Hard Freezes (No BSOD) - Not RAM, how to diagnose?

[Gilbo]

My main desktop has recently become unstable. It hard freezes randomly (apparently). This isn't a BSOD, but a hard freeze. The music keeps sounding, frozen at the moment the freeze happens. The computer doesn't crash or reboot, the screen continues to refresh (doesn't go black), but everything is completely frozen. Needs a hard reset.

Is this absolutely a hardware issue? Could it be a corrupt Windows installation?


I don't think it's:
1. The RAM - Why? It's passed 30 hours of Memtest86+ without an error or a crash. The computer will rarely work for more than 4 or 5 hours under normal usage, but doesn't show any RAM errors even over 30 hours (12 full passes).
2. The CPU Temperature - Why? I've run Prime95 for several hours, and the CPU maxes at 55C (TCase - heatspreader temperature). It'll stay at 55C hours before the freeze hits, so the crash doesn't appear to be heat related, or CPU strain related for that matter.



What should I be looking at? I don't have a lot of spare parts to swap out at the moment unfortunately, so I'm trying to do the best diagnosis possible without taking everything apart. I'm also presently on limited funds while I recover from a rather exuberant vacation (my first in 3 years!). Anything else I can test?


Possibilities in my mind:
1. The PSU - The 12V reads at 6V according to Speedfan, but I'm pretty sure that's an error? It's a Seasonic 650-HT, which is about as high quality as they got when this was built 3.5 years ago.
2. The Mobo - I don't consider this tremendously likely though: It can sustain several hours of 100% CPU usage, and I have no memory errors. If it was a bad cap or decaying VRM I would expect 100% CPU usage to kill the computer faster than less loaded activities, but there appears to be no correlation in my testing so far.
3. Bad SATA cable / bad HDD: Could these cause this sort of behaviour?
4. The CPU - Could it be a fundamentally bad CPU? It doesn't appear to be CPU strain related, but...?


Further Tests I'm considering:
1. Underclocking the CPU - Maybe this will identify a bad CPU. It's at stock right now.
2. Reinstalling Windows - Bleh. Big mission, but cheaper than buying a new computer. At the moment nearly every component is suspect, and I can't reuse the DDR2 .


Any ideas would be greatly appreciated. Full hard freezes like this are pretty unusual; usually, in my experience, you'll get a crash or BSOD rather than this hard freeze.

Does any of this hint at anything out of your collective experiences?



Specs (System is 3.5 years old - I think; maybe 2.5... Whenever this was highend):
*CPU - Intel Q6600
*Mobo - Gigabyte P35-DS3R
*Graphics - NVidia 8800 GTS
*PSU - Seasonic 650-HX
*1x WD 640GB HDD
*4x Hitachi 7K1000.C's.

 

[MaxBurn]

What were you doing when it froze? Not always the same thing?

Very general: reseat everything. While you are in there you could measure the DC volts AND AC ripple on the DC lines, you have to do this with the motherboard powered up so there is a minimum load on the PS so be careful.

 

[Bozo]

Have you looked in Event Viewer?

 

[Gilbo]

Bozo,

No I hadn't. I did though when I saw your post. I was hoping maybe I'd find some sort of driver error near when the crashes happened, but there's nothing with any correlation.

I don't think Windows has a chance to log anything. One second everything is fine, the next, nothing moves and the sound is frozen in the instant.


In fact, it's worth mentioning that there are no slow downs or other issues near when this happens. Everything is smooth as can be until it's not...



Maxburn,

That's a good idea. I'd forgotten that general piece of wisdom.

 

[LunarMist]

I've had bad hard drives lock up computers before.

 

[ddrueding]

Try booting to a liveCD of some kind and stress-test from that. That isolates the hardware from the OS without having to do a complete re-install.

 

[Stereodude]

I'll second the liveCD/DVD suggestion.

 

[Bozo]

Re-install your music player software?

 

[Mercutio]

You have a power supply that's testing bad in a big, big way that's very easy to verify. You really need to double check that first.

And then blame that Western Digital hard drive.

 

[time]

+1. Although Merc left out the nVidia graphics card - always worth a substitution, just in case.

 

[Gilbo]

It's not a bad Windows install.

I ran Prime95 from a live OpenSUSE disc and got a the same hard freeze after only about 10-15 minutes.


So I ordered a new PSU from Newegg.ca (who seem to have consistently lower prices than my good old NCIX these days), a Corsair 750HX. I've seen it rated extremely favourably at X-Bit Labs amongst other places. In fact, it seems to be 2nd only to the new Seasonic X-750 in output power quality and efficiency (also a little behind the Enermax MODU87+ in the latter spec) if one was to split hairs.

I was going to get the Seasonic, but Newegg had the Corsair for $40 off & free shipping so it ended up being nearly $60 less expensive.

I guess it could still be the mobo or the CPU, but at least I'll probably have it narrowed down on Monday or Tuesday.