<< Previous Message Main Index Next Message >>
<< Previous Message in Thread This Month Next Message in Thread >>
Date   : Sun, 24 Jul 2005 19:35:37 +0000
From   : Jules Richardson <julesrichardsonuk@...>
Subject: Re: ARM copros, speech cartridges, real time

On Sun, 2005-07-24 at 19:18 +0100, Jonathan Graham Harston wrote:
> > Message-ID: <4d8ee6b4d4info@...>
>  
> Sprow <info@...> wrote:
> > 2.ARM7 coprocessor board
> >  * 75MHz 32 bit ARM7 with 256k RAM, 256k ROM (expandable)
> >  * Runs BASIC V unmodified
> >  * Simple supervisor prompt skeleton Tube OS
> >  * PCB can be fitted into cheese wedge box, or Master internal sockets
> >  * Cheese wedge plastic cases not available
> >  * Probable cost about `70 each
> >  * Development time about 6 months
>  
> That would be interesting. Take this as a firm order for a Master
> internal version please. As a point of interest, would it use the
> Acorn Tube ULA or something different for the hardware interface?

Presumably the hardware side isn't that difficult, in that it's pretty
much the same for any CPU? (CPU, bus interface, bit of local memory
etc.)  All the effort's really in the software side of it, both the on-
board ROMs for the coprocessor and any application-level stuff kicked
off from the BBC side?

I'm not saying it's easy or anything :) Presumably at that CPU speed
there are all sorts of problems that creep in which wouldn't occur with
a slower CPU (or one more closely matched to the beeb's own)

I'd quite like to see a modular copro system which allowed more than one
expansion CPU. 16 6502 CPUs with a beeb front end doing raytracing in
digital RGB-o-vision or something ;)  (has any bbc-based system ever run
more than one external CPU in a commercial environment?)

cheers

Jules
<< Previous Message Main Index Next Message >>
<< Previous Message in Thread This Month Next Message in Thread >>