[Help] Wife's PC is randomly shutting down.

Tofystedeth

Crisis situation! Diablo 3 comes out in 3 days and my wife's computer is not behaving. It will randomly after a few hours of use simply shut off. Not entirely off, the power LED will stay on and the HDD activity light will blink, and a fan in there is going but that's it. It's done this before, around last November or December. I thought it might be heat related (Speedfan was showing a couple sensors at around 65-70C, which isn't super hot, but hotter than mine which is identical mobo and CPU runs by a little bit), and ended up replacing the CPU fan and heatsink and redoing the thermal grease as it had been stock. Nothing fancy, but it seemed to have done the trick. It's doing the same thing again.
I've blown all the dust out, and checked for bad caps while I was in there. I tried replacing the PSU the other day, which I thought had worked since it had been stable for a couple evenings of Netflix and gaming, but it just did it again a bit ago. I can't find anything in the event log other than Windows complaining the previous shutdown had been unexpected. According to her it seems to do it most often when loading zones in Tera, which does jive with the one just experienced (though it was loading a zone in Path of Exile) which makes me think it might be the HDD. SMART seems to check out (at least according to Speedfan. I'm currently burning a UBCD to check and see what their tools say.)

edit:UBCD's drive diagnostics also checkout. Running Memtest for shits.

If it's not that, I'm in for a headache. It's currently got this mobo, which isn't exactly a spring chicken, anddoesn't support DD3, so if have to replace that or the CPU, I'm going to have to practically replace half the system.

Dratatoo

To me it appears to be RAM related. RAM problems can crash the graphic driver for example or can result in "no signal at boot" issue. Use the Windows memory dignosis or better Memtest. You need to let the memory check run for a couple of hours.

If its a HW failure, the system might crash before Windows is able to generate a log entry.

Otherwise you could try Prime95 and a temerature monitor program. This way you could check memory, temperature and the stability of your system at once.

Tofystedeth

Well, after about half an hour of Prime95 the temps have been stable at 81-82 for the last 15 of them. No errors in the calculations yet, but more importantly, no crashes. It doesn't look heat related, or at least not heat there. I might try and find a disk stress tester.

uean

Sounds like you pretty well got it covered. Everything I would have done you have done. Only other thing I'd check is the PSU and voltages, if there's any spiking or grounding issues, etc. It sounds like a bad mobo or related issue, so power RAM and heat are the things I'd be looking at.

Tofystedeth

Well, this morning after I got home from work it went down twice within minutes of booting doing nothing more strenuous than opening firefox. Ran an hour of scan for artifacts on ATItool with no errors, and let it run MemTest all day with no errors. I think at this point I'm just going to say "Fuck it, Happy Anniversary" and replace everything I haven't already.

taliosfalcon

I've had these exact same symptoms before and it turned out to be a psu issue for me, ymmv though.

Tofystedeth

Already swapped a brand new one in cause that's what it seemed like to me too.