|
|
#1 | |
|
Registered User
Join Date: Sep 2006
Posts: 5
|
Hi there!
I have a strange problem: I had linux kernel 2.6.8-rc2-mm1 with nvidia driver 5336 and everything (2D + 3D) ran smooth! A week ago I upgraded to 2.6.17.7 and nvdriver 8774, but now UT2004 is very jerky, especially with small maps (I think it depends on the number of textures shown). It's a mess to play with this kind of slowness. I had this problem before when I tried any other kernel/driver combination than 2.6.8-rc2-mm1 + 5336. My System: - X7.0 - GeForce FX 5600 (Asus 9560VS) - P4 2.4GHz - 1GB RAM - chipset: SiS645DX Any hints? |
|
|
|
|
|
|
#2 | |
|
Registered User
Join Date: Mar 2006
Posts: 14
|
I get stalls of about two seconds at a time on the 2.6.17 series (...9 and ...11). When they happen I get a Xorg log entry along the lines of:
(WW) NVIDIA(0): WAIT (0, 6, 0x8000, 0x00000ee8, 0x00000ee8, 0) Unlike you, my system is essentially unusable because of it. I don't get as far as running up any games or anything. I haven't bothered making a proper post of it because I haven't got around to running the bug reporting tool. I forgot it existed the first time, and now I can't be bothered rebooting. I should mention, my problem is kernel version related, not driver version related. I had tried 2.6.17.9 with the old driver release, and it didn't work and so I went back to my 2.6.16 kernel, and then when the new drivers were released I had another go with the newer kernel and it still didn't work. I made one last effort with 2.6.17.11, and that failed and I got bored of it. Right now I'm running 8774 on 2.6.16.9 and things are working alright. You could try that kernel, or the latest 2.6.16 kernel maybe. Last edited by Gumboot; 09-06-06 at 04:26 AM. |
|
|
|
|
|
|
#3 |
|
NVIDIA Corporation
Join Date: Dec 2004
Posts: 8,763
|
XenoNGC,
I'd need to see an nvidia-bug-report.log while the problem is present, however considering the significant gap between the kernel & driver versions you're using, the problem could be caused by just about anything. It would be useful to test out different kernel & driver versions to isolate where the problem first occurs. Thanks, Lonni |
|
|
|
|
|
#4 | |
|
Registered User
Join Date: Sep 2006
Posts: 5
|
I tested some, but now I can only test new kernels with nptl support because of my libc.
All of the tested ones were not smooth, but I have the feeling that the last driver->kernel interaction change made it a bit smoother ![]() Attached is my bug report from the script. There my AGP is disabled because I use one X Server without AGP (else my machine freezes after some time, looks like a chipset problem?!). And If I want to play, I start another X Server with AGP enabled. I also have no nvidia-installer.log because I install it via the gentoo portage system. But if it is needed, I can upload the current one! Maybe you could give me a hint which versions I should try first that should work smooth? Regards, Martin |
|
|
|
|
|
|
#5 |
|
NVIDIA Corporation
Join Date: Dec 2004
Posts: 8,763
|
I have a few questions:
0) Is UT2004 the only thing that exhibits the performance problems? 1) If you run a single X server, does the performance problem persist? 2) I'd like to see what "ps -ef" output looks like while the problem is present 3) Does this reproduce if vmware isn't running? Thanks, Lonni |
|
|
|
|
|
#6 | |
|
Registered User
Join Date: Sep 2006
Posts: 5
|
0.: Yes, the only thing I could test. I tried glxgears and neverball, but they don't need much GPU, so they both run smooth.
1.: No, I tried only with my AGP X Server and it runs much better. Not perfect, but good to play with. 2.: Attached! 4 outputs with a 10sec delay while playing ut2k4. 3.: Oh, normally I don't load the vmware drivers, so it doesn't matter as it persists while they are unloaded. |
|
|
|
|
|
|
#7 |
|
NVIDIA Corporation
Join Date: Dec 2004
Posts: 8,763
|
If this doesn't reproduce with a single, AGP-enabled X server, then it sounds like you might be overloading your hardware. Running two simultaneous X servers is not a common use case, and doing it while stressing the hardware is likely to result in degraded interactivity.
You could also try testing with Viewperf from www.spec.org. Thanks, Lonni |
|
|
|
|
|
#8 |
|
Registered User
Join Date: Sep 2006
Posts: 5
|
But it worked perfectly with 5336 + 2.6.8-rc2-mm1 ...
I thought only the running (actual showing) X Server gets the driver and the other one only updates his bitmaps in local storage?! The problem is, it's my work PC, and I have many shells on my default X Server and can't close it. So I have to start another one to keep my work in the default X Server and play with the other one. Also I can't use my default X Server for playing, since it crashes with AGP after some time and without it's useless to play with. BTW: Viewperf is temporarily unavailable for Unix ![]() |
|
|
|
|
|
#9 |
|
Registered User
Join Date: Sep 2006
Posts: 5
|
Maybe there is a driver bug that doesn't release the card correctly on a terminal switch?!
|
|
|
|
![]() |
| Thread Tools | |
|
|
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Boost Your Performance Goals 10x ' This Week on inside* Publications | News | Latest Tech And Game Headlines | 0 | 05-13-12 05:20 PM |
| How to make the GPU stay in Performance Level 2 on battery? | pjman | NVIDIA Linux | 0 | 05-12-12 12:33 AM |
| NVIDIA Unleashes the GeForce GTX 670 Graphics Card ' Performance Perfected (WCCFTECH) | News | GeForce GTX 670 Reviews | 0 | 05-10-12 08:40 AM |
| My UT2003 Tweak Guide | DXnfiniteFX | Gaming Central | 48 | 10-30-02 11:59 PM |
| Low OpenGL Performance | Instinct | NVIDIA Linux | 10 | 08-08-02 03:56 PM |