Ubuntu 9.04, AMD and nVidia: How to determine cause of lockup?

I have a dual-core AMD with a GeForce 9500 card, and have been experiencing random lockups.

I haven't found a way to reproduce them, but the only thing that works is to power off - the machine is completely dead. I've talked to others who have the same problem - it seems to be related to the combination of a dual-core AMD Motherboard and an nVidia video card. I had a similar problem with the on-board graphics (nVidia 7025) except that the screen would corrupt.

I would really like to determine the cause of this. Is there any way to turn on additional logging that might help determine the cause of this?

Update:

I installed Fedora 11, and it's doing the same thing: random freezes. I guess the next step will be to try Win7 to see if it has the same issues.

Solution 1:

Sounds to me like you are experiencing the results of a bad piece of hardware. (since you claim the problem exists even in different OSs).

I have an AMD Dual-Core, Ubuntu 9.04 Desktop with a NVidia Gefore 8800 running Compiz and dual dvi monitors and have never, ever had a lock up like you are describing. This machine runs constantly.

Apparently I am not using the latest NVIDIA drivers. I am running version 180.44.

Edited to add the following suggestion as to how to go about isolating this failing hardware:

If this were my PC I would begin eliminating variables. Switch the video card out with any other one you have laying around. Does the problem go away? If yes, you found the culprit. If the computer continues to lockup in the same way with a totally different video card then you need to isolate a different piece of hardware. Keep in mind a video card may fail only when stressed which would indicate possibly bad cooling or even it being under powered. RAM is also something that can begin to fail over time and cause crashes.

FYI: A lot of computers will stress the CPU 100% when left in the BIOS screen. So as a CPU heat/stability test you could idle in the BIOS' health screen (the one that shows CPU temperature) and see if the computer gets overly hot or crashes there. If it does then I would suspect the CPU and/or its cooling. Are you overclocking anything?

Solution 2:

I've got the same sort of setup with Ubuntu 9.04 on a quad core phenom with the onboard graphics. I found that the system performs 99% better when i disabled compiz. I know its a bummer but it has something to do with nVIdea's drivers for ubuntu. I still get freezes when playing back wmv files every once in a while, but it is till better.