[TUHS] Unix on PDP8?

[TUHS] Unix on PDP8?

[Will Senn]

All,

Was Unix ever ported to a PDP8, or any other 12 bit environment, for 
that matter? If not, why not? My understanding, such as it is, is that 
Unix was created on the PDP7 - btw, thank you very much, Ken Thompson, 
you definitely changed my world :), which is an 18bit machine, and that 
it soon found its first real home on the 16 bit PDP11 series of machines 
(an 11/20), and from there, ever upward or at least ever onward. I'm 
curious about it for historical reasons, of course, but also because 
I've been messing around in the PDP8 emulation world and enjoying the 
excursion into simplified ISA and memory architectures.

-will

[TUHS] Unix on PDP8?

[Clem Cole]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 3103 bytes --]

​below....​

On Tue, Nov 28, 2017 at 2:55 PM, Will Senn <will.senn at gmail.com> wrote:

> All,
>
> Was Unix ever ported to a PDP8, or any other 12 bit environment, for that
> matter?

​Not to my knowledge.




> If not, why not

​My guess ....  is that it was not worth it/did not make economic sense for
anyone.


> My understanding, such as it is, is that Unix was created on the PDP7 -
> btw, thank you very much, Ken Thompson, you definitely changed my world :),
> which is an 18bit machine, and that it soon found its first real home on
> the 16 bit PDP11 series of machines (an 11/20), and from there, ever upward
> or at least ever onward. I'm curious about it for historical reasons, of
> course, but also because I've been messing around in the PDP8 emulation
> world and enjoying the excursion into simplified ISA and memory
> architectures.
>
> -will
>

Ken had borrowed a cast off 18 bit word addressed PDP-7 (which became the 9
for all intents and purposes).   They purchased a 16 bit PDP-11/20.  By
that time of the PDP-7/9 (18 bits) lines are being replaced by the PDP-10
(18/36 bits) and the 12 bit PDP-8 line, is being replaced by the 16 bit
PDP-11.

Ken and company had tried to purchase a PDP-10 and been shot down.   So he
took the PDP-11, which they were able to get funded.   There was nothing
the 8 could do that the 11 could not do as well or better and the it was
not going to cost any more to use 11s.  So there was not incentive at BTL.
  They now have it on the 11 line.

After UNIX get released in to the wild, the same thing is happening.
Unless you had an older PDP-8 you wanted to try to run UNIX on, why
would you have ported it?   TSS/8 was cheap (ney free from DECUS IIRC) so
there was a timeshared system for the the 8s already.   You'd have have to
build new tools which is what we did for other systems, but we always
ported to a system that was giving us something -- often cheaper cycles
than the PDP-11 and getting than what we could have gotten on the packaged
system.  And PDP-11s ankle to run UNIX even at $50-250K were cheap for the
time.

I note that the PDP-11 competitor and the other PDP-8 follow on was Ed
DeCasto's Nova.    And while Xerox clones the Nova for the Alto, I never
heard of a UNIX port to the Nova either.  DG's SW offerings were fairly
good.

So folks like me either bought a PDP-11 from DEC and got UNIX (what most
did) and/or you might have some how need up with a cheap system that 24 or
32 bits and then you thought about porting.   Even the 'NUIX' machine (IBM
Series/1) was a case where the Cleveland State folks got the system for
super cheap.  They had a borrow cycles from Case to do the port because
they did not have an 11.

People like me did not start to "port" UNIX until the 16 bit micro's show
up -- i.e. the 68000, Z8000, 8086 where you could build a 'microcomputer'
that was close to the power of the 'minicomputer' for a lot less money.

Clem
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20171128/6df4e4df/attachment.html>

[TUHS] Unix on PDP8?

[Ron Natalie]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 1679 bytes --]

Ø  People like me did not start to "port" UNIX until the 16 bit micro's show up -- i.e. the 68000, Z8000, 8086 where you could build a 'microcomputer' that was close to the power of the 'minicomputer' for a lot less money.

 

 

 Mike Muuss’s standard answer to any question was that we could put UNIX on it.    This is how he wrested the RSTS-running PDP-11/45 away from the EE department to start with (only proviso is that he had to get Basic+ running on it).    At BRL, he picked up a bunch of PDP-11 variants that were lying around (particularly an 11/34-based RJE system for the Cyber mainframe.    We pitched the card reader but used the printer, the vector general graphics system, and the DQ-11/modem combo).    Then when BRL commissioned the HEP to be built and nobody had a clue what software to put on it, we launched into the port on the HEP.   This was 1982.   It our first foray into a non PDP/VAX platform.    

 

Amusingly, I had a manager show up in my office and tell me that in a few years they would have the performance of a VAX in a little cubic foot box I could have on my desk all to myself and I’d be happy.    It was then I coined “Ron’s rule of computing.”   Our expectations grow with the technology.   Ron always needs a computer “this” (holding my hands out to approximately the width of a 780 cpu cabinet) big.    It worked for many years but I have to admit around 2000 or so, I started being happy with comptuers about the size of 2 drawer filing cabinets.

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20171128/84518f5b/attachment.html>

[TUHS] Unix on PDP8?

[Lars Brinkhoff]

Will Senn wrote:
> Was Unix ever ported to a PDP8, or any other 12 bit environment, for
> that matter?

There are Unix-like systems for 8-bit micros which might possibly be
adapted for PDP-8.  Lunix, UZI, FUZIX...

> I've been messing around in the PDP8 emulation world and enjoying the
> excursion into simplified ISA and memory architectures.

The tiny instruction set is surprisingly versatile!

[TUHS] Unix on PDP8?

[Clem Cole]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 2810 bytes --]

On Wed, Nov 29, 2017 at 1:56 AM, Lars Brinkhoff <lars at nocrew.org> wrote:

>
> The tiny instruction set is surprisingly versatile!
>
​The precursor to today's VHDL and Verilog was a language called ISPL /
ISPS.    There were many ISPS ​descriptions of different systems being
written in those days.  I bring it up because, the PDP-8 - aka 'mini', ISP
was a single 66-line lineprinter page long.  Which was astounding, when you
compared it to the ISP descriptions of the PDP-9/10, the IBM 360 *etc*...
Even the PDP-11 ISP is a many pages because of all the addressing modes.

Remember, Gordon Bell's term 'mini-computer' was not describing a 'small'
computer, but instead it was a 'minimal computer.'   It was only when the
microprocessors were created a few years later that the term was warped to
mean  'small' by the computer press.


As a side note of UNIX history (thank you Will for the reminder) ...
 another piece of my to do list is get ISPS running again.   The original
version was in BLISS-10, then compatible BLISS (Vax) on VMS which Danny
Klein and I worked on.  The BLISS version generated net-lists for DEC
PDP-16 RTM modules, [which I'm sad to say are a lost art and I fear may
have been lost to history.  I may be one of the last groups that ever
designed with them.  It was DEC productization of the 'flip chips'
originally created for the PDP-7 and PDP-8.  IIRC you can read about them
in the DEC 'blue book'].


But the late Ted Kowalski's PhD thesis was a C implementation that ran on
the CMU's V6++ UNIX / PDP-11 [that actually generated moclisp as the parse
trees - very interesting system].   Ted wanted to go to transistors
directly in the back-end.  Working with this thesis advisor at the time,
(Don Thomas) an improved version of Ted's work, would become VDHL - which
Don wrote all the books, in the 1990s.

Besides traditional word processing (troff et al) and C program
development, ISPS was one of the first 'production' use of UNIX I saw.
Trying to generate 'chips' automatically from HW descriptions in the late
1970s.

Anyway, if you find a copy of the Sieworwick, Bell, and Newell's book
'Computer Structures, Reading and Examples', there is a companion volume
that has many of the ISPS descriptions of the machines discussed in the
main text.   As Will's note about HP points out, as historians we should
try to find them all and get them in bitsavers or the like.  I know I have
some of the ISPs for the micro's and the PDP-11 on hardcopy in a filing
cabinet (I just ran into them a few weeks ago when looking for something
else), but I should have them on tape.   It would be a shame to lose those.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20171129/c626a4a4/attachment.html>

[TUHS] Unix on PDP8?

[Robert Swierczek]

> Anyway, if you find a copy of the Sieworwick, Bell, and Newell's book
> 'Computer Structures, Reading and Examples', there is a companion volume
> that has many of the ISPS descriptions of the machines discussed in the main
> text.

It looks like archive.org has a copy as part of the Boston Computer
Museum collection:

https://archive.org/details/computerstructures00bell

including what looks to be the companion volume you described:

https://archive.org/details/designanalysisof0000barb

[TUHS] Unix on PDP8?

[emanuel stiebler]

[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #1: Type: text/plain, Size: 703 bytes --]

On 2017-11-29 07:05, Clem Cole wrote:

> Anyway, if you find a copy of the Sieworwick, Bell, and Newell's book 
> 'Computer Structures, Reading and Examples', there is a companion volume 
> that has many of the ISPS descriptions of the machines discussed in the 
> main text.   As Will's note about HP points out, as historians we should 
> try to find them all and get them in bitsavers or the like.  

> I know I 
> have some of the ISPs for the micro's and the PDP-11 on hardcopy in a 
> filing cabinet (I just ran into them a few weeks ago when looking for 
> something else), but I should have them on tape.   It would be a shame 
> to lose those.

Just send them to bitsavers, to save them ...

[TUHS] Unix on PDP8?

[Ron Natalie]

C itself is going to be difficult on the PDP-8.    Too much ingrained (and
now codified by ANSI/ISO) that requires shorts to be at least 16 bits and
longs to be at least 32 bits, neither of which the 8 supports well.
Even the EAE or FP modules only added limited 24 bit support.    

-----Original Message-----
From: TUHS [mailto:tuhs-bounces@minnie.tuhs.org] On Behalf Of Lars Brinkhoff
Sent: Wednesday, November 29, 2017 1:56 AM
To: Will Senn
Cc: Tuhs
Subject: Re: [TUHS] Unix on PDP8?

Will Senn wrote:
> Was Unix ever ported to a PDP8, or any other 12 bit environment, for 
> that matter?

There are Unix-like systems for 8-bit micros which might possibly be adapted
for PDP-8.  Lunix, UZI, FUZIX...

> I've been messing around in the PDP8 emulation world and enjoying the 
> excursion into simplified ISA and memory architectures.

The tiny instruction set is surprisingly versatile!

[TUHS] Unix on PDP8?

[Dave Horsfall]

On Wed, 29 Nov 2017, Ron Natalie wrote:

> C itself is going to be difficult on the PDP-8.  Too much ingrained (and 
> now codified by ANSI/ISO) that requires shorts to be at least 16 bits 
> and longs to be at least 32 bits, neither of which the 8 supports well. 
> Even the EAE or FP modules only added limited 24 bit support.

You could start with (an early) Minix, I guess; we ran it on a PDT (I 
think) with 2 x RX-05 floppies.  Slow, but it worked, sort of.

-- 
Dave Horsfall DTM (VK2KFU)  "Those who don't understand security will suffer."

[TUHS] Unix on PDP8?

[Warren Toomey]

On Thu, Nov 30, 2017 at 09:11:19AM +1100, Dave Horsfall wrote:
>> C itself is going to be difficult on the PDP-8.

Yes, as noted on this page: http://so-much-stuff.com/pdp8/C/C.php

> You could start with (an early) Minix, I guess; we ran it on a PDT (I 
> think) with 2 x RX-05 floppies.  Slow, but it worked, sort of.

You mean Xinu, perhaps? Minix was always an 80x86 and 680x0 system.

There is also xv6: https://pdos.csail.mit.edu/6.828/2017/xv6.html
with a small monolithic kernel. I've added a BSD library and runtime
to the kernel here: https://github.com/DoctorWkt/xv6-freebsd

Of course, you could always go retro and try Mini Unix :)

Cheers, Warren

[TUHS] Unix on PDP8?

[Dave Horsfall]

On Thu, 30 Nov 2017, Warren Toomey wrote:

>> You could start with (an early) Minix, I guess; we ran it on a PDT (I 
>> think) with 2 x RX-05 floppies.  Slow, but it worked, sort of.
>
> You mean Xinu, perhaps? Minix was always an 80x86 and 680x0 system.

Can't remember; we're talking late 70s / early 80s, and my organic 
associative memory ain't the best these days.  I was busy running a 
network of 11/40s, squeezing as much of V7 into V6 as I could, and talking 
the poxy UT-200 protocol for RJE (the original purpose of our 11/40s) with 
its broken implementation on the Cyber 72.

Damn; I wish I still had those CSU tapes, but they'd be unreadable by now 
(IMHO my "ei.c" driver was a work of art).

> Of course, you could always go retro and try Mini Unix :)

That rings a bell; we're talking about the early UNSW Unix days (CSU, 
AGSM, Basser, Elec Eng, etc).

-- 
Dave Horsfall DTM (VK2KFU)  "Those who don't understand security will suffer."

[TUHS] Unix on PDP8?

[Robert Swierczek]

> There are Unix-like systems for 8-bit micros which might possibly be adapted
> for PDP-8.  Lunix, UZI, FUZIX...

Maybe pdp7-unix could be ported/rewritten.
I'm guessing the available memory would be the biggest issue.

> C itself is going to be difficult on the PDP-8.

Perhaps the B interpreter from pdp7-unix can be be simplified further?

[TUHS] Unix on PDP8?

[Dave Horsfall]

On Wed, 29 Nov 2017, Robert Swierczek wrote:

> Maybe pdp7-unix could be ported/rewritten. I'm guessing the available 
> memory would be the biggest issue.

Wasn't extended memory on the -8 bank-switched via the IOT instruction?

It reminds me of what we did with AUSAM back at UNSW; the current buffer 
"b" (similar to "u") was mapped via KISA5 (?) into whatever was at 0150000 
(?), so you could have as many buffers as you wanted, limited by physical 
memory (I think).

-- 
Dave Horsfall DTM (VK2KFU)  "Those who don't understand security will suffer."