Strange Computer Instability Problems

[[R4A]Bigman]

Hi,
I've been mulling over this one for the past couple of weeks now, and after doing some serious stress testing over the last couple of days I'm pretty much stumped.

Windows (and various components thereof) crashes on a fairly random basis on my current rig. I've had this set up for around a year now, though admittedly the majority of that I was running in Ubuntu. Only recently have I started using Windows for any reasonable period of time again for running games. Occasionally I get other programs crashing as well (such as Apache), but its usually Windows Explorer and Dr Watson. I also have experienced one or two BSOD, but that was whilst playing Crysis and that game is buggy as hell anyway - so I've been more or less ignoring those issues.

The most weird thing about this problem is that it only seems to happen when I'm NOT loading my rig. If I leave it to its own devices for a few hours, I can come back to find it has reboot. Or if I'm doing some light work (word processing, web editing etc) then things start to crash on me. However, I have managed to run OCCT for over 24 hours without any problems at all, Memtest86 for over 27 hours and ATi Tool for around 4.

I initially thought it might be a heat issue (as the fans spin down when not loaded), but I've now got all the fans in my case running at full whack and I still experience the problem.

Rig set up can be found next to my name <<<<<===
I'm running Windows XP Pro (32-bit)

Anything you can suggest at all is greatly welcomed! I have got to the point where I will try just about anything!

[edp33]

Make a note of the temperatures, cpu, gfx when idle and when under load.

Check event viewer in control panel to see if that shows a common list of errors leading up to the crash

Try installing the latest drivers

Could also try unplugging some peripherals, i.e printer etc and see if it's any more stable

Good luck

[[R4A]Bigman]

Thanks for the suggestions

Here goes:

Idle temps:
CPU: 24C
System: 38C
HDDs: 23C
GPU: 52C

Load temps:
CPU: 47C
System: 40C
HDDs: 23C
GPU: 70C

Nothing special seems to be logged in Event Viewer - just the errors themselves. The errors are also miles from anything else happening on the system, so are pretty isolated.

Installed the latest drivers for everything a few weeks back to no avail.

Don't have any peripherals beyond the mouse and keyboard.



I'm wondering whether this is a hardware issue or if its just time Windows got reinstalled?

[Hawker1986]

Well it sounds like youve excluded most of the obvious things, its not memory or graphics, so my money would be on it being an OS problem and if I was you my next step would be a reformat of the Windows partition with the latest chipset/graphics etc etc drivers. Also something to think about is it could be PSU, as that can cause all sorts of odd problems, I dont think its likely, but if you have a working spare around it would be something to try.

Good luck.

Hawker

[jw007]

Temperatures look fine - apart from maybe the GPU (although I'm not running the latest GPU or anything close myself, so that's the area I know the least about).

Any chance it's the PSU??? Left idle Windows will probably start to get bored and do lots of random house-keeping at very random intervals!! This could lead to increased draw on the PSU and cause the system to reboot without causing entries in the Event Log as the event is external to windows. Check from the Event Viewer what it was doing before the report and then try doing the same thing a couple of times and see what happens?!

Thinking outside the box (pardon the pun) could it be an electrical issue? If you can't spot a pattern, it might be that the voltage, etc. to the PSU is dropping momentarily causing the reboot and that would not be reported in the Event Viewer, as it's external to the PC. Might be a dodgey 4-way or socket?! Try changing the socket your using and any 4-ways, etc.

Good luck with your system!

[stevie lee]

it could be speedstep and C1E causing it..

try disabling them so they're turned off..

explanation: these 2 features reduce the speed of your CPU wen you are not using it, they might not let your chip get back to full speed when you or windows put some load onto it - hence the crashes and errors.

[chrestomanci]

it could be speedstep and C1E causing it..

try disabling them so they're turned off..

explanation: these 2 features reduce the speed of your CPU wen you are not using it, they might not let your chip get back to full speed when you or windows put some load onto it - hence the crashes and errors.

Good idea Stevie.

Speed step not only reduces the clock speed of the CPU, it also reduces the voltage. There a supposed to be tables in the BIOS, that say what voltage to use for each clock speed, but it is possible that those table are buggy, or Bigman's CPU does not like having to little voltage.

I would suggest disabling speed step. If that makes the problem go away then Bigman could try to reproduce it, by loading up Linux, then forcing the CPU clock to minimum, and running some Linux specific benchmarks.

Run:

sudo echo powersave > /sys/devices/system/cpu/cpu0/cpufreq/scaling_available_governors

(Repeat for all cpus if you have multi core system.)

If this is the problem it may be possible to fix it by flashing to a more recent BIOS, but that would only work if the mobo manufacturer has found and fixed the bug in the voltage tables.