I have an ATI radeon 32 Mb DDR card, which now works brilliantly (all it
needed was a new BIOS for my motherboard...). I am using AGP 4x and an
aperture size of 128 Mb. Currently the X server seems to allocate 1Mb for
vertex buffers and almost all the rest for textures. The only problem with
this is that the application I wish to run uses no textures at all but
does use an amazing number of vertices. The way it is designed there is a
compile-time value which indirectly controls the number of vertices
required - set it to 32 and it will use 32 display lists each with 32*32*4
calls to glNormal3f and glVertex3f, which I reckon means it needs about 3
Mb to store all the vertices. With this number it performs very well, I
get about 250 frames per second. However, I would like to set it to 64
instead of 32 because at the moment polygon edges can still be
distinguished in high resolutions. At the moment, when I set it to 64 it
is very very sluggish, and my box even starts swapping (which it really
shouldn't, with 256Mb RAM).
Is there a switch in XF86Config to change the buffer sizes? If so what is
it, and if not how hard will it be to add?
Thanks for producing a quality, stable driver for the radeon, it gives me
plenty of ammo against my nvidia loving friends.
echo bzidd@... | tr "bizndfohces" "pwgd9ociaku"