Debian Multimedia :: Proprietary Nvidia Driver - High CPU Usage Xorg
Aug 7, 2015
I recently switched from jessie to stretch on both i encountered a jerky ui in gnome 3. after a longer research i noticed xorg consumes over 80% of my first cpu-core while moving a window. for testing purposes i switched back to the nouveau driver and there is no high cpu usage while moving a window, also the ui is smooth.
i tried 3 different nvidia driver: the one from the debian repository, the current stable and beta driver from the nvidia site. i noticed on all 3 the same problem.
I have a problem that pops up with some games, sometimes: sauerbraten, lugaru, and nexuiz being the ones that pop to mind.
The problem is that when the game starts/loads the mouse "cursor" in the game will not work... the thing is frozen. The "fix" is to jump to a virtual terminal, via alt-ctrl-1, and restart KDM, then I log back into the session and everything is working swell.
This problem does not occur in Osmos, World of Goo, Warzone 2100, or the Linux Ryzom client.
I thought about adding an explicit /etc/init.d/kdm restart in my /etc/kde4/kdm/Xreset file, but that seems too draconian.
This has been a recurring problem on several machines with several 7000, 8000, and 9000 series Nvidia cards running under the proprietary driver, on both 32-bit and 64-bit AMD processors ever since Lenny and up through Wheezy. And it occurs on the following desktop/windowing environments: KDE4.4, icewm, fluxbox, blackbox, E17.
I would guess that it's a driver issue or a driver+xorg configuration issue.
PS: Please don't suggest that I should use the open source driver.
I installed the NVIDIS driver for 64b Linux, and now I face a high cpu utilization (up to 40%) in the XORG when I swith from one application to another. I need the driver because otherwise some web pages are rendered with a something like a watermark
LSPCI 01:00.0 VGA compatible controller: nVidia Corporation G98M [Quadro NVS 160M] (rev a1) (prog-if 00 [VGA controller]) Subsystem: Dell Unknown device 024f Flags: bus master, fast devsel, latency 0, IRQ 169 Memory at f5000000 (32-bit, non-prefetchable) [size=16M] Memory at e0000000 (64-bit, prefetchable) [size=256M] Memory at f2000000 (64-bit, non-prefetchable) [size=32M] I/O ports at df00 [size=128] [virtual] Expansion ROM at f4000000 [disabled] [size=128K] Capabilities: <access denied>
Installing Mythbuntu 10.10, which I finally got installed properly. At first I installed the open-source video drivers just to make sure the installation worked, then I installed the "version current" proprietary drivers using the graphics drivers manager...tool...thing. However, when I restarted the computer, it has a text-mode splash screen and I stay in the first virtual terminal.
If I try to go to the GUI "terminal" [Ctrl-Alt-F7], it appears to be partway through some kind of check:
I ran sudo dpkg-reconfigure xserver-xorg to try to get back to the open-source drivers, but it didn't give any text output and went straight to the next line of command prompt, when I restarted it did the exact same thing. Any tips for at least getting back to the open drivers? I'd like to not have to reinstall again (I'm dual-booting WinXP,). The card in question is a GeForce 6200 AGP.
With the upgrade to Squeeze and the 2.6.32-5-686 kernel, my Dell GX260's 82845 video chip suffers from the documented intel driver bug (freezes within a few minutes). I tried to fall back to the vesa driver, but am only able to get 640x480 resolution. The monitor is an LG L1720P.
I turned off kernel mode setting (via /etc/modprobe.d/i915-kms.conf), and modified xorg.conf -- added Modeline entries (from gtf) and additional Mode lines to try to force resolution. Relevant portions of xorg.conf, dmesg and Xorg.0.log are shown below.
Section "Device" Identifier "Generic Video Card" Driver "vesa"
I just upgraded KDE on my openSuSE 11.2 installation. I have never had any problems doing this in the past, but this time, when I rebooted, I noticed that after a few seconds of idle time, my CPU usage goes sky high. I ran top in a console and noticed the culprit was xorg. I am using an NVIDIA card on an AMD64 3200+ with 1 gig of memory. KDE version is currently 4.5.85. Like I said, I didn't have this problem until the last update. Any ideas as to what could be causing this?
I'm running succesfully Debian 6.0 after first trying Debian 5.0 and ran into missing partitions. This is solved by using Debian 6.0 (Beta 2).
Now it's NVidia's turn: Under Ubuntu (yes...i know it by now...) you had to install a proprietary driver for NVidia to ensure that 3D was supported. What about Debian? There's nothing like this under Debian? How do i know if 3D is supported?
I'm running Ubuntu 10.10 on an old small form factor PC with an AMD Sempron 2400, 1GB RAM, and an nVidia 5200 graphics card 128MB. OK, so a low-spec machine (but that's the great thing about Linux right? Don't need high-end h/w) but it works just fine, except that it can't play full HD (1920x1080) MPEG-4 video. Very jerky and lots of dropped frames. Same in both Movie Player and VLC.
I can't afford to increase the RAM and as it's a SFF I can't just swap the mobo and CPU for something faster so I'm wondering whether getting a higher spec graphics card would make any difference?I'm using the nVidia proprietary binary driver (latest version) and searching the forums I found a post where someone said that the nVidia driver needs at least a 512MB card for HD video.A colleague has a higher spec nVidia card (7600 IIRC) that he'll sell me, but before I spend any money, is this likely to improve things? How much does the graphics card affect performance, or is it simply a case of the machine overall just isn't high enough spec?
I have just started to have a problem with Xorg it is always using at least 30% of my CPU, and the whole system does not run smooth so if I play a video it does not run smooth, it judders, also even if I drag an icon it judders across the screen. Im running Ubuntu 10.10 2.6.35-25-generic x86_64 VGA compatible controller: nVidia Corporation G98M [GeForce G105M] (rev a2)
i updated my nvidia driver im running ultimate gamers edition 2.6 and now it only boots as far as terminal 1. i am a newbie and am completely lost. of course i have searched google high and low and seems to be a common problem however i have tried editing the xconf/xorg cnt remeber the name iv tried using the restore mode which is second on the boot list and have tried using the live cd to boot into graphics mode none seem to do the trick through using the live cd or something as the commands for the terminal are a pain for me also i would put up the xconf file up but i only no how to view it through the cd i dnt no how to copy it to usb to place on here
There is one thing missing (I think) a clear guide to clearing out Nvidia and replacing it with nouveau. For all but hardened gamers, nouveau on 11.4 delivers. It also removes one more barrier to what I think is the intended goad of Tumbleweed.The problem IMHO is not that there are no clear guides. The problem is there are too many. No sooner does one person do a guide (that is clear) and someone else who does not like some point writes another guide that they think is more clear (but in fact is less clear in other aspects). And this goes on ad infinitum.IMHO we have too many guides - many of which are sufficient clear ... but the VAST number only serves to confuse users more.
Having typed that, IMHO this is NOT a Tumbleweed specific issue, but its MUCH WIDER in scope and hence does not belong as a discussion in this Tumbleweed thread.
I just posted about this in this thread, but as the other thread was started by a KDE user then I thought I'd post here as well. I've had high CPU usage for a few months now - probably since trying the 0.9 branch of Compiz then dropping back to the default openSUSE builds (XOrg and gconfd-2 running a Core i5 at about 30% on every core*). I've now finally found a solution after deciding I wanted to fix it once and for all.
Once again, the Ubuntu forums come to the rescue with this thread (I don't like the distro as a whole, but I do find the forums useful!). I'm using Compiz, but it turns out that Metacity was running as well. A quick "killall -9 metacity" and the gconfd-2 process has vanished and XOrg settled down to its normal 1-2% (which is reasonable when I've got a Conky config refreshing every fraction of a second to repaint a sound visualiser!). Now I just need to find out why Metacity starts when I'm using Compiz...
* according to Conky's per-core graphs, although top only reported 15% overall and the Conky "top 3 procs by CPU" reported a measly 3% for each process, so someone's maths was out somewhere!
I did a clean install of Ubuntu 11.04 on my desktop which has a Nvidia GForce 7300LE card. Installation was successful, however, the moment I install Nvidia Current driver the system hangs. The only way I was able to get the system working was by doing a fresh install.
I have used the NVidia proprietary drivers for awhile. Yes, I know about nv and I even know about the prepackaged ones, but I've never minded getting the latest from NVidia, dropping out of X, and running the install which automatically rebuilds everything.
I recently took the synaptic update to 2.6.32-24. It worked fine and -- I guess -- migrated my driver. I didn't think about it. For no particular reason today I tried to build the latest NVIDIA driver (256.53 -- had been on an earlier 256 series). The build failed with some conftest failures. Even trying to rebuild the working driver failed. Reverting to 2.6.32-23 allowed both to be built and they work. So something the NVIDIA installer is expecting headerwise must be different between 2.6.32-23 and -24.
I've been getting a little discouraged with my laptop and I've been finding a lot of machines with gt310m graphics. The driver page last I checked didn't list this as being supported by the proprietary driver, I was just wondering if there's anyone that has tried it, and what the results were.
I currently have an nvidia card (GeForce 8800 GTX) and use the proprietary driver since I game a lot on wine (games like mass effect 2, prince of persia 2008, and some more recent games). I was wondering if using an equivalent ATI card with the free driver would show the same performance as my current on, or if the ATI driver isn't THAT mature yet. Would I be able to play the latest games with it on wine, or am I better of with nvidia and the propietary driver.
(I definitely know nouveau doesn't stand up to it *yet*, i.e., Prince of Persia complains about lack of video features). (note I don't care about a nouveau vs radeon debate, nor for a nvidia vs ati debate, the question is ati+free vs nvidia+propietary).
I'm running debian 2.6.32-5-amd64 and have spent an interesting day trying to get nvidia-glx running...What appears to be happening, is that installing nvidia-glx uninstalls X.I've tried various installation methods (Synaptic, apt-get, module-assistant) and the final result seems to always be /usr/bin/X not found.Re-installing X blows away nvidia-glx.The symptoms look a lot like this post: viewtopic.php?f=7&t=45188Is it possible that this problem has returned?
Everything works great at the moment, no hardware problems or anything like that. But whenever I enable the proprietary nVidia driver 185, it causes the boot screen to not come up, Ubuntu stops recognising my sound-card and refuses to work and when I try to shut down or restart, it just takes me back to the log in screen. When I remove the driver, everything works okay again.
I've been trying to install the driver for my NVIDIA GeForce 6150SE nForce 430 in Slackware 13.1 with no success. I always get the same error report: the module you're trying to build does not match the kernel source or something like that. The result: unable to build module and the installation crashes.
I have tried to:set a custom kernel source path, install it with the slackbuilds driver and kernel, extract the contents and trying to compile it myself, find possibly conflicting drivers or modules, use different versions of the same driver (I've tried installing the versions 256.53, 256.44, 195.36.31 and 173.14.27)recompile the Linux kernel in an attempt to make sure that the tools used to build the kernel were the same used o build the module.
The only time I got a different error message was when I used the slackbuild packages. It built the 'nvidia.ko' module, but it didn't work. I got a version magic notice when booting and, when I tried to start x, a fatal error "no screens were found."
Just to be sure, I made a clean full install of Slackware (only added WICD to be able to download the drivers and ran slackpkg update and upgrade all) and tried again. It didn't work.
I have a CentOS 5 virtual machine (VMware Workstation 7) running under a Windows host, and need the workstation's NVidia graphic card (Quadro NVS 295) to work optimally for my data analysis tools.
When I try to install the Linux driver from NVidia's webpage, I get "You do not appear to have an NVIDIA GPU supported by the 190.53 NVIDIA Linux graphics driver installed in this system". I have found my workstation's graphic card in the list of supported graphic cards in the README.
I suspect this has to do with VMware's own graphics controller having taken over, because when I do "/sbin/lspci" I see: 00.0f.0 VGA compatible controller: VMware SVGA II Adapter
Does anyone out there know how I can let the NVidia driver get installed and take over (it looks like I need its newest version for my software to render properly)?
I've been trying to figure out the cause of system hangs with slack 13 for a a week or so. I get random total freezes with the keyboard LED's blinking at me and the only solution is to do a hard-reset. I was using a self-compiled 188.8.131.52 kernel on 32bit Slack with similar results, but thought it had more to do with several bad starts as I got used to Slack's way of doing things... eventually I decided I'd made enough bad starts with that installation and after reading a bit about _current fixing some stability issues added another installation to my machine using Slackware64 and upgrading to _current. It worked great, then I installed the NVIDIA proprietary driver. The system hangs are back.
Sometimes they happen quickly, sometimes they take hours, sometimes while I'm working, sometimes when the machine has been idle for hours. For a couple of days I've been going back and forth between the NVIDIA propreitary and the nv driver. I've run memtest86 for 4.5 hours and not a single error, I'm running this on a new HD, I even upgraded the case fan (the old one was starting to complain!).
I've also installed different versions of the driver 185.x.x, 190.42.x. I've used slackbuilds, I've used "sh NVIDA.xxx.RUN" I've said "yes" to everything the installer asked, I've said "no".
So far no hangs with the nv driver. I've been leaving one or the other running at all times and no hangs with the nv driver, but within hours with the proprietary driver.
All that is to say I'm confident the issue is with the NVIDIA proprietary driver, OR something in my kernel options that only shows up when I'm using the NVIDIA driver.
Is anyone else running this successfully?
SLACKWARE64 kernel 2.6.32 (same behavior on 32bit with 184.108.40.206) compiled kernel with "make oldconfig" then "make localmodconfig" BEFORE the NVIDIA driver was installed. ext3 filesystem (same behavior with ext4 on previous installation)
I'm having an issue where the proprietary nvidia driver seems to crash my system. the weird thing is that it only happens if i have already started an X session. here is what i mean. i don't use a display manager so i invoke X by running startx. now if i run it for the first time it starts no problem. but if i kill X with ctrl-alt-backspace and then try to run startx again it just hangs and the machine completely freezes up. this only happens with the proprietary nvidia driver. if i use the vesa driver it works fine. the open-source nvidia driver doesn't seem to support my card: GeForce GT220. i never had this problem with slackware 13 x86. i have tried using the 190.42 version from slackbuilds.org as well as the latest version from nvidia's site 190.53. aside from this problem it seems to work fine otherwise.
I've given it the old college try for a couple years, starting with openSUSE 11.1 without success. I'm up to 11.4 now with no change or relief. My openSUSE box with NVIDIA proprietary driver and the default refresh settings of 80KHz/75Hz, has an annoying beat frequency with... something, somewhere, causing an annoying ghostly flicker on my trusty 1280x1024 LCD display. I can run both openSUSE and Windows XP on this hardware and they both have the same annoying flicker at those settings. However, in Windows XP, all I have to do is select a 70Hz refresh, resulting in settings of 74.6KHz/70 Hz, and the annoying flicker is cured... for Windows only, of course. I have tried to change these settings in openSUSE to no affect. In 11.4, I find that the advice is to create modelines using CVT and edit xorg.conf, but despite rigid adherence to instructions, there's no change. The monitor continues to see refresh settings of 80KHz/75Hz and the annoying flicker persists.
I hate to rag on openSUSE since it does so many things well, but there's a number of adjustments I'd like to make, especially the vertical refresh, that simply won't change, even when following documented or testamented procedures. Concentrating on the vertical refresh for now, is there anything that really works?
I'm new to linux, and i have installed linux slackware 64bit..after a complete setup i downloaded the latest Nvidia proprietary drivers, the binary package from nvidia.com..i have a geforce gts250..it's the first time i encounter this issue..i have already installed the driver with my old monitor (an lg flatron with max 1680x1050)and it always worked fine..with this new monitor (lg w2243s with a res of 1920x1080) it seems that every bin package from nvidia don't recognize the monitor...after installing i find a res of 640x480 and i get stucked, i tried to force it by editing the xorg.conf file..but nothing changes..how can i get the max res with nvidia bin package?