Software Results Corp Unibus 68000 board?

From: Ethan Dicks <ethan_dicks_at_yahoo.com>
Date: Thu Feb 17 18:49:26 2000

--- CLASSICCMP_at_trailing-edge.com wrote:
> Hi folks,
>
> I just found a "Software Results Corporation" Unibus
> (hex-height) board with an array of 32 2114 memory chips and
> a big fat 68000 chip on it. There's a COM5025 (UART?) and two
> 40-pin headers on the edge.

Yup. Cool. Out of perverse curiosity, what's the S/N? I can eventually
look it up and tell you who used to own it. I might have even been the
guy that pulled the parts from inventory and tested the finished product.

> Date codes are from early 1984,
> and there's a surprising amount of 54LS (and no 74LS) logic on the
> board, leading me to think that this may have been intended for the
> military market.

Nope... the company got tired of snooping down bad chips. The boards were
a couple of grand, customer cost, so MIL-SPEC chips were not a horrible
cut into the profit margin and were, at the time, individually tested. It
saved enough on labor to be worth the expense.

> Is this, by any chance, a coprocessor type board, or is it a
> "master" CPU?

It is an intellegent synchronous serial card with onboard line-printer
capability. On a RSTS/E system, you could avoid spooling jobs through
the system line printer queue, saving many CPU cycles. We used to hang
LA-180's right off of the printer port or sell a printer adapter for
Dataproducts interfaced printers.

> Am I imagining things, or has "Software Results Corporation"
> been mentioned on this list recently?

Yes. I was hired there in 1984 and eventually bought the name and rights
to the software/hardware/customer list and provided service between 1993
and 1995 to all the customers who bought them in the 1980s (by 1990 sales
were effectively non-existant except for upgrades). Coincidentally, SRC
was the company that took out the full page back ad on "CPU Wars", now
on the web at http://www.e-pix.com/CPUWARS/cpuwars.html

I have all the software, firmware, schematics, wire-wrap prototypes,
*everything* for them. Unless you want to speak 3780 or HASP to some other
device from your Unibus PDP-11 or VAX, that board is useless. The dual 6309
PROMs are only smart enough to feed more complicated programs through a CSR
window into the 16K of SRAM. Even simple diagnostics are downloaded at
runtime,
character by character. In its protocol emulation mode, it uses 16-bit DMA to
feed buffers back and forth.

Unless I am seriously misinformed, this board was the first single-board
DMA device for the Unibus. Software Results pioneered the removal of the
NPR jumper to allow such things to work (prior to this, DMA devices were
one _backplane_ that broke grant between the input and output Unibus cables).
We had to ship dual-height grant cards to accomodate the absence of the board
when it was removed for diagnostic or repair purposes.

More trivia: this board uses SRAM because of two early DRAM problems. Intel
RAM was having higher than expected single bit errors (thought at first to
be cosmic rays, but later proven to be stray alpha particles emitted from
the minute quantities of radioactive materials in the ceramic packages).
Additionally, the first batch of 68K CPU chips (my old boss still has XC68000
S/N 424, yes, there is a serial number engraved on the lid) had microcode
glitches that caused longer than advertised bus cycles. For those designers
that borrowed/copied/migrated existing Z-80 refresh schemes, the outages were
long enough to interfere in some cases, causing catastrophic memory loss. SRC
went with 2114s for reliability.


 

=====
Infinet has been sold. The domain is going away in February.
Please send all replies to

  erd_at_iname.com
__________________________________________________
Do You Yahoo!?
Talk to your friends online with Yahoo! Messenger.
http://im.yahoo.com
Received on Thu Feb 17 2000 - 18:49:26 GMT

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:32:53 BST