Re: SMALL bug

Snow Cat (snowcat@netgate.net)
Sat, 26 Oct 1996 01:33:58 -0700 (PDT)

Colin Plumb once wrote:
>
> The problem is that programming the hardware is not atomic (it takes
> several steps) and if the video hardware starts to display a screen
> while the programming is halfway complete, it will take half of the old
> pointer, half of the new pointer, and put them together to make
> something bizarre which it will then display for 1/60 of a second.
>

As I remember, under MS-DOG the standard practice is to wait until
just after vertical retrace. Then you have the whole frame to update
the register. I guess this would be unacceptable under Linux, because
most video cards don't generate a vertical interrupt and polling would
steal CPU from more interesting tasks.

>From the other hand, using SVGATextMode to set 100MHz refresh rate will
make flicker much less likely :)

>
> Unfortunately, there is no easy way to tell when it's safe to
> reprogram the video card registers, so the penalty for very fast
> screen update is occasional flickers of glitch. I think it's
> a worthwhile tradeoff, myself.

-- 
 _.    	        _       .  
(_ ,_  _ ,  .  / ` _ _L	 | Email: Oleg Kibirev <snowcat@netgate.net>
._)| U(_)\/\/  \_,(_L/L  | Visit http://math.math.CSUFresno.EDU/~oleg/math.html
------------------------'  to get my programs + PGP public key