nvidia, 3d games cause reboot

Asked by Oliver Marks

Could some one give me help diagnosing a problem, i have an asus x59gl which has a C79 [GeForce 9200M G]
 as reported by lshw.

The problem is if i launch any 3d game the system powers off, i think its an overheating problem it works perfectly the rest of the time but launching guild wars hive rise x3 or similar intensive 3d games causes an instant power of.

the nvidia driver i am using is 185.18.36, and the linux kernel is 2.6.31-9-generic.

I am using karmic (i know its in development but i needed the newer kernel, and want to make sure everything works on my laptop.) where can i look to find out if this is a kernel or an nvidia bug. i want to find the cause so i can report it and hopefully see it fixed for the final release.

also this problem exists on other kernel versions and any kernel before 2.6.3 does not work because of acpi issues with the laptop so this could also be acpi related.

Any help you can give to locate (help fix) the cause of this problem would be appreciated.

If you need log files just tell me which ones and i will attach.

Question information

Language:
English Edit question
Status:
Expired
For:
Ubuntu Edit question
Assignee:
No assignee Edit question
Last query:
Last reply:

This question was reopened

Revision history for this message
actionparsnip (andrew-woodhead666) said :
#1

Try the 180 driver or the 190 driver (also beta)

remove the 185 driver with:

sudo apt-get --purge remove nvidia-glx-185; sudo apt-get --purge autoremove; rm ~/.nvidia-settings-rc

Then run:

gksudo gedit /etc/X11/xorg.conf

an change:

Driver "nvidia"
to
Driver "nv"

You can then reboot to the open nvidia driver. You can then run:

sudo apt-get install nvidia-glx-180; gksudo gedit /etc/X11/xorg.conf

and change

Driver "nv"
to
Driver "nvidia"

Then reboot

Try your game, if its no good then remove the 180 driver using the steps above then run:

gksudo gedit /etc/apt/sources.list

and add these lines to the bottom of the file:

deb http://ppa.launchpad.net/nvidia-vdpau/ppa/ubuntu karmic main
deb-src http://ppa.launchpad.net/nvidia-vdpau/ppa/ubuntu karmic main

then save the new file. Close gedit.

run:

gedit ~/key

and put this text in it:

-----BEGIN PGP PUBLIC KEY BLOCK-----
Version: SKS 1.0.10

mI0ESk1KfgEEAN5z1vkd448Af2J0BBQEsoqbIO7u3ofItEFiItH0LxDZatZkDQFUP76BjLtR
VaMkVStZMdgjEe4inXvJFxTlu375sUhOUXsLmzX84AKlMaVOC/d8zWftPG2bYKF0AwBQFdmk
eMIL7S1r59FIQDUoOGNSGiNl7q6ZW81nRMwg7bG/ABEBAAG0H0xhdW5jaHBhZCBOdmlkaWEg
VmRwYXUgVGVhbSBQUEGItgQTAQIAIAUCSk1KfgIbAwYLCQgHAwIEFQIIAwQWAgMBAh4BAheA
AAoJEB2r27TOwGdn+m4EAJ3oVdn7zPgaonFuZmfXuhrLUlSAdx5Iug/Snp3hjeIxAr2WeXzC
7LZ8xB+saDyIFIbTOjDZQI0qruUVx1h2muHY3NS0WoYrEJJ9PPF9Qx3JQlOnEmMjSL9lFOPX
/Vq3ULy4uZuMoZDr0Y0I0s4rKMmSmP6mswJL7pB+4/zNiv7D
=1407
-----END PGP PUBLIC KEY BLOCK---

including the ---- lines, save the file and close gedit

run:

sudo apt-key add ~/key; rm ~/key

Then

sudo apt-get udate; sudo apt-get install nvidia-glx-190

Then re-edit the xorg.conf file back from "nv" to "nvidia"

then save the file and reboot.

Done

Revision history for this message
Oliver Marks (oly) said :
#2

that was a very well written suggestion but alas no joy most i got was about a minute from a game, so still leaning towards power issues acpi or something.

By the way my girlfreind has same laptop and same problem, if i had to hazard at a guess i would say that the fan needs to spin up to cool the gpu and perhaps its only adjusting speed for cpu only but this is just guessing.

Any more suggestions or logs to check to find the cause.

Revision history for this message
actionparsnip (andrew-woodhead666) said :
#3

Try disabling acpi as you say, costs nothing to try:

https://help.ubuntu.com/community/BootOptions

You could also get a monitor for the GPU temperature (of this I do not know how to set up) and see whats going on. You could also open the system to watch the fams.

Revision history for this message
Oliver Marks (oly) said :
#4

okay the temporature went upto 75c and had a cut of at 95c which it never got close to.
I tried disabling acpi and also apm as i though power changes may cause the problem.

I can not really open up the machine because its a laptop so far to fiddly.

thanks for the info so far but any more ideas for tracking this down, there must be a way to capture whats going wrong from some where.

Revision history for this message
Douglas Moyes (aragorn-stellimare) said :
#5

Install the native nVida drivers from nVidia, before installing, be sure do:

sudo rm /lib/modules/*/*/nvidia*.ko
sudo rm /lib/modules/*/*/*/nvidia*.ko
sudo rm /lib/modules/*/*/*/*/nvidia*.ko

That should get rid of any existing nvidia kernel drivers.
next, issue (this will kill X Window and any X11 apps you have running):
sudo /etc/init.d/gdm stop

(or issue "sudo init 1")

log back into the system form the text-only login screen. You'll be greeted with a shell.
Find where you downloaded the factory nVidia drivers,and then run the .run file:
sh NVIDIA-Linux*-pkg2.run

You'll need to repeat this process again when you update/upgrade the kernel package... and on a few occasions, you'll need to edit /etc/X11/xorg.conf and re-add your screen resolutions.

Yes it's annoying, but well worth the improvements. You'll also get a handy new utility:
/usr/bin/nvidia-settings

Always run it as root, otherwise you wont be able to save changes to the X11 file. If you have a dual-display system, you'll need nvidia-settings to properly configure your system.

Revision history for this message
Douglas Moyes (aragorn-stellimare) said :
#6

I should add, these are the drivers from nvidia.com, not from debian/ubuntu repositories (even the "non-free" ones). Get the driver DIRECTLY from nvidia.

The packagers felt inclined to re-order where the nvidia drivers install, so be prepared for update conflicts in the future, or simply don't ever update the kernel/x11 display drivers.

Revision history for this message
Oliver Marks (oly) said :
#7

i have just tried them but get exactly the same problem, just powers off. also nvidia-settings has been around for ages and its in synaptics to install thanks for the suggestion but no joy think i need to find out the cause seems whats going wrong should be logged some where.

Revision history for this message
Douglas Moyes (aragorn-stellimare) said :
#8

It is possible you have a faulty graphics card, especially if the system does a hard reset. I had a video card I purchased from newegg years ago, and it worked fine for a few weeks, and then started causing random reboots when doing anything that required the 3d accelerator (It also had static on the CRT). Try another nVidia graphics card in your computer and see if that solves your problem.

Revision history for this message
Oliver Marks (oly) said :
#9

its a laptop so cant try a different card, it also happens on another identical laptop so unlikely hardware related unless they both went at the same time. I dont get any static either.

Revision history for this message
actionparsnip (andrew-woodhead666) said :
#10

Could try a bios upgrade. risky but may fix it.

Revision history for this message
Oliver Marks (oly) said :
#11

lol, already tried that got one laptop on the very latest bios and one on a slightly older bios, i am still thinking kernel / nvidia / driver bug some where, by the way compiz works fine mainly because its not taxing the gpu.

Still would like a way to find out the cause even if i cant fix so i can correctly report as a bug if thats what it is.

Revision history for this message
Oliver Marks (oly) said :
#12

also extra note i am now on a slightly newer kernel 2.6.31-10 but still having same problems

Revision history for this message
Oliver Marks (oly) said :
#13

going to do some tests later running in windows just to rule out the hardware going on both machines.

Revision history for this message
Oliver Marks (oly) said :
#14

okay done some tests, 3d works perfectly in windows.
also note that the same reboot happens in jaunty with a slightly updated kernel to support the hardware properly.

So its definately a bug in linux i am leaning towards kernel bug at present or incompatibility between kernel and nvidia driver.

both these laptops used to work in linux with games so its definately a recent change to something.

any more suggestions ??

Revision history for this message
Oliver Marks (oly) said :
#15

I tried running cpu burn the other day on my laptop, to see if it was the cpu over heating, but could not reproduce a reboot from just overloading the cpu. it only happens when running intensive opengl games, compiz runs fine by the way.

So still stumped on where else to look

Revision history for this message
Launchpad Janitor (janitor) said :
#16

This question was expired because it remained in the 'Open' state without activity for the last 15 days.