Originally posted by daniell
View Post
Announcement
Collapse
No announcement yet.
AMD Catalyst 8.5 For Linux
Collapse
X
-
DKMS, for the record
"There are several fixes and a few general improvements with Catalyst 8.5, but one of the new features is support for DKMS in AMD's graphical installer. A few months ago, the Ubuntu packaging scripts for the fglrx driver had added DKMS, or Dynamic Kernel Module Support."
For the record, Mandriva has been handling fglrx (and all other modules shipped outside the kernel package) with DKMS since September 2003; nearly five years now.
This new driver version should be in Cooker and /backports for 2008 Spring fairly soon, I expect.
Comment
-
Originally posted by Kano View PostFine, but even I found serveral errors in dkms as soon as I began to use it Maybe all errors are in the Ubuntu part *g*
Comment
-
Originally posted by AdamW View Post"There are several fixes and a few general improvements with Catalyst 8.5, but one of the new features is support for DKMS in AMD's graphical installer. A few months ago, the Ubuntu packaging scripts for the fglrx driver had added DKMS, or Dynamic Kernel Module Support."
For the record, Mandriva has been handling fglrx (and all other modules shipped outside the kernel package) with DKMS since September 2003; nearly five years now.
This new driver version should be in Cooker and /backports for 2008 Spring fairly soon, I expect.Michael Larabel
https://www.michaellarabel.com/
Comment
-
Originally posted by Michael View PostJust to clarify, that statement was made in regards to the DKMS support for Ubuntu through running the driver's downloadable package with the --buildpkg Ubuntu argument.
Actually, I *think* that also has been the case for Mandriva for a while. Though I'm not 100% sure, don't quote me on that one...
Anyway, DKMS is a neat system when it works, so this should be a nice advance for Ubuntu users. Does Ubuntu have an easy way to ensure you have the necessary kernel headers for compilation installed?
Comment
-
What do you trigger on remove of a kernel? On U this is triggered:
dkms uninstall -m $name -v $vers -k $inst_kern -a $arch
But uninstall does not really clean up the binary - which is useless. But when you would trigger remove and it is the last instance then you remove the source too. Basically an option would be needed to keep source when you use remove.
Comment
-
Okay, I finally benchmarked my graphics card (ATI X1600 PRO 512 MB) under Linux and Windows with Nexuiz (tdemo demos/demo1.dem) on 1280x1024 with "High" effects:
Windows: 1910 Frames in 103.4 seconds, 18.5 fps. one second FPS min/avg/max: 7 21 54
Linux: 1910 Frames in 63.6 seconds, 22.8 fps. one second FPS min/avg/max: 10 25 41
This is really incredible, now I can finally say that my Linux box is faster than Windows
However, you see the max. FPS on linux were 14 lower than these on Windows, so there definitely is some room for improvements ;-)
Never though that I'd ever say that, but fglrx rules
Comment
Comment