[TUHS] A PDP-10 used for UNIX just after the PDP-7?

[TUHS] A PDP-10 used for UNIX just after the PDP-7?

[Thalia Archibald via TUHS]

In Mahoney's oral history, he writes that after the PDP-7, "The file system was
then put on a PDP-10". And in Thompson's 1989 interview with Mahoney, he
mentions that "Before the 11/45 was available we bought a PDP-11 that had PDP-10
memory management, KS-1, it was a one of a kind machine".

Are these the same system? I've never seen any other references to a PDP-10 used
in early UNIX development, only that their request for a 10 was denied. Does
anyone know anything about this strange PDP-11 + KS-1?

What was the model of PDP-11 first used for UNIX? What hardware was sold by the
UNIX group to the patent department? I presume it was an 11, since the 7 was
never their property.

Mahoney's words:

    When Thompson realized that the PDP-7 was not powerful enough to implement a
    file system that could offer some of the advantages of Multics he initially
    programmed a bare-bones file system. The file system was then put on a
    PDP-10. Using this enhanced capability, Thompson and the others slowly added
    tools that helped them monitor what the file system was doing. Thompson
    devoted a month apiece to the shell, editor, assembler, and other software
    tools. When asked when he realized that a new operating system was being
    born, Thompson replied,

https://dspinellis.github.io/oral-history-of-unix/frs122/unixhist/finalhis.htm

My summarization of the relevant paragraphs of Thompson's 1989 interview:

    After the PDP-7 was at the end of its life, they needed a new machine, so
    they proposed to buy a PDP-11 model that was about to be announced. The
    proposal was rejected, but the psychology research department decided to buy
    it and give it to them. They placed the new PDP-11 in Osanna's office next
    to the PDP-7 and wrote cross tools. They wrote a PDP-11 assembler on the 7
    in B and ran the paper tape across the floor to the 11. For its first month,
    the 11 was without a disk, so they had a fake in-memory filesystem. Once it
    was delivered, they got UNIX running on it in about a week and, at this
    point, the topology of the directory structure was rather fixed.
    
    When the patent department was about to buy AstroText, a horrific and
    expensive typesetting package, they instead produced a version of
    nroff/troff for their specific formatting needs. They also talked the patent
    department into buying their system, i.e., the physical hardware, and moving
    it out. With that money, they bought an 11/45. Before the 11/45 was
    available, they bought a PDP-11 that had PDP-10 memory management, KS-1, a
    one-of-a-kind machine. This was the first time they were using the machine
    at the same time for program development as typists.

https://dspinellis.github.io/oral-history-of-unix/mike/transcripts/thompson.htm

Thalia

[TUHS] Re: A PDP-10 used for UNIX just after the PDP-7?

[Ken Thompson via TUHS]

after the pdp-7, unix was ported to more
stock pdp-9 and pdp15 (both pdp7 upgrades)

no pdp-10 at bell labs.


On Thu, Jan 15, 2026 at 10:54 PM Thalia Archibald via TUHS <tuhs@tuhs.org>
wrote:

> In Mahoney's oral history, he writes that after the PDP-7, "The file
> system was
> then put on a PDP-10". And in Thompson's 1989 interview with Mahoney, he
> mentions that "Before the 11/45 was available we bought a PDP-11 that had
> PDP-10
> memory management, KS-1, it was a one of a kind machine".
>
> Are these the same system? I've never seen any other references to a
> PDP-10 used
> in early UNIX development, only that their request for a 10 was denied.
> Does
> anyone know anything about this strange PDP-11 + KS-1?
>
> What was the model of PDP-11 first used for UNIX? What hardware was sold
> by the
> UNIX group to the patent department? I presume it was an 11, since the 7
> was
> never their property.
>
> Mahoney's words:
>
>     When Thompson realized that the PDP-7 was not powerful enough to
> implement a
>     file system that could offer some of the advantages of Multics he
> initially
>     programmed a bare-bones file system. The file system was then put on a
>     PDP-10. Using this enhanced capability, Thompson and the others slowly
> added
>     tools that helped them monitor what the file system was doing. Thompson
>     devoted a month apiece to the shell, editor, assembler, and other
> software
>     tools. When asked when he realized that a new operating system was
> being
>     born, Thompson replied,
>
>
> https://dspinellis.github.io/oral-history-of-unix/frs122/unixhist/finalhis.htm
>
> My summarization of the relevant paragraphs of Thompson's 1989 interview:
>
>     After the PDP-7 was at the end of its life, they needed a new machine,
> so
>     they proposed to buy a PDP-11 model that was about to be announced. The
>     proposal was rejected, but the psychology research department decided
> to buy
>     it and give it to them. They placed the new PDP-11 in Osanna's office
> next
>     to the PDP-7 and wrote cross tools. They wrote a PDP-11 assembler on
> the 7
>     in B and ran the paper tape across the floor to the 11. For its first
> month,
>     the 11 was without a disk, so they had a fake in-memory filesystem.
> Once it
>     was delivered, they got UNIX running on it in about a week and, at this
>     point, the topology of the directory structure was rather fixed.
>
>     When the patent department was about to buy AstroText, a horrific and
>     expensive typesetting package, they instead produced a version of
>     nroff/troff for their specific formatting needs. They also talked the
> patent
>     department into buying their system, i.e., the physical hardware, and
> moving
>     it out. With that money, they bought an 11/45. Before the 11/45 was
>     available, they bought a PDP-11 that had PDP-10 memory management,
> KS-1, a
>     one-of-a-kind machine. This was the first time they were using the
> machine
>     at the same time for program development as typists.
>
>
> https://dspinellis.github.io/oral-history-of-unix/mike/transcripts/thompson.htm
>
> Thalia
>

[TUHS] Re: A PDP-10 used for UNIX just after the PDP-7?

[Thalia Archibald via TUHS]

On Jan 15, 2026, at 23:57, Ken Thompson wrote:
> after the pdp-7, unix was ported to more
> stock pdp-9 and pdp15 (both pdp7 upgrades)
> 
> no pdp-10 at bell labs.

Hi Ken,

So the "PDP-10" that Mahoney refers to is the PDP-11 with KS-1 memory
management? Seems like a strange machine.

And did you only ever use the one PDP-7 from the graphics group, and the other
PDP-9s and 15s were used by other groups?  Warner Losh writes that the PDP-7
UNIX “Total install base was 4 (1 pdp-7, 2 pdp-9 and 1 pdp-15)”.
    https://papers.freebsd.org/2020/fosdem/losh-hidden_early_history_of_unix/

Thalia

[TUHS] Re: A PDP-10 used for UNIX just after the PDP-7?

[Ken Thompson via TUHS]

except for porting and testing, i didnt use the
pdp-9s or -15s. the pdp-7 could support 2 people
while the others only had tty33. also the pdp-7
had the filesystem with all the current work.

the ks-11 was the first memory protected pdp-11.
we needed it to pretend like it was a real time
sharing system. ks was the dec designation
for a one-of-a-kind. dec did a lot of special
system designs for customers. there is a
computer, i think pdp-2, that was a completely
custom built  computer.

On Thu, Jan 15, 2026 at 11:12 PM Thalia Archibald <thalia@archibald.dev>
wrote:

> On Jan 15, 2026, at 23:57, Ken Thompson wrote:
> > after the pdp-7, unix was ported to more
> > stock pdp-9 and pdp15 (both pdp7 upgrades)
> >
> > no pdp-10 at bell labs.
>
> Hi Ken,
>
> So the "PDP-10" that Mahoney refers to is the PDP-11 with KS-1 memory
> management? Seems like a strange machine.
>
> And did you only ever use the one PDP-7 from the graphics group, and the
> other
> PDP-9s and 15s were used by other groups?  Warner Losh writes that the
> PDP-7
> UNIX “Total install base was 4 (1 pdp-7, 2 pdp-9 and 1 pdp-15)”.
>
> https://papers.freebsd.org/2020/fosdem/losh-hidden_early_history_of_unix/
>
> Thalia
>

[TUHS] Re: A PDP-10 used for UNIX just after the PDP-7?

[Warner Losh via TUHS]

On Fri, Jan 16, 2026 at 12:12 AM Thalia Archibald via TUHS <tuhs@tuhs.org>
wrote:

> On Jan 15, 2026, at 23:57, Ken Thompson wrote:
> > after the pdp-7, unix was ported to more
> > stock pdp-9 and pdp15 (both pdp7 upgrades)
> >
> > no pdp-10 at bell labs.
>
> Hi Ken,
>
> So the "PDP-10" that Mahoney refers to is the PDP-11 with KS-1 memory
> management? Seems like a strange machine.
>
> And did you only ever use the one PDP-7 from the graphics group, and the
> other
> PDP-9s and 15s were used by other groups?  Warner Losh writes that the
> PDP-7
> UNIX “Total install base was 4 (1 pdp-7, 2 pdp-9 and 1 pdp-15)”.
>
> https://papers.freebsd.org/2020/fosdem/losh-hidden_early_history_of_unix/


I wrote that because I found references, I think in the 0-edition manuals,
that had those
numbers. I got the impression from somewhere, and I can't quite find it
this morning, that
those machines were 'cast off' that the department used on an interim basis
until they could
get a pdp-11. But the evidence was thin at the time, and I included it
because it was more
than 0, which was a surprise to me.

Warner

[TUHS] Re: A PDP-10 used for UNIX just after the PDP-7?

[Phil Budne via TUHS]

Thalia Archibald wrote:
> On Jan 15, 2026, at 23:57, Ken Thompson wrote:
> > after the pdp-7, unix was ported to more
> > stock pdp-9 and pdp15 (both pdp7 upgrades)
> > 
> > no pdp-10 at bell labs.
> So the "PDP-10" that Mahoney refers to is the PDP-11 with KS-1 memory
> management? Seems like a strange machine.

https://gunkies.org/wiki/KS11_Memory_Protection_and_Relocation_option

describes an MMU much like the KA10 (first PDP-10 CPU): Two sets of
base and limit registers (highseg and lowseg).  I think the second set
of registers was an option on the KA (*).  The PDP-10 was a follow-on
for the PDP-6, which only had one base/limit pair.

(*) Way WAY too much info for the TUHS list:

The PDP-6, DEC's first machine larger than 18 bits had been regarded
as an overreach for the then small company (only 23 were sold).  Among
other things it used large circuit boards that had to be soldered in,
and the largest (core) memory box sold was 16K.  One customer who
wanted to run time sharing using only DEC memories returned the system
because it was too flaky, but shops that bought larger memory boxes
from 3rd parties had no problems.

To get buy-in to making a follow-on to the '6, they went to ridiculous
ends to make it possible to price the base system around $100,000 (the
bare PDP-6 CPU listed at $176K) including making the following
options:

KE10	KA10 byte instruction option ("extended order code")
KM10	KA10 16 Word 0.21 microsecond fast memory (ACs) option
KP10	KA10 power fail/restart option [I/O Bus & 7 level PI?]
KT10	KA10 single protection/relocation reg [never sold]
KT10A	KA10 dual protection/relocation regs

Without "fast ACs", the registers (visible as the first 16 addresses)
used external core (was also an option on the PDP-6).

[TUHS] Re: A PDP-10 used for UNIX just after the PDP-7?

[Noel Chiappa via TUHS]

    > From: Thalia Archibald

    > in Thompson's 1989 interview with Mahoney, he mentions that "Before the
    > 11/45 was available we bought a PDP-11 that had PDP-10 memory
    > management, KS-1, it was a one of a kind machine".

In trying to find out more about the KS11:

  https://gunkies.org/wiki/KS11_Memory_Protection_and_Relocation_option

I queried Ken, and he was nice enough to reply, and explain to me that it
operated like the memory management hardware on the KA10 model of the PDP-10.

I think that's how the 'PDP-10' confusion started.


    > Does anyone know anything about this strange PDP-11 + KS-1?

The KS11 was the very first hardware memory mangement available for the
PDP-11. So, it was either that - or a bare macine (-11/20) with _no_ memory
mangement.

Bob Supnik told us a bit about the KS-11: "The KS11 created a PDP-10 (KA)
style memory management system for the 11/20: that is, high-segment
(instructions, in theory, and shareable, in theory) and low-segment (data)."

No documentation on the KS11 has ever been located. There is KS11 emulation
included in Bob's MIMIC PDP-11 simulator; in theory someone energetic could
dig into that, and work out the details of how it works. More here:

  https://gunkies.org/wiki/Talk:KS11_Memory_Protection_and_Relocation_option

if anyone is interested. MIMIC was written long ago (the header says "March
31, 1971"); it's in PDP-10 assembler (it long pre-dates SIMH), so it wouldn't
be trivial.

There was also support for the KS11 in UNIX V2, from which it would be easier
to produce a programming manual for the thing. Alas, that version of UNIX has
been lost (although a bit about how the KS11 worked can be deduced from the
V2 manual, which is extant).


    > What was the model of PDP-11 first used for UNIX?

The PDP-11/20. Dennis' "The Evolution of the Unix Time-sharing System":

  https://www.nokia.com/bell-labs/about/dennis-m-ritchie/hist.html

indicates that they got it _very_ early in the life of the -11: "the PDP-11
was so new a product that no disk was available until December". A later
comment there ("a single .5 MB disk") indicates that they probably got an
RF11:

  https://gunkies.org/wiki/RF11_disk_controller

Not sure why they didn't get an RC11.

    Noel

[TUHS] Re: A PDP-10 used for UNIX just after the PDP-7?

[Phil Budne via TUHS]

Another PDP-10 footnote, which brings things back around to Ken
and UNIX:

BB&N had done a VM LISP for the PDP-1, but wanted a larger, timeshared
system with VM.  They looked at the PDP-6 as ideal (a 36 bit word
could hold two 18 bit pointers for the CAR and CDR of a LISP list
element), but it was cancelled, but then they heard about development
of the PDP-10.  BB&N tried to interest DEC in building in hardware for
paging, but as I previously noted, the DEC's emphasis at the time was
NOT making another overblown white elephant; the 24-bit SDS 940 (the
commercial version of the Berkeley modified SDS 930) had limited
address space, so BB&N picked the PDP-10 and built their own external
MMU (paging box). The project was led by Daniel Bobrow, and was named
TENEX.  PDP-10s were a major share of systems on the early ARPAnet,
and most of them were running TENEX.

Here's the cruicial bit of UNIX co-history:

TENEX was designed in 1969 (and was up and running in under a year?),
and took feedback from the Multics team on complicated stuff they
should throw out, as well strong influences from the Berkeley "Genie"
time sharing system; The term for a process in TENEX is "fork").
Ken had used the Berkeley system before joining the Multics team.
(See DMR's "incomplete history of QED"(*))

In Ken's paper "Reflections on Trusting Trust" on the receipt
of the Turing award, he wrote:

    I thank the ACM for this award. I can't help but feel that I am
    receiving this honor for timing and serendipity as much as technical
    merit. UNIX swept into popularity with an industry-wide change from
    central mainframes to autonomous minis. I suspect that Daniel Bobrow
    would be here instead of me if he could not afford a PDP-10 and had
    had to "settle" for a PDP-11

Dan Murphy (who had created TECO for the PDP-1 at MIT) who had worked
on the TENEX VM system, and later went to DEC (first to port TENEX to
the KI10, the second PDP-10 CPU, which BB&N again failed to convince
DEC to implement their VM design in, but DID implement a much simpler
paging system, using the same page size as TENEX) to turn TENEX into
TOPS-20 has a page on TENEX, and whose reminiscences I have likely
mis-remembered above:

https://opost.com/tenex/

And includes a version of his history published by IEEE (in an issue
that also includes an interview with Gordon Bell including some
history of the PDP-6):

https://opost.com/tenex/ahc_20150101_jan_2015.pdf

And a longer version, written for a never-published book on PDP-10
history/lore, perhaps for the 1984 20th anniversary of the PDP-6?

https://opost.com/tenex/hbook.html

(*) https://www.nokia.com/bell-labs/about/dennis-m-ritchie/qed.html

Ken wrote a version of QED for the MIT CTSS time sharing system on the
7090, that the Multics team was using for bootstrapping, then QED for
Multics.

[TUHS] Re: A PDP-10 used for UNIX just after the PDP-7?

[segaloco via TUHS]

On Friday, January 16th, 2026 at 10:00, Noel Chiappa via TUHS tuhs@tuhs.org wrote:

> There was also support for the KS11 in UNIX V2, from which it would be easier
> to produce a programming manual for the thing. Alas, that version of UNIX has
> been lost (although a bit about how the KS11 worked can be deduced from the
> V2 manual, which is extant).
> 
> Noel

I presume everything would be in the kernel that touches hardware memory management stuff?  Reason I ask is there is plenty of documented EAE use in V2 userland stuff:

https://gitlab.com/segaloco/v2src

For those who haven't perused it, thats my slow-burn restoration of the V2 userland from Dennis Ritchie's "s2-bits" tape.  EAE use dates it to the PDP-11/20 system and presence of no, V1, and V2 a.out magic numbers puts them at V2.  By comparison, the V2 "cmd" path on the UNIX tree (s1-bits recovery) is later versions, probably V3, as they are still assembler but any EAE usage has been replaced with PDP-11/45 arithmetic operations.

If there is any compelling thought on whether any V2 userland code would *know* about this memory subsystem, I'm happy to go disassembling and looking around for clues.  We don't know of any other V2/V3 artifacts beyond manuals and s1/s2 bits tapes do we?

- Matt G.

[TUHS] Re: A PDP-10 used for UNIX just after the PDP-7?

[Al Kossow via TUHS]

On 1/16/26 10:24 AM, Phil Budne via TUHS wrote:

> Ken wrote a version of QED for the MIT CTSS time sharing system on the
> 7090, that the Multics team was using for bootstrapping, then QED for
> Multics.
> 

The 1965 manual for Genie QED
http://bitsavers.org/pdf/sds/9xx/940/ucbProjectGenie/30.60.30_QED_Nov65.pdf

[TUHS] Re: A PDP-10 used for UNIX just after the PDP-7?

[Al Kossow via TUHS]

On 1/16/26 10:00 AM, Noel Chiappa via TUHS wrote:

> Bob Supnik told us a bit about the KS-11: "The KS11 created a PDP-10 (KA)
> style memory management system for the 11/20: that is, high-segment
> (instructions, in theory, and shareable, in theory) and low-segment (data)."
> 
> No documentation on the KS11 has ever been located. There is KS11 emulation
> included in Bob's MIMIC PDP-11 simulator; in theory someone energetic could
> dig into that, and work out the details of how it works. More here:
> 
> 

I thought someone had the segmented kernel running in SIMH, which is why the
KS11 support was added.

[TUHS] Re: A PDP-10 used for UNIX just after the PDP-7?

[Noel Chiappa via TUHS]

    > From: Matt G.

    > I presume everything would be in the kernel that touches hardware
    > memory management stuff?

I expect so. We know from Dennis' "A hardware story":

  https://www.nokia.com/bell-labs/about/dennis-m-ritchie/odd.html

that the KS11 took almost everything in the I/O space on the UNIBUS out of
the user's address space. The I/O space was where the KS11 registers were.

So, any fooling with the memory management by the user would have to have
been done by a system call - hopefully documented in the V2 manual, which we
have. I do't see any such system call (although break() may have played with
it).

	Noel