View previous topic :: View next topic |
Author |
Message |
Joseph_sys Advocate
Joined: 08 Jun 2004 Posts: 2712 Location: Edmonton, AB
|
Posted: Sun Feb 24, 2013 7:59 pm Post subject: |
|
|
Gusar wrote: | Joseph_sys wrote: | Yes, rebooting is simple. But my question is how system knows which module to load, how the selection process goes? |
It's all in this thread - through blacklists and xorg.conf |
SOLVED.
I got it, finally :-/ maybe I was trying take too much control over the process (my mistake).
The procedure actually is very simple.
CORRECT PROCES TO FOLLOW to switch between nvidia and nouveau.
(assuming the kernel is using "nvidia") to switch to nouveau:
Code: | mv /etc/modprobe.d/blacklist.conf_nvidia /etc/modprobe.d/blacklist.conf
eselect opengl set xorg-x11
mv /etc/X11/xorg.conf_nouveau /etc/X11/xorg.conf
reboot |
Assuming: Code: | cat blacklist.conf_nouveau
blacklist nouveau
cat blacklist.conf_nvidia
blacklist nvidia
cat xorg.conf_nouveau
# uncomment when the card gets IN
Section "Device"
Identifier "nouveau"
Driver "nouveau"
EndSection
at xorg.conf_nvidia
Section "Device"
Identifier "Nvidia card"
Driver "nvidia"
EndSection |
Interesting fact about the two drivers when it comes to frame-rate as tested with "glxgears" The difference is big.
glxgears - noveau Code: | 4780 frames in 5.0 seconds = 955.883 FPS
7108 frames in 5.0 seconds = 1421.445 FPS
6846 frames in 5.0 seconds = 1369.169 FPS
5874 frames in 5.0 seconds = 1174.488 FPS
5095 frames in 5.0 seconds = 1018.948 FPS
7378 frames in 5.0 seconds = 1475.425 FPS
7177 frames in 5.0 seconds = 1435.272 FPS
7599 frames in 5.0 seconds = 1519.792 FPS
7364 frames in 5.0 seconds = 1472.701 FPS
7473 frames in 5.0 seconds = 1494.570 FPS
7057 frames in 5.0 seconds = 1411.307 FPS
5618 frames in 5.0 seconds = 1123.416 FPS
4612 frames in 5.0 seconds = 922.264 FPS
4643 frames in 5.0 seconds = 928.554 FPS
5882 frames in 5.0 seconds = 1176.399 FPS
5763 frames in 5.0 seconds = 1152.490 FPS
5834 frames in 5.0 seconds = 1166.588 FPS |
Playing video with nouveau is almost impossible, the picture display is barely moving, like a slow animation.
glxgears - nvidia Code: | 301 frames in 5.0 seconds = 60.196 FPS
300 frames in 5.0 seconds = 59.946 FPS
300 frames in 5.0 seconds = 59.947 FPS
300 frames in 5.0 seconds = 59.947 FPS
297 frames in 5.0 seconds = 59.348 FPS
299 frames in 5.0 seconds = 59.744 FPS |
With nvidia the movie frame rate is fast. So it seems the smaller the number the better. |
|
Back to top |
|
|
Gusar Advocate
Joined: 09 Apr 2005 Posts: 2665 Location: Slovenia
|
Posted: Sun Feb 24, 2013 8:04 pm Post subject: |
|
|
*facepalm*
glxgears is not a benchmark! In this particular case, inform yourself about vsync.
The video playback thing is weird. Nouveau supports textured video, so it should play video fine. |
|
Back to top |
|
|
depontius Advocate
Joined: 05 May 2004 Posts: 3509
|
Posted: Mon Feb 25, 2013 2:13 pm Post subject: |
|
|
Gusar wrote: | *facepalm*
glxgears is not a benchmark! In this particular case, inform yourself about vsync.
The video playback thing is weird. Nouveau supports textured video, so it should play video fine. |
And you know what the real problem is here... We have no simple OpenGL benchmark. I know OpenGL is a complex thing, a maze of turny, twisty passages, all different, and that an application suite is what's really needed. But that's really hogwash! We need a simple benchmark, with the understanding that it's just a simple benchmark. I keep my Gentoo systems up-to-date, and probably the single most variable performance aspect of those systems is video. So when I build a new kernel or see an update from mesa or x11-drivers, I'm generally curious about what it did to my video performance.
We're after an OpenGL buglight here... No real change... It got a little better... It degraded a little... It's an absolute disaster... It's wonderful... We're not after detailed performance metrics - if so we'd emerge the Phoronix benchmark suite and run that.
Lacking a buglight, people run glxgears, and then other people say, "glxgears is not a benchmark." Maybe I should emerge the Phoronix benchmarks and see how bad they are tor run, just for sake of argument. _________________ .sigs waste space and bandwidth |
|
Back to top |
|
|
Gusar Advocate
Joined: 09 Apr 2005 Posts: 2665 Location: Slovenia
|
Posted: Mon Feb 25, 2013 2:54 pm Post subject: |
|
|
The problem is, even for such a buglight you'd need something with not-trivial geometry, high-res textures with normal maps and parallax maps and other maps, plus high-poly models, and some sophisticated visual effects. Nothing else will properly stress the shader units. You could make something custom that wouldn't come in a huge package, but on the other hand, why not just download one of the Unigine demos? Unigine Sanctuary is only 28MB.
Or, there's always .kkrieger . But I don't think it stresses today's GPUs much. |
|
Back to top |
|
|
|
|
You cannot post new topics in this forum You cannot reply to topics in this forum You cannot edit your posts in this forum You cannot delete your posts in this forum You cannot vote in polls in this forum
|
|