| 1 |
Dale wrote: |
| 2 |
>> root@smoker / # eselect set opengl nvidia |
| 3 |
>> !!! Error: Can't load module set |
| 4 |
>> exiting. |
| 5 |
>> root@smoker / # |
| 6 |
|
| 7 |
Bah, I just gave you the command in the wrong order. `eselect opengl set |
| 8 |
nvidia` was the right one. Need to tell it which module, before the |
| 9 |
action to use from that module. I see you figured that out below. |
| 10 |
|
| 11 |
> I assume this ain't good right. O_O OK, I logged out, stopped the GUI, |
| 12 |
> rmmoded nvidia and reemerged all the nvidia stuff. Reloaded nvidia and |
| 13 |
> ran the command. Now this is what I get: |
| 14 |
> |
| 15 |
>> root@smoker / # eselect opengl show |
| 16 |
>> nvidia |
| 17 |
>> root@smoker / # |
| 18 |
> So that works and then I get this: |
| 19 |
> |
| 20 |
>> root@smoker / # glxinfo |
| 21 |
>> name of display: :0.0 |
| 22 |
>> display: :0 screen: 0 |
| 23 |
>> direct rendering: Yes |
| 24 |
|
| 25 |
Sweet! |
| 26 |
|
| 27 |
> That looks good to, still no worky and my screens are really slow. |
| 28 |
> Sometimes even the mouse is slow. I also noticed that if I go to |
| 29 |
> configure desktop and try to play with the screen saver settings, it |
| 30 |
> hogs up a LOT of CPU time. Almost all of it yet it does not work at all. |
| 31 |
|
| 32 |
Odd. |
| 33 |
|
| 34 |
> Are you out of ideas yet? I'm not sure what to do. The emerge -e world |
| 35 |
> is still going though. 309 of 866 and counting. kdelibs went by a bit |
| 36 |
> ago. It's on swig right now. I dunno what it is. |
| 37 |
|
| 38 |
This could be a problem, particularly if you have preemption off or |
| 39 |
don't set PORTAGE_NICENESS. If you're running that in a screen session, |
| 40 |
try a ^Z (ctrl-z) to suspend it for a bit and see whether that helps, |
| 41 |
then `fg` to foreground it again. |
| 42 |
|
| 43 |
> Any clue? Am I doing something wrong? Is this new xorg that bad? |
| 44 |
> Should I unmask the newer xorg? I read the nvidia drivers are not ready |
| 45 |
> yet. |
| 46 |
|
| 47 |
At this point it's not real clear what the problem is. You could try |
| 48 |
running oprofile to see where all the time is being spent, then rebuild |
| 49 |
that package with debugging support (Add -g to CFLAGS, remove |
| 50 |
-fomit-frame-pointer) to get info on where in that program the time is |
| 51 |
spent. |
| 52 |
|
| 53 |
To get oprofile going, build the kernel module (it's included in the |
| 54 |
main kernel, just enable it -- CONFIG_PROFILING in Instrumentation), run |
| 55 |
`make vmlinux` in /usr/src/linux, emerge oprofile, then run this little |
| 56 |
script like this: |
| 57 |
./oprof someapp |
| 58 |
|
| 59 |
It will start profiling, and stop when you quit that app. Make sure you |
| 60 |
run it long enough so that the app's startup stuff isn't the primary |
| 61 |
thing you profile; you want to profile the runtime slowdowns. |
| 62 |
|
| 63 |
Thanks, |
| 64 |
Donnie |
Archives