0

I've recently bought my desktop a PNY Geforce GTX 650 TI Boost so I could play Steam games on Ubuntu. I've also bought a brand-new 500 W power supply unit because my older one wouldn't be able to handle the extra work and didn't have a PCI-e plug.

After carefully installing the hardware, I proceded to install the drivers. In order to do so, I've followed Luis Alvarado's tips from here. Baiscally, it involved installing the nvidia-325 package from the xorg-edgers PPA then running sudo nvidia-xconfig to create a new xorg.conf.

The instalation seems to have gone without errors or even warnings, and my computer was finally able to run resource-intensive games. However, every now and then the computer freezes. To better describe the problem:

  1. The screen always freezes;
  2. The sound usually goes into a loop (rarely it keeps runnning, indicating that only the screen had freezed);
  3. There's absolutely no mouse or regular keyboard feedback. 90% of the time I have to Alt+PrtScrn+R+E+I+S+U+B my way through a restart, but sometimes not even that works;
  4. It seems to occur at random intervals, usually when I'm running Steam games, but sometimes when I'm on Unity, working on spreadsheets and statistics software. What intrigues me is that sometimes it occurs even before the system boots (at the first POST screen, which presents the Geforce card). This makes me wonder whether the problem isn't actually in the hardware itself (which would be a shame, because it costed me a small fortune and I can't return it anymore).

So far, none of these has solved the issue:

  1. Reinstalling the 325 driver;
  2. Installing older drivers;
  3. Installing nvidia-settings to see if there were any problems such as overheating (there isn't: the card runs at a comfy 45 Celsius even after extensive work);
  4. Installing steam-login in an attempt to bypass any issues the card might be having with my onboard Intel graphics.

Since the problem sometimes occur during POST, I guess configuring something on my My CMOS might help. The problem is I'm kind of lost. Here's some settings I think might help (from "Advanced Chipset Features"):

  1. Intel EIST: Enabled (GV3);
  2. Intel XD Bit: Enabled (when disabled, force the XD feature flag to always return 0)
  3. Intel VT: Enabled (when enabled, a VMM can utilize the additional HW Caps. provided by Intel Virtualization Tech)
  4. Memory Hole Remapping: Enabled (allow remapping of overlapped PCI memory above the total physical memory)
  5. Video Memory Size: 32MB (other options: 64MB and 128MB)
  6. DVMT Mode: DVMT (other option: fixed)
  7. DVMT/Fixed Memory Size: 256MB (other options: 128MB and Maximum)

My computer is plugged to a 24' LG TV through an HDMI cable, instead of a regular monitor. Could this be the source of the problem (or worsen it)? Unfortunately, I don't have a computer monitor here to test it. All I know is that even though I feel like I've already read all the "My computer is freezing" questions on Ask Ubuntu, I've run out of ideas on how to diagnose and fix the problem.

Community
  • 1
Waldir Leoncio
  • 1,895
  • 1
    This is mostly the case with chipsets, which contains CUDA (sometimes with Optimus) - for this case this solution here might help: http://askubuntu.com/questions/311151/how-can-i-install-nvidia-driver-gt-520-and-cuda-5-0-in-ubuntu13-04/311154#311154 – dschinn1001 Oct 04 '13 at 11:21
  • @dschinn1001, I haven't tried that (I thought CUDA and Optimus were for laptops or desktops with two cards -- onboard Intel + PCIe Nvidia doesn't count as a hybrid system, does it?). But I confess that answer to which you linked got me kinda confused, I was thinking about following this instead. Should that give me the same result? – Waldir Leoncio Oct 04 '13 at 12:02
  • 1
    (in case of CUDA and Optimus) some chipsets are hybrid of NVIDIA (so are some of ATI too). In case of CUDA it does not matter, if it is for notebooks/netbooks or for Desktops. You could try this driver of CUDA_SDK_1.1-beta - I have it installed too on my notebook and it is a big difference ... if you have CUDA try my link first. If this does not work out, then try the other link. Nothing can happen seriously. – dschinn1001 Oct 04 '13 at 16:59
  • 2
    I'm voting to close this question as off-topic because this is a hardware problem, unrelated to Ubuntu. – Fabby Jan 27 '16 at 21:44
  • @Fabby, fair enough, but maybe I should just delete it already. – Waldir Leoncio Jan 28 '16 at 06:12
  • @WaldirLeoncio: Meh, leave it in... It's a nice Q&A... It'll still be there but nebies will not be able to answer once it's closed... – Fabby Jan 28 '16 at 22:00

2 Answers2

1

As you stated. The crashes even come in while booting. You clearly have a hardware problem. You should fix those before trying anything on the software side.

I have never had trouble with HDMI or heard that this can cause crashes. You have already ruled out heating problem.

You should check if your graphics card and the power cables are all properly connected. Try switching one component after the other. Try a different old graphics card or, if available, the on-board graphics card.

  • Maybe you can test the Nvidia graphics-card and the PSU in a different computer?
  • Also, did you test the CPU temperature? Maybe your CPU heat-sink got a little lose and that is causing the unstable system?
  • Did one of your RAM components turn bad? If you have more than one, try to remove one and test again.
  • If all this still doesn't single out all the component that runs unstable you will finally have to test the motherboard, which will inevitably mean a lot a work.

Edit added the additional info from the comments.

MadMike
  • 4,244
  • 8
  • 28
  • 50
  • thanks for your input. I've already checked the power cables; my PSU has two PCI-es, but using either doesn't seem to make a difference. I'm honestly unwilling to try a different PSU, since that would require me to buy another one just for testing. I've though about trying some different BIOS setting, but I don't know where to start. How would I go about "switching one component after the other"? – Waldir Leoncio Oct 03 '13 at 13:35
  • What about testing with your old graphics card? Or maybe you have a on-board graphics card you can turn on and remove the current one. – MadMike Oct 04 '13 at 11:41
  • Actually, I never turned off my onboard graphics card (I don't even know how to do this). Could that be the problem? I just installed the Geforce and plugged my HDMI to it instead of using the HDMI port that came with the motherboard. Unfortunately, I don't have other graphics card. – Waldir Leoncio Oct 04 '13 at 11:44
  • You would need to access the BIOS to switch of your on board graphics card. But I don't think this causes the problems. Instead try to remove your new nvidia-card and report if it still crashes. There is also the possibility that your motherboard is damaged so ruling out the graphics card is important. – MadMike Oct 04 '13 at 11:50
  • Sounds like a plan. The only problem I see is that without the card I won't be able to run the games that tend to incite the crash, so I'll see what I can do. BTW, I've posted part of my CMOS configuration. I'll try to semi-blindly venture myself into tweaking those. – Waldir Leoncio Oct 04 '13 at 12:00
  • The only application I personally know to stress test ubuntu is handbrake. Google tells me to use stress – MadMike Oct 04 '13 at 12:09
  • Thank you once again, Mike. I'm noy familiar with handbrake, but I'll study the option and get back to you. – Waldir Leoncio Oct 04 '13 at 12:27
1

So after 1,5 years of testing it seems I've finally solved this problem. All it took was sheding some light into the issue. Literally.

This is what I was only able to see after putting some bright, direct light into my PCIe slot:

PCIe

I'm guessing this happened when I first inserted my video card into that slot. The card may have pushed a couple of pins upward, shorting the circuit and causing all those freezes.

The only tool I had to reach that narrow space and try to unjam those wires was a pocket knife. I was half desperate, half okay with the fact that I'd have to get another motherboard if I wanted to game in my machine, so I just went ahead and started shoving that knife in there and pushing those wires back and down, to see if I could put them back in their places. I do not recommend this technique.

My technique had everything to go wrong. However, it did work, and my system's been freeze-free for over a month.

Waldir Leoncio
  • 1,895