Problem - Turns off and won't turn back on for a few minutes

[harchong]

I've had my prebuilt Cyberpower desktop for a few years and this is a recent problem. Every time when I'm playing a game on my computer it will eventually just turn off with no warning and then after a minute or two will turn itself back on. I get the warning screen on bootup about not turning it off correctly so I'm pretty sure it's a hardware problem.
I've seen some other threads before that suggest this problem could be the PSU or the graphics card or overheating. I've already did major dusting and found I don't do that often enough.

Windows 7
MB: Gigabyte GA-P55A-UD4P
GPU: 2x Nvidia Geforce GTX 480
PSU: Coolermaster Pro 1250 RS-C50-EMBA-D2
CPU: i7 K875
I have 6 fans, 2 of which are used by the Enermax liquid cooler on the CPU. I can't be more specific because I can't find it on their website. A cylinder on the CPU has tubes to a block between two fans.

Oddly I get turned off more often on Star Trek Online than Mass Effect 3. When I turn my computer on there is no beep, that could mean a PSU problem but I read online that Gigabytes beep codes are for very broad reasons.
I've used SpeedFan to log my temperatures and voltages to try and figure this out myself. But I can't find any website to explain the chips sensors and which voltages are normal and which are not.

Here are my current Idles (ambient is 22C)
Chips:_____IT8720F____________LM75_ADT7473________________________________Intel core
GPU_GPU__Temp1_Temp2_Temp3_Temp_CPU___Local_Remote 2__CPU__Local__Remote 2_Core 0_Core 1_Core 2_Core 3
51.0_54.0__46.0__25.0___35.0___2.0___45.0__43.0__45.0 48.75_46.25_48.75_45.0 44.0_45.0___45.0___1.26___1.58
And here is right before a crash
88.0_82.0__60.0__25.0___51.0___2.0___82.75_68.25_83.0 76.25_68.0__76.25_62.0 58.0_57.0___56.0___1.25___1.58

In trying to find out what means what; Everest thinks Temp1 is my MB and Temp3 is my CPU. But HWMonitor thinks Temp1 is my PCH and Temp2 is my CPU. Nvidia says my cards are good to 105C. Intel says my CPU's Tcase is 72.7C. So which probe is the correct probe? the IT8720F doesn't show any danger numbers but the ADT7473 might mean I'm in trouble. Maybe my discontinued liquid cooler failed?

Now for my power numbers
Idle
IT8720F________________________________________|ADT7473
Vcore1_Vcore2_+3.3V_+5V__+12V_-12V_-5V__+5V_Vbat_|Vcore_+3.3V_Vcore_+3.3V
3.3____4.97___3.39__-1.09_-8.12_3.58_3.26__3.0_3.21_3.0___3.23
Crash
3.26___4.92___7.17__-1.58_-7.07_3.58_3.26__3.0_3.16_3.0___3.18

That +3.3V is freaking me out, it's got the largest jump and since I was in ME3 when I crashed I assumed it was for a graphics card, but I managed to find a website that explained what voltages are in what wires and the graphics cards don't get 3.3v, the MB and SATA (rarely used) do. http://www.playtool.com/pages/psuconnectors/connectors.html (Those last 2 empty columns are not a typo, SpeedFan gave me nothing for them.)

 

[Raiddinn]

When a computer is working for a long period of time and quits working out of nowhere, I would hazard a guess that 95% of the time the PSU died.

You may want to get another PSU of about equal wattage. I would suggest Enermax instead this time. They are the highest quality brand in the 1000w+ space.

If you can borrow another PSU of about equal quality, that could help confirm this.

It could also help confirm if you take one of the 480s out and just use one. If it is indeed a problem that the PSU has lost its ability to supply enough power for the two 480s then it may still have enough power capability to power one without problems.

 

[mojorisin23]

i agree wholeheartedly with Raiddinn... i would guess PSU. go to a computer store with a good return policy (open box), keep everything, and use it and see if that works.

 

[harchong]

My cmstorm sniper case has a mesh side and a 200mm intake fan there. On the front is a 200mm intake, bottom as a 120mm intake. Top has a 200mm out, and back has a 120mm out (liquid cooler here).
I've had a box fan pointing at it for the past few days and it hasn't crashed. I've dropped speedfan and am now using cpuid monitor which is giving me much easier data to look at. I guess I had a simple overheating problem, sorry for making such a deal out of it.
In case someone googles this I'll just say I dusted, added a fan between my two graphics cards to expel out the back (which I don't think has impacted my gpus temperatures at all) since apparantly my 480s are notorious for heat production. And I partially disassembled my Asetek LC510 liquid cooler, just removed the brick between the two fans and used canned air through it.
I might replace those fans with something of higher cfm, I can barely feel any air moving out the back. I think I'll replace my side mounted 200mm with two 120s for more cfm as well. Maybe even cover up the side grill excepting the fan to improve wind tunneling. As it is, the air could simply come out the side around the fan. That sound like a good idea? What about replacing the innermost 120 around the liquid cooler to push more air through it.
I've searched tomshardware for articles about air flow, is there one? All I find are forum posts, which I will read.

 

[Raiddinn]

http://www.tomshardware.com/reviews/cooling-airflow-heatsink,3053.html

I can tell you that the side fan is one of the fans with the absolute biggest impact.

In general, people usually advise that 1x a bigger fan is preferable to 2x a smaller one.

It wouldn't be the worst idea to cover up the holes on the side while your 200mm is still on it.

 

[harchong]

Best answer selected by harchong.