🚀 go-pugleaf

In article <38...@network.ucsd.edu> c...@coast.UCSD.EDU (Charles Coughran) writes:
>Our 1130 gave many years of bullet proof service.  We finally had to
>give it up after IBM quit servicing the beast.  The 1800`s lasted
>longer because we had several to cannibalize.

Speaking of the 1130/1800, does anyone know of any other architectures
that are binarily compatible but use different "word" sizes?  If I
remember correctly, the 1130 used 16-bits while the 1800 used 18-bits(?).
But they both used the same instruction set and software.

Anyone else care to comment on these architectures?

>Ahem!  I take objection to that!
>The 1130 is a fine machine for learning machine architecture.
>It had a very thoughtful interrupt structure and a front panel that
>let you see that machine's state all over.
>
I'm pleased to see someone remembering the good points of a solid
old machine.  It`s always easy to dump on old technology.
I too learned on an 1130 in the late sixties - assembly and Fortran
IV.  (I never have figured out what`s wrong with a Goto.)

Our 1130 gave many years of bullet proof service.  We finally had to
give it up after IBM quit servicing the beast.  The 1800`s lasted
longer because we had several to cannibalize.

--
Signature?  I don't got to show you no stinking signature!
ccou...@ucsd.edu (Internet) ...!ucsd!ccoughran (UUCP) CCOUGHRA@UCSD (BITNET)

OK, after listening with great amusement to all you other old hacks, I 
feel obliged to recount a few of my own stories.

First the IBM 1800.  Rummage, rummage through the archive file cabinet.
I quote from File No. 1800-01, Order No. GA26-5918-8,
IBM 1800 Functional Characteristics.  Ninth edition, August 1970.  
(This is formatted exactly as in the manual.)

    CORE STORAGE

    The 1800 system main storage uses magnetic cores 
    for data and program instruction storage.  Each ad-
    dressable core storage location contains 18 bit posi-
    tions and is called a word.  One of the 18 bits in a 
    word is used for storage protection and one bit is
    used for parity checking.  The remaining 16 bits in
    each word are data bits.

So the 1130 and 1800 machines were strictly compatable, except for
the 1800 instructions to do storage protection and process i/o.

Some stats on memory:
    min   4096 words, 2 or 4 micro-second cycle time
    max  65536 words, 2.25 micro-second cycle time


Now the obligatory anecdote.  I was a systems hack-nerd in 1972
while a high school student.  The 1800 belonged to the Chem Eng
department at the University of Alberta.  It was used for 
process control research.  The machine had oodles of process 
i/o (analog, digital, varying speeds) and 27 interrupt levels!

The console printer was the print mechanism of an IBM selectric
typewriter.  Every so often it would burble and print out some
status message such as
  Over temp warning on evaporator, shutting down.

Really important messages would be printed in red.

One error message, I forget the details, went something like
  MPX01029 - Parity error at ....

When you went to the error manual, there was a penciled note that
said:
  Open core box door and follow instructions.
If you didn't know what a core box was, you stopped right then.

When you opened the door on the box (about the size of a fridge)
hanging before you was a rubber mallet.  On one of the boxes was
a yellow X.  The instructions said 
  hit the yellow X firmly, once.

Sure enough, that would shake up the insides enough that the 
problem would go away for another few months.   I don't remember
if the customer engineer ever fixed that one.


As for error messages, one of the devices the 1800 controlled 
was a gycol evaporator.  There was this great process operator's
console with a 3 line display made from 15 segment nixie tubes.
You could push buttons to get and modify various control parameters
for the evaporator control loops, and read off things like temp and
flow rates.  Being full of flashing lights and buttons, the console
was a perfect example of what in law is called an "attractive 
nuisance".  Passers by the lab were always playing with it.

The syntax for commands to the console was somewhat
baroque, so if you just pushed buttons at random it would usually
respond with "INVALID INPUT".  But if you did this three times in
a row it would reply "YOU DON'T YOU WHAT YOU ARE DOING, DO YOU?"
and lock out for 30 seconds.


Prof. Jim Hoover                   | Office +1 403 492 5401 or 5290
Dept. of Computing Science         | FAX    +1 403 492 1071
University of Alberta              | hoo...@cs.ualberta.ca
Edmonton, Alberta, Canada T6G 2H1  | 
--
Prof. Jim Hoover                   | Office +1 403 492 5401 or 5290
Dept. of Computing Science         | FAX    +1 403 492 1071
University of Alberta              | hoo...@cs.ualberta.ca
Edmonton, Alberta, Canada T6G 2H1  | 

>Speaking of the 1130/1800, does anyone know of any other architectures
>that are binarily compatible but use different "word" sizes?  If I
>remember correctly, the 1130 used 16-bits while the 1800 used 18-bits(?).

The 1130 had, as I remember, 16 bits per word plus two parity bits. 
(I'm certain about the 16 bits per word.)  As I remember, the 1800 had
16 bits per word plus one parity bit plus a "write protect" bit, which
could be set (or maybe cleared, it may have been a "write enable" bit)
to prevent attempts to write to a location.

The registers were 16 bits long on both machines, the instructions were
16 or 32 bits long, and all data items were 16 bits long or a multiple
thereof.  I.e. you had 16 data bits per word on both architectures;
MAXINT was 32767 on both machines.

>But they both used the same instruction set and software.

Well, not exactly; the 1800 had some additional instructions and ran a
different OS.

>Ahem!  I take objection to that!
>The 1130 is a fine machine for learning machine architecture.
>It had a very thoughtful interrupt structure and a front panel that
>let you see that machine's state all over.

I liked the 1130 too, don't get me wrong.  It was useful, and a lot of
fun, I wish I had one in my basement.  For one thing, I could still
program in FORTRAN II that way!  :-)

When I said it ultimately belonged in the junkyard, I meant that it got
sadly left behind by the technological revolution that followed.  When
they should have done was retrofit it with a RISC CPU!  Or a 68020.
Wouldn't you love to be the only programmer on Earth writing 68020 code
on Hollerith cards??

-- 
A doubled signature is the devil's work. ** Tom Neff <tn...@bfmny0.BFM.COM>
-- 
A doubled signature is the devil's work. ** Tom Neff <tn...@bfmny0.BFM.COM>