[Simh] Older Model VAX simulators, what can you do with them and how do they work?

[Clem Cole]

The key point for a simh person - while Mark and Bob allow you to
​put together ​
configurations in simh that may or may not have never been realized and
often they will "just work" --- you really need to match not just the HW by
the timeframe of the release of that HW with the timeframe of the release
of the SW that supported it.   UNIX, and BSD in particular, was amazingly
fluid; but some of the earlier editions are not going to run on all HW
systems or support all the different devices that came on later systems.

On Sun, Oct 9, 2016 at 2:43 PM, Mark Pizzolato <Mark at infocomm.com> wrote:

> My first suggestion would be to try and dig up the original instructions
> for installing that particular OS on a real VAX 11/750.  This may not have
> even been an OS that ran on a 11/750.


​Right - even if you get past the boot strap, the 750 did not exist when
3BSD was made available from UCB's CSRG.  While it is a lot like a 780,
there were definitely differences.    IIRC, Pike does not take delivery on
the first 750 at Murray Hill until after 4BSD came out (the Research VAX
kernel ran on it first - Joy brought the changes back to UCB after a summer
in NJ -- I'm guessing 81 or 82​).

To contrast for the "modern" user.    At this time, the PC/AT was just
appearing at about $3K per system, running MS-DOS and wowing some people.
I remember Rob describing the 750 @ about $50K as a great "Personal
Computer" at a USENIX and thinking how great to work for company that could
spend that kind of money per engineer on a computer.   Because the 750 was
a "real computer" with megabytes of memory, VM so it could address even
more, real disks etc, ran UNIX really well.   Us UNIX types considered the
AT a toy and even with it running Microsoft's "Xenix" - did not have the
BSD stuff so it was pretty unimpressive and we turned our nose up to it
 (not seeing the disruption coming).


Back to the question, the  4.1BSD release does support the 750 and probably
ran on more 750's then anything else but that may have been a patch after
the original release (I'd have to check my tapes).  We did have a lot of
them at BSD and one -- thing the 750 has was a funky tape drive that
replaces the 780's floppies. I do remember work went into getting the
standalone system to deal with that tape system.   Stretching my memory a
bit, the 730 may have used that same front end.  But I think it was
similar, although slightly different so there were changes needed for that
also.

I've forgotten now, Horton or Armando might remember better than I did, but
4.1BSD >> might have later<< supported the 730 "out of the box"; but I do
not think so.  Again - that would have been a much later tape and much more
likely a patch available on the BSD ftp site if it was available.
Frankly, I do not remember anyone that ran BSD on a 730 (I do not even
remember any of them at UCB but their could have been - I remember lots of
780s and 750s).   Basically, by the time DEC released 730 the 68K based
systems had already been available for about 2-3 years from the Masscomp,
Apollo, and eventually Sun.   Those systems were faster and significantly
cheaper, hence the start of the UNIX migration to same.

As a (humorous?) side bar; at the time, you could tell how serious the HW
vendor was in their advertisements by which DEC system they used as their
benchmark standard (780, 750 or 730). *i.e.* we used to joke, that the 730
was a greatest gift to the UNIX HW biz, because it allowed anyone to
benchmark against a "VAX" and demonstrate their system was faster - even
with an 8Mhz 68K and a very simple (untuned) UNIX port.      Those
benchmark games and the question of what defined a MIP *etc.* - begat the
SPEC benchmarks.


As for BSD itself, between the years of small differences in CPU models
from the PDP-11, then the 780/750/730 making life even more interesting,
with what would become 4.2 (I think it was cut in 4.1B, but I have to check
tapes and SCCS), Sam would write the BSD kernel configuration program to
generate the HW specific kernel code.

So by the time BSD4.2 comes, out you have different models being recognized
in the base base system and bootstraps being published widely.   Also, TIG
in Merrimack had begun to push specific HW support "upstream" to UCB (thank
you Armando).   Certainly besides more CPUs, 4.2 had more later model
device support than 4.1 did.


Clem
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.trailing-edge.com/pipermail/simh/attachments/20161010/b8a9bbf1/attachment.html>