"Power loss" while playing games - how to narrow down the failure ?

hi

Got an Intel Core I7 950, NVIDIA GTX295, ASUS P6T Board, 6BG RAM (3 channels), 750W power supply and a 30" HP screen (2560x1600). The HW is quitea few years old, bought it more or less when it was released first. No HW component was ever over clocked. Using constantly the currently latest WHQL drivers (285.62 right now) on a win7 64bit which gets it's updated almost daily.

Issue:
Since a few month (2 - 3), the PC crashes (immediately shuts down, exactly if you'd disconnect the power cord) and reboots afterwards. This ONLY happened when playing PC games. This crash however does not necessary happen every evening I play. Still I'd say if I play more then 2 h it has a very high chance to occure.
So far I have only noticed this "crash" when I play grafic intensive games like Skyrim,... Games like Wow were so far not affected by this crash. If a crash happens and the PC is rebooted, I have an almost 100% chance that 10 - 30 sec after the game is started again the PC crashes again. If crashed once, the crash will happend very repeatable.
So far I could avoid these "crashes" by playing the games not at max resolution but something like full HD which worked over 1 - 2 month for Skyrim and other games very well (0 crashes in a time frame I'd have expected to see ~20 - 30 crashes ==> def. no coincidence). Up to now this worked but since today even wow crashes and it did do it also when playing at only full HD resolution.


What I did (noticed) so far:

1) As the crash looks exactly like a powerloss and very related to the GPU, I checked the temperature of the GPU with GPU-Z. I have a few logs where the PC crashed and in no case the GPU temperature was above 51° C (GPU load usually is slightely below 20%... Seems low to me ?). Even the PCB temperature is very low ==> not really an indication of a too hot PC

2) Missing a tool for the CPU, tried the same with CPU-Z for the CPU but it does not seem to have temp sensors for the CPU

3) Oh by the way, since quite a few weeks, it is VERY dry in the PC room, often below 40% (30 - 40%)


Question:

- Anyone an idea in what direction to look ? Any tools I could log some conditions to narrow down what component is causing this ?
I'd assume a temperature issue resulting in a power loss, but where ? My tendency is the power supply ?

- Since I build my PC 's since ~25 year on my own, playing with HW is not really new to me. And obviously I'd have changed HW piece by piece to see which component is playing bad. But exchaning a GPU for a 30", CPU, MB, RAM, Power supply is not really cheap just "to try through what is causing the crash".

- BSOD are not generated. At least the "BlueScreenView"er from NirSoft tells me there is no BSOD. Not really surprising if it is a power problem, how should one be written...

Legaia said:

Thought have you tried to stress test one component at the time?

So far I was not aware of a stress test which would really point me down to one specific component.

I would recommend using OCCT, Includes CPU, GPU and PSU stress tests. I'm not sure how the PSU test works thought but....

OCCT, that is new to me. I'll give it a try.

Beside that it looks as if there is no way around a PSU change.
I'll also visually check the condensor on the MB and GPU this evening.

You could also check your GPU out to see if you have dust build up in the fan, or in the system altogether. If you are reasonably certain that your computer isn't overheating, then I would turn my attention to the PSU. It is likely failing, or if it is a generic one, it isn't able to handle the loads gaming puts on it, so it is starting to die.

Be sure to follow DeaconFrost's suggestion first, then the PS check.

Side bar:
You posted "3) Oh by the way, since quite a few weeks, it is VERY dry in the PC room, often below 40% (30 - 40%)" to which I respond - "Send some my way! 30%-40% out here is called MUGGY."

Regards,
GEWB