Computer restarts when playing graphic intensive games

System Specs:
Motherboard: Gigabyte GA-970A-UD3 revision 1.2 F7
Ram: G.Skill Ripjaw 16GB 1600 DDR3
Graphics card: EVGA GTX 760 superclocked with ACX cooling
CPU: AMD AM3+ 8150 Zambezi 8-core with cooler master hyper 212 plus
PSU: Rosewill ARC-M650 modular single rail
Hard drive: Seagate Barracuda ST1000DM003 1TB 7200 RPM 64MB Cache SATA 6.0Gb/s
Keyboard and mouse: CM Storm Devastator - LED Gaming Keyboard and Mouse Combo Bundle
Operating System: Windows 7 home premium

I have removed the 2 new sticks of 2x4gb ram and began trying to test the PC.

I ran memtest on each of the 2 remaining sticks of ram for 12(first one overnight) and 9 passes without finding errors.

I have ran the furmark test for over 20 minutes and my temps never ran higher than 83(which is about what the card is supposed to do)

However when I try to run Prime95 in any test with 8 threads is almost instantly crashes. I was able to run the small fft test with up to 6 cores without a problem, but as soon as I added a 7th worker is causes a crash within minutes. I have not played around much with the thread number for the other tests, I just know it causes a restart without BSOD every time I try to run the 8 worker torture test.

I have also tried running OCCT tests. With the OCCT: CPU test in small data set a reset happens after about 2-3 minutes. Large data set doesn't run for more than 5 seconds before restart. And the OCCT: Linpack test has about the same results.

I have monitored the temperatures with the CPUID HWmonitor and haven't noticed the temps going out of the ordinary while running the stress tests, but I did notice that the -12v value drops pretty drastically(leveling itself out usually shortly after)

At this point I'm not sure what the problem is. I just know that every stress test I have tried to run on the CPU causes a restart.

Please help.

P.S. none of these restarts create a dump file(and I have it set to not restart under power options).
It's also running in the high performance profile.

I do not currently have access to my home computer as I'm at work, but if anyone has any idea's I can try them when I get home.

 

500W is fine for a GTX 760 /Ti, with a >20% overclock on that 8-core, I'd want another 100W+ to feel safe though.

That Rosewill PSU is made by ATNG (www.atngpower.com.tw), it might be anywhere from rubbish to very good - can't find a review of it so I'd assume the worst.

Reset any overclocks to default, if you hit the same snag, try to borrow a known good 500W+ PSU from a friend. Same for the GPU - swap parts and test one change at a time.

CPUs are usually the strongest part of a system by a big margin, if it's not the PSU at fault, the motherboard would be next on my list.

 

So I picked up a new power supply which has more amps than I could need and it's one of the better ones(Rosewill diamond series).

I hooked everything up and ran the OCCT Large data set stress test....then about a minute in my computer restarts....again...

 

Hmm, suspect the 'board next - trying to test that theory will be one big challenge

You did revert any overclocking to defaults in the BIOS, yes?

Uninstall all 3rd party 'utilities', including those from the graphics/'board makers, many of those are buggy to some extent.

 

My bios is up to date(F7) and everything is set to "optimized default".

 

Good. Before you remove all the utilities, can you check for -12v value drops again?

 

Well I just ran the OCCT GPU burn in test to check to see if my video card was the issue(I've previously ruled it out, but wanted to check again). Everything held pretty constant, but the -12 rail was all over the place in voltage ranging from -11.2 to -14.6. Not sure if that helps.

At this point I just dished out money for a new power supply, I can't really afford to keep replacing things if they're not responsible...

 

I tried to edit my old post to add this on, but forum restrictions stopped me. Sorry.

I also just ran a OCCT Linpack test, which causes my computer to restart after about 3 minutes. I was watching the temperatures and they never went over 48C The CPU should be good up until about 61C. So I don't think it's a ventilation or overheating problem.

 

I just remembered a topic elsewhere with a similar problem, 8-core CPU + the same 'board: "4 phase power that isn't heatsinked running an octo core... Glad its back to perfect! Make sure that your PC isn't throttling under load with the weak motherboard."

 

So you're saying it's likely it's the Mobo? Because I do have a heatsink on it...

 

You have a heatsink on the VRM section of the 'board? does it have good airflow over it?

 

Yeah I have the cooler master hyper 212 pro. I've been running this PC for 2 years now, only recently had these issues. The fan is directed out the back of the PC, there's also the front case fan.

 

The fan's directed out of the rear, which way is the 212 blowing?

If the PSU is above the CPU, which way does that blow?

 

The PSU is at the bottom of the case with a fan blowing up, the heatsink is above the CPU right in front of the back ventilation holes blowing out the case, then there's the case fan blowing out the front bottom of the PC. The Graphics card has 2 fans blowing down basically right into the PSU(as there really is no other way to put the card in).