[Coco] Multi-Processor 6809 Computer System
Brian Blake
random.rodder at gmail.com
Wed May 1 13:11:13 EDT 2013
On 5/1/2013 11:14 AM, Gene Heskett wrote:
> There is a 3rd 400lb gorilla here too. Since the GIME is in charge of
> the memory access, interleaving video access with cpu access, the
> speeds of the memory would soon be found to be a limiting factor also.
> Probably before even the 63C09's nominally 4MHZ maximum clock has been
> reached. Any thoughts of a new GIME at 20MHZ and you are looking at
> replacing the memory with 20ns memory.
The new RAM upgrade from C9 appears to be 70ns... Not as quick as 20ns,
but, getting closer.
> Even if we did have the design decoded, and wanted to redo the GIME in
> cmos, to run at 20+ MHZ, the NRE (Non Returnable Engineering) costs at
> whatever fab house could do it, would likely be in the region of half
> a million. So with our limited numbers to spread that cost around,
> obviously that will never happen. The closest we will ever get
> practically would be some sort of an FPGA version with a base that fit
> the GIME socket, and which would need its own wall wart power supply.
An FPGA/Propeller/whatever on a board that could plug into the GIME
socket would seem to be the most practical solution. Not a perfect one,
but, if a GIME goes, so does the CoCo3 since there are no replacements
around these days.
> There are so many basic hardware limits to be overcome that I don't
> believe the idea is at a practical. The whole coco3 or 4, in one of
> the emulator boards like the DE-1 or similar 400k gate device, with an
> adapter that makes it fit the coco3 case, is a far more do-able
> project. But even that would be a labor of love due to the startup
> costs and restrictions placed by the suppliers of such devel kits.
And I would certainly agree with this. Gary's CoCo-X is a perfect
example of the above - along with unwanted drama...
More information about the Coco
mailing list