KMC11/DMC11 folllow up

[Dave Mitton]

Date: Tue, 1 Feb 2022 18:46:46 -0500
From: Bob Smith <bobsmithofd at gmail.com>
Subject: Re: KMC11/DMC11 folllow up
Message-ID:
	<CAHtNYbXBRwrOQm2cH+T+CDn5nq3sZN+PQRa9fEMVhqQ=pAMDSw at mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"

Yes! thank you! Nice, smart, and did good work despite the magateer, I
mean marketing dweeb.
I lost track of Harvey, Bob wen ont to start a company over in west
concord, can't recall the name, used some of his gear for projects in
the mid 80s.
I was designing a comms system in late 80s and heard Len Bosak had
started a company, had a couple of meetings with him, and went with hs
gear. Man those were fun post DEC projexcts.
bob smith (PDP8 engineering, DecComm (Stockebrand, VInce et al), small
systems, then of toe LCG for 2080...

On Tue, Feb 1, 2022 at 2:43 PM John Forecast <john at forecast.name> wrote:
>
> On Feb 1, 2022, at 8:20 AM, Bob Smith via cctalk <cctalk at classiccmp.org> wrote:
> >
> > KMC11 - Paul K cited the docs. It was a bit different from DMC CPU board
> > in both cycle time and in the use of ram versus prom.
> > Both boards/products used the 4bit Alu but I don't call that bit
> > slice, as the 2901 is more of a bit slice.
> > KMC and DMC are Harvard architecture based devices, as is the 11/60 CPU.
> > DMC and KMC benefited from the microcode work of Harvey Schlesinger,
> > Bob Rosenbaum, Richie Larry, and I think Clarise joined the team in
> > 77. Can't recall her last name.
>
> Patton? Harvey, Bob and Clarise joined the DECnet-RSX development team sometime in 77/78.
>
>   John.
>
> > DMC had (when I left the project and it had been shipping for a year
> > or two) a 300NS cycle time, while the KMC had a 240NS cycle time
> > thanks to the instruction register I had suggested to remi as we were
> > thinking of a RAM based device because PROMS were a royal pain with 2
> > and 3 code changes a day. This change allowed the machine to begin to
> > access the next instruction as one was executing - there are no
> > interrupts in either board.
> > bob
>
I’ve been watching this thread go by, and never finding time to contribute….

I started in Aug 1977 and finished the CommIOP products from Bob and Harvey.  Clarise was no longer involved.
It was basically done, I just did the QA on them and released it.  I did find some bugs in the debugger we provided.

I basically did KMC software support, on and off, going forward.  When the KMC-B came out (I think Remi did that)
I rev’ed the tools.    I also did a KMC Tools package for VMS.   I tossed into that package a VMS line printer driver, I wrote as an example and POC, that ran multiples LP11s at significantly better speed.   We ran that in our lab.  The MIT LCS lab loved that.

My memory is weak on things like the “DMX” and other things that CSS did.   The Lab Products group built some successful data collection products around the KMC.

Attempts in the Comms HW group to do an updated UNIBUS DH never got off the ground.   We (NAC) did pitch the idea of using the CommIOP-DZ to VMS, as a way to off-load character terminal loads, and they would have nothing to do with it.  For whatever reasons, they did not like the idea of smart devices.

That led to the attempt to build a “universal” terminal concentrator based on an 11 networked into the system.  That project was complicated by what it tried to integrate, and took too long to build.  It was overtaken by the simpler LAT based products that DEC went forward with.

George Conant, Bob Rosenbaum, and Pete Nesbeda left the company and founded Xyplex that fielded successful products in this space.

I could go on, but …   I do have copies of the KMC Tools doc and maybe the CommIOPs, but no KMC hw docs.
Dave.

Sent from Mail for Windows