On Mon, Jul 21, 2008 at 2:23 PM, Chris Bannister <[EMAIL PROTECTED]> wrote: > On Fri, Jul 18, 2008 at 07:34:34PM +0200, Mathieu Malaterre wrote: >> As a side node I could never get the module-assistant auto-install >> nvidia to work, even explicitely setting gcc to an older gcc 4.1 >> version, it would fail with an empty log. I had to run the shell >> script from nvidia. > > Mmmm, should work. Just used it yesterday. > > m-a prepare nvidia > m-a build -t nvidia (so you can see the screen output.) > m-a install nvidia > > There was one issue but I think I was mislead: > > xdriinfo reckoned the libGL was to old. And cause I was only sometimes > seeing the nvidia splash screen I thought there was a problem. > > I don't know if an ln -s libGL.so to the nvidia libGL did anything or > not, but I installed mesa-utils and glxgears[1] was showing about 1800 fps > but xdriinfo still reckoned the libGL was to old. glxinfo showed dri > was on and using nvidia; not mesa. > > Stellarium worked a treat.
Sorry my posts must have been interchanged. So I finally got the nvidia stuff installed properly using m-a but I had to use stuff from unstable 173.x version (from the top of my head). To get the libGL.so symlink I installed: nvidia-glx-dev > [1] Not a good benchmark though. At some point it has a very funny command line(*) option to keep reminding people about that, I never really look into why that was a bad benchmark. -- Mathieu Ps: something like -iacknowledgethisisnotabench -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
