8-bitters and multi-whatever - test-drb@ccmp.vtda.org

11 Sep 2007

...

Subject: Re: 8-bitters and multi-whatever
   From: "Chuck Guzis" <cclist at sydex.com>
   Date: Mon, 10 Sep 2007 13:51:12 -0700
     To: "Roy J. Tellason" <rtellason at verizon.net>,
 "General Discussion: On-Topic and Off-Topic Posts" <cctalk at
classiccmp.org>

On 10 Sep 2007 at 16:24, Roy J. Tellason wrote:

  I remember seeing that in some sales literature
and it always did strike me as 
 being more spendy than I wanted or could afford to get into.  :-) 
For the purist, Multibus has a lot more going for it than any of the 
"hobbyist" buses.  If you look at the signal layout, it appears that 
some thought actually went into the design.  AFAIK, Multibus cards in 
some incarnation are still being produced or at least sold.

And the MDS-800 was built like a battleship. 

Thats and understatement.  If you ever moved a MDS800 it was a battleship.

...
 re:  CP/M Networking:

  What does that take on the software side of
things? 
A CP/M add-on called CP/NET.  The downside is that it takes valuable 
memory. 
CPnet was less than half the software.  You still needed a nios, network 
bios which was often unser created for the specific IO used to network.

...
 PC's had a number of "cheap" networking
setups.  I've got one here 
called "The $25 Network", basically run through serial (maybe 
parallel) ports.  There were others, some with low-cost cards, such 
as "The Invisible Network".  I wonder if a ring could be set up using 
the old DOS Interlink. 
Many of those "cheap" PC networking schemes were found in the 8bit world.
The interlink like scheme I'd done for going between my S100 crates back 
in '81 to solve the problem is limted disk and more than one computer 
needing a disk.

CP/M or most other OSs for non 8080/z80 could easily be fooled into 
redirecting disk IO to a serial port. Though CP/M was modular enough
and most widespread it was most often hacked that way.  There is 
however no rule that says a OS must talk to a disk as storage and once 
that is clear then it's easy to cobble up a packet protocal that transmits
the needed data across a serial or parallel port to a willing and enabled 
host. Some systems like the big S100 crates or multibus running MPM used
the bus and some common memory so that multiple CPUs typically z80 with 
128k ram, rom and serial IO plus a bus interface and memory manangement 
for off board memeory.  MPM would be the server and CP/M would be the 
local cpus that users interacted with.  It wasn't seen often as it was
expensive to ahve the hardware and the average hobbiest at the time 
rarely had more than one fully functional system and maybe a SBC
of the KIM-1, EVK68, AIM65 or SDK85 level.

By '81 I had a NS* Horizon, Netronics explorer8085, an Altair, and two 
NEC PDA-80(sorta s100) for the 8080/8085/z80 realm alone. Only two had 
a disk controller but they all ran CP/M. When the NS* got a hard disk
since that was a $1K investment at the time sharing that resource with  
the deprived systems was important as back then I didn't have a swarm 
of floppy drives. So having read and seen networks in use I figured
sensing all that over a fast serial port was not unreasonable and 
after some thought and a few tries it worked.  These days people use
PCs for that but I personally would rather program z80 than 80x86.

...
 MS-DOS has had some flavor of networking
"hooks" for a very long 
time.  CD-ROM access is implemented as a networked device.  I've 
implemented a number of foreign filesystem drivers using networking 
where file naming conventions or oddball block sizes weren't amenable 
to normal DOS filesystem conventions. 
Thats PCs.  ;)

Allison