Talk:Control Data Corporation

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Companies  
WikiProject icon This article is within the scope of WikiProject Companies, a collaborative effort to improve the coverage of companies on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.
 ???  This article has not yet received a rating on the project's quality scale.
 ???  This article has not yet received a rating on the project's importance scale.

This article has comments here.

CDC alive still going strong?[edit]

I would like to know if Control Data Corporation is Alive today. Thanks.

Yes. It is no longer called Control Data Corporation (August, 2003). Control Data Corporation changed its name twice in the 1990s. The first time, they changed to Control Data Systems. The second time, they changed to Syntegra, when Control Data Systems was bought out by British Telecommunications, PLC. Now, according to Charles Babbage Institute, Control Data is operating as Syntegra. Paul G.
I disagree; the brief answer definitely should be "no". CDC sold off some business units in the early 1990s, and the remainder in 1992 split into 2 parts, neither named "Control Data"-anything today, and neither manufacturing computers--the basis of CDC's fame. Although the reality is different from--and more complex than--the description above by "Paul G.", I've created a WikiWiki page at Control Data that should clear things up for readers arriving at this page with questions about the fate of CDC. --ClayPhipps 23:06, 22 Oct 2004 (UTC)

Also see:

RJBurkhart 10:58, 18 January 2006 (UTC)

IBM Model 92 FUD or not?[edit]

Can IBM's non-existent Model 92 really be referred to as FUD, shorthand for "Fear, Uncertainty and Doubt"? Isn't paper launch a more appropriate term?

--Drhex 07:59, 5 October 2005 (UTC)

Perhaps, but considering CDC successfully sued IBM based on the argument that it was FUD (although as the article notes this term was not common), I think it's suitable to refer to it this way. According to what I have read of the case, IBM salesdrones were telling potential customers to ignore the CDC machines because the 92 was going to be out "Real Soon Now" and would outperform it. Neither of those claims turned out to be true, as history noted. Maury 15:32, 22 August 2006 (UTC)

CPI & Centronics[edit]

I wrote articles on Computer Peripherals Inc and Centronics some time back that might deserve a mention here. CDC is not my area of expertise, so if someone thinks it fits and can find a good way to fit it in, please do so. --Gadget850 13:22, 20 October 2005 (UTC)

Control Data Institute[edit]

Is this company related to the "Control Data Institute" that used to advertise their vocational education program rather heavily?

Yes it is. I attended Control Data Institute (CDI) in 1965. I was in the first graduating class. --Dennis Fernkes 21:26, 20 February 2006 (UTC)
I attended CDI (Long Beach)in 1970. Retired in 2009 after 39 years in IT. — Preceding unsigned comment added by (talk) 04:22, 28 January 2014 (UTC)
I attended CDI in 1973. I was in the last class to graduate from the Long Beach, CA campus. My final project was a 'paperless' timekeeping system design. No punched cards. Data was entered via a terminal. The instructors didn't think a system like that would work! 20:34, 7 August 2007 (UTC)


I would like to begin creating a section about PLATO ("Programmed Logic for Automatic Teaching Operations"), the computer-based education system that CDC co-developed during the 1960s-70s and, I think, sold off in the 80s. It was, according to them at the time (my father worked in PR there, which is the source of my memories), the first computer system that attempted to be user-friendly, using as much street language as possible for commands. Its greatest significance, however, may be that it created the formation of the first online community, at the same time ARPANET was being developed. The first personal email and discussion groups were evolved on PLATO as well, at least according to this page, which has many details:

"PLATO: The Emergence of Online Community," by David R. Woolley

This site, by Brian Dear, also has a lot of information: PLATO People: A History Book Research Project

A quote from the above: "Before Microsoft. Apple. The Web. AOL. The Internet. Before everything, there was PLATO: the first online community. The network that time forgot. The birthplace of instant messaging, chat rooms, MUDs (multi-user dungeons), personal publishing, screen savers, flat-panel plasma displays, one of the first spell-checking/answer-judging mechanisms, and countless other innovations."

I participated on PLATO as a teenager and everything I've read on these websites ring true to me, though I had nothing to do with programming and so was and am ignorant on those issues. I do know, however, that the PLATO story has been lost to the (silicon) sands of time, and that Wikipedia is the perfect place to return it to its rightful place in computer and online history.

One of the rare mainstream press articles about PLATO: "PLATOfest to Celebrate First Online Community," by Steve Silberman

I'm new on Wikipedia and made the mistake of editing large parts of another entry without discussing it first (out of ignorance, not pure rudeness), and didn't want to make the same mistake with this, so here I am, on the Discussion page.

MaudB 3:06 pm, 14 June 2006

Have you looked at the wikipedia article PLATO? It ought to be linked from this article though. EncMstr 19:19, 14 June 2006 (UTC)
That is so weird; I looked for a Wikipedia article on PLATO before posting in here, but didn't see it.

Apologies; I'll look it over. MaudB 4:44, 17 June 2006

The Wiki search engine is extremely poor, IMHO. Changes in capitalization, spacing (etc) result in no hits. The same term typed into the "search" field and then GO'ed works fine. It's a real problem. Maury 15:32, 22 August 2006 (UTC)


The section on the STAR is a bit light and completely ignores the contributions of Neil Lincoln who assumed project leadership after Jim Thornton went on to other things. In particular, Neil was a very big part of ETA systems.

Also, the section omits quite a bit of history in the design and evolution of the STAR. The first incarnation of the STAR was the STAR-100, which had an incredibly huge instruction set with instructions for everything from approximate masked key searches to BCD arithmentic. A number of potential customers had expressed interest in Thornton's design, but all with very diverse requirements. Given that the STAR was heavily microprogrammed, an attempt was made to accomodate all of them, with disastrous results.

As an aside, there were two other STAR models at the same time; the STAR 65, developed in Canada, with about half the speed and less than half the cost of the 100, was scrapped when the Canadian operation was discontinued. As far as I know, only one of these machines ever existed. The other machine, the STAR-1B, was really a low-speed emulator that could make use of STAR-100 peripherals and stations, but otherwise probably ran at the speed of a slow microprocessor. All extant examples of the 1B were scrapped sometime around 1976.

The STAR-100 used its long pipelined vector units for everything, including the simplest scalar operation (memory fetches and stores were performed in 512-bit "superwords" or SWORDs). This meant a very large latency for instruction startup with sometimes startling effects. For example, for N<1000, a serial vector search was substantially faster than a binary search. On the other hand, the ability to handle large sparsely-populated vectors ("sparse vectors") using a bitmap to indicate populated positions represented a welcome feature.

The STAR-100A incorporated a separate scalar unit and began the process of pruning away some of the ridiculously large instruction set. (At one time, it was a mark of distinction if you could think of a use for some of the more esoteric ones, or even knew that they existed.) I believe this became the CYBER 201.

The STAR 100B and 100C models were released as the CYBER 202 and 203, respectively, with incremental improvements in architecture (e.g. semiconductor memory, additional vector pipes). In this sense, the STAR was an evolutionary product, not a revolutionary one.

-- 21:14, 6 August 2006 (UTC)Chuck G.

Chuck, please go nuts in the STAR article! I wrote what I could based on the books I had (The Supermen being the only useful one, and only marginally so) and lots of "overview" descriptions in print and online (mostly copies of print material, obviously). I realize the article is extremely thin and needs work, but I have exhausted my own knowledge on the issue. If you can help, please, feel free to re-write! Maury 15:32, 22 August 2006 (UTC)
This section on the STAR, and other sections, would be greatly improved by citing sites. The STAR burned the careers of friends, both mentioned in Supermen and not for security reasons. Attempting to locate artifacts for the preservation of computer history makes this doubly important. (talk) 23:47, 13 December 2007 (UTC) --enm

Operating Systems and Networks[edit]

There should be some discussion of CDC's software and networking, which was excellent for its time. Sometimes around 1976, CDC built the MERITSS time sharing system, which supported 2000 terminals located in the highschools of Minnesota. In terms of operating systems, they has a batch system called SCOPE, and later KRONOS and NOS, which were time-sharing systems based in part on paradigms from the Dartmouth Timesharing system. CDC was the king of timesharing. Having use the hideous IBM and Univac TS systems, I can say CDC really did get it right. DonPMitchell 03:11, 11 August 2006 (UTC)

One should mention that some of CDCs better products were "bootleg" efforts. KRONOS, for example, used to be known as MACE, a product of Greg Mansfield (with Dave Calllender) done mostly in the wee hours on the QA floor at Arden Hills.

SCOPE was primarily a PP-resident operating system, with only the memory-move function be CP-resident. One could run SCOPE wihtout the CEJ/MEJ feature enabled, whereas KRONOS required it.

SCOPE was updated a bit to compete with KRONOS and re-named NOS/BE (batch-environment).

There were other operating systems in use for Special Systems projects, such as TCM (Time Critical Monitor), a heavily CP-oriented OS with a response time better than 100 usec.

7000 SCOPE bore no resemblence to 6000 SCOPE. 7000 SCOPE was mostly resident in the CP, and used a clever scheme of nested privledge levels, with each higher-level including the lower ones in its RA and FL. SCOPE was also the first CDC operating system to introduce the idea of a Record Manager. ChuckGuzis 04:26, 18 August 2006 (UTC)

CDC_and_the_PC (cmd-click)">'''CDC and the PC'''.

CDC had some chances to change its focus but could not see the need. The upper management had lost its willingness to take chances as the team grew older. Personal computing was introduced as a concept for funding in the 1970s (around 1973). Even though CDC understood the technology and was manufacturing many of the required component parts, it was felt that this was too risky ($250K R&D) and a concept that would never take off. These ideas were revisited circa 1985 when it was suggested that the then mainframe operating system (NOS VE) could and should be ported onto personal computers. Again personal computers were seen as no more than expensive adaptable mainframe terminals by the teams that set company direction.Gr11zzly 22:51, 11 November 2006 (UTC)

CDC_and_the_internet (cmd-click)">'''CDC and the internet'''.

A small team of systems architects saw the need for distributed communications to rid the mainframes of what was termed the ‘front end bottle neck’. A product was proposed and implemented called CDCNET. CDCNET was a network of intelligent routers and bridges that in 1989 had some 10,000 active users in 16 countries In 1985(?) a CDCNET prototype was demonstrated at a New York trade show where CDC showed 16 different applications running in 16 windows with each application on geographically diverse systems including IBM, Burroughs, Honeywell and CDC mainframes. The user was able to cut and paste data from a machine in Belgium to a machine in San Francisco from the terminal in New York. It had the potential to be winning product but for a few poor decisions that caused the delivery to miss the market window. The protocol chosen was not TCP/IP but an up and coming ECMA standard. And the implementation language was changed midstream from C ( with off the shelf available code modules) to Cybil (a home grown Pascal based language) These decisions added 5 years to the delivery.Gr11zzly 22:51, 11 November 2006 (UTC)

NPOV and References[edit]

Much of this article sounds as though it were written by a disgruntled employee ruminating on the "glory days" of CDC. I have no idea whether or not this is truly the case, but the tone of this article needs a massive overhaul. Sentences like "For most of the 1960s they built the fastest computers in the world by far, only losing that crown in the 1970s to what was effectively a spinoff" and "CDC was well known and highly regarded throughout the industry at one time, but today is largely forgotten" are all well and good with citation, but as there are no citations throughout MOST of this article, I am forced to add both a Citations Needed and an NPOV tag.--Icetitan17 20:08, 3 September 2007 (UTC)

Well, for most of the 1960s they did build the worlds fastest computers, by far. And they did lose their crown shortly after their chief computer architect, Seymour Cray, left to start Cray Research in 1972. And CDC was very well-known, and was highly regarded in the industry in its day. (For example in 1988 they had revenues of $3.6 billion/year, had over 30,000 employees, and was the 5th largest mainframe manufacturer.) However they never had a presence in the microprocessor market, so they are very unknown to most young 'uns today. But many have heard have heard of Commercial Credit, Arbitron, Ticketron, portions of Seagate Technology, and dozens of other spinoffs that are still with us today. --Wws 04:25, 7 September 2007 (UTC)

I disagree with the originator of this section, and I believe that the NPOV tag should be removed. When I read the article, I too thought that it might have been written by a CDC technical old-timer, because the technical detail is so good. I worked for CDC from 1980-91 in the division that was formed (years before I joined) to fight the IBM antitrust case. We used the history of our division, Control Data Litigation Support Services, as a marketing tool. For this reason, I have seen the IBM internal marketing documents relating to "pre-announcement" of a system to compete with the 6600 that were discovered during the case. The use of FUD was (and is) common in markets where there is a dominant player. IBM was hugely dominant in the computing market during the 1960s. It is unfortunate that the original author did not include references, but I know quite a bit about CDC history, and I did not see a single factual error. I would hope that a moderator/editor would remove the NPOV tag. (talk) 19:20, 6 December 2007 (UTC)

I agree with you, however I arrived here to read the summary and ended reading most of article plus Cray's bio only to clear up wich spinn-off is alluded. Well, also because it's nice reading... (talk) 12:58, 25 August 2008 (UTC)

I side with Wws on the NPOV, but I am not certain that I would consider the term "disgruntled." For instance the early year section mentions Parker, a name which I could guess and later confirmed on the ERA wikipedia page, and doesn't mention Norris until later down on the page and completely leaving out another significant founder. I think that the NPOV should be kept and increasingly I've noticed how the ETA pages have become increasingly glowing. (talk) 23:02, 13 December 2007 (UTC) --enm

Most of this article, and ERA, was originally written by me (it has grown significantly and in very positive ways since then). I am not a former CDC employee, and am too young to have even seen one of their machines in operation. I do, however, have a good basic understanding of the technical details and am relatively well read on the topic. I was quite flattered to see two people think I may have worked there :-)
The original argument is that the terminology is too glowing. This may well be the case, but of the two examples given the first is simply a statement of fact. The 6600 was ten times as fast as the next fastest machine when it was released, and they kept this performance crown until the 1970s. They only lost it because Cray left in the aftermath of the 8600, and went off to build a machine that simply crushed all comers. Forget CDC: the key issue here is that Cray was the key element in machines that held the performance lead for twenty years. How can one state this very important point without sounding "glowing"?
Watering down statements about the effect this company had on the industry does every reader a disservice. Maury (talk) 20:48, 14 April 2008 (UTC)
It's time to remove the NPOV tag. As an architecture and OS specialist, I can attest to the speed of the 6600 series. Terms that it was a factor of 10 faster than IBM machines is actually conservative. Many batch jobs that would take minutes on an IBM mainframe took seconds on a CDC. The prose may be a little flowery, but it's not inaccurate. Even IBMers had great respect for Cray and Norris.
--UnicornTapestry (talk) 22:36, 11 July 2009 (UTC)

Wikimedia Commons and other images[edit]

I added Category:Control Data Corporation to Wikimedia Commons so that any images pertaining to Control Data could be found from there, because a cursory search for just 'CDC' yields results that mention Centers for Disease Control and Prevention in one way or another. — Which is why all CDC computer images are welcome to be categorized to the aforementioned category. I also added an external link to Commons. -Mardus 03:53, 11 November 2007 (UTC)

The entry on Control Data Corp. contains the following sentence: Thus SPIN explored all of these approaches, and eventually delivered a very large 28" fixed disk of extremely high performance, much better than "hard drives" of today with their overly complex giant magnetoresistance design, as well as a smaller multi-platter 14" removable disk pack system as well. I suggest that this sentence be removed or rewritten to remove the reference to today. Although that old SLED (Single, Large Expensive Disk) fixed-disk design was fast for the time, the performance of storage systems and servers -- as of 1987-1990 -- began getting much faster with a quite different architecture (RAID), and today, drives far exceed anything available from CDC's heyday. As of 2008, certain enterprise-caliber corporate data storage arrays contain solid-state disk that perform at nearly pure memory speed.

CDC 3800[edit]

CDC also made a model called the 3800. It was a CDC3600 with the CDC6000 memory modules. These modules took 5 cabinets for 4K of memory. Each memory location was 52 bits long, i.e. 48 data bits with 4 parity bits and took a 1 ton refrigeration unit to keep the modules cool. I went to school on them at the Los Angeles CDI school in 1967. I was then transferred to another city in California where the military had 6 of them along with about 10 CDC 160A computers.

I was in the last class CDI taught on this computer. Even CDI in Minneapolis did not teach it any longer.

The mainframe of this computer was made from "discrete" components, i.e., no integrated circuits as was the CDC 160A. Since CDI no longer taught it and we needed people trained, I developed a component level course on this computer. It was 6 weeks long, 8 hours per day and 5 days per week.

I knew of only one CDC3800 that was not owned by the military. This one was owned by a company named "Interaccess". I had to work on their system one time. —Preceding unsigned comment added by BikerLen (talkcontribs) 23:19, 22 July 2008 (UTC)

One 3800 on display at the NASM Annex at Dulles outside Washington DC. 00:25, 28 January 2011 (UTC) —Preceding unsigned comment added by (talk)

Cyber 18 and minicomputers made by CDC[edit]

I worked with a Cyber 18 minicomputer in the 1979-1981 timeframe. I do not know much of its characteristics other than it had a 16 bit processor and around 256kilobytes of RAM. I worked with two installations, one at the University of Georgia that was used for FORTRAN programming classes and one for the UN World Meteorological Organization in Saudia Arabia where it was used as a message switch as part of the WMO weather data collection network. The Cyber 18s I worked with came with assembler, FORTRAN and RPG II programming languages. It had a multi-tasking/multi-user operating system and supported several simultaneous users with VT100 type terminals.

The one I used in the university had a small card reader, a tape drive, and two disk drives that were of the old stacked platters discs that were removable. The actual drives looked a lot like small top loading washing machines and the platter stacks were stored in what looked like cake carriers when they were not in the drive itself.

At the university some of the student employees were interested in C programming language and the UNIX operating system at the time and added the RATFOR preprocessor for FORTRAN to the Cyber 18 installation.

Several articles mentions the Cyber 18

Here is something about the disk packs

I kind of wonder if the Cyber 18 was basically CDC just redeploying one of the periperal processors used in the CDC mainframes in order to get into the minicomputer market as quickly as possible. I don't very many were sold. Richardelainechambers (talk) 17:48, 9 August 2009 (UTC)

info about Magnetic Peripherals Inc.[edit]

I have some info for you on Magnetic Peripherals Inc. (aka: MPI) I thought perhaps you might want to add some of this info to create a page for MPI, linked to the CDC page. This should help with expanding the MPI info and the products they handled.

I used to work there in 1980 and 1981 in Oklahoma City. I still have my ID if I need to verify it to someone in charge. I was a technician back then and troubleshot and repaired most of the peripherals we handled.

I worked on the 14 inch cartridge disk, break boards, controller cards, etc, which we troubleshot using a very large diagnostic station called a "Genrad." I rewrote one of the diagnostic software programs to improve the Genrad diagnostics. Genrad is a contraction for "General Radio", a Massachusetts electronics firm well-known for quality test equipment over several decades.

I also did troubleshooting and repair on the 8" Shugart floppy drives and control boards which were typically found in IBM systems such as the PDP series... if I remember correctly, a couple of those systems were the PDP 1128 and 1133... but I'm not positive on the numbers. but I worked on those systems which had the shugart drives for a prior employer.

I also worked on the Winchester drives and control boards and phoenix control boards.

I was there when the new Tandy 5 1/4 floppy drives were introduced and started coming into the repair dept. These were our first sandwiched drives with a double layer of circuit boards so the drive could be physically smaller. I believe these were the very first 5 1/4 floppy drives made in the industry at the time. No one had ever seen a drive that small before.

We had quite a problem with CMOS chips .... especially on the Winchester drives until they learned about static damage. I had called the upper managers in 1983 after I moved and explained how the static was damaging the chips and sent them info and photos I had from JPL on this problem I had special training in after moving. MPI implemented a new program on preventing static damage.

I also remember the Teledyne test station... I knew the person handling those boards but did not use it myself.

I also remember seeing the 10" stacked platter drives there but did not work on those.

I also have a very nice, large photo of the entire staff management and technical crew in the repair dept from that time... it it a high angle photo so it shows almost the entire repair dept and was signed by the crew when I moved. I would be willing to make the photo available for the page when I find it. I also had smaller photos of various members of the crew, etc.

Last I heard, their name was changed to Seagate. I may have some other info squirreled away somewhere. Let me know what you need. I will defer to someone else who wants to add this to the appropriate pages since I'm not familiar with the page publishing rules.

ExecPE (talk) 10:09, 26 October 2008 (UTC)

Shark award[edit]

The article contains mention of a Shark Award for salesmen. It doesn't seem to be in an appropriate section and it's not clear to me that it's appropriate for the article. Thoughts?

--UnicornTapestry (talk) 22:47, 11 July 2009 (UTC)

Since there were no objections and the Shark sales award appeared to have nothing to do with the topic of super-computing, I removed that paragraph.
--UnicornTapestry (talk) 15:41, 31 August 2009 (UTC)

"Card punch" link incorrect[edit]

The link "Card punch" redirects to "Keypunch", which is not the same thing at all. This probably affects a lot of computer-related articles. There doesn't seem to be a relevant wiki article for card punch. The "card reader" article at least lists "punched card reader" (should that be "punched-card reader?", although it doesn't describe one. Peter Flass (talk) 14:03, 31 August 2009 (UTC)

Peter, you're absolutely right. I created a card punch stub.
--UnicornTapestry (talk) 15:28, 31 August 2009 (UTC)

Challenging Author's Performance Assertions[edit]

As a technical person who was active in the business in the 60s, I'm struck by the naivety reflected in this treatise, and I doubt that some of the comparisons are accurate. When talking about computer performance, we first have to answer the question, "Doing what kind of work?." Real customers at the time (the 60s) were complaining about machines that were overly optimized for commercial work (character machines) or scientific work (floating point, or "word" machines. The author makes the unsupported statement that the CDC 6600 (1964) was 10 times as fast as its competition. Doing what kind of work? Even limiting the comparison to Fortran-only workloads (i.e., predominately floating point) the 6600 made about 3 mflops (1 to 9 mflops, depending on which source you cite) and IBM's System 360/Model 91 (1966) at 5.5 mflops [ref:].

The fact that neither machine could really be considered a commercial success because of their high cost and limited sales volume is particularly telling. It was for this reason that IBM had decided earlier on to shelve its high-performance program (ACS, or system 8000) in favor of System 360. It became clear early on that designing a very fast machine could be accomplished if the design could be carefully tailored to the workload, and then, to what end? In fact, CDC recognized this even during the 6600 era, opting, instead, to focus its sales activities on the much less costly 6400 which used far fewer specialized components. For the System 360 program, IBM diverted more of less all of its development resources with the goal of a complete line of compatible machines, suitable for both commercial or scientific (engineering) work. — Preceding unsigned comment added by Atraveler1 (talkcontribs) 20:53, 19 April 2011 (UTC)

Current owners of KRONOS/NOS and NOS/VE code[edit]

I tried locating who are the current owners of the old OSes but it seems no one really knows. Tried mailing to the maintainer of the "dtcyber" simulator site, but there is still no reply. If someone knows whether it is BT or Ceridian or any other entity who owns the rights (and has) the binaries on tapes and/or the code, I think it would be of interest to add this as well.

As far as I know BT was indeed the last owner of the old CDC OSes (that was circa 2004?). I think they even licensed the OSes for hobbyist use at the time.
Maybe try the mailing list. Also the guys at have a couple of operational Cybers maybe they can help you as well.
Bitsavers has some Cyber18 stuff [1] but nothing for CDC mainframes.
--Jside (talk) 12:33, 4 July 2013 (UTC)