Weekend RadeonHD Reading

[phoronix]

Phoronix: Weekend RadeonHD Reading

From Revenge to stacks being called workspace areas in AMD's documentation was discussed today on the RadeonHD IRC channel. If you're interested in the latest development information on the RadeonHD driver and open-source AMD, the IRC channel is certainly worth monitoring...

http://www.phoronix.com/vr.php?view=NjMxNg

[d2kx]

Exciting times Can't say it often enough.

Hopefully RadeonHD will be fully capable of the KDE4/KWin4 desktop effects by the time KDE v4.1 releases (but I think it should).

[chrisr]

I hope that's "R300 DRI with bug-fixes"

It's all very well using the R300 DRI as the starting point for R500 3D support, but I use the existing R300 driver a lot (World of Warcraft under Wine, with a Radeon 9550) and can assure everyone that it's hardly bug-free! Just try running the "rubik" xscreensaver OpenGL hack to see what I mean. Will the release of the tcode allow these problems to be addressed too?

[bridgman]

Yeah, the r300 3d support was largely reverse engineered, so it was a lot harder for the devs to make progress and some bugs required significant RE effort to even understand so they just didn't get addressed.

Alex feels that release of docs will make a big difference for earlier ASICs as well, since a lot of the code will be common across the generations. The shader opcodes change going to 5xx, and the hardware which *runs* those shaders is significantly different (Ultra-threaded dispatch and all that) but from a programming point of view we didn't change that much from 3xx to 5xx.

It also appears that *for the existing functions* the programmers view didn't change much for 6xx, but we need to get code running before we will be sure there are no "gotchas".