Ben Harris wrote:

> On Sat, 28 Oct 2006, Mike Pumford wrote:

> I've fixed the problem in revision 1.30 of vidc20config.c by not trying to 
> pick the best FIFO preload and just picking a safe value.  This might 
> cause problems if you use a very high-bandwidth mode on a machine with 2MB 
> VRAM.
> 
Thats got it. Sensible output restored.

I can test the high bandwidth stuff as I have 2MB of VRAM and a monitor 
than can cope with anything the RiscPC can throw at it. > There's still 
a problem with 32bpp modes, but I don't think it's a new
> one.
> 
Agreed. At least not if the comments are to be believed.

Thanks for tracking it down.

Mike