You are not logged in.
I have weird performance issue in OpenGL. This is how glxgears looks normally:
$ vblank_mode=0 glxgears
ATTENTION: default value of option vblank_mode overridden by environment.
19515 frames in 5.0 seconds = 3902.839 FPS
19736 frames in 5.0 seconds = 3947.021 FPS
19843 frames in 5.0 seconds = 3968.372 FPS
19794 frames in 5.0 seconds = 3958.615 FPS
19706 frames in 5.0 seconds = 3941.096 FPS
19810 frames in 5.0 seconds = 3961.937 FPS
and this is how it looks when I pass along
LIBGL_DEBUG=verbose
:
$ LIBGL_DEBUG=verbose vblank_mode=0 glxgears
libGL: OpenDriver: trying /usr/lib/xorg/modules/dri/tls/radeonsi_dri.so
libGL: OpenDriver: trying /usr/lib/xorg/modules/dri/radeonsi_dri.so
libGL: Can't open configuration file /home/paladin/.drirc: No such file or directory.
ATTENTION: default value of option vblank_mode overridden by environment.
libGL: Can't open configuration file /home/paladin/.drirc: No such file or directory.
libGL: Using DRI2 for screen 0
22099 frames in 5.0 seconds = 4419.569 FPS
21492 frames in 5.0 seconds = 4298.308 FPS
21524 frames in 5.0 seconds = 4304.647 FPS
21520 frames in 5.0 seconds = 4303.923 FPS
22356 frames in 5.0 seconds = 4471.042 FPS
21727 frames in 5.0 seconds = 4345.382 FPS
22176 frames in 5.0 seconds = 4435.077 FPS
22268 frames in 5.0 seconds = 4453.411 FPS
Noticed anything weird? Yes, the version into which I pass env variables causing more output to happen has 10% HIGHER performance. Would anyone know why it is like that and how can I fix it?
Last edited by gray_-_wolf (2016-08-11 22:57:33)
Offline
glx gears is NOT in the slightest meant to be a benchmark. Ignore everything it says about fps and such. ALL it is is a test that your OpenGL sub-system is working, nothing more. For an actual test launch something that uses OpenGL like a game.
Offline
yeah, I know
glxgears -iacknowledgethatthistoolisnotabenchmark
but I'm not arguing the actual number, what I find weird is that debuging version has higher performance..
Offline
Just ignore it, use some OpenGL software that does real stuff for a real purpose as a benchmark instead of glxgears. That is your answer.
Offline
I tried Unigine benchmark and even there seems to be 1 or 2 frames higher, but that below reasonable detection, so good enough for me
Offline
Please remember to mark your thread as solved by editing the first post and prepending "[Solved]" to the title without the quotes.
Offline