Hiya Mike,

On Mon, Feb 04, 2002 at 10:49:32PM +0000, Mike Pumford wrote:
> Write back caches and a write buffer do make a rather nasty difference in this 
> case. 
> 
> It would probably make sense to to it actually in the routines that actually 
> put stuff on the screen. A cache flush on every n'th vsync is probably going 
> to be too much of a hit. Since my SA RiscPC does Firewall/Mail/News/NAT/http 
> Proxy service I'll probably notice unless you can do some sort of partial
> flush.

Euhm... i think you've misread my proposal .... i was proposing only a
(partial) cache flush, when possible offcource but _only_ if there was
printing done ... i.e. when nothing is printed on the screen no cache flush
or whatever will be done.... The purpose of the counter is to reduce the
number of updates that are nessisary. I.e. if a lot is printed the counter
won't reach zero so no flushing is done. If in an configured amount of
vsync's nothing is printed the cache is flushed only once until the timer 
reaches zero again after printing.

Cheers,
Reinoud