I have read claims that someone patrolled with a shopping cart of vacuum tubes, but current reports of tube failure make that seem unlikely, if anyone has a definitive statement I would be interested. Jgrudin (talk) 11:22, 15 November 2009 (UTC)jgrudin

Electronic Numerical Integrator and Calculator?

There have been several attempts today to alter the acronym expansion in the first paragraph of the article, replacing "..and Computer" with "...and Calculator." I'm not sure if this is out of a desire to denigrate the ENIAC, placing it in a lower status than later stored-program machines, or if the editor came across this expansion in bona fide contemporary sources (of which I'm sure several examples exist); in either case, it is assuredly incorrect. The earliest documents refer to the machine as "...and Computer," or in some cases, "...and Computor," which seems to be a misspelling preferred by Mauchly, and the one that made it onto the plaque affixed to the ENIAC, a fact that I don't think bears mentioning in the article proper. In hopes of remedying the vandalism I have added a footnote reference to the Goldstine book. While Goldstine was wrong about several things, on this matter he should be considered authoritative, as he was present at the BRL meeting that proposed the machine and had a clear memory of the device's name (whether or not it originated with Col. Gillon). Robert K S 02:23, 16 August 2006 (UTC)

Someone tried this again today. I checked the cited reference (Goldstine, Herman H. (1972). The Computer: from Pascal to von Neumann. Princeton, New Jersey: Princeton University Press. ISBN 0-691-02367-0.), which unvaryingly uses "computer". For example, "At this meeting, Gillon named the proposed machine the Electronic Numerical Integrator and Computer and gave it the acronym ENIAC." (Page 150). Goldstine also cites many earlier documents that all use "computer". For example, Goldstine, H.H; Goldstine, Adele (1946). "The Electronic Numerical Integrator and Computer (ENIAC)". Mathematical Tables and Other Aids to Computation. Vol. II. pp. 97–110.. —Mark Dominus (talk) 14:23, 1 July 2010 (UTC)

Speed

Does anyone know how fast the ENIAC was in MHz, or a comparable unit? Hey, my warranty didn't run out after all! 20:42, 14 January 2007 (UTC)

That information is in the third paragraph of the description. The ENIAC machine cycle speed cannot be compared directly to modern CPU clock cycles because (for the problems it was designed to run) it was so much more parallel than today's typical architectures. Possibly the closest comparison would be to the Cell microprocessor found in the PlayStation 3, which uses a master Power Processor Element to control, in parallel, its eight Synergistic Processor Elements, which is broadly analogous to the way the ENIAC's master control panel could control its many accumulators, high-speed multiplier, divider/square-rooter, etc. Robert K S 21:00, 14 January 2007 (UTC)

Then how about math operations per second? Any kind of measurement is better than none at all. —Preceding unsigned comment added by 68.3.63.127 (talk) 04:37, 10 November 2010 (UTC)

Tube failures again

"Most of these failures, however, occurred during the warm-up and cool-down periods, when the tube heaters and cathodes were under the most thermal stress"

Warmup failures are usually addressed with slow-start circuitry. Today, silicon regulators and timing relays for tube gear are cheap, and back in the fourties they could use variacs for slow-starting. Fairly big variacs indeed but was still practical.

I'm sure that they had to consider variacs - but what happened then? Did they actually try it, and if yes - did it improve anything?

Cheers, East of Borschov 08:42, 19 November 2010 (UTC)

(1) I've never heard of them doing this, but I can't answer definitely. (2) How do/did Variacs work? "Tube gear" today tends to mean amplification, wherein the amplifiers have one or a few tubes. Here we are talking about counting and switching circuits with many tubes. Would implementation of Variacs still be practical? Are Variacs devices that are likewise susceptible to breakage and would thus introduce just yet another point of failure? Robert K S (talk) 22:53, 19 November 2010 (UTC)

If you haven't already, try searching for Variac. We're speculating here, we'd need a reference that variable autotransformers were used, or not used, with early computers. I've seen a variac as big as 30 kva, which would run a large number of 6SN7 heaters. --Wtshymanski (talk) 01:29, 20 November 2010 (UTC)

Coders or Programmers

I actually met two of these ladies who worked on Eniac. I was studying computers, and they were introduced in a lecture. They said they were Coders and that there was no such thing as Programmers at the time. They said jokingly that Programnming was invented in the 50s so that men could do computing as well. They explained that women were far more intelligent and it was they who did coding. The coding consisted of series of 0s and 1s, whereas programming was issuing instructions which were translated into 0s and 1s. Wallie (talk) 11:50, 14 April 2009 (UTC)

Given that it was 6 women who were writing code for quite some time for ENIAC, there is very little about them in the article. How can the 6 people who used it day-to-day be so marginalized? Wtf people? Seems sexism is alive and well after all...

Cite it and write it. If you know this to be a problem with the article, then you can fix the article. --Wtshymanski (talk) 14:06, 24 January 2011 (UTC)

When was ENIAC first shown to work?

I came to this article hoping to find out when ENIAC was first working but I'm confused by the intro. Can someone help clarify this?

First it says ENIAC was built by the Army in 1942. (Does that mean it was working then?) Then it says it was commissioned in 1943. (There's a link from "commissioned" to the explanation of the meaning of "putting a ship into service". But does this mean it was put into service; or is it the other meaning of "commissioned", that an order was placed?) Then it says it was constructed by Penn's Moore School of Electrical Engineering from mid-1944. (Does this mean it was working in 1944?) Then the intro says it was "unveiled" in 1946. (Does this mean it was working for the first time in 1946?)

If anyone can come up with an answer that would be a great help? —The preceding unsigned comment was added by 82.35.173.37 (talk • contribs) 15:14, June 6, 2006 (UTC)

Some errors recently crept into the article. ENIAC was built for the Army by engineers at the Moore School of the University of Pennsylvania. The 1942 date was incorrect. The article should have said that the project was approved for construction in mid-1943. Design work took about a year, and the computer was built over 18 months from mid-1944 to the end of 1945. Final assembly was undertaken in the fall of 1945. For example, the divider and square root modules were installed in October 1945. Parts of it were working prior to that, but you could say the full machine was not functional until the last couple of months of 1945, with formal acceptance at the February 1946 public event. --Blainster 10:46, 7 June 2006 (UTC)

Incidentally, the "February 1946 public event" Blainster references above took place on February 15, not 14 as cited by McCartney and possibly several other secondary sources. Primary sources, such as the event invitation, show that nothing was scheduled for the 14th. Robert K S 15:48, 2 August 2006 (UTC) I've revised my thinking on this since my remark (stricken, above). The formal dedication was certainly the 15th but there may have been a demonstration event on the 14th, since that is the date remembered by the women computers, according to Kathy Kleiman, who extensively researched and compiled their stories. To my knowledge, no written evidence of an event scheduled for the 14th has surfaced, but I would be interested in it should someone come across a written schedule of dedication events. Robert K S 20:29, 23 October 2007 (UTC)

The issue of when ENIAC was first shown to work was an important one in Honeywell v. Sperry Rand, and Judge Larson made findings on this matter (see especially 1.1.3). At this point I cannot find any contemporary or primary source which supports a Feb. 14th unveiling date. Rather, Feb. 14, 1946 was the date on which ENIAC's programmers worked frantically to get all of the bugs out of the system for the public dedication the following day. The program for the unveiling, which was entered into evidence in the patent case and is therefore available in the microfilm record, shows that the date was February 15. The judge found no disclosure activity took place on February 14th. Robert K S (talk) 16:04, 31 January 2011 (UTC)

Here is a New York Times article indicating that the machine was "announced" on the evening of February 14, 1946, even though the formal dedication and public demonstration was not until the following day. Robert K S (talk) 16:29, 31 January 2011 (UTC)