Single MOST important Question about Nv40 & R420...

MasterBaiter said:
digitalwanderer said:
"Derek Smart!
Derek Smart!
Derek Smart!"

Nooooo!!!! :oops:

You must not say "He who's name must not be spoken" ! :devilish:

Me wonders why he doesn't have his own little spot in the developers forum. :LOL:

Because we'd all get sued by the end of the week. Either for posting something he didn't like, or because we weren't all kissing his ass in his forum non-stop.
 
archie4oz said:
Yes I was referring to palleted textures... And yes they do work on FX hardware because I've used them via GL_EXT_palleted_texture and GL_EXT_shared_texture_pallete on a G5 with a 5200FX and 12" Powerbook with an 5200FX Go... Haven't tried it on a PC though...

Actually poking around Nvidia's developer site it looks like it's been dropped from the NV4x lineup... :(

Wow, that's weird. Considering Cass Everitt said this:

cass said:
Palletted textures will not be supported on GeForceFX. While the functionality obviously has uses, it consumed a disproportionate amount of real estate relative to the number of applications that made use of it.

in this OpenGL.org discussion topic: http://www.opengl.org/discussion_boards/cgi_directory/ultimatebb.cgi?ubb=get_topic;f=3;t=009234

Maybe Apple emulates them somehow? But wouldn't using dependant texture reads instead be expensive if you wanted to do filtering after the indirection? Anyway, this is all OT...

edit: Nevermind, after some further investigation it seems that the silicon for this feature was always there in the FX, they just disabled the extension on some consumer cards for certain driver versions for a while. Looking at the extension database at Delphi3d seems to verify this as there are lots FX cards with support for the extension with newer driver revisions.
 
Ingenu said:
Because people said two different things, I'd like to confirm that:
Z-Buffer uses Z
W-Buffer uses W = 1 / Z

W-Buffer is more evenly distributed, thus lowering Z-Fighting problems.

[edited for Xmas]
A W-buffer uses the W-component of the vertex coordinates in clip space. Which is not 1/Z.

D3D documentation on Depth Buffers said:
When a 3-D scene is rasterized with depth buffering enabled, each point on the rendering surface is tested. The values in the depth buffer can be a point's z-coordinate or its homogeneous w-coordinate?from the point's (x,y,z,w) location in projection space. A depth buffer that uses z values is often called a z-buffer, and one that uses w values is called a w-buffer.

W buffering is not mentioned in the OpenGL spec, but the Glide docs say the same (well, they say the W-buffer stores "q/w", but I didn't find out what q is...)

Whether the distribution is even or not depends on whether you use fixed point or floating point, too.
 
GameCat, there's also the possibility that the drivers emulate paletted textures by simply expanding them to RGBA. Defeats their purpose, of course, but at least makes them work.
 
Back
Top