6

In the late 60s, the Incompatible Timesharing System ran on a PDP-6 without any paging hardware (i.e. an MMU in modern terminology). See the Jargon entry for BLT, for example, which states (my emphasis):

... one resource-intensive memory-shuffling operation done on pre-paged versions of ITS, WAITS, and TOPS-10 was sardonically referred to as "The Big BLT".

It was later moved onto a PDP-10 and, sometime around 1970, a paging system was installed and ITS was modified to use virtual memory.

The MIT Dynamic Modeling group adopted ITS for their PDP-6 and PDP-10 (named MIT-DMS). Initially, there was no disk to provide swapping store, so they ran a non-paged ITS until around 1973.

There is a copy of ITS for MIT-AI and MIT-ML (but not MIT-DMS) from 1973, but there is no trace of support for running on a PDP-6 or any possibility of running without virtual memory.

I believe the majority of software development in the non-paging era was done using DECtape so, if an pre-historic ITS was to be found, it would probably be on such a tape. Another possibility is paper tape, since this was often used to load software for booting.

Does anyone know where this (a non-paging ITS) could be found?

  • 3
    I loved ITS. I think the AI ITS system had a passing box by 1969. It was made by the "systems concepts" company and had a small cache of mappings. An indirect reference chain could easily be inevitable. The page table cache was reloaded by system software -- no hardware page table walks. I didn't know that Dynamod didn't have paint hardware. – cmm Sep 12 '18 at 2:13
  • 1
    I worked in DM during part of that time frame. I remember a KA-10 with a small disk connected. A lot of source code was kept on DECtape, if only for backup. The KA-10 had dual protection and relocation, which is not the same as paging. I have no recollection of DM ever having its own PDP-6. It's more likely that they used the AI PDP-6. – Walter Mitty Oct 24 '18 at 15:52
  • 1
    Please elaborate. The earliest ITS version I know is 724 from 1972, and the binary file is 44K words. What length paper tape is that, and how long would it take to load? I would assume the PDP-6 ITS would be much smaller. Let's pull a number from thin air: maybe 10K words? – Lars Brinkhoff Oct 26 '18 at 7:54
  • 2
    FWIW, I bumped into a page which says "You won't find ITS sources to run on a PDP-6 or non-paged KA. Such sources don't exist any more". I don't know whether that's authoritative. Link: inwap.com/pdp10/cpus.txt – another-dave Nov 16 '18 at 23:48
  • 1
    It was written by Mark Crispin, so it is pretty authoritative. But I refuse to give up. – Lars Brinkhoff Nov 18 '18 at 20:36
2

This is a frame from MIT AI film #43. I'm not sure what year it's from, but it seems to be from the PDP-6 era. The code is very similar to old ITS source code, so it's likely this is a tiny fragment of PDP-6 ITS.

MIT AI film #43

Transcription, including a few more lines visible in the film:

        LDB B,[MLU,,MEMBLT(A)]
        DPB B,[MLO,,MEMBLT(E)]
        LDB E,[MLU,,MEMBLT(A)]
        CAIE B,1777
        DPB E,[MLU,,MEMBLT(B)]
        SOS MEMFR
        JRST POPJ1

IMEMR:  MOVE B,MEMFP
        DPB B,[MLO,,MEMBLT(A)]
        DPB A,[MLU,,MEMBLT(B)]
        MOVEI B,1777
        DPB B,[MLU,,MEMBLT(A)]
        MOVEI B,75
        DPB B,[MUR,,MEMBLT(A)]
        MOVEM A,MEMFP
        AOS MEMFR
        POPJ P,
↓₁₄TIBL=20
TOBL=20
NTTYS=4
        TYICLS
TYI:    SKIPG TICC (A)
        PUSHJ P,UFLS
        MOVEI B,TIOP(A)
        CAMN B,TIBEP(A)
        SUBI B,TIBL
        HRRM B,TIOP(A)
        ILDB B,TIOP(A)
        MOVEM B,(C)
  • How can you be sure that the code is very similar to old ITS source code? – Wilson Feb 12 at 12:48
  • It is customary on StackExchange to transcribe images into text, because image sources disappear, some people have visual impairments, and others want to be able to copy/paste. Can you please do that? Then you may use that text to address Wilson's concern. – Dr Sheldon Feb 13 at 2:02
  • Wilson, first grepped all files I have collected from the ITS systems. The monitor itself was the only good match, so I compared it against ITS 785 which is the oldest source code I know of. Symbol names and labels match, though the code has been rearranged. – Lars Brinkhoff Feb 13 at 6:28
  • So you have a copy of the ITS source code? I thought I'd read somewhere that it had been lost to obscurity. – Wilson Feb 13 at 10:24
  • 2
    Oh, not at all! See here: github.com/PDP-10/its – Lars Brinkhoff Feb 13 at 12:25

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.