[Coco] A Technical Question

[Mark McDougall]

On 20/05/2012 5:10 AM, Fedor Steeman wrote:

> That sounds like an awesome idea! I hope this would become a reality some
> day. Given the CoCo3's capabilities, what would be the highest
> resolution/color depth you could push it up to through VGA? Perhaps
> 800x600x32 ???

In theory, there's no real limit to the resolution or colour depth if you're 
re-generating the screen in an FPGA. However, you quickly run into 
practicality problems of how much graphics data the 6809 can push around.

Besides, my aim wasn't really to produce a new Amiga, but rather enhance the 
current hardware with perhaps a few more colours and sprites etc. We still 
want 100% backwards compatibility, and still have the new modes 'look like' 
a Coco...

My thoughts were much along the same lines as the (Apple II) Carte Blanche, 
and have the cartridge "snoop" (video) memory accesses. It then re-generates 
the VGA video from an internal copy of the memory, whilst being able to 
overlay sprites, for example. The sprite controller would be mapped into 
Coco memory space somewhere suitably unobtrusive.

The main problem I can see with this approach is the fact that the VGA video 
generation is completely asynchronous to the GIME, even more-so for a PAL 
Coco. This precludes any "bare-metal" programming of the GIME that relies on 
horizontal raster interrupts, for example.

I'm really not sure how to solve this issue. One possibility would be to 
connect the RGB sync output to the FPGA on a separate cable, and derive the 
video timing from there. Another is to somehow (partly?) 'disable' the 
on-board GIME and have the FPGA implementation 'master' the system. Don't 
know, haven't thought it through so I'm thinking out loud here...

But, as I said, this is something that can be prototyped wholly within an 
FPGA! It would involve taking a complete Coco3 implementation, and adding 
the 'cartridge' logic wired via the bus (complete with 2nd GIME) and 
extended graphics functionality. Would be an interesting project!

One could even get real fancy and provide advanced functions to rotate the 
video memory mapping, or maybe provide software programmable resolutions. 
This would make porting of the Williams arcade games and Tutankham, for 
example, much much easier! ;) Actually there's a point; maybe a blitter 
could be added too!

Then again, this is also something you could do with an 'enhanced GIME' that 
simply replaced the original one. I really like the idea of using a stock 
Coco as far as possible.

Regards,

-- 
|              Mark McDougall                | "Electrical Engineers do it
|  <http://members.iinet.net.au/~msmcdoug>   |   with less resistance!"