Talk:PDP-10

From Computer History Wiki
Jump to: navigation, search

Category

The reason I put this page in Category: DEC Systems was that it covers a number of different machines; KA, KI, etc. So it's not really a PDP-10 system, but it covers them all. But I can see the point of putting it in Category: PDP-10 Systems, and since that cat goes in 'DEC Systems', it's still findable, so it's OK there too. Jnc (talk) 15:22, 5 July 2019 (CEST)

PDP-10 emulators

There is a bit of a difference between the functionality of the PDP-10/X and PCjs's PDP-10, much larger than hardware vs Software.

David Conroy's, advanced beyond the PDP-10s architecture, ran faster, and could use IDE disks, and with a software modification run ITS on a disk larger than 32MB, ( and ran on a 500MB disk ), as well as a full DEC Net protical layer, which allowed him to clone disk installations over networking.
He had some problems emulating the Floating point instructions, specifically mentioning FDIV
PCjs is "WARNING: PDP-10 emulation is still in development, so functionality will be limited until further notice."
It gives this funny init warning. "PCjs Machine "testka10"

ForOldHack (talk) 03:38, 25 January 2021‎

PDP-10 Buses

I'm toying with the idea of moving the PDP-10 bus material to a separate article. Actually, maybe two separate ones: PDP-10 memory buses, and PDP-10 I/O bus? What do people think? Jnc (talk) 18:56, 25 September 2022 (CEST)

Are you planning on expanding the text? If so, sure, separate pages would make sense. If no, I don't think the current text bloats the article much. Larsbrinkhoff (talk) 22:02, 26 September 2022 (CEST)
Ah, sorry; I should have explained what the motivation was. It would not be size; it would be to make it easier for other articles which wish to link to detailed content on various PDP-10 buses to do so. At the moment they have to link to PDP-10#Busses, which is not very clean. One or two, OK; but there are lots of pages that want to link to PDP-10 buses. Jnc (talk) 01:05, 27 September 2022 (CEST)
Yes, I know it's not the most iron-clad case (as the expression in English goes), because one could just use a 'section link' (as above), but I think the two things are important enough to warrant the existence of PDP-10 I/O Bus and PDP-10 Memory Bus as top-level articles. Jnc (talk) 12:50, 21 October 2022 (CEST)
Well ok, if you think they are important, sure why not? Now that I have a READ IN switch to play with, I think it would be interesting to describe hardware read-in in detail. Larsbrinkhoff (talk) 13:41, 21 October 2022 (CEST)

KL20?

The CACM paper "The Evolution of the DECsystem 10" contains the following interesting sentence:

The cost-reduced KL20 evolved by integrating the Massbus controllers and PDP-11 interfaces onto a single high-speed, synchronous bus.

but I've never heard of a KL20, or any system that could be described that way (below). OTOH, two of the authors (C. G. Bell, A. Kotok, T. N. Hastings, and R. Hill) were in a position to know authoritatively. Perhaps one of the other two added this erroneously, and the two main authors didn't catch it? Or perhaps this is a reference to a later-cancelled machine?

The 'KL20' can't be the KS10, as that has no PDP-11's, and its Massbus controllers are on a UNIBUS. (Although the main memory bus might be synchronous, I don't recall the details). Equally, it probably can't be the later TOPS-20 KL10 machines (such as the 2060); although they look quite different, and have only an internal memory bus, other than that they are internally basically identical to the earliest KL10's, and also can't really be described as a "cost-reduced" machine. Jnc (talk) 12:24, 27 October 2022 (CEST)

I checked the paper, and the most likely explanation I can see is that what it refers to as "KL20" really is the KS10. The paper is from 1978, just about when the KS10 went to market, so presumably the name may have been in flux. Under this theory, the phrase "integrating the ... PDP-11 interfaces onto a single ... bus" would refer to the KS10 having Unibus rather than the legacy PDP-10 buses. Note that it does not really say the KL20 has a PDP-11, just the interfaces of a PDP-11. No, the Massbus isn't integrated, so that's a point against. Note also the list of five implementations: "PDP-6, KA-10, KI-10, KL10 and KL20", where the KL20 sits exactly where you would expect the KS10. It certainly fits the theory that the KS10 is massively cost-reduced over a KL10. Larsbrinkhoff (talk) 13:31, 27 October 2022 (CEST)
Very good point that it doesn't say 'PDP-11s', but "PDP-11 interfaces". And maybe the "integrating the Massbus controllers and PDP-11 interfaces" refers to the fact that they all talk to the KS10 memory through a UNIBUS interfaced to the memory.
So maybe it really is the KS10; that certainly is cost-reduced over the KL10.
If I get a chance, I'll read the KS10 tech manual and see if its main memory bus is synchronous. Jnc (talk) 14:27, 27 October 2022 (CEST)
It is. (Actually, it's the main system bus; but that's a detail. Jnc (talk) 21:49, 2 November 2022 (CET)
I think we can conclude that the "KL20" indeed refers to the KS10. Larsbrinkhoff (talk) 07:22, 3 November 2022 (CET)

Add a 'PDP-10 Users' category?

I thought of this while doing Compuserve; they had a mid-sized fleet of PDP-10's. I have this vague memory that another service bureau had a lot of them as well - or am I forgetting that it was Compuserve I'm thinking of? And of course there are also BBN, ISI, SRI, SAIL, MITAI, LCS etc . . . who else?

I think Tymshare fits the bill. Larsbrinkhoff (talk) 19:01, 8 January 2024 (CET)
At first I wasn't at all sure that was correct, but after a lot of checking, I think you are right. I had this distinct memory of a lot of content about replacing the old, inefficient power supplies, at some place that had a lot of them, and I didn't think it was Tymshare (which I also had a bit set used SDS machines - and neither Tymshare oral history page talked about PDP-10's much), but: Eric Smith's PDP-10 page talks about the power supply thing, and it's about Tymshare; the PDP-10 Serial Numbers page lists a lot of machines at Tymshare. QED.
What about Category: PDP-10 Users? Is that worth having? Jnc (talk) 21:34, 8 January 2024 (CET)
Right, they started as an SDS shop and Ann Hardy turned the Berkeley Timesharing System into a hardened production system called TYMCOM. I think when Tymshare bought CCC in Seattle, they got their PDP-10s and programmers (some SAIL alumni), which turned them onto the 36-bit path. Tymshare apparently also had IBM 370 machines. I read "The Tym Before..." but I don't recall all details. They also brought in Engelbart's ARC, see e.g. August. Anyway, many PDP-10s. Like CompuServe licensed to build their own Systems Concepts machines, Tymshare licensed from Foonly and made their own versions, e.g. http://bitsavers.org/pdf/sri/arc/augment_engine/26KL_Brochure_Oct83.pdf
I wouldn't mind a category. It would be handy if you want to search around for some PDP-10 shop. I guess we might then also need similar for PDP-11, VAX, IBM, CDC, etc. Larsbrinkhoff (talk) 07:11, 9 January 2024 (CET)
Argh, I'd never heard of the Gregory book; you should have added it to the Tymshare article! Never mind; I have done so now (and ordered a copy for myself).
The thought of having a Category: VAX Users freaked me out so badly (it would be like having a Category: IBM 360 Users - everybody would be in it) I almost regretted suggesting Category: PDP-10 Users! I think I have recovered, though, and it might be a useful tool, so I will set it up. Jnc (talk) 01:15, 10 January 2024 (CET)