Computer shuts down when increasing vcore FX 8350

[snix]

I'm trying to OC my FX 8350. I have followed a OC guide carefully. I have disabled the turbo boost, C1E, Cool & Quiet, and C6. I increased the clock multiplier by .5 each time and ran load tests. I get to 22 (4.4) and ran IBT (10 loops) and prime95 for an hour with no problems. My max CPU temp via HWMonitor is 62 C while running Prime95. When I go to 4.5 IBT returns a negative number; from my understanding this is when I need to increase the vcore. When I increase the vcore by .025 (1.429), windows boots fine but 20 to 30 seconds running IBT my computer shuts down. I have to turn the PSU swith off and back on to boot up again. I also tried .05 increase to vcore with same results.
- my temps are still in the 50's c when the computer shuts down.
This is my exact PSU:
http://www.newegg.com/Product/Product.aspx?Item=N82E16817163109

Do you think the PSU is the culprit or am I missing a setting? FYI - My PSU has had zero issues running this system for the last few months without the overclock, however it is an older PSU.

 

[photonboy]

Your problem is likely the motherboard supplying insufficiently stable voltage, OR it's your CPU. Just because the CPU is below the Tj threshold doesn't mean it will be stable at a high frequency.

Just accept that you've reached the point of instability and back off the overclock until you're good and stable. It's never a good idea to push things near the limit.

It's not like that's a bad overclock.

My advice is stick with 4.4GHz since that appears pretty stable. Going from 4.4 to 4.5 is a theoretical 2.3% gain but in reality will be less than that.

 

[Calculatron]

It could be your motherboard. What are you using?

EDIT: I'm not implying your motherboard is bad, I am saying that your motherboard might not be made for such an overclock. It should have a decent amount of phases for its VRM, and preferably have a heatsink.

 

[snix]

My motherboard is: Gigabyte 990FXA-UD3 Rev. 3

 

[TomTom111]

My motherboard is: Gigabyte 990FXA-UD3 Rev. 3

I have that exact motherboard, but i have the FX-6300.
I was able to get my CPU up to 4.7 stable (this was the highest i could go before experiencing issues like what you are describing).
The only difference between our CPUs is, core count obviously, and the 8350 is a 125W CPU where the 6300 is 95W.
So i agree with the two previous posts, this may be a motherboard limitation.
The PSU is good, the CPU is awesome and the mobo is solid.
It may just be a limitation the MOBO has, because when pushing the electronics past their operating guidelines, they experience instability.
So throw in the added power needed, and it's just another variable in the equation.
You can tinker with the the VCore settings, NB voltage, and every other setting, but ultimately, you may be at the limit of the MOBO.

4.4 GHz is a respectable OC for an 8-core!

 

[Calculatron]

The motherboard has a decent number of phases for its VRM, and it has a heatsink.

Have you messed with Load-Line Calibrations (LLC)?

What kind of case do you have? (With how many fans, where?)

 

[snix]

No sure why but the 990fxa-ud3 mobo has no LLC setting which may play a role. I have a older falcon northwest case, 120mm intake fan in front, 170mm exhaust fan on top, and 2 80mm fans in the rear. I also have 2 80mm fans cooling my video card, and the evo 212 cooling my CPU. It seems odd to me that this set up can't handle a 0.025 increase in voltage.

 

[Calculatron]

Cooling set-up should be fine.

The lack of LLC is a bit odd. Maybe it's under a different name?

Beyond that, it seems like everything should work. I have heard of cases where giving too much voltage can cause stability issues (somehow, I don't understand it yet) but those are off-handed mentions that are few and far between. So maybe try a .0125 voltage bump?

Other than that, I am not sure what is up. The PSU should be fine. The Motherboard should be fine (emphasis on *should*). The silicon lottery usually doesn't play out like this.

Have you tried cross-referencing your CPU Core temperature with different programs, like Coretemp, or (if you have an AMD GPU) AMD Overdrive?

 

[snix]

I have used HWMonitor and Speedfan at the same time to cross reference.

Question: Is HPC (high performance computing) suppose to be enabled or disabled? I have read different takes on this.

All the tests above i.e. 22 multiplier, 4.4, and a stable system was with it enabled.

I just disabled it and ran IBT. It passes IBT with lower temps than with it enabled however my Mhz/Voltage jumps all over the place and it seems to be down throttling however it passes the test.

 

[Calculatron]

Try Coretemp, to triple check.