Talk:PDP-1

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia

PDP-1 article[edit]

It would be nice of have a summary of the features

  • CPU register bit width
  • Memory size
  • External storage medium
  • Clock frequency

(Sorry to answer here instead of on the page, but i'm not 100% sure of the answers.

Not sure it had a clock. A lot of early DEC machines used asyncronous logic. I think that includes the PDP-1. Speed was probably of the order of 100 thousand instructions per second.

Probably no external storage medium was bundled with the thing, because it was aimed at OEMs and others who would build their own system.)

Hannes Hirzel

The basic register length was 18 bits. There was a central clock that fired off every 5 microseconds. Most commands took either 1 or 2 cycles to complete. My two systems (#35 and #53) each had 16 KB (Yes, KILOBYTES) of hand constructed magnetic core memory. Most of the programmers used the DEC assembler. There was DECAL, which was DEC's version of Algol. PDP-1guy (talk) 18:55, 28 July 2015 (UTC) PDP-1guy[reply]

How about more facts

  • What year was it made?

(Circa 1960)

  • How much did it cost?

(Circa US$100,000 . For a computer in those days, that was a bargain.)

Peder Persson

The machine at MIT was modified to support time sharing. It may have been the first time sharing machine. Each user got a 4k memory image of their own, swapped from drum memory. I could be wrong, but I think TECO was written for the PDP-6. ET was the PDP-1 editor I remember. I think there was an assembler called "possible", so the error messages could say "not possible". I think DDT was also a PDP-1 innovation, an improvement over FLIT on the TX-0. AJim 04:38, 25 Mar 2004 (UTC)

My recollections:

  • First time-sharing system: It wasn't the very first, at least I don't think so. There was a timesharing project at BBN also hosted on a PDP-1 which I think preceded it. The CTSS (compatible timesharing system) of Corbato et al was in the same time frame and, I believe, a bit earlier, but it probably depends on when you judge it to have been truly in operation. Experiments on an IBM 709 preceded Implementation on an IBM 7090 and general use by people at the Institute.
  • TECO: No, it was implemented originally on the PDP-1, primarily by Dan Murphy. I'm very, very, very sure of this one. The PDP-1 version had no screen display. Expensive Typewriter was friendlier and easier to use, and was improved by Peter Deutsch to include search-and-replace (!), so as with vi-versus-emacs more people used ET. Teco didn't take off until the PDP-6 implementation with screen display.
I thought I remembered writing more about TECO. It's in the TECO article. Dpbsmith 13:26, 25 Mar 2004 (UTC)
Dan "the man" Murphy, on WTBS (now WMBR). What do you know. Another first for the PDP-1. I think DEC gave MIT that machine; they sure got a lot back. I know I have seen the PDP-6 story before. What I remember most, now that I think about it, was ID (Invisible Debugger), which I think of as sort of the shell for the timesharing, but I think that must have come later. I heard the "first" claim from Herb Teager, but I agree there was a lot of stuff going on at the same time. Flexowriters, yes, I knew the mention of selectrics could not be right. Thanks. AJim 20:42, 25 Mar 2004 (UTC)
Yes, you're right about ID. And it was an intrinsic part of the timesharing system. If memory serves, that was done largely by Michael Wolfberg. And yes, Digital did give that machine to MIT. Typewriterish things connected with the building 26 second floor RLE PDP-1: there were offline Flexowriters, which were made by Friden and combined a tape punch and reader with an IBM electric typewriter mechanism. That was the old-style IBM electric typewriter, with typebars (not a Selectric). Flexowriters were well-designed, well-built, and reliable. The PDP-1 console terminal was also an IBM typebar-type electric typewriter, with a bunch of switches and solenoids kludged into it by a company named Soroban. It was a piece of s--t. Frequently missed case shifts or jammed on case shifts. Occasionally parts of it literally dropped out of it. At one point, circa 1965 or 1966, they did have some kind of IBM Selectric (golfball) attached to that PDP-1. Professor Jack Dennis treated that PDP-1 as a kind of sandbox; the various instructions that were added to it for timesharing kept expanding. At one point, but this was after I graduated and wasn't in close touch any more, I believe they even added some kind of index register to it. Dpbsmith 21:46, 25 Mar 2004 (UTC)
  • DDT: Yes, this was first written for the PDP-1. FLIT was claimed to stand for "Flexowriter Interrogation Tape" and DDT for "DEC Debugging Tape." DEC later rechristened it "Dynamic Debugging Technique." Dpbsmith 13:26, 25 Mar 2004 (UTC)

Perhaps it would be worthwhile to mention that although the prototype PDP-1 used a Flexowriter for the system console, the production machines used an IBM Model B electric typewriter modified by Soroban Engineering to be used as an I/O device. According to Steve Russell, and also correspondence from DEC to Soroban (now in the Computer History Museum archives), the the Soroban typewriter was notoriously unreliable. Flexowriters were still used as additional I/O devices, though I can't find any reference to what DEC option was used to interface them to the PDP-1. --Brouhaha 22:34, 17 Nov 2004 (UTC)

The Soroban was indeed unreliable. There was literally a part that kept falling out of one on the second floor at MIT's building 26. I don't think Flexowriters were directly interfaced to PDP-1s. They were used offline to prepared punched paper tape, and to print output that was produced on punched paper tape. [[User:Dpbsmith|Dpbsmith (talk)]] 23:51, 17 Nov 2004 (UTC)
There was an option used to interface additional terminals to the PDP-1; perhaps they were Teletypes rather than Flexowriters. --Brouhaha 01:36, 3 September 2005 (UTC)[reply]
The article has said, for a while:

The PDP-1 used an 18-bit word size and had 4096 words as standard main memory (equivalent to 9,216 eight-bit bytes, though the system actually used six-bit bytes), upgradable to 65536 words. The magnetic core memory's cycle time was 5 microseconds (corresponding roughly to a "clock speed" of 200 kilohertz); consequently most arithmetic instructions took 10 microseconds (100,000 operations per second) because they used two memory cycles: one for the instruction, one for the operand data fetch.

which covers "CPU register bit width", "Memory size", and "Clock frequency".
It has also said:

The PDP-1 used punched paper tape as its primary storage medium.

and

In later years, DECtape drives were added to some PDP-1 systems, as a more convenient method of backing up programs and data, and to enable early timesharing. This latter application usually required a secondary storage medium for swapping programs and data in and out of core memory, without requiring manual intervention. For this purpose, DECtapes were far superior to paper tapes, in terms of reliability, durability, and speed. Early hard disks were expensive and notoriously unreliable; if available and working, they were used primarily for speed of swapping, and not for permanent file storage.

which covers "External storage medium", so the original questions have been answered a while ago. Guy Harris (talk) 19:36, 28 July 2015 (UTC)[reply]

Flexowriters and IBM electric typewriters[edit]

Brouhaha, you say, without explanation, "Flexowriters were NOT based on IBM typewriter mechanisms." All I can say is, when I was a PDP-1 hacker, I looked at the insides of a Flexowriter and I looked at the insides of an IBM electric typewriter of the same period and they certainly looked identical to me. And identical to the typing guts of the Sorobans. Whereas a Royal, for example, looked completely different. These are old memories and perhaps unreliable.

Why do you say the Flexowriters did not have IBM electric typewriter mechanisms?

Do mean they were mechanically different? Or do you mean they were manufactured by Friden rather than IBM, which could well have been done under a license or something...

I am, of course, talking about pre-Selectric IBM electric typewriters. Dpbsmith (talk) 12:22, 2 September 2005 (UTC)[reply]

  • http://www.blinkenlights.com/classiccmp/friden/bmast.txt , a Friden reminiscence by one Bob Mast, says: "The Flexowriter was first manufactured by IBM , during WWII, to be used as an automatic letter writer. After the war several IBMers bought the rights and formed Commercial Controls, Inc. They manufactured same in the old IBM Electric typewriter building in Rochester NY. In the late fifties, Friden bought Commercial Controls."
I personally own a Flexowriter from the early 1960s, and also an IBM Model B electric typewriter (the model the Soroban was based on), and although the mechanisms work on the same general principles, they are not identical. You're welcome to come to the Computer History Museum and inspect the PDP-1 console Soroban and the Flexowriters that the museum has. --Brouhaha 01:32, 3 September 2005 (UTC)[reply]
Well, I recently edited the article and put Bob Mast's comments in as a footnote. If you want to tweak it further, to say that the mechanism was "similar to" instead of "mechanically the same as" an IBM typewriter, be my guest. I have a vague idea that the Flexowriters resembled the _crufty older_ typewriters that _less senior_ secretaries used to use in the mid-1960s and that the Sorobans resembled the _spiffier newer_ ones. I'm not a typewriter maven though. As to the Computer Museum, well, I live near Boston, and I visited it when it was in Marlboro and I visited it when it was in Boston--both when it looked like a computer museum and when it looked like Sesame Street, but I don't get out to Mountain View very often. Does it still have the COBOL tombstone? Dpbsmith (talk) 02:17, 3 September 2005 (UTC)[reply]
Yes, the COBOL tombstone is still there. Maybe someday it will be accurate. The Computer History Museum is much more focused on preserving the artifacts, stories, etc. than the Computer Museum was in its later years. Anyhow. let me know if you do get to the west coast sometime, as we'd love to have you visit, see the running PDP-1, and tell us about your own PDP-1 experiences. --Brouhaha 06:49, 3 September 2005 (UTC)[reply]

Software[edit]

Several new articles started last summer have been tagged merge because of what appears to be lack of input (I have contributed only what I learned from the restoration video). What do you think?

Merge into new article PDP-1 Software:

Merge with Spacewar

Merge with TX-0

Thank you for your thoughts. --Susanlesch 12:32, 28 December 2006 (UTC)[reply]

  • I will remove the merge tags since there is no support. If you should have comments later feel free to leave them on my talk page. Best wishes. -Susanlesch 05:58, 2 January 2007 (UTC)[reply]

First computer to be used in time-sharing (networking)[edit]

I can't remember where, but I read somewhere that the PDP-1 was used for the first time for demo-ing time-sharing (networking) by some guy that was the top-brass of ARPA in the 60s.

This of course eventually lead to the Internet. It might be important enough to note this on this article. --Zybez (talk) 20:14, 7 April 2008 (UTC)[reply]

PDP-1-based time-sharing systems were implemented at MIT and at Bolt, Beranek, and Newman, but I'm not sure whether they antedated a time-sharing system that was implemented on an IBM 704 which became the antecedent of CTSS. CTSS was much more influential because it had remotely connected teletypes... including one miles away in a researcher's home... and was much more network-like than the PDP-1 systems.
The "top-brass of ARPA" you're probably thinking about might be J. C. R. Licklider, but I don't think it was a case of anyone convincing Licklider; it was the other way around, it was an idea Licklider was personally invested in and wanted to make happen.
I think. Dpbsmith (talk) 01:31, 8 April 2008 (UTC)[reply]
Yeah, most of this is in the J. C. R. Licklider article. Dpbsmith (talk) 01:33, 8 April 2008 (UTC)[reply]

Alan Kotok comments in "current status"[edit]

I don't agree with the removal of these comments by user Asenine. The comments by Kotok put the PDP-1 into historical perspective, and the fact that they are being made by a PDP-1 pioneer gives them some weight. Dpbsmith (talk) 23:09, 10 August 2008 (UTC)[reply]

Memory size[edit]

The article claimed that it had the equivalent of 9 kilobytes = "9000" bytes. This seemed dubious to me, especially since redefining pibinary prefixes wasn't even tried before 1999, and especially memory sizes used, and use to this day, binary kilobytes rather than drivemakers' ones. Somehow, nearly every single reference refers to "kilowords" without spelling out the number. After viewing a few dozens of pages, the only reference I found was [[1]] which describes the address space which might be different from physically installed memory. Still, it would be strange for these two to differ without a need, and it's consistent with what "kilobyte" meant for all those years. Thus, I didn't spend more time and replaced it in the article.KiloByte (talk) 09:34, 2 March 2011 (UTC)[reply]

I think you're missing the point of the byte equivalence. It's just to provide a comparison to modern computers. In terms of number of bits, it's exact: 4096*18 = 9192*8, but since the word organization is different such comparison has limited value. We should also be clear that we are talking 8 bit bytes.--agr (talk) 20:16, 3 July 2011 (UTC)[reply]
Regardless, someone needs to fix the egregious error of "six-bit bytes". That's a contradiction in terms. What is probably intended, is "six-bit words". 222.154.87.53 (talk) —Preceding undated comment added 03:19, 8 December 2015 (UTC)[reply]
Given that words on the PDP-1 were 18 bits long, clearly "six-bit words" was not what was intended. The system was not byte-addressable; the 18-bit word was the smallest addressable unit of memory, but it didn't use 18-bit characters, so the smallest addressable unit of memory was not the same as the unit of memory used for a character. Guy Harris (talk) 04:16, 8 December 2015 (UTC)[reply]
A byte is not the same thing as 8-bits. A byte is just a chunk of data that can hold a small amount of data. Computers in the last 30 years have standardised on 8-bit bytes but in the 1960s bytes were often 6, 7, 8 or 9 bits long.  Stepho  talk  04:35, 8 December 2015 (UTC)[reply]
The later PDP-1s had the LCH and DCH instructions, which addressed 6-bit bytes. See <http://archive.computerhistory.org/resources/text/DEC/pdp-1/DEC.pdp_1.1964.102652399.pdf>. John Sauter (talk) 05:11, 8 December 2015 (UTC)[reply]

New main pix?[edit]

I'm inclined to replace the main photo for this article with File:Steve Russell and PDP-1.png. I cropped it from a more distant shot and I think it's a better picture of the PDP-1, while avoiding the fair use issue with the current shot. I'm raising the issue here first since the current shot could be deleted from Wikipedia if we remove it from this article.--agr (talk) 20:16, 3 July 2011 (UTC)[reply]

Did it.--agr (talk) 02:40, 6 July 2011 (UTC)[reply]

Type 30 CRT[edit]

A recent edit to the caption of the top picture claimed that the Type 30 CRT was a vector display. It wasn't—it was a dot plotter. To program it, you loaded the AC and IO registers with the X and Y coordinates of the dot, then issued the display IOT instruction. I am putting this note here rather than just changing the article to avoid an edit war. If there are no objections I will correct the caption. If there are I will try to find a reference. John Sauter (talk) 13:29, 17 May 2012 (UTC)[reply]

Here's a reference, from an early PDP-1 manual. See pages 8 and 9: PDP-1 Input-Output Systems Manual F-25. John Sauter (talk) 13:44, 17 May 2012 (UTC)[reply]

Well, the vector monitor page says that a "vector monitor" is
...a type of CRT, similar to the oscilloscope. In a vector display, the image is composed of drawn lines rather than a grid of glowing pixels as in raster graphics. The electron beam follows an arbitrary path tracing the connected sloped lines, rather than following the same horizontal raster path for all images. The beam skips over dark areas of the image without visiting their points.
and the Type 30, from the description, is definitely a display where "the electron beam follows an arbitrary path ... rather than following the same horizontal raster path for all images". Perhaps the Type 30 is different from other "beam-steering" displays, but it's definitely different from a "vector graphics" display in the sense that much of the vector graphics page talks about - that page is mainly about graphical objects described by "vector"-format descriptions of images that are turned into raster sequences by CPU or GPU hardware or both and displayed on a raster display:
Computer displays are made up from grids of small rectangular cells called pixels; the term comes from "picture elements". The picture is built up from these cells. The smaller and closer the cells are together, the better the quality (resolution) of the image, but the bigger the file needed to store the data. However, modern storage devices and working memory have gigabyte, even terabyte capacities, so there is less need for particularly compact forms of data.
Modern displays and printers are raster devices; vector formats have to be converted to raster format (bitmaps – pixel arrays) before they can be rendered (displayed or printed). ...
If you want to refer to it in the caption as a "point plotter oscilloscope" display, or something such as that, I would have no objection. Linking to vector graphics in the caption would be a mistake, as it would be misleading. (And, yes, the vector graphics page is itself misleading when it talks about beam-steering displays in the beginning of the "Applications" section. There really should be pages about "vector displays", i.e. beam-steering displays, and "raster displays", i.e. CRT displays where the beam does TV-style scanning and non-CRT displays such as modern LCD etc. displays, and separate pages about "vector graphics", whether implemented on vector or raster displays, and "raster graphics". If the Type 30 was not a "vector display", perhaps a third display type needs to be described, as it's definitely not a raster display.) Guy Harris (talk) 16:28, 17 May 2012 (UTC)[reply]
What you (User:John Sauter) are saying is consistent with what I remember from seeing the display (a long time ago), and you have a good ref. I suggest that you add a short paragraph about the display, with the footnote reference, in addition to fixing the caption. The distinction between a dot plotting display and a true vector display is very important, and shouldn't be muddled or glossed over. The topic probably deserves a small section in the Vector graphics article, or its own small stand-alone article.
I once interfaced a dot plotter display to a PDP-11 for our own homebrew version of "Spacewar"; we certainly knew what a vector display was, but couldn't afford the mini-fridge sized vector processor hardware (mid-1970s). The PDP-11 was fast enough to draw a decent Spacewar image without too much flicker, good enough for a lab Open House family day. --Reify-tech (talk) 16:49, 17 May 2012 (UTC)[reply]
If the dot-plotting display vs. vector display deserves a discussion, it would belong in vector monitor, not vector graphics, or in an article about dot-plotting displays. I remember some Slashdotter confused about this, who thought that, with more use of "vector graphics" in the "high-level description of graphic objects isn't raster-based, and gets rasterized" sense, "vector graphics" in the sense of "beam-steering displays" would come back; I had to explain that the two aren't related these days. Guy Harris (talk) 19:20, 17 May 2012 (UTC)[reply]
You're absolutely right, it either belongs in vector monitor, or in a separate article about dot-plotting displays. --Reify-tech (talk) 20:59, 17 May 2012 (UTC)[reply]
There are a bunch of articles that overlap on this topic and most lack info on early technologies. They include:
They all need to be straightened out. I would attack the History article first. :)-agr (talk) 21:28, 17 May 2012 (UTC)[reply]

I've corrected the caption and added info about the display from the PDP-1 handbook. It was indeed point mode but an optional line and curve generator is mentioned, with no details.--agr (talk) 17:42, 17 May 2012 (UTC)[reply]

Looks good. --Reify-tech (talk) 20:59, 17 May 2012 (UTC)[reply]

Display Names: It might be interesting that the Type 30 display seemed to have been originally called "Visual CRT Display", while the name "Precision CRT Display" was reserved for the Type 31 display (4096 x 4096 px). See: PDP-1 Handbook "Programmed Data Processor Manual" (F-15B), 1961; p. 27/28 [2]

Later DEC Materials (apparently 1963 and later) refer to them as "Precision CRT Display (Type 30)" and "Ultra-Precision CRT Display (Type 31)" See: PDP-1 Handbook "Programmed Data Processor Manual" (F-15C), 1963; p. 33/34 [3]

On the type of display: The Type 30 was a point plotting display (not a vector display). According to Steve Russell, the vector-like flicker effect to be observed in videos showing "Spacewar!" is caused by an unstable frame rate, phasing between 12 and 15 frames per second: "The display is part of the main loop, and the time take depends on what's going on. There is a crude attempt to keep the frame rate constant, but its only good enough for humans, not movie cameras." (Personal e-mail, no further reference.)

The Type 30 display exhibits visible pixelation as can be seen in this screenshot of the PDP-1's character-set displayed on the CHM's restored PDP-1 [4]

(NN - you can either use clear names or expose the IP, but not both) — Preceding unsigned comment added by 80.108.245.78 (talk) 16:46, 2 February 2013 (UTC)[reply]

What's with the gloves the gentleman in the pictures is wearing[edit]

Ebaychatter0 (talk) 20:40, 6 September 2012 (UTC)[reply]

The oil in fingerprints is slightly acidic. So it is common in museums (the caption says the picture was taken in a museum) for anybody handling the artefacts to wear gloves.  Stepho  talk  22:46, 6 September 2012 (UTC)[reply]

A 4th example...[edit]

If anyone wants to do some garbage mining, there is a 4th example of the PDP-1 at the dump in Sudbury. This machine originated at Chalk River where it was used for years until they could no longer supply it with the germanium transistors. They cut all the internal cables and shipped it to Science North where it was the centrepiece of a display on information technology in the 1980s. SN dumped it some time in the 90s, the guy I emailed wasn't sure, and it ended up in the dump. Maury Markowitz (talk) 21:00, 8 November 2014 (UTC)[reply]

1978 Trade of PDP-1 and PDP-7[edit]

In 1978, the MITERS (MIT Electronics Research Society; the electronics club) possessed a PDP-1 and a PDP-7. I believe the PDP-1 was the original one given to MIT by DEC. A trade was then engineered with DEC: They were thinking of starting up a Computer History Museum in Boston, which later occurred. The '1 and the '7 were traded for about 40 pallets of scrap computer equipment, including some PDP 11-34's, perhaps part of a PDP-11-45 (?), some 11-03's, at least one 300 lpm printer, and various other computer hardware. A Senior in MITERS at this time was Jordin Kare. Perhaps this is him: https://en.wikipedia.org/wiki/Jordin_Kare Sluefoot (talk) 04:45, 5 October 2015 (UTC)[reply]

Operating System[edit]

The newly-added sidebar says that the operating system for the PDP-1 is the BBN Time-Sharing System. This is somewhat misleading. DEC shipped an assembler, a debugger, some utilities and diagnostic software on paper tape. They all ran stand-alone. As far as I know, the BBN software was only used at BBN. At Stanford we wrote our own time-sharing system, which was only used at Stanford. I expect other customers wrote their own software based only on the software shipped by DEC. John Sauter (talk) 19:34, 29 August 2020 (UTC)[reply]

This was one OS that was available/created for the PDP-1. If there are others, please add them to the infobox. Thanks. Ghettoblaster (talk) 19:58, 29 August 2020 (UTC)[reply]

byte addressing[edit]

In response to a comment on a recent edit saying that the PDP-1 did not have byte addressing, the PDP-1D at Stanford had the lch and dch instructions, which manipulated strings of 6-bit characters, packed three to a word, with the first character in the high-order six bits. Surviving documentation for this is on bitsavers, at http://www.bitsavers.org/pdf/dec/pdp1/F15_1D-45_PDP-1D_Upd_Jun64.pdf . John Sauter (talk) 15:40, 3 December 2020 (UTC)[reply]

The document says it was added by BBN; were multiple PDP-1Ds made, or did BBN either 1) send somebody out to Stanford to modify the machine or 2) send a modified machine to Stanford?
And was this an inspiration for the PDP-6/PDP-10 byte manipulation instructions? Guy Harris (talk) 19:17, 3 December 2020 (UTC)[reply]
Stanford's computer came directly from DEC. There is a good reference here: http://simh.trailing-edge.com/docs/pdp1d.pdf . In addition to the differences between toe two machines described in that document, Dave Poole disliked Ring Mode, so disabled it by cutting a wire. I think it interfered with time sharing. Also, Stanford did not use the Type 630; instead we used video terminals from Philco. I don't know if the lch and dch instructions inspired the ldb and dpb instructions on the PDP-6, but I wouldn't be surprised. John Sauter (talk) 20:01, 3 December 2020 (UTC)[reply]
So there were two PDP-1Ds, both custom built with different feature sets. The bitsavers document you mentioned may have been for serial #45, the BBN machine, as it mentions BBN. Guy Harris (talk) 20:34, 3 December 2020 (UTC)[reply]
The simh document actually mentions both #45 and #48, the Stanford machine. Both had the character instructions, but there were differences between them. The Stanford machine had four 4,096-word memory banks. If I remember correctly, banks 0 and 1 were occupied by the monitor, bank 2 was the display buffer, and bank 3 contained the currently-executing user program. The tracks on the swapping drum were 4,096 18-bit words long, could be addressed on any word boundary, and could read one track while writing another. Context switching thus took one drum revolution. Core memory uses destructive readout, which normally requires a word that has been read to be re-written. Instead, the drum would read a word from core but write back the word read from the other track. John Sauter (talk) 21:28, 3 December 2020 (UTC)[reply]
Yes, the simh document mentions both - the document I was referring to was the bitsavers document, titled "PDP-1 Supplement (PDP-1D-45)", and which starts out saying "This supplement describes special instructions added to PDP-l D-45 at Bolt Beranek and Newman.", so that document appears to be for #45. Guy Harris (talk) 22:09, 3 December 2020 (UTC)[reply]

First minicomputer?[edit]

While many computer history sites describe the PDP-8 as the first minicomputer, this article lists the PDP-1 as a minicomputer. While it seems DEC starting using the term minicomputer only in the 60's and thus by their standards the PDP-8 was their frst minicomputer, it also seems we have classified the PDP-1 as a minicomputer so maybe the idea that this is the first minicomputer should be included in the article unless another contender existed in 1959. Notcharliechaplin (talk) 03:19, 2 September 2022 (UTC)[reply]

It's a bit of a grey area. minicomputer says that (compared to a mainframe of the same time), a minicomputer must be cheap, small, general purpose instruction set (ie not specifically accounting or scientific instruction sets) and able to connect to many 3rd party peripherals (ie, not tied to only products in its line). The PDP-1 comes close except possibly for its size - small compared to many mainframes but still large compared to something like the PDP-8 or even the PDP-5. That may or may not be excused by the time difference between the PDP-1 and PDP-8 (newer = smaller).  Stepho  talk  05:37, 2 September 2022 (UTC)[reply]
The PDP-1 was also expensive compared to the later mini computers. It was like the CDC-160A but with 18-bit memory and built into standard racks. I think it deserves to be called an ancestor of the mini computer. 10:57, 2 September 2022 (UTC) John Sauter (talk) 10:58, 2 September 2022 (UTC)[reply]
True, it was more expensive than later minicomputers (eg 1965 PDP-8 for $18,500). But it was also a lot cheaper than mainframes from its own time. 1959 PDP-1 for $120,000 vs 1957 IBM 709 for $1,200,000 .  Stepho  talk  11:44, 2 September 2022 (UTC)[reply]
Yes, and it therefore had a place in the market. However, I don't think it can properly be called a minicomputer, considering how that word was defined. John Sauter (talk) 18:52, 2 September 2022 (UTC)[reply]
Really? Considering how a minicomputer is defined (see my first comment), it looks to be a perfect fit - compared to mainframes of its time it is cheaper, smaller, general purpose and easily connectable to outside equipment. Also, it only required single phase mains power and air conditioning was not required.  Stepho  talk  01:11, 3 September 2022 (UTC)[reply]
According to the definition described in the Minicomputer article you cited, the DEC PDP-1 is too expensive to be a mini computer, even though it is less expensive than the IBM 7090. The article says a minicomputer costs less than $25,000 whereas the PDP-1 costs $120,000. John Sauter (talk) 13:19, 3 September 2022 (UTC)[reply]

The $25,000 figure came from a New York Times survey in 1970:

In a 1970 survey, The New York Times suggested a consensus definition of a minicomputer as a machine costing less than US$25,000 (equivalent to $196,000 in 2023), with an input-output device such as a teleprinter and at least four thousand words of memory, that is capable of running programs in a higher level language, such as Fortran or BASIC.[1]

so what's relevant there is the price in 1970 dollars. "How much would a PDP-1 have cost if you built it, in 1970, from the same components as were used in 1959" would be tricky (and what if the transistors in question weren't being made any more?), but if we just use 1959-to-1970 inflation, the Wikipedia template says that US$120,000 is equivalent to $160,000 in 1970, so it's a little bit more out of the price range for a minicomputer in 1970.

That also puts the PDP-5 out of that class, although not by as much; it was US$27,000 in 1963, which is equivalent to $34,000 in 1970. The PDP-8 successor, however, was US$18,500 in 1965, which is equivalent to $23,000 in 1970, so it squeaks in under the wire.

Using the 1970 number means that, as time goes on, it becomes easier to fit into the minicomputer price category, as technology improved.

It Would Be Nice if there were a term for machines in this category, but I'm not sure there was one at the time. Midrange computer implies that they're in the middle of a range, but the PDP-1 wasn't in the middle of DEC's range of computers, it was the only machine in the range. By the time the PDP-8 came out, there was the PDP-10 at the top and the PDP-7 below it and above the PDP-8, so maybe, if the 18-bit PDP-7 didn't fit into the minicomputer class by price, it could be a midrange. Guy Harris (talk) 21:11, 3 September 2022 (UTC)[reply]

References

  1. ^ Smith, William D. (April 5, 1970). "Maxi Computers Face Mini Conflict: Mini Trend Reaching Computers". The New York Times.
Retrieved from "https://en.wikipedia.org/w/index.php?title=Talk:PDP-1&oldid=1204387912"