[TUHS] PDP-11 questions

Clem Cole clemc at ccc.com
Mon Jan 25 04:49:21 AEST 2016


On Sun, Jan 24, 2016 at 12:37 PM, Mark Longridge <cubexyz at gmail.com> wrote:

> Ok, I got a few questions about PDP-11.
> First, I was wondering when Bell Labs got that first PDP-11/20 what
> software (if any) came with it? I assume when one bought a PDP-11/20
> you would get some type of OS with it.
​SW?? We don't need no stinking SW - we do our own..

Seriously, SW was an option you paid for. The system came with paper tapes
to run diagnostics to prove the system worked.   And when they got their
first 11, there is no disk and DOS-11( it's 1st OS) was not yet released
IIRC.  ​

Have to ask Ken, but my guess and I think one or more of the BLTJ articles
back it up, the would have purchased it as it.   Maybe an standalone
assembler running from paper tape or DEC tape.  But whatever it had, it
would have been very, very limited.

But that was not an issue, they had other systems and could write their own
tools and cross assemble or compile them as needed (which is what they

> According to the folks at alt.sys.pdp11 the PDP-11 computer doesn't
> have anything equivalent to a PC's BIOS.
​Sigh... Kids these days .... more in a minute....​

> ​... ​
> Wouldn't the DRAM
> ​ ​
> on the PDP-11/something need to be initialized too?
​What's DRAM -- early 11's had core.   That said, Intel was selling the
1101 (1k x1 bit) chips which were being consumed at a pretty good rate for
the memory systems for PDP-10s​.    DEC would not release a DRAM board for
the 11 for a few more years.   It was 2K x 16 bits (with ECC IIRC).    And
they were pretty pricy.   I seem to remember the had faster access times
that the core boards, but I'm hazy on that.

When UNIX starts to leak to the Universities in the mid to late 70's many
(??most??) of us are using DRAM on our 11's but we would often by the
minimum config from DEC and the use after market DRAM boards.   We have a
very early serial # 11/34 under 10 IIRC in the EE Dept at CMU (One of my
claims to fame was bring UNIX up on it for the first time - by hacking the
11/40 support - although I think Noel and few others did it in other places
too there after).

Anyway - that system had 24K words (48K bytes) of DRAM memory on it to
start with.  We made some memory board for it ourselves to max it to 128K
words and National Semi memory chips.   I remember the value of the memory
chips was greater than the CPU at that point.

​Anyway - core machine you did not want to init.  You usually left the OS
or whatever in place.
It's going to be interesting to see if this becomes the new norm with the
Crystal Ridge (Xpoint 3D or whatever marketing is calling it).

Perhaps an older
> PDP-11 doesn't have DRAM but surely the later models did?
Sure by why the pre-boot have to do it?  Init of the memory system is done
by the OS.   Look at the code in V6 and V7 that is called very early and
prints out the size of the memory it finds.   It's working backwards until
it find memory that responds and clears it out.

There are a few parts/functions to the BIOS rooms and fear you may be
mashing the together.
Init of the memory system is not done in the original PC BIOS the way it is
done now.  There reason is because today we have memory controller chips
with lots and lots of different options.
The firmware BIOS is used to set up that controller (pre-boot).   The 11
(and the Vax for that matter) did not have such an idea.   What we call the
memory controller was just part of the logic in the CPU.

> ​... ​
> It
> ​ ​
> seems though that there should have been a PDP-11 based desktop and as
> far as I can tell that didn't happen. Instead we got a bunch of micros
> with 8080, z80 and 6502 cpus, but nothing that could run Unix, at
> least not a Unix v7 with source code.
​I can see from an later observers view you might fall into a trap thinking
this, but a bunch of it is actually not true.

1)  There were small form factor PDP-11's that did appear late in the
PDP-11's life.   Some based on the LSI-11 and some even on the F-11/J-11
(single chip 11).   DEC had a line called DEC Processional series.  But
they really were not super popular.  But as other point out, DEC was slow
to recognize this as a market. In fact a professor at Harvard business
explain the problem at DEC and coins a term for the behavior when the 8 bit
and 16 bit micros appear [Clay Christiansen's book "The Innovators'
Dilemma" - the term is call "disruptive technology."]

FWIW:  At this time, Wang Labs (Dr. Wang invented the core memory at IBM
BTW, was selling a system for secretaries / admin that ran originally on an
8080, later a Z80.   It was quite popular for what it could do... which was
allow them to edit letters/documents.   But it was very much focused on a
specific market/task - which interestingly enough was the original task
UNIX had ;-)

Also by the time DEC did try to build a workstation (after Masscomp,
Apollo, Sun et al had taken many of their engineers) it was too little too
late.   The ship had sailed and they never recovered that market.

2) Economics was really the reason.  Please understand in the 1975 dollars,
a 11/34 with 2 RK05's, 24K words of memory and a single serial interface
cost about $45K.​    If you want to a 9-track tape drive that was another
4-8K, 200M RPxx style disk, another $15K, as I said the chips to make the
memory was $45K.  Much less, serial ports, a printer etc....

Using an 11 for "personal" computer was not cost effective.   You would
need to get the prices of memory, large storage, down and size/speed of the
processors in single chip form before you really do it.

That said with birth of the IBM PC, Andy Tannebaum wrote a really good Unix
V7 clone for the 16 bit 8088 - called Minux.   And V& itself began to move
to more capable micro's    But DEC was making a huge amount of money
selling Vaxen.   So the workstations and small capability systems were not
interesting (see Christiansen for why).

3)  Some people actually did get UNIX or close to unix functionality
running on the 8-bit machines.  The Guy who wrote BDS C (Brain Damaged
Software) brought an 8" floppy disk based Z80 system that originally had
been running CP/M to a Usenix in the late 1970's/early 1980s and showed a
couple of us including Dennis his OS.   I remember Dennis being pretty
impressed and stating that was it was fast and as good as he remember early
UNIX.   He was quite encouraging.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://minnie.tuhs.org/pipermail/tuhs/attachments/20160124/69dfb2d6/attachment.html>

More information about the TUHS mailing list