Talk:Digital Equipment Corporation
|This is the talk page for discussing improvements to the Digital Equipment Corporation article.
This is not a forum for general discussion of the article's subject.
|This article is of interest to the following WikiProjects:|
- 1 Why did DEC fall?
- 2 website
- 3 Polycenter?
- 4 Soviets
- 5 Intel processors inspired by DEC PDP-8?
- 6 VMS++ = WNT
- 7 Microprocessors influenced by PDP-8?
- 8 Logo
- 9 Personal computers
- 10 450 or 350 million dollar?
- 11 The word "Computer" banned.
- 12 Mis-use of ARCHITECT as a verb
- 13 Writable control store
- 14 T11?
- 15 The word "computer"
- 16 Name of the Company
- 17 Details...
- 18 Other innovations and achievements
- 19 Structure of article
- 20 ftp.digital.com
- 21 Closing DEC's business
- 22 "DEC alphabet"
- 23 Citation for Assertion about the X Window System
- 24 Famous people who worked at DEC
- 25 You did not like my coffee-mug?
- 26 64 bits binaries are bigger and adressing 64 bits was unnecessary
- 27 Major edits made during March 2010
- 28 Proposed major restructuring of article
- 29 unsourced/exaggerated synopses of notable people
- 30 Linux not DEC-accomplishment
- 31 DEC
Why did DEC fall?
One reason that DEC fell was its inability to hold its senior executives accountable for financial performance after the abolition of the Product Lines in the early 1980's. Ken Olsen was very proud of creating the Product Lines after the problems with the PDP-6 that almost killed the company in the early 1960's. Each Product Line had responsibility for a computer system (PDP-8,9/15, 11, DEC-10's) and the GM/Managers were held accountable for profit, revenue growth, and to a lesser extent customer satisfaction.
It knew early in the mid 1980's that it needed to cut its cost structure by about 1/3 if it was to compete in the 1990's. However, Ken Olsen absolutely forbid layoffs, so little happened until the situation became so desparate that he had to relent.
Instead it attempted to grow its way out of the problem through a massive re-deployment of resources to the field that wasted hundreds of millions of dollars.
Another major reason the company failed was that it did not adjust to the restructuring of the computer industry from full service highly integrated systems suppliers to areas of horizontal expertise - Intel for micros, Oracle for database, etc. DEC had the technologies to compete in many areas but the corporate strategy was to only sell systems.
It was a great company that could not adjust to a dramatically changing business environment that meant tightening its belt and changing what it sold and how it sold... until it was too late. —The preceding unsigned comment was added by 220.127.116.11 (talk • contribs) 20:45, January 14, 2006.
- When I worked at DEC in the mid-80s, there was an article in a business magazine (Forbes? Fortune?) about how DEC was trying to sell computers to farmers with a slogan that described the company as the "Second-largest computer company in the world" - to which farmers would ask, "so what's the biggest computer company?" I think it's funny that the article repeats this mistake: "At its peak in the late 1980s, Digital was the second-largest computer company in the world, with over 100,000 employees."18.104.22.168
- One might argue that DEC failed because "cheap" always trumps "good". DEC products were good, PCs were cheap, the rest is history. (One might also wryly observe that this was exactly how DEC built its business against competitors like IBM, but I'd like to believe that many DEC products were both cheap and good.)
- Atlant 15:10, 26 January 2006 (UTC)
- Agreed. DEC Failed to embrace microprocesor based systems early on. In the late 70's early 80's they had superior operating system software for small computers and could have blown IBM away had they introduced their own uP based system with their software at the time the PC was introduced. DEC's products were 'cheap' and 'good' compared to commercial mainframes at the time, but the PC was a game changer. Produced for the mass market, it was very cheap but its software was not very good. From what I've read Ken Olson would not support those in the company that wanted to bring to market a low cost uP based system. His lack of vision is what killed a top notch company.22.214.171.124 (talk) 14:55, 12 November 2009 (UTC)
Didn't DEC go bankrupt due to its own inept sales force pushing Pentium machines over the AlphaPC? I once heard someone complain that DEC was a company you couldn't give money away to (his words).
- I can tell you why they fell into disfavor at one NASA center: they were too expensive for what they were providing. The fact they could be heavily hardware-configured in days when hardware was very expensive indeed was a great advantage, but it was an advantage that whittled away when cheaper "general purpose" computers came out, because then buyers could afford to buy components which they really were not going to use. Also, of course managers loved the idea that whole computers could be "plug compatible" the way employees were held to be. I.e., if something goes wrong with a project, a manager can shuffled in or out computer equipment -- putting a lot less emphasis on long-term planning. (And...hiding management miscalculations.)
- The incredible cost of configuring digital computers is shown by a statement, after giving their paper at an AI conference, by one of the DEC speakers that the AI configuration program cost $1,000,000 a year to maintain. Not to write, just to maintain. 126.96.36.199 (talk) 09:33, 1 September 2008 (UTC)
Well, let me tell you a story which says why DEC lost one customer. In the mid-1980s, the company I was then working for had bought 20 PDP 11/73s, delivered in pairs over ten months. We also got a tape with the operating system (RSTS/E) and other software with the first pair of systems. Unbeknownst to us, halfway through the period, there was a hardware upgrade in the CPU, which was subtly incompatible with the version of RSTS we had. After we had been running for well over a year, I was able to track down some errors that we were having on some machines to this incompatibility. The old hardware was no longer available, and we were told that we would need to get ten new CPU boards and a new software tape to make all the hardware the same and to have a good version of the OS. The kicker was that DEC insisted that we pay $8000 for each board and $2500 for the software upgrade, for a total of $82,500. We maintained that since the problem was caused by them, they should eat the cost. We might have come to some arrangement for splitting the cost, but they were adamant. We had to pay it all. They would not come down even on the price of the tape. Since even the properly working systems were slow and the application software difficult to maintain, we determined that we would start over again on this application, using a proper DBMS (the old one used ISAM files), C (the old one used BASIC+2), and UNIX and faster hardware. However, at the conclusion of the meeting where this was decided, the CIO stated "Digital Equipment Corporation will not be invited to bid on this contract." Thus, DEC lost out on a contract worth over a million because they insisted that we pay $82,500 for a problem that was really their fault. Jhobson1 (talk) 13:34, 12 April 2011 (UTC)
Why Did DEC fall? Why did Apple Succeed? Marketing!!! As a DEC OEM I fought long bitter and costly battles trying to get DEC to understand that "digital" was a kind of watch in the minds of every single prospective Customer I came in contact with. My greatest sales hurdle - building up the digital image in the minds of my prospective customers. DEC had promised me for years it was going to enter into cooperative marketing agreements to help cover my marketing costs and promote the digital name. They never fulfilled those promises. So we started looking at the emerging microcomputer as a means of controlling our costs. And in the process we ported our software to several platforms including CP/M MP/M DOS and others. When IBM introduced the PC we were ready. It took us about a week to port our software to the PC - Of course everyone knew who IBM was - we didn't have to sell IBM. So it was "Good bye" to DEC & Ken Olsen's promises and Hello to IBM. Our PC solutions sold like hot cakes. — Preceding unsigned comment added by StumpLifter (talk • contribs) 21:20, 1 October 2011 (UTC)
The reason I came to this page was because I read (on a web history page) that DEC put up the first commercial website. In this article it says they were amongst the first .com websites. Were there any .com websites before DEC that weren't commercial? I thought if it was one it would be the other as well, but it'd seem strange to refer to the first ever as one of the first... sheridan 03:30, 31 January 2006 (UTC)
I believe one of the other items DEC divested was their management software known as Polycenter. This software went to Computer Associates. Yes it was one of the first generation of enterprise management offers in the market.
Is it worth noting that an 11/782 was seized in 1983 as part of a contraband shipment to the Soviets? --Gadget850 ( Ed) 13:48, 27 May 2006 (UTC)
Intel processors inspired by DEC PDP-8?
Without a citation, I don't buy the idea that any of the early Intel 4-bit or 8-bit microprocessors were "inspired by the PDP-8". There is no significant architectural similarity between the PDP-8 and the 4004, 8008, or 8080. On the other hand, there are very clear similarities between the PDP-8, DG Nova, and HP 2116 family (of which the HP 2100 is a member). If no citation is forthcoming in the near future, I will remove the 4004 and 8080 references in that section. --Brouhaha 18:43, 9 June 2006 (UTC)
- That also leads to some questionable assertions in the next sentence: "Machines based on the PDP-8 can be characterized by a small number of accumulators (such as AX or BX) rather than a series of regular registers (such as R1-R8) like the PowerPC and PDP-11, so it might be said that it was the descendents of the PDP-8, not the later PDP-11, that inherited the dominant processor status of the 2000s." If the 4004 and 8080 don't resemble the PDP-8 that closely, than the 8080's x86 descendents can't be said to descend from the PDP-8 either.
- The PDP-11 and its successor, the VAX, were CISC processors, while most modern processors are RISC-based. The difference between CISC and RISC designs probably accounts for more architectural differences between the PDP-11 and today's processors than the differences between the PDP-8 and PDP-11 do. -- CWesling 01:17, 23 June 2006 (UTC)
- I don't buy it, either. The PDP-8 was, at least from the point of view of a programmer, a beautiful, clean, economical design. I did quite a lot of programming in PDP-1 assembly language and LINC assembly language, both considerably richer in instructions than the PDP-8. Yet the PDP-8 seemed to be just as expressive and easy to program.
- For several years, I did quite a lot of programming about equally divided between the PDP-8 and the PDP-11. It always astonished me how easy the PDP-8 was to program, and how little you seemed to be cramped by the seemingly severe limitations (only six addressable instructions). When I first saw the PDP-11 instruction set I thought I had died and gone to heaven. Well, when I actually started writing code, sure, the PDP-11 was nicer than the PDP-8, but not nearly as much nicer as I expected. Who'd have thought that eight silly little autoincrementing memory locations could be at least 3/4 as effective as real index registers? And I could swear that it always took significantly more core to write the same program on a PDP-11 as on a PDP-8.
- In contrast, virtually all the microprocessors were ugly, asymmetrical, ad hoc monstrosities. No doubt it all had to do with how they could run the microscopic data paths on a plane, or whatever. The Motorola 68000 does "feel" a little like a PDP-11, but the 8080 and the 6502 never felt the slightest bit like a PDP-8 to me. Dpbsmith (talk) 01:36, 23 June 2006 (UTC)
- dpbsmith, see orthogonal instruction set for an explanation of why your PDP-11 programs always seemed bigger than your PDP-8 programs. It's the nature of the beast that a highly-orthogonal instruction set (such as the PDP-11's) is always far less bit-efficient than a non-orthonal instruction set (such as the PDP-8's). (And I say this as an experienced assembly-language programmer on both systems.) Both systems were joys to program, though, compared to the baroque 8080 and x86 architectures.
- Atlant 12:03, 23 June 2006 (UTC)
VMS++ = WNT
The article discusses how one can get the acronym "W/NT" by incrementing each letter in "VMS". Someone just edited in "Cutler later verified this."
AFAIK, Cutler has never actually made any affirmative statement such as "Yes, I did that intentionally." Instead, he's always said something coy like "Did you just notice that?". (That is, he makes it clear that it was deliberate while not actually saying so.)
Have I got this right?
If so, we should edit the article.
Atlant 16:28, 14 June 2006 (UTC)
- That's how it looked to me, too, when I Googled for this. So yes, we should probably reword that a little. -- CWesling 23:21, 16 June 2006 (UTC)
Microprocessors influenced by PDP-8?
I still don't buy it. "Machines based on the PDP-8 can be characterized by a small number of accumulators?" No, this is hardly a defining characteristic that makes an architecture PDP-8-like. Broadly speaking, small numbers of accumulators (and small numbers of registers with somewhat specialized, asymmetrical functions) are found in any situation in which registers aren't cheap. Many, many early machines have an "accumulator" and an "accumulator extension," for example. I think the (original) Illiac did. Certainly the PDP-1 did.
The advantages of having many registers are obvious, and they appeared whenever the economic balance for that particular machine permitted it. Conversely, any machine that was cheap or otherwise constrained to have minimal hardware, had a few.
Here are a list of PDP-8-specific characteristics. I'd like to see an architecture match a handful of these before I'd call it "PDP-8-like." I'm not very familiar with the 8080, but I don't think it matches any of them.
- Absence of any provision in hardware for moving more than one word at a time.
- Fixed word length.
- Absence of any hardware support for a stack.
- Addressing architecture in which the address space is divided into pages, and every addressable instruction can address: directly within the current page; directly within page 0; indirectly within the current page; indirectly within page 0.
- Absence of any conditional branches; that is, branches that have a target address. All conditional operations are implemented by a single "skip" instruction, which transfers control to either PC+1 or PC+2, on conditions microprogrammed by individual bits within the skip instruction.
- Extreme economy of instruction set; only six addressable instructions in the case of the PDP-8
- Word length that is not a power of 2.
Any similaries between microprocessors and the PDP-8--other than the Intersil 6100, of course!--are simply the result of having to solve the same problems, i.e. build a very small computer with limited hardware resources. The PDP-8 and 8080 are no more similar than the PDP-8 and the LINC, or the PDP-8 and the CDC-160.
This is quite different from the situation with respect to, say, Digital operating systems like OS/8 and CP/M and MS-DOS. If you compare these three, you notice things like
- the strange use of CTRL-Z to mark the end of a text file, particularly strange because according to the ASCII spec "end of medium" is CTRL-Y, not CTRL-Z--probably a mistake on Digital's part, but certainly an idiosyncrasy
- the use of the CR-LF pair to separate lines
I think this was typical for many computer models using dumb printers / terminals / teletypewriters as peripherals in the 1970s. Bold characters and underlines on a printer were the consequence of using CR and overprinting the line. A new line REQUIRED an LF. An LF without a CR continued the text on the next column in the next line. Unix was the first place I encountered the simplification to a single character. Hrcohen (talk) 02:00, 9 October 2008 (UTC)
- the use of file names consisting of a name and an extension separated by a dot
This is a convention also followed in Honeywell/GE GCOS and Burroughs MCP mini-computers and I think was incorporated in the conventional ASCII tape format. I would also note that the file name never required an extension. Only that extensions were desirable for several reasons.
- the prohibition of space characters within file names
All of these characteristics are shared by OS/8, CP/M, and (early) MS-DOS. None of these characteristics are shared by the Apple ][ DOS 3.3; only one of them is possessed by UNIX.
Thus, I think it is very reasonable to see evidence of Digital OS influence in microcomputer OSes. I do not see any obvious influence of the PDP-8 on microprocessors. Dpbsmith (talk) 20:51, 23 June 2006 (UTC)
P. S. And, regardless of what I or anyone else might have to say, in any case the verifiability and no original research policies mean that anything said on this topic, however well-founded, must be traceable to a published source, not stated without attribution. Dpbsmith (talk) 20:54, 23 June 2006 (UTC)
- I can assure you that the 8080 and PDP-8 ISPs (and their respective assembly languages) have damned little in common, aside from the facts that they're both (basically) accumulator-oriented load/store architectures.
- Atlant 22:53, 23 June 2006 (UTC)
Just thought I'd add a pointer to a query about the logo here. I'm probably wrong, but anybody know for certain? I'm intrigued now, and don't have any DEC logos handy to leer at. Chris 23:06, 25 July 2006 (UTC)
- The question of the "dots" verses the "squares" above the "I's" in the red logo is correct.
- When Bob Palmer took over, the advertising company he hired made the change. They changed it to red and replaced the squares with dots. It was something like DB Neadham agency. They also tried to give DEC a modern feel by playing a fast changing graphic commercial to Lenny Kravitz, Are you gonna go my way. —The preceding unsigned comment was added by 188.8.131.52 (talk • contribs) .
DEC did sell Intel-based Windows compatible computers under their logo (for example, Prioris, Celebris and Venturis product lines). --184.108.40.206 16:59, 7 August 2006 (UTC)
450 or 350 million dollar?
"AR&D later sold its investment in Digital for approximately $450 million, certainly the best VC return ever at the time"
"When Doriot sold his share in 1972, it was worth $350,000,000"
The word "Computer" banned.
It is quite interesting that Digital excluded the word "Computer" from their product and brand names, both, as an early way to get funds, so that investor were not afraid to give money to a company that would compete against giant IBM (besides other large computer companies were losing money at that time), and as an early marketing strategy to sell their products to engineering teams in companies where the accountants would reject computers.
Another version of this story, is that because computers were on one General Services Administration schedule and electronic equipment on another, that it would be easier for government-funded agencies and schools to buy DEC products if they were called something other than computers for purchasing approval. patsw (talk) 18:50, 15 January 2008 (UTC)
Mis-use of ARCHITECT as a verb
I dispute the use of the word 'architect' as a verb in this otherwise well-written article, because it is not a verb. http://en.wiktionary.org/wiki/architect http://www.google.com.au/search?q=define%3A+architect There are plenty of other verbs that can be used in place of 'architecting' and 'architected': construct, design, create, devise, invent, plan, produce. JeffreyMeunier 01:05, 8 October 2006 (UTC)
Writable control store
Atlant removed the information on writable control store, and in the edit summary asked which models supported it. WCS was optional on the 11/780, and standard on the 11/785, 11/750, 8600, 8650, 9000 series, and possibly other models. Microcode development tools were offered as a product for the 11/78x. Most of the later VAX implementations based on VLSI processors had small a on-chip WCS intended to patch microcode bugs. --Brouhaha 00:35, 12 December 2006 (UTC)
Could someone add information on this chip? Drutt 08:35, 15 May 2007 (UTC)
The word "computer"
In the '70s I was a DEC customer. I was told that the PDP line (Programmable Digital Processor) and the DEC name (Digital Equipment Corporation) avoided the word "computer" because government procurement procedures for "computers" were far more complex than the procedures for almost anything else... by avoiding the word "computer" procurement of DEC products was easier. Could be urban legend?
RobertTaylor21 21:00, 13 June 2007 (UTC)
- If you can find a cite for this, we can include it in the article --rogerd 21:24, 13 June 2007 (UTC)
- The original venture captitalists wouldn't fund a company that intended to make computers, believing that there was no room in the market for another competitor to IBM and the "BUNCH". So Digital built "Programmed Data Processors" instead of computers. I think Digital at Work contains this story; I'll try to remember to look.
- Atlant 13:59, 11 September 2007 (UTC)
Name of the Company
I was hired by Digital just before its acquisition by Compaq, and all the legal documents that we changed over to "Compaq Computer Corporation" had used the name "Digital Equipment Company" [not Corporation]. Digital was indeed a corporation, but the formal name needs to be verified -- especially since "Corporation" is used in the title of the article. --NameThatWorks 18:43, 10 September 2007 (UTC)
- Well, I'm looking at the "2211 Memory Exerciser Manual", copyright 1962, and it very clearly says "DIGITAL EQUIPMENT CORPORATION". (It's a cool manual because it has the old bi-color "|d|i|g|i|t|a|l|" logo where the d, g, t, and l are white on black rectangles while the i,i, and a are on grey rectangles.) Even earlier than that, I'm holding in my hand a "103" Laboratory Module (yes, the original product) and it says "digital equipment corporation" along with two big, bold "dec" logos.
Of course, by the time you were hired, GQ Bob had been running the joint and it wouldn't surprise me one bit that he'd didn't even know what the name of the company was.
- Atlant 22:40, 10 September 2007 (UTC)
- I guess I wouldn't rely on the company name in marketing pubs or tech manuals from 1962. The acquisition by Compaq occurred in 1998, and a lot of restructuring happened between 1962 and then. As I said, ALL the legal documents that were included in our bids and proposals gave the legal name as Digital Equipment Company. --NameThatWorks 17:36, 27 September 2007 (UTC)
I can't be sure at this late date, but I think when I started work for DEC in the UK (1977) I was working for Digital Equipment Company, a subsidiary (or something) of Digital Equipment Corporation. --dave, 2009-07-01 22:20-0400 —Preceding unsigned comment added by 220.127.116.11 (talk)
I noticed a few details that maybe should be corrected/cleared up. First of all, I don't believe the DECtape was created for the PDP-10. I think the DECtape actually predates the PDP-10. If I remember correctly, the DECtape was pretty much based straight off from the LINKtape, with a few changes. It definitely exists for PDP-8, PDP-10 and PDP-11 systems (since I've observed those), but I think it was used on 18-bit systems as well.
Second, the C language didn't originate on the PDP-7. It might just be that I'm reading too much into that section, but to me it appears as if that is claimed. Unix ran on the PDP-7, yes, but that was written in assembler. The same is true for the early versions of Unix on the PDP-11. It was only rewritten in C after a few iterations on the PDP-11 had already been done.
Under the list of what happened to various DEC produects, I'm missing the disk manufacturing. Sure, DEC retained StorageWorks, but the actual disks was sold. If I remember, that went to Seagate.
Oh, and could someone find a proper logo? Blue, that is, not the (pretty recent) burgundy or whatever the color was called... :-)
- Our DECtape article claims that it dates back to the PDP-6. It was certainly sold on at least some members of the 18-bit family, probably as old as the PDP-9.
- The bulk of the storage group (i.e., "Colorado Springs") was sold to Quantum (which itself was subsequently sold).
Other innovations and achievements
Having just finished reading Ed Schein's book on DEC (DEC Is Dead, Long Live DEC - Schein 2003) I wonder whether this article might be enhanced by some additional emphasis on DEC's role as one of the first learning organisations. As well as the technology legacy, Schein observes that DEC was one of the first companies to explore democratic management, and made significant innovations in community relations and corporate social responsibility, affirmative action, people development and human resource practices, leadership and management, team building (including virtual teams supported by the internal network Easynet), sales methods and practices, and even marketing innovations such as DECworld. Redwaterjug (talk) 21:06, 8 February 2008 (UTC)
- Weak arguments for the following reasons...
- 1960, and 'the sixties', are at least about 200 years after corporate 'learning' began in North America.
- The list of positive highlights you present here are extremely subjective.
- The list of positive highlights you present here led to the company's failure while the market was exploding.
- The article already skirts the 'neutrality' guidelines for Wikipedia.
- markB (talk) 05:16, 2 January 2013 (UTC)
Maybe include an example of DEC culture.
From around 1984 onwards DEC employees used Notes-11 (later called VaxNotes) to conduct what oday is called Web 2.0 Social Networking. From 1985 until I left in 1993 I used Vaxnotes in much the same way I use LinkedIn Professional Groups today 25+ years later. VaxNotes didn't have the invitation to join my network feature however.
See http://mikeg.typepad.com/perceptions/2009/01/why-history-is-relevant-to-the-future-of-collaboration.html The section titled What the story about VxxNotes tells us...
- I request more objectivity and less redirection from this encyclopedia to a commercial product's marketing page.
- 'Type Pad' reference links are commercial promotion. Wikipedia is not the place for such.
- Implying that this company invented social networking and web 2.0 is ridiculous.
- markB (talk) 05:34, 2 January 2013 (UTC)
Ridiculous sounds quite emotional to me. Can you explain why you think the idea is ridiculous? I was a regular VAXnotes user from '89 to '95. And indeed, it was used very much of a social network. Just without the images and the information pushing. Fungee (talk) 13:36, 12 March 2014 (UTC) Fungee
Structure of article
This article severely needs to be restructured in chronological sequence. Grouping products by the architecture type does not really accurately reflect how this company developed, unlike, say, Intel. Also, one tidbit that should be added -- DEC produced some very hard drives for a long time. During the late 1980s and early 1990s, they were among the best (if not the best) in the business. As part of the dismantling of the company in the early 1990s, that group was sold to Quantum. The Quantum Atlas line was the result. This facility passed through Maxtor, and then to Seagate. The DEC Shrewsbury facility still operational. 18.104.22.168 (talk) 05:27, 24 February 2008 (UTC)
It seems that HP has retired the server that runs ftp.digital.com. The announcement can be found here: http://h18002.www1.hp.com/alphaserver/options/asgs1280/asgs1280_options.html
The ftp service is now inaccessible. The ftp.digital.com link(s) at the bottom of this page should be replaced with another link that provides the same information if possible, or removed. Rilak (talk) 08:08, 28 February 2008 (UTC)
Closing DEC's business
The sentence "However, Palmer was unable to stem the tide of red ink." has been bothering me for some time, and I was wondering if anybody had any comments about this: I know we need to keep away from anything that's perhaps largely based on opinion and possibly rather subjective, which is why I'm not inclined to change anything myself since I'm not certain the exact facts of the matter, but wasn't Palmer actually responsible for much (most?) of the red ink in question? I do recall that the number of vice presidents doubled (or more) under Palmer and the number of managers appeared to rise exponentially during the swathes of layoffs; and Palmer's own bio here on Wikipedia does indicate his fondness for restructuring, something that was a constant blight on the company to the point where many people were simply unable to do their jobs. And apart from anything, "was unable to" seems a curiously disempowering statement for the man who was in total control of the organisation.
I'm aware that we must be neutral and not allow personal feelings to negatively influence what's documented, but I think that the article goes much too far the other way, it's far too favourable and actually exonerates Palmer of his responsibilities. While I think that Wikipedia's account is far less inaccurate than other versions of DEC's demise, some of which are just blatantly untrue (though I figure it probably isn't helpful to comment further!) I think that this section could probably do with some attention. —Chris (blather • contribs) 15:29, 23 June 2008 (UTC)
- I agree and disagree with you.
- The entire sentence 'However, ... ink' is subjective. It could read '.. Yet, losses continued'.
- You allude to having detailed historical knowledge of this company by means of mentioning inaccuracies of numerous and presumably disassociated commentaries regarding the company's demise. You remind me of just how difficult it is to remain neutral and objective. I wonder if you struggle with the same concept. Wikipedia is not the place to champion for redress or vindication. You mention Palmer in the light of one who likely escaped blame for something, and this blatantly indicates you are probably championing for another individual indirectly. Was he 'fond of restructuring' or did he often restructure. There is a difference.
- markB (talk) 06:16, 2 January 2013 (UTC)
I think it's incorrect to credit DEC with the idea of not using letters which are homoglyphs or near-homoglyphs. I expect there are several examples of prior art - one that springs to my mind are United Kingdom military aircraft serials, which have omitted the letters C, I, O, Q, U and Y since 1940. Letdorf (talk) 09:56, 9 October 2008 (UTC).
Citation for Assertion about the X Window System
Besides tons of messages in mail archives, I gave an invited talk at Usenix some years ago to cover (some of X's) history. I leave it to someone who knows wiki markup better than me to add it to the main article JimGettys (talk) 21:31, 19 November 2008 (UTC)
Famous people who worked at DEC
You did not like my coffee-mug?
On 2 apr 2009 10:55 User:Badger Drink removed my original/historical Digital coffee-mug?
- Ken Olsen's primary concern about customers and employees "Our Employees are our greatest Asset" was distributed on a coffee mug, to encourage all employees.
I do like your coffee mug picture, as evidence of Ken Olsen's attitude towards DEC employees. I suggest moving it to the biographical article about him, which is sadly in need of good pictures, including a picture of the man himself. Reify-tech (talk) 20:15, 21 March 2011 (UTC)
64 bits binaries are bigger and adressing 64 bits was unnecessary
I added this sentence concerning the DEC Alpha but it has been suppressed because of lack of citations:
- A 64-bit CPU was also too early in the sense that 64-binaries were about 3 to 4 times bigger than 32-bits binaries and memories and hard drives were costly at that time. The ability to address 64 bits was also not seen as a priority even for scientific applications.
But there are no citation for the other parts concerning DEC Alpha and I remember very well discussing with DEC people and testing their first 64 bits binaries. It was obvious for researchers involved in computer science at that time. We have worked on Univac machine whichs were 36 bits, they did not add so much in comparison with IBM 32 bits for scientific computations. Also when the VAX came, its main advantage was virtual memory and its ability to swap its memory on hard disk. And thus adressing full 32 bits was just possible with a VAX, and it made its success. But adressing 64 bits was not a real need. I remember Digital saying: Yes, it is not the current need for scientists now, but it will come and we are offering you 64 bits machines at the price of 32 bits machines. And it wasn't true, because we needed to buy many more memory which was very costly. And thus if somebody find citations, the above sentence could be reinserted.--Nbrouard (talk) 16:50, 13 July 2010 (UTC)
- "3 to 4" is questionable, unless you can find some authoritative source which gives that, e.g., based on instruction set efficiencies of RISC vs CISC, with the numbers coming from the source. Otherwise, it's just hazy recollections Tedickey (talk) 19:23, 13 July 2010 (UTC)
- I do not think that the content in question is accurate or relevant. Firstly, why should a 64-bit binary be approximately three to four times larger than a 32-bit binary? Apart from the larger pointers and possible padding, the binaries should be similar in size if the 32-bit ISA and 64-bit ISA were similar. Am I missing something? I think that the point itself is inaccurate. If the previous version of the binary was compiled for VAX, and the subsequent version was ported to Alpha or translated, then the binary would be significantly larger. This is however not a consequence of the Alpha being 64-bit, but rather VAX being CISC and Alpha being RISC.
- Yes, both effects were interacting. And the RISC versus CISC was probably the main effect over larger pointers. Having an IBM/RS6000 as my first RISC machine, I remember much smaller binaries than on our Alpha 64 bit. But again, it could come from other reasons that I never investigated. My main point was that the cost of memory was very high when the Alpha came and it was a very sensitive issue: 1 Mb of memory was 5,000 $ in 1977 and 6$ in 1994 ().--Nbrouard (talk) 09:58, 24 July 2010 (UTC)
- Secondly, I do not think that the point regarding 64-bit addressing being not required when the Alpha was introduced is relevant. Alpha was designed to last 25 years per the paper by Richard L. Sites published in Communications of the ACM. If I remember correctly, the inclusion of 64-bit addressing capability was a direct consequence of predictions showing that by the mid- or late 1990s, the ability to address more than 32 bits of memory was nessecary in both high-end commercial and scientific applications.
- Everyone agreed on the utility of addressing 64 bits in the future. Altavista was probably the first example of such use. But for researchers, moving from mainframes to workstations, it was not so urgent. Even today, I still prefer 32 bit Linux to avoid many unsolved problems. Mac OS/X just implemented its first 64 bit OS in late 2009.--Nbrouard (talk) 09:58, 24 July 2010 (UTC)
- Ironically, if I remember correctly, the first version of the Alpha ISA did not support a full 64-bit address space, several bits were reserved, and these could be used to implement a full 64-bit address space in the future. In addition to this, none of the Alpha implementations have ever supported 64-bit addressing in virtual or physical form.
- As Tedickey has said, the content in question is questionable and requires citations to authorative sources if it is to be included in the article. Additionally, the lack of citations for the rest of the section does not necessarily mean that it is inaccurate, and it does not mean that questionable claims should be inserted. If there uncited claims that should be scrutinized, then they should be pointed out here. Rilak (talk) 11:48, 19 July 2010 (UTC)
Major edits made during March 2010
In March 2010, a series of major edits were made that introduced major changes into the article. One of the changes made greatly concerns me. It is in regards to DEC's PC product lines during the 1990s. If one looks at the revision from 25 Feb. 2010, one will find that DEC's PC product lines of the 1990s are mentioned. This is very important because DEC's Alpha business only constituted less than a third of the company's revenue, IIRC. The other major hardware product line were PCs. The omission of these product changes how the article reads. Before, it was: "DEC failed in early its PC attempts, but it eventually got out some PC product lines during the 1990s." Now it reads like this: "Stupid DEC failed in its early PC attempts, it never tried to get back into the PC market, and left only with silly overpriced closed VAX systems, DEC failed. What a bunch of dinosuars! LOL!" Not NPOV is it?
I would fix it myself, if there was a section that I could put it in. But the article structure was altered significantly, and it is not much better from the previous structure. The issue remains the same. Trying to tell DEC's history by organizing it into sections around product lines mixed with out-of-place sections about specific happenings doesn't work. Why? Because DEC's product lines overlap each other. It also encourages work on the products, not the history. And it is a problem as people for some reason don't seem to realize that Wikipedia has articles on the products, so they add detailed information about products into an article that was not meant to be about the products.
The article needs to be rewritten entirely. The present version is poorly structured and it misses the point. For example, if one were to look in Electronics, the articles from the late 1980s say that DEC got successful from one ISA, one OS, one product line from workstations to superminicomputers, networking, and VAXclusters. And then it began to decline because of client/server computing, open platforms, and RISC microprocessors. The current article doesn't say this explicitly, it barely even mentions these points, and it implies that what caused DEC's demise were "business microcomputers". In regards to structure, the history should be organized by decade, not by product. This will allow the article to read more smoothly and allow it to show relations between events and products that are presently disjointed.
Proposed major restructuring of article
I think Rilak's comments above have considerable merit. A better overview of the evolution of DEC should be presented in this article, beyond a simple chronicle of different product lines. For example, the long-standing influence and then departure of Gordon Bell at the highest levels of DEC management was a significant factor in the rise and fall of DEC. Bell reputedly was one of the few people within DEC who could tell Ken Olsen to his face that he was completely wrong, and still be listened to. His involvement in multiple product lines was more than incidental, in that he guided much of the "big picture" strategic technical vision of the company. As Rilak points out, DEC continued to make products for the PC market into the 1990s, and these products were an influential presence in the market, even if they failed to achieve market dominance.
IIRC, the book by Edgar Schein et. al. "DEC is Dead, Long Live DEC" was chronologically organized. I don't have a copy at hand right now, but I remember that the book's coverage correlated well with my experience as a long-time DEC user and acquaintance of many DEC employees. The book is already listed as a reference for many DEC-related articles, and is a must-read for anybody interested in better understanding the rise and fall of DEC, as well as its continuing legacies.
Rather than the daunting (and possibly controversial) project of reorganizing the entire article at one fell swoop, perhaps it would be better to start a new section organized chronologically, as a parallel overview of DEC's history. It could absorb some of the existing non-product-line sections of the article, and also host new content such as the history of DEC's PC efforts in the 1990s. If and when it becomes apparent that the product-line oriented approach has been superseded, the remaining content could be subsumed into the chronological narrative, or broken out as separate subsidiary articles.
- Please remember that any new material should reference reliable sources. Schein doesn't go into a lot of technical detail - better technical history material can be found at Gordon Bell's website  particularly the DEC timeline . Regards, Letdorf (talk) 22:18, 21 March 2011 (UTC).
- Agreed. It's easy for articles to become list of individual things, and completely miss the larger picture. What brought me here was the use of Template:Hard_disk_drive_manufacturers when hard disks are never mentioned in the article! Did DEC actually manufacturer these at any point? I know they were big with tape, and sold DLT to Quantum. --Juventas (talk) 20:39, 19 June 2011 (UTC)
- DEC did manufacture its own disk drives, as well as buying them from other manufacturers (e.g. Memorex for RP series, CDC for RM series). IIRC, they had a disk and tape division in Colorado Springs (Colorado Memory Systems Division??), which developed hard disk systems (removable and fixed), and also developed the DLT (Digital Linear Tape) systems. I think the disk division started with the RK05 (14-inch single platter, removable cartridge, 2.5MB). They managed to keep up with the rest of the industry until the late 1980s, when DEC's financial problems started to impede their ability to invest in advancing the technology. They also had reliability problems with the RA90 hard disk. When DEC started to founder, they sold the entire disk/tape division to Quantum Corporation to raise quick cash. All this is written from memory, as I don't have my old DEC literature at hand (it's in storage). Anyway, the whole disk/tape division story is a saga in itself, significant but a sideshow in the whole story of DEC. Reify-tech (talk) 00:52, 20 June 2011 (UTC)
The DEC article as it stands now isn't that bad, but it suffers from large gaps in its coverage (e.g. the disk/tape division) and needs better integrated "big picture" coverage of the mid 1980s and later. Perhaps the article should have a — I hesitate to say this — "matrix" organization? I mean by this that a historical/chronological approach should be built up, spanning the many DEC product lines and research efforts. This would complement the existing product-oriented coverage, without requiring it to be completely torn apart and rebuilt on the fly.
Unfortunately, I don't know of any secondary-source definitive history of DEC that really fits the bill. The books by Schein, Bell/Mudge/McNamara, and Ceruzzi are the best sources I know of so far, and they all have their limitations. Any other suggestions?
I'm not up to taking the lead on this whole thing, though I could help out in areas I have some experience with. I'm just reluctant to start tearing the current article apart, without the certainty that something better will immediately be built in its place. That's why I'm suggesting a parallel build with a chronological framework, while leaving most of the existing structure in place. Reify-tech (talk) 00:52, 20 June 2011 (UTC)
unsourced/exaggerated synopses of notable people
- Apologies for the sloppy edits; I was in a hurry at that time. Of course there were many other people working on those projects. I've modified the summary descriptions, and hope they are more accurate and acceptable. You're certainly welcome and invited to improve them further. --Reify-tech (talk) 01:52, 24 May 2012 (UTC)
Linux not DEC-accomplishment
The couple of unsourced comments about Linux don't belong in the section on DEC's accomplishments. Further, the two sentences don't belong together, since they apparently equate events that took place several years apart. TEDickey (talk) 01:15, 27 July 2014 (UTC)
- Yeah, DEC has a lot of accomplishments to its credit, but Linux is definitely not one of them. Reify-tech (talk) 02:48, 27 July 2014 (UTC)
- Compaq was actively developing core modules for Linux!
- I do not agree... I would propose to again add the paragraph, and to refer to http://www.informatica.co.cr/linux/research/1999/0302-a.htm where it clearly is indicated that Compaq was actively participating during the period 1994-1999 into the Linux development... both with hardware and software Geertivp (talk) 20:36, 31 July 2014 (UTC)
- And here you have the reference for Red Hat 7.2 http://www.informatica.co.cr/linux/research/2002/0108.htm (January 8, 2002). Also the Gentoo distribution was ported to AlphaServer... So, please add again the deleted paragraph about Linux on AlphaServer... Geertivp (talk) 21:22, 31 July 2014 (UTC)
- It would have been nice if your change comment had reflected that you modified the content (reverting is different). By the way, your source is a press release, which isn't really that good in itself, since largely those are promotional and not close to fact. TEDickey (talk) 22:44, 31 July 2014 (UTC)
I still fail to see how either could be fairly categorized as a DEC accomplishment. It was first developed for x86 and then ported to many systems; what makes this port a DEC accomplishment particularly when it is not clear that DEC did the porting? Likewise, its hard to see how LINUX on an alphaserver after DEC disappeared is a DEC accomplishment, Compaq, maybe, but even there is a port an accomplishment worthy of Wikipedia inclusion. If everything done on a DEC system is a DEC accomplishment then the list needs to be a lot longer, so I still support removing the two cites. Right now there does not appear to be a consensus - let's see what others say. Tom94022 (talk) 23:50, 31 July 2014 (UTC)
- The given change and its source are promotional rather than encyclopedic, and lacking genuine third-party sources, there's not much to discuss. TEDickey (talk) 00:02, 1 August 2014 (UTC)
- I would propose to move the paragraph to the 32-bit MIPS and 64-bit Alpha systems section... still it is a worth-full fact, even if you would not consider it a huge accomplishment...
- Evidence: Compaq released multiple kernel modules under General Public License (GPL) and assigned full-time engineers to help the Linux community to port and optimize Linux on the Alpha platform.
- Geertivp (talk) 06:25, 1 August 2014 (UTC)
- you are not responding to other editor's comments. here for instance is a copy of the original press release. As noted, your suggested sources are sketchy, promotional, and based on a timeframe after the range when DEC was an independent actor separate from its acquisition by Compaq. TEDickey (talk) 07:51, 1 August 2014 (UTC)
- In May of 1994 Linus Torvalds was financed by DEC's Unix Marketing Group to attend DECUS in New Orleans at the urging of the Chair of the Unix Special Interest Group. Mr. Kurt Reisler. There Linus gave two talks about Linux and showed an Intel 32-bit distribution based on his V1.0 kernel to Jon_Hall_(programmer)#Career, who then went back to Digital and arranged for Linus to receive a DEC Alpha machine. Jon_Hall_(programmer)#Career saw a great use for Linux on a 64-bit, RISC architecture to be used for computer science research in large address spaces. Jon_Hall_(programmer)#Career also found a small group of programmers in the DEC Semiconductor group headed up by Andy Riebs, Jim Paradis, Jay Estabrook and David Rusling (among others) who did significant amount of engineering work to help with the porting of the kernel. David Rusling wrote a Free Software bootloader for the Alpha called "Milo", because the boot-loader that came with the Alpha was not Free Software. Jon_Hall_(programmer)#Career also arranged for the Digital Semiconductor group to have access to another DEC Alpha processor for their work. DEC Alpha machines were very expensive in 1994, not as inexpensive as many computers are today, and many DEC engineers did not have access to an Alpha for their own use on operating systems like VMS and DEC Unix. In addition, DEC helped the port by making technical documentation available to members of the Linux community. In November of 1995 Digital sponsored the production of 4000 CDs of Red Hat 2.1, the first version of Red Hat to run on the Alpha, and gave these away at trade shows and conferences around the world. Many engineers in the Digital Unix Group of DEC were already using Linux on their laptops to do work, and contributed to the port. This port was also the first that Linus Torvalds actively participated in. Before this the ports to other processors were forks copied off the Linux kernel tree that other people were porting. For this port, Linus created an "N" way (as opposed to two-way) source code tree so other ports in the future would not be as difficult. It was also the first time that Linux was positioned as either a 32-bit or 64-bit kernel. Later, after distributions of "Alpha/Linux" started to stabilize, DEC also formed a group called "Computer Special Systems" that sold Beowulf Clusters made of DEC Alpha Systems. This was a fairly successful business. It was January of 1998 when Compaq announced it was buying Digital Equipment Corporation, and fully a year after that before Digital was assimilated. During that time we found that some Compaq engineers and groups were also supporting Linux. Maddoghall (talk) 23:05, 20 November 2015 (UTC)