Question Question in regards to overclock stability testing

mjbn1977

Reputable
Aug 20, 2015
443
10
4,915
37
I have a air cooled 8700k (cooler is DARK ROCK 3 PRO) which I ran for over 1 year at 4.8Ghz at 1.25V. I just recently successfully delidded the CPU and applied Conductonaut which dropped the temps by around 15C to 20C depending on core and type of stress test. The applications I mostly use on my system are Games, Unreal Engine 4 Editor, and a little bit of Blender. So there is some AVX stuff happening on my machine once in a while (rendering, building lighting etc), so I need a stable overclock for AVX.

Currently I am searching for the lowest Vcore that can run my system stable at 5.0Ghz. For thermal and stability testing I'm pretty much following the guides from our fellow forum members CompuTronix (https://forums.tomshardware.com/threads/intel-temperature-guide.1488337/ ) and Darkbreeze (https://forums.tomshardware.com/faq/cpu-overclocking-guide-and-tutorial-for-beginners.3347428/ ). By the way, thank you for those fantastic guides. They are very helpful.

Nevertheless, while doing the testing I came up with a few questions which I would like to ask Darkbreeze and CompuTronix, but please, everyone else feel free to comment as well.

So, I tried several different Vcores with the RealBench stress test and found out that the lowest possible Vcore to run RealBench is 1.32V. This runs successfully without any problems for full 8 hours and passes. Any lower voltages (I tried 1.28V, 1.3V, 1.31V) will have blue-screens within the first hour of the Real Bench stress test. I haven't done the prime95 v26.6 blend 8hr run (I plan this for tonight), but out of curiosity I ran prime95 current version small ftps and worker on my second core basically stop within the first couple of minutes.

Question 1: I now that prime with AVX is unrealistically hard on the CPU and unrealistic for most applications, but would you consider my overclock at 1.32V really stable (assuming that it will pass the prime 8hr blend test tonight), after passing real bench 8hrs but not prime95 small fft's for even 2 minutes?

Question 2: I am planning to also come up with a overclock that uses an AVX offest of -2. How do I check for stability in that case. Since Real Bench stress test always uses AVX it will never test for the higher non AVX core clocks. Lets say, hypothetical I want to check if my CPU can do 5.1Ghz with AVX -1, how do I stress test for the higher non AVX value?

Oh, by the way, temps seem to be fine. I am well below 85C. During the 8hr stress real bench stress test it averaged mid 70s (but the hottest core maxed at 80C). Ambient temp was 74F. All 6 cores also seem to be very close together temp wise, around 2-3C. The prime95 v26.6 small FFTs test had lower temps than Real Bench.

Thank you!
 
Last edited:

ASK THE COMMUNITY

TRENDING THREADS