I have been lurking on this list for many months, watching the developments as AMD/ATI's documentation gets put to good use. I made an unfortunate decision to only watch the 'radeonhd' mailing list (instead of watching 'radeon' as well) -- because 'radeonhd' was running better on RS690 at the beginning of the year than 'radeon' -- but now that KMS has arrived and is stabilizing, I will now have to play catch-up in learning the status of 'radeon'. The GeForce 7950GT card on my desktop machine recently died, so I replaced it with a Radeon HD 4850 so that I could use open source drivers. (The RS690 chipsets are on 2 "server" machines I built, but I didn't really want to experiment with drivers on those... in case I did something stupid to the filesystem by accident.) I run Debian unstable on my desktop, and didn't want to mess up my package management by running 'make install' and getting a bunch of untracked files on my system, so I decided to borrow the efforts of the Debian X Strike Force (their 'debian' directory from their source package of radeonhd 1.2.5) and use it to build the 1.3.0 tarball from ftp://ftp.freedesktop.org/pub/individual/driver/ That built, installed, and runs just fine. Many thanks to the radeonhd team! (And the various users who have contributed bug reports, fixes, comments, etc.) I found that the build dependencies specified by the Debian team (for 1.2.5) were sufficient to allow my build of 1.3.0 to succeed, and I ended up with nice DEB files which allow me to easily add/remove different versions of radeonhd without leaving untracked cruft in my filesystem. Of course, Debian's radeonhd 1.2.5 package only worked using software rasterization on RV770; for my first attempt to build 1.3.0, I went with defaults just to make sure I had the correct dependencies installed. This resulted in a build which does not support DRI or 2-D acceleration using EXA. Now that have a successful start from which to continue, I am moving on to the next step: enabling some acceleration. I realize that I have to use an experimental git branch of DRM in order to be able to have any 2-D or 3-D acceleration on RV770. However, I decided to add "--enable-exa" and "--enable-dri" to my ./configure options when building, just to see what would happen. (The build process doesn't know whether radeonhd can support acceleration on my hardware, right?) This is (part of) what I got: [...] checking for DRI... yes checking GL/gl.h usability... no checking GL/gl.h presence... no checking for GL/gl.h... no checking whether to enable DRI support... no [...] NOTE: DRI support is disabled [...] Does this mean I have some missing dependencies that I need to correct before I move on to trying the experimental DRM version? Or is this entirely explainable by the fact that I was using a non-experimental version of DRM (Debian unstable has version 2.6.14)? [I had presumed (wrongly) that the necessary GL files would be in 'glproto-1.4.10' (Debian calls this x11proto-gl-dev).] If I meet with some success on this next step, I'll move on to kernel mode setting with a kernel 2.6.32 release candidate. Words cannot express how glad I am for this work being done by the radeonhd, radeon, and X.org teams! I'm so glad I don't have to mess with proprietary drivers like 'nvidia' or 'fglrx' any longer! Thanks, Dave W. -- To unsubscribe, e-mail: radeonhd+unsubscribe@opensuse.org For additional commands, e-mail: radeonhd+help@opensuse.org