On Wed, Jan 15, 2020 at 2:12 AM Jon Steinhart <jon(a)fourwinds.com> wrote:
we tried to use LSI-11s
but those turned out to be useless because of the way that DEC did the DRAM
refresh; it wasn't interleaved, they just stopped everything every so many
ms and refreshed everything. Non-starter for real-time systems.
Be careful as to who you denigrate, my friend. 😂
Very interesting history, IMO. Yes, DEC sold the LSI-11, but Western
Digital designed it. DEC (KO specifically) had just put Ray Ball and Ken
O'humundro's CalData out of business for cloning the PDP-11/45 with a
Unibus on his Caldata 500
<http://www.bitsavers.org/pdf/calData/CalData_Brochures_1974.pdf>. At the
time, WD had developed and started to sell to the systems manufacturers a
new set of bit-slice chips the MCP-1600
<https://en.wikipedia.org/wiki/MCP-1600>, to compete with AMD's 2900 and
Intel 3000 series (plus they were already a chip supplier to DEC for UARTS).
So WD designs and builds a few LSI-11 as a sales demo of what you could
do with their new bit-slice chip (*i.e. *as those things often go, the
board, bus, and memory was a quick and cheap hack).
It's important to note that the way DEC nailed CalData was the *same
instruction set on the same bus*, WD did their own bus for their demo.
Also, please remember that at the time, WD was in the chip business. KO's
reaction this time was different. Instead of suing, DEC got the design and
started to build and sell them. WD took the board design, wrote a new set
of microcode based on the USCD Pascal-P machine
<https://en.wikipedia.org/wiki/UCSD_Pascal>, then sold that as a 'system'
called the Pascal MicroEngine
<https://en.wikipedia.org/wiki/Pascal_MicroEngine>, but primarily used it
is the sales demo.
I remember seeing one of the WD Pascal-P systems once when we were at Tek
(along with my favorite named workstation, the Modula based Lilith
<https://en.wikipedia.org/wiki/Lilith_(computer)>). But I do not think the
Pascal-P (nor Lilith) was very successful. Also, AMD ultimately won the
bit-slice chip business, as most designers at manufacturers like DEC,
Masscomp, FPS, *et al*. designed their new systems or at least the FP/AP
coprocessors with the 2900 series.
BTW: this is also why a few years later when Ken O'Humundro created another
full computer board with a 68000 on it with his new Able Computer Corp, he
put it on the QBUS which DEC could not lock up because they did not create
it as WD had.