whea_uncorrectable_error on Overclocked PC after 1.5 months without any problems

[rozhabrev]

PC was working hard and without a single issue for about 1.5 month. I overclock CPU only. In 100% load it shows 90 Celsius per core max and 81 per core stable with Kraken x61 cooling system on it.

Suddenly whea_uncorrectable_error starts to appear on RealBench Stress Test and on typical video encoding work.

Windows updates were installed. New Nvidia drivers were installed. NZXT Cam update were installed. New LAN switch were installed in office. Thats all.

I've tried to Update BIOS and reinstall Windows on Clean. I've tried to Recover Acronis' Windows copy which worked stable from the day 1.

It seems there is no WHEA error on not overclocked CPU but I didn't tried it for long. In 4.5GHz RealBench shows Error in about 14 minutes. On 3.0GHz it holds up one 15 min round.

AIDA64 don't show me WHEA error at all (about 20 mins tested).

Also some strange USB behaviour appears. Connections not working properly every time. Not sure it linked somehow.

I raised up Core Voltage from 1.3 to 1.35 without a result. Raised up again to 1.4. Temp also rised to 102 Celsius per core max but it looks like system became stable again at least for 30 minutes RealBench test.

Spend 3 days trying to understand what's going on. Hope you can help.

All PC info in Speccy:
http://speccy.piriform.com/results/TiXiOvOQMBToQiXul60vLWi

Dump file:
https://yadi.sk/d/hEDogEZgn9i8w

 

[getochkn]

4.5 is too high of an Oc for that CPU I think.

http://www.hardwareluxx.com/index.php/reviews/hardware/cpu/33127-oc-guide-bringing-haswell-e-to-its-limits.html

Also, you don't jump voltage 1.3 to 1.35 to 1.4. According to that, 1.375 should be the max and you should go up, small, small, small steps. 1.301 kind of steps.

I think though 4.5 is too high for that CPU. It's not a super overclocking beast

 

[rozhabrev]

Understand. But it worked well for more than a month with 100% load 80% time.

When I build my PC I use http://rog.asus.com/365052014/overclocking/rog-overclocking-guide-core-for-5960x-5930k-5820k/ as a starting point and get stable results with 4.6 and 1.35v. Than I lower it to 4.5 with 1.3v and PC started using it for hard video works. It passes AIDA64 and RealBench stresses for all night with 4.5 1.3v. What could possible go wrong now?

 

[rozhabrev]

Seems like PC working stable when the room temperature is lower. Interesting because CPU temp and GPU temp remains the same with different room temperature. So it's not CPU or GPU overheating. What temps should I check so?

 

[rozhabrev]

In super cold room (about 5 celsius) WHEA disappears completely on every Stress Tests: RealBench, Intel XTU and AIDA64. RealBench failed with different errors than:

UNEXPECTED KERNEL MODE TRAP
DRIVER CORRUPTED EXPOOL

Windows Reported twice something like Nvidia drivers failed and were restarted (right corner notification) due RealBench test.


PS: Tried different RAMs - same result
Tried more powerfull PSU (1200 W) - same result

Any ideas please?

 

[bgunner]

Your OC/voltage is too high on the CPU causing heat to be an issue. If lowering the ambient temp helps stabilize the CPU then heat is an issue.

GPU drivers crash if the OC is to high on the card so if you overclocked the GPU lower the clocks. IF you have not Overclocked the GPU then try uninstalling and doing a clean install of the GPU driver again.

whea_uncorrectable_error is usually caused by the CPU being unstable/bad BIOS setting.