Talk:PDP-11

Digital sold UNIX for the PDP-11?

Can anyone verify that Digital sold UNIX for the PDP-11? I'm tempted to move UNIX to the third-party section.

--Rochkind 8 July 2005 04:41 (UTC)

I don't think Digital resold Unix directly, but Unix was absolutely available for the PDP-11; this was Thompson and Ritchie's second platform after the PDP-7. (See the picture at the top of this article!) DEC's "Telco" product Line (TPG) was essentially devoted to the Unix business. By the time of Ultrix on the VAX, Digital was directly reselling Unix.

Atlant 8 July 2005 12:49 (UTC)

To put it more precisely: At the time of Ultrix-32 they also sold Ultrix-11 for the PDP-11. Ultrix-11 was essentially BSD-2.9 IIRC. See PUPS archive at http://minnie.tuhs.org/PUPS/ --Glasreiniger 07:29, 11 July 2005 (UTC)

Thanks for the info! --Rochkind 18:29, 15 July 2005 (UTC)

When I was at Bell Labs in the early 1980s, my group had a PDP-11/70 that ran PWB Unix 2.0 (and updates) JHobson3 (talk) 14:51, 28 December 2018 (UTC)

Yes, but that wasn't purchased from DEC; you got that from another part of Bell Labs. This section is discussing the UNIX sold by DEC for the PDP-11, i.e. Ultrix-11. Guy Harris (talk) 18:51, 28 December 2018 (UTC)

That paper tape picture

This picture, while a very nicely done picture, suggests that the PDP-11 was stuck in the Stone Age. The picture of the old PDP-11 with DECtape drives doesn't help either. In fact, later models of the PDP-11 worked with a wide variety of MASSBUS peripherals including RP06 disk drives. Does anyone have some pictures of such? Jeh (talk) 01:02, 22 May 2008 (UTC)

A picture of newer/larger peripherals, or perhaps a picture of a large 11/70 system, would certainly be nice. Then again, the existing pictures are also appropriate, reflecting the first few years of the product line. Paul Koning (talk) 16:47, 21 August 2008 (UTC)

Splitting off architectural details

Following the example of the article on the Atmel AVR, I have split off former Section 5, Architectural details, into a separate article PDP-11 architecture. This section was about half of the total PDP-11 article, which had gotten big, and would be of exclusive interest to techies, while the remaining text is of interest to persons studying the PDP-11 as a phenomenon.

The only changes I've made during this pass is that I didn't think the various optional instruction-set packages deserved separate mention in the Table of Contents; they are all now in Section 3.7, Optional instruction sets. I've changed capitalization in some of the section heads to be more like Wikipedia, and created links between the two articles. For now, I won't delete any of the See Alsos; other editors who know the works pointed to may decide they belong in one article and not the other. --Spike-from-NH (talk) 00:16, 15 February 2009 (UTC)

Afterthought: I've also moved the entire section on "PDP-11 lore" to PDP-11 architecture as well. It belongs there (if it belongs anywhere!) as it denies a "folk myth" on the impact of the instruction architecture. The section on MACRO-11 might deserve the same thing, but I'll wait to see if there is buy-in or outrage at this surgery. --Spike-from-NH (talk) 00:37, 15 February 2009 (UTC)

In fact, there has been no comment at all, so the section on MACRO-11 is now moved. --Spike-from-NH (talk) 22:04, 27 August 2009 (UTC)

Memory-size timeframe

The term "common" doesn't appear to fit, since multiple-megabytes for a minicomputer were arguably rare in the late 1970s when the VAX was introduced. Suggest that the comment be toned down to reflect that Tedickey (talk) 21:00, 25 August 2009 (UTC)

A comparable assertion (which doesn't assert that megabytes were "common") is made in the first paragraph of Section 3, "The Decline of the PDP-11." A separate problem with the sentence in the intro is that it asserts a fatal problem with the PDP-11, but instead of explaining it, directs the reader through the link "64k barrier" to the 16-bit article, which nowhere explains the problem or the "barrier," or relates 64k to 16-bit addressing except at the end,in examples involving the Intel architecture. I recommend you simply delete the sentence. --Spike-from-NH (talk) 04:12, 27 August 2009 (UTC)

The basic design of the PDP-11 was sound

A recent addition in the 'Decline' section was 'The basic design of the PDP-11 was sound, except for its 16 bit logical address space' (addition is underlined). While this contributed to its decline in the late 1980's, I would say that making it have 16 bit logical addresses was a quite reasonable decision for circa 1970. Memory was expensive and the PDP-11 was always meant to be an affordable machine. In fact, a larger logical address would probably have required further expensive hardware to implement and would not fit well into 16 bit general purpose registers and would not fit well in the instruction format. Thoughts?  Stepho  talk  05:14, 7 June 2011 (UTC)

I took it out, it's at least a little ungracious criticism of a design that lasted for (a couple of decades...hey, there's no HISTORY in the article. When was the first PDP 11 sold? When was the last one sold? How many were sold?). If you want to criticize a short-sighted design, someone should write the article "Blunders in the design of the IBM PC" - if only IBM's team had known they were going to sell more than 10 or 20 thousand of the beasts... --Wtshymanski (talk) 14:17, 7 June 2011 (UTC)

It's depressing to do a Google search for "Digital Equipment Corporation history PDP-11" and have the *second* hit be the Wikipedia entry you just updated 3 mintues before...--Wtshymanski (talk) 14:34, 7 June 2011 (UTC)

(actually I usually find my Wikipedia entry is the number one result. Not depressing but scary to me!) W Nowicki (talk)

It wasn't a sound design, even in 1970. Memory was selling for around $1000/kilobyte then and dropping fairly fast. Affordability has nothing to do with it. Putting $64,000 worth of memory in a minicomputer based system was not unheard of and Unibus was designed from the beginning for 18-bit addresses (256KB). Designing something that would max out in a couple of years was a foreseeably bad decision. As for longevity, the PDP-11 came out in 1970; The Motorola 68000 in late 1979. By then no one in their right mind was designing new systems around the PDP-11. I had several arguments with DEC people about this in the 70s, including one with Gordon Bell. They were all hardware people and their argument was that any software bigger than 64K should be segmented. Later on VAXen were sold, and priced, as midrange systems, and that kept DEC alive for another decade, until PCs got powerful enough to compete. If the PDP-11 had been designed with a larger address space, we'd probably still be using them. Saying "the basic design of the PDP-11 was sound" is pure POV, and if we can't agree on wording, it shouldn't be in the article.--agr (talk) 15:02, 7 June 2011 (UTC)

Depends on if you're a comp sci weenie or running a business, I guess. They sold 170,000 of them in 10 years - that was a *lot* for those days. The comany went from a few thousand employees to 100,000. They made a bucketload of money selling PDP-11s. And when they stopped selling PDP-11s, the company essentially died! Could we find some citation somewhere that says the PDP-11 design was UN-sound? --Wtshymanski (talk) 15:11, 7 June 2011 (UTC)

In fairness, we should compare the PDP-11 to other contemporary designs of the time. The PDP-11 was based on a 16-bit word length and byte addressing, so it made logical and economic sense at that time to have a 16-bit address space. The extra 2 bits on the Unibus were viewed as enough "future-proofing", compared to other competing architectures. When DEC realized they needed still more address space, they added memory mapping hardware to extend it to 22 bits as an interim measure, and began working in earnest on the 32-bit VAX architecture. Gordon Bell and others commented extensively on the PDP-11 design tradeoffs in DEC's "Computer Engineering" book (Bell, Mudge, and Gray??? — I don't have my copy handy right now). The article needs more footnotes, and this book is a rich source, as well as the Cerruzzi book.

In the retrospective light of history, we should be careful not to harshly criticize an original design by rigorous modern standards from decades later. The mark of a "sound design" is its ability to be extended and adapted to future needs without requiring too much baroque and convoluted "kludging". For example, compare the "protected mode" and address extension implementations of the much-later IBM PC designs. The sheer amount of engineering time and end-user-visible kludging effort in extending the early PC designs stands in marked contrast to the relatively smooth transitions DEC offered its longtime customers, throughout the PDP-11 family, into the VAX family, and into the Alpha series.

The PDP-11 was announced as the "16-bit minicomputer architecture for the 70s", and the VAX was billed as the "32-bit superminicomputer architecture for the 80s" (DEC marketing brochures, again not at hand). DEC's 64-bit Alpha architecture was delayed by the departure of Gordon Bell (early 90s?) and managerial indecision, which put DEC off its pace, eventually dooming the company. Reify-tech (talk) 16:08, 7 June 2011 (UTC)

Obviously "bucketload" is not encyclopedic. Come on folks, there have been articles written on the PDP-11, so cite them. Gordon Bell's papers and books are a start. The policy clearly states any information not supported by a citation can be removed at any time. This whole section needs to be rewritten in encyclopedic language and be more precise. For example, by "design" do you mean architecture or implementation? Orthogonality of the instruction set as well as mass production are already discussed. And if you mean "commercially sucessful" then those words should be used, with citations to actual sales figures ro at least reliable published estimates. W Nowicki (talk) 16:19, 7 June 2011 (UTC)

We could always make a list of all the computer architectures introduced in 1970 that were still being sold 20 years later...shouldn't be that long...--Wtshymanski (talk) 16:28, 7 June 2011 (UTC)

This was a fight between hardware designers and programmers. The former thought it was just fine for the logical address space to be smaller than physical memory. The latter knew better. IBM was the one company at the time who got it and their 360 architecture is still around 45 years later. The 68000 and the '386 are also still with us, 31 and 26 years later. The 16-bit address space sank DEC, DG and every other computer company that built its architecture around it. Other than that it was a great success.--agr (talk) 21:39, 7 June 2011 (UTC)

PDP-11's weren't in the same niche. An apt comparison would be IBM 1130, which from your comments one might expect to see a large user community still around. TEDickey (talk) 22:53, 7 June 2011 (UTC)

We need to think in the context of 1970 minis. The 360 was a mainframe, rather than a mini - mainframes were expected to be costly while mini's sacrificed some capability to be more affordable. The 360 architecture survived for so long partly because its instruction format was relatively sparsely packed. IBM could add new instructions and new addressing modes to allow for higher memory requirements. But the 11 used very tightly packed instruction formats (probably to keep lower instruction decoder costs and memory access time), which allowed very little room for expansion. That may have doomed them in the long run but in 1970 the economics of building affordable machines would have dictated these sacrifices.

Memory getting cheaper is obvious to us with 20/20 hindsight but wasn't so obvious back then. Fully loaded systems (eg the 11/70) were generally run as multitasking systems. An 18 bit unibus address with only 16 bit addresses per application made sense for multitasking many small/medium size programs. Higher demands usually called for a machine in the next higher niche (eg one of the small mainframes).

But no matter what my ruminations are, if we want to say 'sound' or 'not sound' then we need some solid references.  Stepho  talk  05:15, 8 June 2011 (UTC)

I am thinking in the context of the 1970s. I spent the entire decade dealing with 64KB addressable machines and it was a pain. You'd develop whatever, customers liked it, they wanted more capabilities, happy to pay for them, then you'd hit memory limits, have to reconfigure into separate core loads, performance plummeted, everything got harder. Fixing a simple bug could involve a major rewrite if that coreload was out of memory. (I remember one case where we had already changed the manuals to say the bug was a feature when I figured out how to shoehorn in the fix.) The designers of the PDP-11 were aware memory was getting cheaper, hence the 18-bit initial physical limit. Your claim that "fully loaded systems (eg the 11/70) were generally run as multitasking systems" is not the way I remember things. DEC sold a lot of PDP-11 boxes for turnkey and embedded applications, such as CAD, typesetting, process control, communications switching, etc. Timesharing, though popular at universities, was not their biggest market by any means. I'm not saying the folks at DEC were idiots, just that they made a mistake limiting PDP-11 logical memory to 64KB. They had a great run with the PDP-11, but in the end I think that limit killed the company. I'm not sure I understand your IBM 1130 reference. It had a 15 bit address space with 16-bit words, so it had the same problems. In any case, IBM killed the follow on to the 1130 because it was too capable (according to a former boss who ran the project -- I believe it was code named Bimini.) Bottom line, though, I agree with you that we should not be rendering a sound or unsound judgement without a reliable, neutral source. --agr (talk) 23:38, 10 June 2011 (UTC)

Your experience is both greater and earlier than mine, so I will defer to you.  Stepho  talk  08:50, 11 June 2011 (UTC)

On the other hand, Arnold's experience is about on a par with mine. Reliable sources are what we're after, not reminiscences TEDickey (talk) 10:04, 11 June 2011 (UTC)

I think it's worth pointing out that history is complex, but I quit agree that article content should be based on reliable sources.--agr (talk) 23:51, 12 June 2011 (UTC)

The PDP-11 was revolutionary at the time. It was a radical departure from any minicomputer that came before it. It had a clean instruction set (if you ignore "MARK", which I recommend). It wasn't intended to replace the PDP-10. I worked at DEC from the summer of 1981 until 1986. Memory was expensive and not very dense. It may have been "future proofing" to make the Unibus address 18 bits, though many people I worked with suspected that it being half the word width of a PDP-10 had more than a little to do with it starting at 18 rather than going to 24 or even 32 in the MMU segment mapping. Many microprocessor designs that followed were influenced by the PDP-11. The PDP-11 was an elegant design, but it hit an architectural dead-end. Filker0 (talk) 00:27, 9 June 2018 (UTC)

Sorry, POV only. When discussing the Unibus memory address limitations it should take into account the PDP-11/20 which was the first system in production was not microprogrammed. Therefore real estate available in the CPU for memory control was constrained. Core memory could be 4K words and was implemented with multiple modules (3) on the Unibus backplanes. Also Unibus was designed to be compatible with some of the earlier PDP-8 generation minicomputer boards. Due to the module backplane connectors used the Unibus and modified Unibus backplanes had a limited number of address lines available. Therefore backplane physical capacity for additional memory required expansion boxes, plus 18-bit addressing was future proofing.

The PDP11/10 (alternate designation was PDP-11/05) was the second system in production. This CPU was microprogrammed but the early production CPU boards did not have the physical capacity to extend memory control past 16-bits. !8-bits memory remained future proofing for Unibus.

Pdp11.caps11 (talk) 20:30, 21 March 2019 (UTC)

Reorganization

This article should be reorganized to be inclusive of ALL PDP-11s, then separated by bus implementation, Unibus and Q-Bus implementation differences. As currently organized, it almost suggests the Q-Bus systems were not PDP-11s. In fact, the 11/83 & 11/84 used the identical CPU board and private memory interconnect.

Ggdr Gerry G 16:13, 16 July 2011 (UTC)

Not sure what your specific proposal is. It seems to be mentioning the Q-Bus systems already? In fact, there is a whole section on the LSI-11. My question is if "LSI-11" was an official product name, an internal project name, or the name of the chip etc. I would say it is a bit odd to have the lists of models in the middle of the article. Would this work better in, say, a table? or just a simpler list, and more prose to give a narrative version? I would be interested in the years each one was announced or delivered, for example, as well as how they technically related. And of course any sources you can dig up would be greatly appreciated, thanks. W Nowicki (talk) 18:58, 16 July 2011 (UTC)

Mention of PDP-11 simulator on "Super 8 (film)" viral marketing site

The article can be found at Super 8 (2011 film). —Sladen (talk) 11:28, 2 December 2017 (UTC)

Ok, Spike-from-NH (and possibly others), I am in no way affiliated with that site. Why the hell would I care about the site? It's the fact that this was mentioned that is important. The fact that more people became aware of it. I don't know why you stubbornly keep on assuming that I have an ulterior motive for my addition. You can't just assume that people want to add things in because of "marketing" or "personal" agendas. All I want to do is put that people were able to know about this PDP-11 because of this usage of its interface or some other thing. In fact, that's how I came upon this article - because I read about it in the article on Wired, and went and searched it up, so I'm adding the fact that this was referenced as a sort of "back-reference" to the fact that this subject was mentioned, if you know what I mean. (This subject itself is so obscure and known to only a small number who have actually owned/operated one or are interested in it, so is there a problem with mentioning that people became aware of this computer because of a reference from something?) It's a big deal if this subject was referenced because so few people know about it. The content I added was supposed to document that fact that it was referenced, not for the links, which are only for verification and required citation. Maybe I might be weirdly and stubbornly clinging to the fact that this was referenced somewhere else in which more people found out about it, but still, a single sentence with a proper citation doesn't need to be removed. It's a freaking good-faith addition for goodness sakes! Why do you keep pushing it off and blocking? Comments and reasons appreciated, (and I did not understand a single bit of that last sentence you wrote in the edit overview summary, Spike. Please re-explain it in a clear manner.) - M0rphzone (talk) 21:48, 4 April 2012 (UTC)

I can say with confidence that no one using this article to study the PDP-11 cares that a viral video 20 years later used it as the theme of an advertisement, whereas the people producing the film and the ad campaign care greatly. The nature of this paragraph and its citation in this article is essentially trivia, which is discouraged (see Wikipedia:Handling trivia and Wikipedia:"In popular culture" content). I do not care who you are or aren't affiliated with. In your repeated attempts to plant this material into a Wikipedia article--and the above flame--you do seem to be using Wikipedia as a tool of a viral (def.: needing to attract eyes using any tactic) ad campaign. Spike-from-NH (talk) 23:47, 4 April 2012 (UTC)

Look here, the term "viral marketing" is a label that was used by the Wired article. Stop accusing me of having ulterior motives. I don't understand why you went for some personal attack by repeatedly saying I was attempting to market the site. Jeezus, all I was doing was using the same adjective! I don't care about marketing and I don't have a job yet or anything. Maybe I might be obsessing over a trivial thing, but still... I noticed the reference and the fact that this subject is relativity obscure. And seriously, why would I even care about any greedy monetary gain or any viral marketing. - M0rphzone (talk) 05:48, 5 April 2012 (UTC)

Hmm.... it seems that you possibly don't want certain info "tainting" the article or you want this article to have nothing to do with references from outside sources. Maybe that's why.... who knows. - M0rphzone (talk) 06:18, 5 April 2012 (UTC)

I'm willing to believe that M0rphzon is acting on good faith and is not related to the viral website. He has a point that the PDP-11 was mentioned on the website and that it introduces the PDP-11 to a new generation. However, WP:TRIVIA points out that unless that mention was important to the PDP-11 or public awareness of the PDP-11 or changed the world in some significant way then it should be discounted as useless trivia. Looking at the website, I see what appears to be a PDP-11 simulator booting. It shows boot messages containing 'RT11' in them and fragments of MAC-11 assembly code. But it also shows fragments of Unix C code. This leads me to suspect that it is only a simple text emulation controlled by a simple script, not a true binary emulator (although my last use of a PDP-11 was 1985 and my memory is a bit fuzzy). Also, is it important that it is specifically a PDP-11 or could it have equally well been an IBM 370, 8080 CP/M system or an Apple II? To me it looks like it could have been any computer from that era for the purposes of the movie and the viral ad. Lastly, has it really raised awareness of the PDP-11 among today's crowd? M0rphzon has obviously found his way here but my guess is that most people visiting that website didn't stay interested in the PDP-11 very long - if at all. The Super 8 (film) article has a fleeting mention of the PDP-11 and a link to here - that seems about the total amount of attention it really needs.  Stepho  talk  04:26, 5 April 2012 (UTC)

Thanks for attempting to verify and determine why it was referenced instead of stubbornly reverting. Maybe I shouldn't have been stubborn myself, or tried to have my way, but I thought this little bit could add to the article. I have to admit, it is a bit trivial, but I already took the effort to propose it and find the sources for it. And Spike's accusations didn't help change my thought; in fact, it just made me more stubborn and annoyed. - M0rphzone (talk) 06:02, 5 April 2012 (UTC)

One of my main reasons why I wanted to add this in was to reference the topic back to the reference(wikilink) from the Super 8 (film) article. - M0rphzone (talk) 06:21, 5 April 2012 (UTC)

Agree with Stepho. It doesn't matter whether you're affiliated with the site or not. For what relation the chatter that appears on "screen" at the Super-8 marketing site has to a real PDP-11 booting RT-11, there is nothing about either that compelled the use of these particular bits of old tech (had there been, you'd have more of an argument). The site could have equally well used the chatter from any similarly unfamiliar and cryptic OS.

Thus the note and the link do not provide information that adds to the reader's understanding of the PDP-11 or why it was notable. That is, in general, the standard of inclusion for material in WP articles. This fails to meet that standard. It's trivia, not encyclopedic, and WP leans strongly against "trivia" and "in popular culture" material (that's what the likes of Wired magazine are for).

The "back link" idea is not a compelling reason to ignore this policy. "Every outside site that links to a Wikipedia page should be linked back" is not a policy. Nor are internal Wikilinks encouraged to be symmetric. PDP-11 might be interesting to a small number of readers of Super 8 (film) but that does not necessarily mean the opposite is the case. People come here to learn about a computer. If they came here via internal or external articles about the film's marketing site, then they already know about that. Get it?

This may seem like an awful lot of argument time spent to keep this little item out of the article. But the reason WP:TRIVIA exists is that without it, trivia and "in popular culture" sections proliferate. Heck, they proliferate even with it. And WP is not written for seekers of trivia... not supposed to be, anyway. For the target audience of PDP-11 (people who want to learn about the computer and why it was notable) this information and the reference link would be a waste of their time.

All of the above would go double--heck, squared, at least--if we were talking about a simple external link (see WP:ELNO).

Sorry about the time spent to find a source, but that won't carry any arguments either. Jeh (talk) 17:23, 5 April 2012 (UTC)

Meh, I already knew it was trivial. Thanks for providing a thorough explanation for opposing it. That's what I want to see/know. The only reason I even started edit-warring was because of Spike's inadequate reasons/explanations and "resentful/hostile" attitude. He continued to insist that I was attempting to game the system or use Wikipedia as a tool. I would've liked an apology, but forget it. - M0rphzone (talk) 22:29, 5 April 2012 (UTC)

How fast were the various models?

20 Mhz, 16 Mhz, 12 Mhz, 10 Mhz? Ebaychatter0 (talk) 14:00, 19 November 2012 (UTC)

Still looking for a nice compact table - but cycle times I've seen in ads mention 500 ns down to 300 ns - 2 Megahertz up to 3 1/3 MHZ. I think the LSI models ran faster but still not in the double-digits range. Still looking for a better source than "Computerworld" ads. --Wtshymanski (talk) 17:49, 19 November 2012 (UTC)

But the LSI models were microprogrammed. They might have "run faster" but the time to execute an instruction did not improve on the discrete-logic models. The Western Digital chipset was the most compact implementation of a PDP-11 to date, but not the snappiest. Can't help with specifics, but it would be a good addition to the article, for completeness as well as historical perspective. Spike-from-NH (talk) 19:52, 19 November 2012 (UTC)

I came to the article looking for the PDP-11 clock speed. Read the entire article, and now this Talk. All very interesting, but I am still none the wiser about clock speeds. — Preceding unsigned comment added by TropicalCoder (talk • contribs) 14:37, 31 December 2022 (UTC)

I have a PDP-11/70 Processor Handbook from 1975 that has an Appendix C on instruction timing. It's complicated. The timing had up to three components, Execute/Fetch time and the time required for the source and destination. The last two components depended greatly on which addressing mode was used. One example given was an ADD A,B with addressing mode 6 (indexed). In that case, source and destination times were 0.6 microseconds each, plus an Execution/Fetch time of 1.35 microseconds for a total instruction of 2.55 microseconds. And that assumed all cache hits. Each cache miss added 1.02 microseconds. Floating point was even more complex, since there was some overlap, but in general it was significantly slower. Any thoughts as to what to include in the article? I can supply more data, if needed.--agr (talk) 01:52, 20 November 2012 (UTC)

Wow--and in my experience, the 11/70 was the prevailing big-daddy multi-user machine until the VAX came along. The timing algorithms will be different for different models; for a good sample, the article might include three or four standard instructions of your choosing, with timings for the original model (11/20), the minimalist 11/05, the LSI-11, and the 11/70 or perhaps the fastest PDP-11 that DEC made. Spike-from-NH (talk) 02:24, 20 November 2012 (UTC)

A short table showing three representative early, middle and late models, with one or two different instructions, with all the special cases (cache miss, address modes) would be illuminating and would help explain to the reader just how complicated the idea of instruction timing really is. The type of person interested in megahertz drag racing will be a little disappointed that she's not getting a simple number, but a simple number is wrong anyway. A more thoughtful reader might even wonder why modern machines with thousands of times faster processors and larger memories seem not to be doing things thousands of times faster, on the whole. But that's off topic for this article. --Wtshymanski (talk) 15:04, 20 November 2012 (UTC)

Agreed that some instructions take longer than others. But if you take a 6502 speced at 1Mhz and a PIC controller spec-ed at 20Mhz, you can at least conclude that the present day PIC is 20 times faster than the vintage 6502 *at least on register-to-register move* since both can do register to register move in 1 clock cycle. Ebaychatter0 (talk) 18:50, 20 November 2012 (UTC)

Just looked at the 6502 TAX instruction cycle and it takes 2 cyles (moving accumulator to x register takes 2 cycles) , therefore a 20Mhz PIC is roughly 40 times faster than a 1Mhz 6502 in term of register to register moves. Ebaychatter0 (talk) 18:56, 20 November 2012 (UTC)

We could make a table showing how fast every processor ran a NOP - this would be abou as meaningful as any other single instruction for megahertz drag racing. NOPs have the advantage of only one addrssing mode. --Wtshymanski (talk) 20:01, 20 November 2012 (UTC)

That instruction timing is complicated is not of interest to the reader of an article on a machine from decades ago. What is of interest is that these "drag racers" are not street-ready: That they are much slower than current machines, with an attempt to quantify this, in which lack of exactitude is not a problem. In view of this, we don't need examples of actual PDP-11 instructions, just a memory cycle time. I agree that the efficiency with which modern users use all that free computing is off-topic. Of those users of the "big-daddy multi-user" machine, none were doing anything more than arithmetic and processing of (usually all-uppercase) text. No videos or MP3s. Spike-from-NH (talk) 21:23, 20 November 2012 (UTC)

SRI was doing machine vision research on PDP-11 in the late 1970s, and Cognex Corporation's first product was a PDP-11-based machine vision character recognition system in the early 80s. You can see pictures here: [1]. PDP-11s were also used for computer aided design, process control, word processing (ASCII supports lower case just fine), telephone switching and a zillion other applications. I think some information on speed is appropriate to give a proper historical perspective, though it probably belongs in the PDP-11 architecture article, since that is where the instruction set is explained. I'll take a stab at it.--agr (talk) 04:07, 25 November 2012 (UTC)

"Mostly upper case"?? Where do you get such an idea? A company called Data Processing Design made a good business out of a software product called "Word 11", which implemented a full-featured (for the day) word processing system that ran on PDP-11's. Worked on your choice of RSX or RSTS, as I recall. Upper-and-lower-case printed output was available on daisy wheel printers or on line printers with a 94-character print drum. Nor was this capability new with the PDP-11, for Word 11 was a close emulation of "WPS-8", a word processing package that DEC sold for years. It ran on PDP-8 CPUs and their emulators. As for more sophisticated users, I used to work on and around a pair of optical character readers that were based on Data General Nova 2's, so the notion that a PDP-11/70 wasn't ever used for more than "arithmetic and processing of usually-all-uppercase text" is pretty silly. Heck, some early AI research was done on the PDP-1! Jeh (talk) 10:14, 25 November 2012 (UTC)

Your point is valid, and long before Word 11, PDP-11s were doing not just word processing but typesetting too. My point was simply that, despite these first forays into "multimedia," 'Seventies computing was not comparable to modern computing but much more primitive. (Word processing on a multiuser machine often featured distracting delays in the echoing of keystrokes.) Usually computing was not so much a quest to put everything on-line or model reality but to solve specific problems. Even refreshing the screen dynamically to show changes to data was avoided. Spike-from-NH (talk) 13:01, 25 November 2012 (UTC)

Microsoft Word running on a multi-gigahertz Intel Core Duo often exhibits "distracting delays in the echoing of keystrokes." And there were plenty of sophisticated computer applications that were done on PDP-11s and other machines in that class. You may have had experience with one type of application but that does not mean others did not exist. We should describe what actually happened based on reliable sources. --agr (talk) 18:03, 25 November 2012 (UTC)

PDP-11 was a 16-bit machine and used 8-bit ASCII characters (with parity IIRC). Perhaps the editor who thought it was an upper-case-only machine was confusing it with the PDP-8, a twelve-bit machine that used a 6-bit upper-case-only character encoding, IIRC. As for annoying delays, anything on a time-shared system was subject to delays when the machine was under high load. Yworo (talk) 18:16, 25 November 2012 (UTC)

The typical character encoding used on a processor has little to do with the character sets that application software can implement. Although the PDP-8 did "natively" use six bits per character, WPS-8 and WPS-78 both supported the full 96-character set of VT52 and LQP01 printers. Heck, at the paper I used to work for, they had even used a PDP-8 for typesetting, this requiring not only upper and lower case but a number of additional characters and format codes (flush left, etc.). Similarly, although the "natural" character size on a CDC Cyber 170 was six bits, programs could use upper and lower case via a mode in which twelve bits were stored per character. And on our desktops we have machines that store eight bits per byte... but, there is Unicode. Jeh (talk) 05:17, 3 December 2012 (UTC)

@Spike: It is difficult to reconcile your comments with my experience. From 1978 through 1984, I heavily used several PDP-11s for software development. The first was an 11/45 running RSX-11M, the second an 11/40 running Unix V6, and then several LSI 11/23s running RT-11, TSX-11, or an operating system I wrote. All of these systems used VT102 terminals at 9600 bits/sec. On RSX-11M and RT-11, my editor of choice was KED, a wysiwig text editor which used the VT100 capabilities to great effect, especially the "application mode" of the keypad (same as EDT). I never waited for key echo on any of these systems (except when the O/S was crashing). Furthermore, the editors' macro abilities provided significant capability to transform a document in complex ways. Only when the operation involved thousands of steps, even in the midst other heavy usage (these were all timesharing systems a dozen or so users), would the Working... message appear, meaning it took more than three or four seconds. Once when I was developing a low level monitor, I computed the number of loops that a busy-wait for character output would take at 960 characters per second. On the lowly 11/23, it was over a thousand iterations! Even when I had background compilations running, key echo was essentially immediate. —EncMstr (talk) 20:47, 25 November 2012 (UTC)

I come here neither to bury the PDP-11 nor to praise it (and I concede that my memories of it might not be representative) but merely to argue that old-time computing is not to be confused with modern computing, and to agree that some mention of processing speed is warranted. ArnoldReinhold did indeed add suitable text to PDP-11 architecture. Spike-from-NH (talk)

TRAX: a DEC operating system

TRAX is listed under "third party operating systems". It was not; it was implemented by DEC, on the RSX11M (or possibly M+, I forget!) kernel. — Preceding unsigned comment added by 173.76.121.238 (talk) 02:26, 21 October 2019 (UTC)

Yup. Guy Harris (talk) 06:53, 21 October 2019 (UTC)

PiDP-11: RECREATING THE PDP-11/70 (Second Attempt)

I've just undone this deleted entry as it is not an advertisement and it is relevant to improving this article. This whole entry was deleted without an answer to the question which still stands. I'll restate this as: Would it be appropriate to document the PiDP-11 and may be other systems that are being implemented in hardware?

What is given below is an indication that there have been significant sales i.e. over 1,000 and the hardware is still on back order. Then I documented in this talk page that there were other PDP-11 systems implemented in hardware. The first was a CPU module that can be inserted into DEC's Unibus backplanes.

Next hardware implementations of PDP-11 systems were based on Field Programmable Gate Arrays (FPGAs). This is a similar concept to the VAX-11/750 CPU except these days technology has advanced to the point where the whole CPU can be contained in one device. It is probably impossible to estimate how many PDP-11 systems based on VHDL have been implemented.

The text below would not be added to the article in this current form. It would definitely need to be edited with an axe. I have no relationship with any of these makers except that I have a PiDP-11 on order. My background included hardware maintenance of PDP-11 and VAX systems, plus using these systems as home PCs (insert photo of VAX-11/750 on seagrass matting). Hence my interest in hardware implementations. It may be appropriate for somebody with knowledge of VHDL and FPGAs to review the references to determine the relevance of these PDP-11 systems implemented in silicon.

I deliberately raised these topics on the Talk Page to isolate the text extracted from maker's web sites. I have an email from the maker of the PiDP-11 kit stating that to date he has supplied just over 1,000 of these systems. Demand for these systems is so high that they are still on back order e.g. I was sitting at the end of the order queue in October 2018, was too late with my March 2019 order confirmation, and I should receive my PiDP-11 in the next batch, probably in April 2019. This was only included here to show there were significant sales and this section may not be included on the article page.

Buried in the PiPD-11 web site are references to other implementations of PDP-11 systems in hardware. The production volumes of these PDP-11 implementations is unknown. `

The makers web site links to Unibone a Linux to UNIBUS bridge.

The makers web site links to PDP2011 which is PDP-11 systems implemented in VHDL on FPGA boards. The PDP2011 web site stated: ″The VHDL of the PDP11 core is not vendor specific. I regularly use both Xilinx and Altera FPGA’s, and the VHDL works unchanged on both."

While the PDP2011 Overview states: "This website describes PDP2011 – a re-creation of the well known series of PDP-11 computer systems in VHDL. Everything that is needed to run a PDP-11 system is included; you can run a complete Unibus PDP-11 system with console, disks and other peripherals on a simple low cost FPGA development board. The original V5-V7 versions of Unix, BSD 2.11, and the original DEC operating systems for the PDP-11 work.

Current status

The CPU model is configurable, from the smallest 11/10 up to the biggest 11/70 and J-11 systems. Model-specific exceptions are implemented to the point that many of the original XXDP MAINDEC test programs run without error, and the original model detection code used by the operating systems in most cases correctly identifies the configured model.

The memory management unit handles 18- or 22 bit translation and supports separate instruction and data spaces, as well as kernel, supervisor and user mode. For the 11/45 and 11/70, the 3-bit ACF is implemented including memory management traps and A- and W-bits; all other models have the later, simpler version of memory management based on a 2-bit ACF. The special maintenance mode is implemented – this makes it possible to run the original XXDP test programs to verify that the MMU works correctly. For the 22-bit processor models the Unibus map is included in the memory management unit. Model-specific optional instructions, like SPL, MARK, MxPx, MxPS, MFPT, CSM etc. are present according to the CPU model.

J-11, 11/45 and 11/70 have the dual general purpose register set.

The EIS instructions: MUL, DIV, ASH, ASHC, and XOR are available if the CPU model supports them.

The floating point processor is complete, and is included by default with the CPU models that it could optionally be installed in. It can be disabled too, to save FPGA resources.

RK11 and RL11 disk controllers work and use SD cards to store the disk data on. Up to 8 RK05 or 4 RL02 disks can be emulated on a single SD card.

If the RK or RL disks are too small, then there is the option to configure a RH70 (for 11/70) or RH11 controller with a single RP04, RM05, or RP06 disk. The disk data for the RM or RP disk is also stored on a SD card.

A KW11L line clock is present; by default it is configured at 60Hz, but 50Hz and 400Hz are also possible.

Up to four serial lines for console and terminals can be configured.

A DEUNA/DELUA compatible core is available; the DEUNA core is a front end for a ENC424J600 ethernet chip, which does most of the ‘real’ ethernet work. Digilent’s PMODNIC100 contains the ENC424J600 chip, and can be easily connected to FPGA boards.

If the FPGA board includes ps2 and vga ports, a simple terminal can be added to the FPGA core so you don’t need a pc with hyperterm or minicom to run a PDP-11. The CPU that runs the software for the terminal is of course also a PDP-11.

Everything is verified in hardware, on both Altera and Xilinx boards. The VHDL is not vendor specific, so it should be possible to make it work on any FPGA that is large enough with only minor changes in the top level VHDL to map the board resources like connectors, blinkenlights and memories to the core. Several top level VHDL files and tool chain project definitions are available for popular development boards from Terasic and Digilent.

Oscar Vermeulen’s PiDP-11 console can be connected to the FPGA. At present this is still somewhat experimental and requires lots of wiring, but the logic works, and we’re developing a PCB to hook up the FPGA board to the console.

A DR11C can be configured; this is one of the simplest versions of a ‘universal peripheral’ that allows interaction between the cpu and a ‘user device’."

The description of the PiPDP-11 isn't relevant. Text extracted from maker's Web sites belongs on the maker's Web sites; you can just link to that site rather than just copying and pasting a big block of stuff.

"The text below" isn't going to be added to PDP-11 at all; it simply doesn't belong there. The PDP-11 page doesn't go into that level of detail for real PDP-11's; if the PiDP-11 is worthy of more than a mention in the Emulators section, or another section at the end, that would belong in a "PiDP-11" page.

The "Emulators" section describes software simulators.

The PiDP-11 is a combination of one of those simulators (SIMH) running on a Raspberry Pi with a reimplementation of the front panel; it sounds as if the peripherals are simulated. I'm not sure whether they're suggesting that the Unibone card could be used to provide a real UNIBUS for a simulated PDP-11 or not. A case can be made that the PiDP-11 might deserve to be an item in a separate section, as there's more to it than just the simulator.

The PDP2011 is a PDP-11 implemented in FPGA, so I think it would definitely belong in a separate section at the end.

But, again, there's no reason to give all those details on the PDP-11 page; if either the PiDP-11 or the PDP2011 is notable, in the Wikipedia sense, they should be given their own page, with relevant details mentioned. Guy Harris (talk) 03:31, 22 March 2019 (UTC)

I agree, this is a very interesting and relevant development, and perhaps deserves its own Wikipedia article, with a cross-ref from this one. Until/unless such an article exists, it is appropriate to discuss it here. Reify-tech (talk) 15:56, 22 March 2019 (UTC)

This might not yet be notable, but it is worth discussing on this talk. There would need to be a lot more notable coverage to create a separate article. I would be inclined to rename the emulator section here to something more generic like legacy. I notice that the article does not mention that the VAX-11 had a hardware assisted PDP-11 compatibility mode. Then the emulators text. Perhaps add a single sentenace that there are ongoing attempts to simulate the 11 with a link to The Register article. I wouldn't mention the company or the product by name. I don't see any notable mentions of the FPGA projects, but there is some recent academic interest in a PDP-11 on a chip.[2] There's an interesting reference to an 11 simulator being used to teach machine-level programming.[3] I wouldn't consider any one of these projects to be WP notable itself, but there have been/are numerous efforts to replicate the 11. --mikeu ^talk 16:46, 22 March 2019 (UTC)

References

DEC Datasystem

DEC Datasystem 350

All DEC Datasystem 350 models have PDP-11/10 CPUs.

DEC Datasystem with PDP-11/34

PDP-11/34 decdatasystem. (talk) 11:10, 4 November 2022 (UTC)

References