Talk:IBM 650

702
The 650 was not IBM's first business computer as was previously stated in this entry; that was the 702 (not the 701 that I mentioned in the edit summary of the main page. --Brouhaha 02:06, 24 Aug 2004 (UTC)

BLIS not BLISS
I definitely remember the name as BLIS - BLISS would have stuck in my memory. I have written the guy at Murdoch whose web site shows BLISS to try to find out where s/he got the name from. I guess we have to leave it as BLISS until I get an answer, but I used it briefly, and I am 99% it was BLIS! Ah! I see from his references - but I still think they are wrong! This is supported by [], [] Jpaulm 15:18, 18 September 2006 (UTC)

There also appears to be a language called BLISS - maybe this is contributing to the confusion! Jpaulm 17:34, 19 September 2006 (UTC)

I have the same recollection as you, although I never actually used it at the time. Changed the name here to match Murdoch on the assumption that I was wrong (it has happened more than once). Not sure you'll get an answer from Murdoch; I have the ACM CACM in my garage but getting to it is almost impossible so I can't check that reference. I'll change it back to BLIS and note the question.

The other BLISS is a DEC language, I've actually got that manual in my bookcase -- no idea why. 69.106.254.246 20:27, 19 September 2006 (UTC)

A Google search "Bell Laboratory Interpretive System" returns a few Japanese documents where "BLIS" can be seen in the Google search result. 69.106.254.246 20:53, 19 September 2006 (UTC)

Other IBM-650 programming tools
Runcible was interesting as a Fortran forerunner, it had just 3 types of variables: In, Jn & Xn where was a number from 1 to 50 (as there were 50 words per drum "track"),

There was also a Statistical Bell Interpreter, which instructions for statistical handling.x

Event: mexico is celebrating 50 years of computing as an IBM-650 was installed at the University of Mexico (UNAM) in 1958, I did program it, for seismic research.

MEXICO -- AGS -- --Dagofloreswi (talk) 05:05, 8 November 2008 (UTC)

IBM 650 picture of front panel
In the article on the IBM 650 the first photo doesn't appear for me, although it does for Mike who put it up there. All I get is a box containing the following text: "IBM 650 front panel, showing bi-quinary indicators", and 2 overlapping squares. If I click on the squares, I just get text, including what I guess to be Esperanto. Is anyone else having this problem, or am I the only one?! BTW I get it in both Netscape and IE. Thanks. Jpaulm 20:55, 17 October 2006 (UTC)

Revert edit re "Error Sense"
This text "One of the 650's most significant features was that, if processing was interrupted by a "random processing error" (hardware glitch), the machine could automatically resume from the last checkpoint instead of requiring the operators to restart the job manually from the beginning." has been reverted.

The 650 could be set, "Error Sense", to execute the instruction set in the console switches when an error was detected. That's it; anything else that happened was determined by the application's programmer - there were no 650 provided checkpoints (as the quoted text implies) and no automatic resumption. "Error Sense" was NOT "One of the 650's most significant features". 69.106.243.72 (talk) 02:45, 15 November 2008 (UTC)

Nomenclature
Is it appropriate to mention the common term "one and a half address computer"? "Sloptimization" (assigning the next instruct to an address 5 word times away instead of looking up instruction timings)? Shmuel (Seymour J.) Metz (talk) 20:36, 14 May 2010 (UTC)


 * Does anyone know a reliable source that uses the term "sloptimization"? If so, please add that source as a reference to the article "optimum programming". Thank you. --DavidCary (talk) 18:10, 13 December 2015 (UTC)

First mass produced computer in the world
The article claims that the 650 was the first mass produced computer in the world. It has a reference to back it up but it's rather weak (vendor claim, mentioned in passing with no further support). The IBM 701 article also claims that it was the first mass produced computer, although with no supporting reference. Which one is right?  Stepho  talk 21:50, 18 January 2021 (UTC)


 * According to Pugh, the 701 shipped before th 650; the distinction might be "mass produced" - there were only 18 701's shipped. Tom94022 (talk) 00:14, 20 January 2021 (UTC)


 * True, 19x 701 vs 2000x 650 according to each article. I'd hardly count 19 as "mass produced" . Our articles should at least be consistent among themselves.  Stepho  talk 23:09, 20 January 2021 (UTC)


 * Why don't u then correct the 701 to eliminate "mass produced." You might footnote this article noting 2000 as "mass produced." Tom94022 (talk) 23:30, 20 January 2021 (UTC)


 * I can. I just preferred to gather some facts/opinions first, in case I had missed something.  Stepho  talk 22:09, 21 January 2021 (UTC)
 * Perhaps the 701 was listed as "mass produced" because it wasn't a one-of-a-kind machine (it wasn't the first mass produced computer, but IBM 701 doesn't say it was, it just says it' was IBM's first mass produced mainframe computer).  Perhaps there's another term for that - I'd suggest "series produced", but series production redirects to mass production.... Guy Harris (talk) 22:24, 21 January 2021 (UTC)


 * Done. Nobody objected, so I changed the IBM 701 page to say 'series production'.  Stepho  talk 21:26, 26 January 2021 (UTC)
 * What about UNIVAC? Why doesn't that count as the first mass produced computer. Because IBM said so?--Zebbie (talk) 15:29, 22 June 2021 (UTC)

Removing entire claim and citation, since the source cited refers back to this very article as it's source. See WP:CIRCULAR. Note also that extraordinary claims require extraordinary evidence (WP:EXTRAORDINARY). Please do not add this back in without providing a WP:RELIABLE source. Johnnie Bob (talk) 15:59, 22 June 2021 (UTC)

It was IBM's first mass produced computer according to Computer History Museum. Doesn't make it first mass produced in the world but its a start. Tom94022 (talk) 18:11, 22 June 2021 (UTC)

Phister, Data Processing Technology and Economics, (c) 1979 shows IBM 650 as 80% of market 1955 thru 1959 (Fig 1.31.1) with Univac at 10% (Fig 1.31.2) but he doesn't go back before 1955. He has the first installation of the 650 in 11/54 and has 240 total installations by year end 1955, 700 by YE56 and peak of 1554 in 1959. However, he does note the Univac 1 "established a handsome lead for itself in the early fifties."

Eighteen Univac 1 installations are identified from 1951 to 1954 with 12 occurring in 1954. The US Army BRL report, March 1961 identified 48 Univac 1 computers delivered.

So the question does become does limited volume of the Univac 1 amount to "mass production" when compared to the that of the IBM 650. I don't think so but the above may be OR. Tom94022 (talk) 19:03, 22 June 2021 (UTC)

Picture of EMMA - side panel
The side panel is not missing; instead, it has been slid left (to the back) to open the box and offer an inside view. Zechenhund (talk) 07:45, 28 August 2021 (UTC)


 * I looked carefully at the full-size image and I agree. I have replaced "an exterior side panel is missing" with "an exterior side panel is open".  Stepho  talk 09:27, 28 August 2021 (UTC)

650 vs 701/etc
IBM had two competing groups in the early 1950’s.

The 650 was a fully IBM archetype, and it was mass produced for several years. While the 7080 replaced it in the product line, it was an incomparable new machine. Thus the end of the architecture.

The 701 was created as a mass produced version, admittedly in its first form, based on VonNeumans machine at the center of advanced studies at Princeton. That group put out a public status report, that included discussions of alternatives, rationale for the evolving design and the design and status of the first machine. It was being closely tracked by a number of companies. There is a good history book that came from researching the Princeton archives and interviewing people.

IBM, probably under Tom Watson Jr’s direction, hired VonNeuman, to make a once a month trip to Poukeepsie, for a day, briefing them on the status report and sharing ideas.

Several universities and companies were reading the reports and produced single machines, that differed as those engineers felt they had better ideas, and were different. Some were built, in a single copy, and run for several years.

The first chief engineer suffered from a condition common to engineers, of being imaginative, thinking of new ideas, and not able to select the best of his children. It looked like it would never complete, and he was replaced by a Northwestern University Engineering graduate, James Pomerene, who made the hard choices, and created the first working machine.

Princeton Advanced Studies decided it was basically a think tank and not a builder of things, and did not plan to continue. Jim Pomerene was offered a position at IBM, as well as at least two others that became major figures at IBM. I suspect, but do not know for sure, they went on the 701 team.

While IBM produced a limited number of 701 machines, it was followed by a machine with a superset of capabilities, mostly index registers, the IBM 704, on about 1958 the IBM 709 was created, using the same vacuum tube electronics. The main improvement was dual ported memory, and a data synchronizer. The data synchronizer could do simultaneous compute and I/O. The Christiansen patent, issued in about 1968, was seminal, and IBms most important and valuable for 20 years.

(The 709 had a compality mode for running as a 704. This compatibility mode existed a decade later, while the system 360, model 65, was run in 704 emulation. In about 1970 I met a programmer, at an oil company, where a programmer could not figure out the math in a 650 program for analyzing seismic data from explosions. He wrote a 650 simulator. A decade later a model 65, in 7094 emulation, would run in 709 compatibility mode, running in 704 compatibility mode, interpretively running the 650 program.)

The next step in the evolution was the implementation of the 709 in transistors, was the 7090. During the next years the index registers went from 3 to 7 and double precision hardware added, as the models went from 7094 to 7094 model II.

Thus the 701 was the first in a series of upward compatible versions, replacing flip flop technology, but keeping the same organization through 15 years, and creating the first mainframes.

The 650 was a dead end. — Preceding unsigned comment added by 2601:643:8880:81A0:C410:7EA1:7A84:400C (talk) 20:50, 18 February 2022 (UTC)


 * You placed your comment as a reply to a comment from 2008 about 650 programming tools. Yet your long comment didn't mention anything about programming tools. So I shifted it to the bottom of the page, where new topics are conventionally placed.
 * Reading it through again, I can't see what your point is. It's a long essay that wanders across many points but doesn't ask any actual questions or make any explicit points about what you want changed.  Stepho  talk 23:30, 18 February 2022 (UTC)
 * I think he is responding to the discussion above which seemed to downplay the 701, and he is quite right in the sense that the 701 was the progenitor of a long line of IBM scientific computers (he forgot to mention the 7040/7044). The 650 was a dead end, both in terms of architecture and electronics technology, single tube modules from the 604 vs multi-tube modules used in the entire 700 series. The successors to the 650, the 7070 and 1620, both decimal, had entirely different architectures, as there was no need for an address of the next instruction with core memory. I remember trying to wrap my head around that idea during a presentation at my high school by IBM on the then new 1620, back in 1961. My teachers were despondent that all the programs they had written for the 650 at Watson Labs would not run on the new machine we were getting. In my opinion the lasting impact of the 650 was introducing programming to thousands of people who would soon be needed as the computer industry exploded.-agr (talk) 15:51, 20 October 2023 (UTC)

Slower?

 * You reverted my edit saying the 650 was marketed as a less expensive but slower machine compared to the 701 and 702. saying: "Need supporting references. I find it hard to believe that IBM would put the word "slower" in any marketing material."

The first way IBM communicated the slower speed was in the machine's name "IBM 650 Magnetic Drum Data-Processing Machine." People in the field understood that magnetic drum was much slower than Williams tube or core memory used in the 700 series. But the text of IBM's announcement makes it clear that the 650 was in a different class:

"The following is the text of an IBM press release distributed on July 14, 1953. International Business Machines Corp. announced today the introduction of a new commercial electronic decimal calculator designed to meet the vast accounting and computing requirements in areas between those now served by its "giant brains" and the widely-used smaller machines such as the 604 Electronic Calculating Punch and the Card-programmed Calculator. ... All of the calculator's arithmetic operations are controlled through a program which may be entered either automatically from punched cards or manually from the operator's console and stored in the form of magnetized spots on the surface of a drum only 4" in diameter and 12" long, spinning at 12,500 revolutions a minute. The calculator's arithmetic unit operates at electronic speeds. It can: accumulate 10 digit numbers to form a 20 digit total at the rate of 200 a second; multiply 10 digit numbers by 10 digit numbers to develop a 20 digit product at the rate of 60 a second,"

By contrast the announcement of the 704 (the 701 had a short marketing life) shows the much higher speed. "The Type 704 multiplies or divides in 240 microseconds, or approximately 4,000 operations per second. The doubling of speed in arithmetic, plus the use of index commands, makes the Type 704 a very versatile and capable data processing machine.

The price difference was also large $4000 per month for a basic 650, $32,000 per month for a basic 704. The current text suggests the the 650 was in the same class as the 701 and 702, just "general purpose." That just isn't true. --agr (talk) 16:37, 20 October 2023 (UTC)


 * I agree with Arnold that "slower and less expensive" are appropriate in this context. Whether IBM used the terms is irrelevant; they are factual and supported by IBM's use of "big" 701 to relatively describe the 650 in the original 650 announcement. Tom94022 (talk) 18:54, 20 October 2023 (UTC)


 * Agreed that the 650 was slower than the 701 line - probably as a speed/cost trade-off. But you can see in the ad copy that you presented above that IBM shows the speed of each processor but does not compare them to each other. It is bad advertising practice to ever say that you have a slower processor for sale. You do not have a fast processor and a slow processor. But you can have a fast processor and an even faster processor. So my complaint about the edit was that you said that IBM marketed one as slower - which was never true. IBM marketed a fast one and a faster one but did not compare them to each and definitely never used the word "slow". I'm am happy if you reword to say that the 650 was slower and/or more affordable but we have to be careful to not say that IBM marketed it as slower. Marketed as more affordable certainly, but not marketed as slower.  Stepho  talk 00:19, 21 October 2023 (UTC)


 * OK, how about IBM "offered it" as a slower and less expensive ... ?Tom94022 (talk) 05:20, 22 October 2023 (UTC)


 * Excellent - I'm quite happy with that rewording.  Stepho  talk 06:06, 22 October 2023 (UTC)
 * Works for me. I've made the edit as I understand it.--agr (talk) 16:04, 23 October 2023 (UTC)