> actually i never had problems with the x stuff (i did have to remove
> the sense == 0 code since my 17" monitor tickles that bug)

This code is #ifdef'd out since it doesn't actually work.

> >I'll post another patch sans the fast video stuff.
> 
> i am interested in getting the fast video stuff working.  you can
> never have enough performance :)

Well, it looks to me as though it needs to be turned on and off.
The bus is proably put into some special state with the bit on
thus the video doesn't show until after the block bit is set off.
Like I said earlier, it is also possible that you need a full
compliment of VRAM for it to work as well.

For this to be worth doing, the feature would have to be
incorporated into the X server's block copy functions or
the console's block copy functions. The question is, exactly
how much speedup would we get out of doing this?

Part of my new internal video interface is for the video drivers
to pass on ranges of memory that contain acceleration registers.
If the top level driver continues to pass them along to the
user then it should be possible to write accelerated X servers.