[Coco] OS-9 80-column driver is ready

[Ciaran Anscomb]

"Luis Antoniosi (CoCoDemus)" wrote:
> On Tue, Sep 24, 2013 at 10:53 AM, Ciaran Anscomb 
> <cocomalt at 6809.org.uk>wrote:
> 
> > Not for long, fast mode also inhibits RAM refresh!
> 
> Even with the newer SAM chips ?

Wouldn't think so - I mean check the data sheet, but to do that it would
need to alternate between clocking the CPU in those extra timing slots
and not clocking it, depending on HS.

> Because if I poke my coco2 (didn't try on my coco1) i have the crazy video
> but I can repoke to slow down again and still is working.

Yeah it'll last a short while unrefreshed.  This is why I reckon it's
almost certainly generally safe to be in fast mode outside of the active
area, and back to slow only during the active area - 192 active lines
is > the 128 addresses is strobes to refresh, so you definitely get a
full refresh 50 (60) times a second.  You can get away with much longer
than that, but unless the RAM access pattern of your code *happens*
to refresh everything, not indefinite.

I guess you could squeeze some performance out of it (if not consistently)
by having an IRQ service routine toggle the speed :)

..ciaran