MULTIPROCESSING FOR THE IMPOVERISHED Part 1: a 6809 Uniprocessor

[Norman Jaffe]

I built a dual-6809 in the late '70s - it was a brand-new, exciting part - and we used the E part for exactly that reason. The system used memory that had an access time that was better than the 4x clock, so that each processor could run at full speed. 

From: "cctalk" <cctalk at classiccmp.org> 
To: "Liam Proven" <lproven at gmail.com>, "cctalk" <cctalk at classiccmp.org> 
Sent: Tuesday, August 6, 2019 12:19:55 AM 
Subject: Re: MULTIPROCESSING FOR THE IMPOVERISHED Part 1: a 6809 Uniprocessor 

On Fri, Aug 2, 2019 at 11:15 AM Liam Proven via cctalk < 
cctalk at classiccmp.org> wrote: 

> 1993 article on building a multiprocessor 6809 box. 
> http://www.bradrodriguez.com/papers/6809cpu.htm 
> 

I disagree very much with the author's advice to use the MC6809 rather than 
the MC6809E. With the E version you have to supply a quadrature clock, but 
all that's required to generate that is a single-phase 4x clock (which you 
need with either the E or non-E part) and a single 74HCT74. If you feed 
your single-phase 4x clock into multiple 6809 (non-E) parts, their E clock 
phases won't match, but will be off by arbitrary multiples of 1/4 cycle, 
which makes the shared memory design unnecessarily difficult.