Why does my PC crash every couple of hours?

pickarooney

A ridiculously vague question, but I can find absolutely nothing definite which contributes to these system freezes that can occur at any point during the day, while performing any activity.

Running an AMD64 3400+ with 1GB RAM and a GeForce 6200 Turbocache PIC-E graphics card, onboard sound, no idea of the motherboard brand. The only other added hardware is a PCI TV tuner card (Hauppauge WinTV Express) and an ethernet modem. I've been round the block several times trying all manner of different versions of the nVidia drivers, but there's no change whether I used older or newer ones (and I make sure to remove one set before testing another). The crashes just happen seemingly at random, the whole shebang just freezes up and I have to reboot via the power button. One weird thing is that every time I reboot after a crash, all photos and videos take on this washed-out, low-colour appearance, and I have to reset X to fix it.

/var/log/messages has yet to come up with anything helpful.

If it's an overheating problem, what's the best way to test this?

Any ideas otherwise?

Hagar

Run your system with the case open, taking all due precautions regarding electricity, and note whether the CPU fan is spinning properly or not. The bearings may be gone and it will cause the cpu to overheat, behave erratically and shutdown. There may also be another small fan on a chip on the motherboard, Northbridge I think it's called, if that has stopped working you will get problems too.

If you do have to replace the fan get a decent quality one and don't forget a liberal smear of thermal paste when fitting.

ethernet

I'm thing processor too -- over-heating.

You may also want to run a memory test to rule out faulty RAM.

Khannie

I found the IVTV drivers for my hauppauge card caused me the occasional hard crash in the old days. Try updating them (the latest are not always the most suitable. Have a read around a bit).

pickarooney

Would bad driver versions for a TV card have any effect if I never use the card (currently the case)? AFAIK I'm not using IVTV drivers, just basic BTTV ones.
I ran a memtest but it didn't make a whole lot of sense, just about a hundred lines of Unable to malloc 3698327552 bytes and then Allocated 3685744640 bytes...trying mlock...Killed

If there's overheating going on (in January) it might be because I have more hard drives than slots and so one is just hanging alongside another. Perhaps I should try insulating them from one another somehow.
I installed gkrellm and I'm getting constant temps of 51 degrees for the CPU and 37 for the hard-drives, with very little activity going on. Seems maybe a little high but hardly problematic.

ethernet

pickarooney wrote:

... I installed gkrellm and I'm getting constant temps of 51 degrees for the CPU and 37 for the hard-drives, with very little activity going on. Seems maybe a little high but hardly problematic.

I've an Intel P4. The manual says temperature should not exceed 40 degrees. 51 seems quite high for a processor. You may well end up replacing the thermal grease between the heatsink and the processor [or resort to water cooling ].

Regarding the X issue, did you install any drivers for your graphics card? I have the same GPU [PCI-E 16x]. It works very well in openSUSE.

pickarooney

On closer inspection, the temperature displaying (51) is actually that of the GPU. The CPU temperature doesn't display, and sensors-detect gives a load of errors, so perhaps there's no physical sensor on the motherboard.

WRT the GFX card, I've tested with several different nvidia drivers, installing cleanly each time, but performance is identical regardless of the version.
I'm able to provoke a crash now, just by using google-earth for two minutes. The system invariably hangs when skipping from one location to another.

ethernet

Is Google Earth set to open gl or what? Or am I getting all mixed up by Windows!

What distro are you using? I wonder what libraries Google Earth uses.

Have you installed any new hardware recently? A simple USB hub prevented me from installing SuSE 10.2 a few days ago.

pickarooney

The problems started a few weeks back when I finally got internet access. Up to then, I hadn't been using the PC for extended periods, but since then I've
- added an ethernet modem
- added (and removed as of today) a TV Tuner
- changed OS from Ubuntu 32 to Kubuntu 64
Not sure if Google Earth uses OpenGl or GLX or what.

ethernet

If it's not because of over-heating, I'd put it down to running a 64-bit OS. I don't think they can really compare with 32-bit kernels in terms of stability *right now*. Of course, 64-bit whips 32-bit's puny a$$ at the end of the day!

Perhaps the drivers for your tv card caused some hassle. Connecting a modem [router] shouldn't cause any problems if connected by ethernet. If you're using a USB modem, I'd be suspicious though.

What do you reckon -- is a re-install in order?

jmcc

Just a wild guess but is the motherboard one of those dual channel memory ones and your RAM is two sticks of 512MB? Sometimes you can get such boards to boot but they can lock up after a while due to mismatched sticks of memory. Download an iso of Memtest86 and test the memory. (I think the ordinary version of Memtest has probs with AMDs and there is a http://www.memtest.org site that has an apparently updated version though I have not tried it.)

As to the video problem - it could be the video card but it also likely that the monitor is on auto-detect.

Regards...jmcc

Zombrex

pickarooney wrote:

If it's an overheating problem, what's the best way to test this?

Firstly what Unix are you running? Does your machine crash in any other OS (either Windows or another version of Linux/Unix). If you don't want to wipe your OS try a Live CD or something, left running for a bit. If it only happens with a specific distribution of Linux then it is most likely a software problem.

If it does happen with other operating systems and distrubutions it could be a problem with your memory. I had the same thing, my system would just freeze up or crash after about 30 mnutes or an hour, in Windows and in Linux. I had 1GB of RAM spread over 2 512MB boards. I tried it with just one 512 MB RAM in on its own and it worked fine. When I tried the other 512MB on its own the system wouldn't even start. That is when I knew that that 512MB was busted. Replaced it and everything works fine again.

pickarooney

jmcc wrote:

Just a wild guess but is the motherboard one of those dual channel memory ones and your RAM is two sticks of 512MB? Sometimes you can get such boards to boot but they can lock up after a while due to mismatched sticks of memory. Download an iso of Memtest86 and test the memory. (I think the ordinary version of Memtest has probs with AMDs and there is a http://www.memtest.org site that has an apparently updated version though I have not tried it.)

As to the video problem - it could be the video card but it also likely that the monitor is on auto-detect.

Regards...jmcc

I have memtest86+ installed, but I can't figure out how to use it.
I just found the following message in a log using something called System Log - APIC error on CPU0: 40(40). Don't know if that's important and google results for the error are a bit hazy.
The memory is one stick, the original 1GB that came with the machine.
I could test with a live CD, but I'm not sure how conclusive that would be considering all the drivers would be crap and I wouldn't be able to do anything remotely intensive. Then again, the machine crashed while executing a sed command yesterday...

Khannie

pickarooney wrote:

I have memtest86+ installed, but I can't figure out how to use it.

Should be a grub option at boot time (you may need to configure this). Once it kicks in, you just sit back and watch it. I'd suggest you get this sorted (I think you can download a livecd with it too, but it needs to be run at boot time).

pickarooney

I added memtest to grub and ran it for 7 hours, with 0 errors found.

scojones

Undo any new hardware changes you have made.
Switch back to 32 bit.
See if you still have the issue.
Did you build this machine yourself? I'm guessing not because in your original post you didn't know the motherboard brand. If it's a Dell it will come with a utility partition from which you can run a diagnostic utility to test all the hardware in your machine, I recommend running this. You can run it by holding down the Ctrl key and pressing f11 when you see the Dell splash screen when booting.
Just because your memory is fine, does not mean your gfx card isn't. It could simply be a driver issue, or your gfx card could be on the blink.

To me it doesn't sound like an overheating issue. Usually when a computer overheats, the CPU reaches PROC_HOT and totally powers off, in your case it is freezing and staying powered on - not very good for a computer that's nearly melting.

Zombrex

pickarooney wrote:

I added memtest to grub and ran it for 7 hours, with 0 errors found.

Memtest+ didn't find my problem either. It only became clear when I swapped out each of the memory modules in turn.

Khannie

pickarooney wrote:

I added memtest to grub and ran it for 7 hours, with 0 errors found.

Well....at least you can rule out a bogey memory stick then.

Edit: saw wick's post after: WTH?

pickarooney

sjones wrote:

Undo any new hardware changes you have made.
Switch back to 32 bit.
See if you still have the issue.

Seems a little extreme; I'd prefer to know what's going wrong first. I did have a couple of crashes in the week before I switched to a 64-bit OS, so I'm not convinced there's a connection. As for hardware changes, I've removed the TV-tuner, and it hasn't stopped the rot. The only other change was the addition of a modem. It's not a bog-standard modem, however, one of these jobbies, which functions as a router.
If I could get a LiveCD containing Google Earth and usable gfx drivers I could test with that. Otherwise, pulling out the gfx card and using the on-board ATI card is a last-ditch option, though spending weeks trying to get a Radeon to work properly in Linux is not such a joyful thought.

Hagar

I've set up several Liveboxes and none have caused the problems you have so I would be inclined to think that it was not an issue.

/ edit what wattage is your PSU?

pickarooney

Hagar wrote:

I've set up several Liveboxes and none have caused the problems you have so I would be inclined to think that it was not an issue.

/ edit what wattage is your PSU?

I don't know off-hand. Is there a quick way of checking without running down to the cave to read the specs or powering down to look at from inside?

Hagar

Usually it's on a label but you have to take off the side panel to read it. If you are careful you won't have to shut down.

pickarooney

It says "total output shall not exceed 250 Watts". Sounds pretty low.
It just crashed for the second time in a row trying to burn a DVD

Also, I borrowed some RAM from work, booted up, ran google earth and the system hung within ten seconds.

Any other ideas?

[edit]Looks like I may have a hardware conflict between the MB chipset and the gfx card.