100% CPU usage under CentOS 6.8/NVIDIA 367.xx/GTX 660 idle desktop/no composite
Hi, After I upgraded from CentOS 6.7 to 6.8 and updated NVIDIA drivers from 364.12 to 364.19 I now experience 100% CPU usage under a completely idle X.org graphical session with no applications running other than my DE. This continuous CPU usage doesn't happen instantly but it occurs after a minute of two of running my graphical DE. Please look into this problem ASAP. I've also filed a [url=https://bugzilla.redhat.com/show_bug.cgi?id=1340357]bug report[/url] at RedHat's bugzilla.
Hi,

After I upgraded from CentOS 6.7 to 6.8 and updated NVIDIA drivers from 364.12 to 364.19 I now experience 100% CPU usage under a completely idle X.org graphical session with no applications running other than my DE.

This continuous CPU usage doesn't happen instantly but it occurs after a minute of two of running my graphical DE.

Please look into this problem ASAP.

I've also filed a bug report at RedHat's bugzilla.
Attachments

nvidia-bug-report.7z

Artem S. Tashkinov
Linux and Open Source advocate

#1
Posted 05/27/2016 03:53 PM   
364.19 is not stable. If you haven't seen it yet, you'll get screen pixelation/tearing later on. Not sure what's going on with that driver, but it's been deprecated for 367.18. https://developer.nvidia.com/vulkan-driver If you can, I'd stick with 361.45 for now. And remember, the bird is the word.
364.19 is not stable. If you haven't seen it yet, you'll get screen pixelation/tearing later on. Not sure what's going on with that driver, but it's been deprecated for 367.18.

https://developer.nvidia.com/vulkan-driver

If you can, I'd stick with 361.45 for now. And remember, the bird is the word.

#2
Posted 05/28/2016 04:00 AM   
This is definitely a problem/regression with NVIDIA drivers, because I'm running 367.18 beta drivers now and everything is OK. So, 364.12 work fine 364.19 bug 367.18 work fine
This is definitely a problem/regression with NVIDIA drivers, because I'm running 367.18 beta drivers now and everything is OK.

So,

364.12 work fine
364.19 bug
367.18 work fine

Artem S. Tashkinov
Linux and Open Source advocate

#3
Posted 05/28/2016 08:11 AM   
Yeah, it seems a little like 367 should've been released as another 364 revision because it seems to mostly contain bugfixes for the last 364 release rather than new features, but in any case you should go to 367.
Yeah, it seems a little like 367 should've been released as another 364 revision because it seems to mostly contain bugfixes for the last 364 release rather than new features, but in any case you should go to 367.

#4
Posted 05/28/2016 05:51 PM   
I'd keep an eye on 367.18. I've been seeing reports of monitors not being detected, sli not working, and general brokedness(that's not a word). But like that girl you think is hot but really not, if it works for you, that's all that matters.
I'd keep an eye on 367.18. I've been seeing reports of monitors not being detected, sli not working, and general brokedness(that's not a word). But like that girl you think is hot but really not, if it works for you, that's all that matters.

#5
Posted 05/30/2016 05:44 AM   
I have managed to get this bug on 367.18 drivers but it has happened just once in the three days that I've been using these drivers. Seems some kind of a race issue.
I have managed to get this bug on 367.18 drivers but it has happened just once in the three days that I've been using these drivers.

Seems some kind of a race issue.

Artem S. Tashkinov
Linux and Open Source advocate

#6
Posted 06/02/2016 10:47 PM   
Could NVIDIA please investigate this issue? It's really bad. The latest stable drivers are affected as well.
Could NVIDIA please investigate this issue? It's really bad. The latest stable drivers are affected as well.

Artem S. Tashkinov
Linux and Open Source advocate

#7
Posted 07/18/2016 04:20 PM   
Scroll To Top

Add Reply