Talk:IAS machine

Early comment
IAS machine words could store numbers as well.--Cmd353 18:30, 29 December 2006 (UTC)
 * as well as what? Perhaps the confusion is that the numbers it stored could be interprested as either data or instructions? W Nowicki (talk) 22:13, 26 May 2011 (UTC)

MAY NEED TO EDIT ARTICLE: This IAS Wikipedia article says IAS machine used two's complement arithmetic, while this source specifically states, "the actual [IAS] machine did fixed point calculations, not two's complement integer calculations" Jim.Callahan,Orlando (talk) 20:00, 15 November 2013 (UTC) http://www.cs.unh.edu/~pjh/courses/cs611/03spr/ias.html

IAS Machine
What does the acronym "IAS" serve for? —Preceding unsigned comment added by 130.127.82.252 (talk) 12:21, 4 September 2010 (UTC)


 * Institute for Advanced Study as given in the lead.

IAS architecture computers
Why BESM is mentioned here? Any sources to prove it was an IAS machine? -komap (talk) 10:51, 30 January 2011 (UTC)


 * The IAS history page seems to be the source. Obviously a citation addition opportunity. W Nowicki (talk) 22:13, 26 May 2011 (UTC)

Dubious

 * The computer was built from 1942 until 1951...

Alas, one Smithsonian page http://americanhistory.si.edu/collections/object.cfm?key=35&objkey=31 says this, but seems incorrect. Most other sources, including the other Smithsonian page and the IAS page, make it clear von Neuman worked at other places during the war and did not start the IAS machine until later, about 1945 or 46. It was probably confusing this with the ENIAC? W Nowicki (talk) 22:56, 26 May 2011 (UTC)

George Dyson, Turing's Cathedral
George Dyson's book "Turing's Cathedral" has been reviewd in the UK Sunday paper The Observer by John Naughton on 26 Feb 2012. This affects this article. It now seems essential to be able to show sources predating late 1945. Waffent (talk) 00:35, 27 February 2012 (UTC)

Derivatives
The list of derivatives is long, the requirements for being a derivative - or not - are not obvious. For example, if the IBM 701 is a derivative, then are not the IBM 704, 709, 7090, ... derivatives? What are the requirements to be, or not to be? In http://ds.haverford.edu/bitbybit/bit-by-bit-contents/chapter-five/5-8-the-ias-computer/ is: "The influence of von Neumann and the wisdom of his approach was such that the IAS machine – a stored-program, parallel processor – became the paradigm of modern computer design, and most computers built since the late 1940s have been “von Neumann” machines."

The IAS and "Von Neumann" machines are parallel at the bit level, but serial at the word level. So, all the bits in an integer would be added in parallel, but the operations above that level would be serial (the von Neumann bottleneck). This is explained in "Turing's Cathedral". Von Neumann architecture also refers to mixing data and programs in one shared memory instead of separate memories for programs and data (as in Harvard and dataflow architecture computers). Sharing the memory was very important when the cost of memory was a million times more costly than it is today (compare cost of 1 or 2 kilobit (Kb) Williams tube in 1950s dollars with cost of modern 4 gigabyte (GB) RAM memory card in 2010 dollars). Jim.Callahan,Orlando (talk) 19:28, 15 November 2013 (UTC)

"The 704 also marked the transition to magnetic core memory." IBM 701 magnetic cores (magnets strung on wires) are VERY DIFFERENT (physically) from the Williams tubes (vacuum tubes) used on the IAS and 701 computers. Williams tubes were clever, but short lived innovation in computer memory that are characteristic of the IAS computers and close copies. Jim.Callahan,Orlando (talk) 19:00, 15 November 2013 (UTC)

Text states "Plans for the IAS machine were widely distributed to any schools, businesses, or companies interested in computing machines, resulting in the construction of several derivative computers referred to as "IAS machines," although they were not software compatible in the modern sense." That seems to say that receiving a copy the IAS plans was sufficient to classify constructed machines as derivative. Is that correct?

The list Von Neumann architecture has almost the same members as the derivative list here. Should they be the same?? Should this small article be replaced by a redirect to Von Neumann architecture? Special:Contributions/99.65.176.161|99.65.176.161]] (talk) 16:10, 15 November 2013 (UTC)

In the broadest sense, every modern computer (tablet and phone) inherits from the work of Kurt Gödel/Turing/Atanasoff–Berry Computer / Claude Shannon/ENIAC/EDVAC and IAS. Jim.Callahan,Orlando (talk) 18:32, 15 November 2013 (UTC) In the narrower sense of being a close copy, any computer (outside of the UK where the Williams tube was developed), including the IBM 701 that had the peculiar design feature of using "Williams Tubes" for main memory and used 36 bit words could be considered derivative of the Institute for Advanced Study (IAS) computer at Princeton. The Williams tube was a CRT display tube used in oscilloscopes (larger CRT tubes were used for the TV screen of standard definition TV sets) that was repurposed to be used as a computer fast/short term memory storage device. This memory technology is different from the mercury delay lines proposed/used at the University of Pennsylvania, Moore School ENIAC and EDVAC computers and the magnetic core memory used in the MIT_Whirlwind computer and later computers until the development of transistorized memory. The IBM 701 had a 36 bit word size, like the IAS computer and unlike word sizes of the later IBM 7030 Stretch (64 bit) and IBM_System/360 line of computers (32 bit). So, yes the IBM 701 resembles the IAS computer a lot more than the later IBM Stretch or IBM 370 that were genuine, major innovations by IBM. IBM's innovations on the 701 were the power supply, physical layout, external switches and the cabinet which basically commercialized the initial laboratory prototype IAS at Princeton. See for example, "Turing's Cathedral" and "The Computer from Pascal to Von Neumann".^^^^


 * Above is "any computer (outside of the UK where the Williams tube was developed), including the IBM 701 that ... and used 36 bit words could be considered derivative of the Institute for Advanced Study (IAS)" and "The IBM 701 had a 36 bit word size, like the IAS computer". But IAS machine states "The IAS machine was a binary computer with a 40-bit word" and Bashe ..., describing the 701, IBM's Early Comuters p.138 states ...IAS computer ... departed from it ... word length would be 36 bits instead of 40.  Leaving me confused about word lengths being a criteria for "derivative".


 * There are several comments about technology change, especially memory. A statement such as "All computers built using Williams tubes are derivatives of the 1st such machine" would lead to parallel - and unreasonable - statements about vacuum tubes, transistors, etc. :Can a statement such as "Computers having 4 or more of the following 6 features are considered derivatives of the IAS machine"? be written?


 * Better, why not delete the list entirely? The list Von Neumann architecture meets most needs.  And deleting the list would get back to my other question - should this article be merged into the Von Neumann architecture article? 99.65.176.161 (talk) 04:19, 18 November 2013 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 2 external links on IAS machine. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20120604211339/http://www.computer50.org/mark1/new.baby.html to http://www.computer50.org/mark1/new.baby.html
 * Added archive https://web.archive.org/web/20120604211339/http://www.computer50.org/mark1/new.baby.html to http://www.computer50.org/mark1/new.baby.html

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 00:17, 10 November 2017 (UTC)

Used for
[http://www.dtic.mil/docs/citations/AD0694608 DIGITAL COMPUTER NEWSLETTER. VOLUME 5, NUMBER 2]: "[...] large number of meteorological forecasts, several hydrodynamical calculations, a number of astrophysical calculations, some work on the determination of quantum-mechanical wave functions, a mathematical application of biology, and some determinations of the eigenvalues and associated functions of symmetric matrices."

[http://www.dtic.mil/docs/citations/AD0694609 DIGITAL COMPUTER NEWSLETTER. VOLUME 5, NUMBER 3]: "[...] some problems in physical chemistry, and a variety of other calculations." --89.25.210.104 (talk) 19:06, 24 May 2018 (UTC)

Dubious priority dissection
The general organization is called Von Neumann architecture, even though it was both conceived and implemented by others.

I've been reading Turing's Cathedral by George Dyson, whose account is certainly authoritative on this matter, as far as it goes.

At the very least, von Neumann contributed an early and thorough grasp of Turing's foundational paper, and it seems that he maintained a steady hand on the vision for the machine as a whole, as well as making vital contributions to the ongoing bull sessions concerning just about every key design element. He also showed up every day to interrogate and lucidly debug the debuggers (who almost invariably found their focus on the challenge at hand improved by this brisk interaction).

On the other hand, von Neumann was certainly a force-of-nature gadfly around the edges (definitely not your standard oversight tooth fairy), and there's no shortage of credit to go around concerning almost every key contribution; it became quite a matter of contention later to disentangle priority on any aspect of its conception.

The problem here, is that the collective did not operate at the stately speed of priority-centric academic collaboration. The typical time delay between bull sessions and soldering irons was measured in days, not months; often circuits were wired (at least partially) before parts arrived, so even the procurement process was frequently speculated. And I'm sure many guesses were reversed and revised as theory advanced upon practical reality's main force (last paragraph).

This analysis of the priority situation violates the fundamental law of computer science: that design X and Y and Z in isolation, does not yield a finished complexity of X+Y+Z. Successfully interconnecting napkin(X) to napkin(Y) to napkin(Z) is the other 90%. Only that's not even the whole story, because at this point in the design and implementation of computers, transmission line artifacts, induced electromagnetic interference, and (crappy) statistical tube behaviour envelopes were not yet codified for (nearly) flawless digital operation.

I came away from Dyson's book with the impression that we now know this as the von Neumann architecture because von Neumann had the deepest and clearest understanding of Turing's seminal paper, and that for this reason, he was a steady hand in steering the many bull sessions to a minimal generality (expedience often generates ill-conceived design carbuncles you regret later). So they were also battling the as-yet unnamed second law of computer science: legacy happens.

It's completely ridiculous to argue priority here by who achieved a tidy napkin drawing first, that being perhaps 1/3 of what was actually at stake here, in raw daring and innovation.

Probably as important as the architecture itself, the main concern was to implement the machine such that it didn't choke to death on its own electrical noise—almost certainly the second driving principle behind maintaining a posture of architectural minimalism at all cost (beyond Turing's theoretical work). Otherwise noise(X) + noise(Y) + noise(Z) could have easily achieved critical mass, and the whole thing might have wound up in the same mass grave as Babbage's irregular tooth friction.

If this picture isn't grasped in full, this article can only end up two letters short of a VWXYZ. And that's why von Neumann was so critical and charismatic to this venture: man universally recognized by his peers as least likely to end up two letters short of VWXYZ in the first half of the 20th century. Even Turing said, of the various American efforts, "only the IAS group has a real clue", which also reflected upon their outward communication, largely on the back of the gadfly express.

Third law of computer science: the team with the clearest priority chain will always finish second. &mdash; MaxEnt 15:56, 28 May 2018 (UTC)

IAS derivative(?) - Circle Computer
Circle Computer

THE CIRCLE COMPUTER

CIRCLE

Circle Computer: A General-purpose Digital Computer for Science & Engineering

Circle Computer: The low-cost general-purpose computer for science and industry

[http://www.bitsavers.org/pdf/onr/Digital_Computer_Newsletter/ DIGITAL COMPUTER NEWSLETTER. VOLUME 6, NUMBER 1], Jan 1954

[http://www.dtic.mil/docs/citations/AD0694613 DIGITAL COMPUTER NEWSLETTER. VOLUME 6, NUMBER 3] --89.25.210.104 (talk) 01:49, 30 May 2018 (UTC)