Overclock stable on i7 8700k but experiencing problems
Overclock stable on i7 8700k but experiencing problems
This setting indicates a 200 Watts output, meaning your 8700K is operating with AVX/AVX2 support, which Prime95 v29.8 permits you to turn off.
Prime95 v29.8
Even though transcoding tools like HandBrake utilize AVX instructions, the high variability in load and power draw usually stays below the consistent 100% workload seen in Prime95 Small FFTs when all AVX tests are disabled.
Click on the AVX test selections that aren't grayed out to ensure all three AVX boxes are selected, as demonstrated, then retry.
Hi,
After disabling AVX, prime95 performs exceptionally well maintaining temperatures around 77 degrees and consuming less than 160W. The results are solid with an OC of 4.8GHz and an offset of -30, resulting in a VCore of 1.164V.
Something new to note: you haven't finished with just Prime 95. It only addresses CPU cooler thermal stability.
We haven't found evidence that your OC is voltage stable. Prime 95 doesn't handle both well; its workload is too consistent for a proper voltage check.
Execute Cinebench R20 'Infinite Loop'. To configure the 'infinite loop': File > Preferences > Minimum Test Duration. Input an unrealistic value.
Proceed with the test for an hour – the longer it runs, the better.
And there it is. Great work!
Using AVX in transcoding applications demands far less processing power compared to AVX in P95, which can be a demanding 130% workload. Disabling AVX in P95 results in power usage and heat output very close to that of transcoding—what we anticipate.
To keep P95 stable with AVX, you’d need a substantial rise in Vcore, which naturally boosts power draw and temperature... or
... a reduction in clock speed from 200 to 300MHz, which negatively impacts transcoding duration.
You might consider adjusting the AVX offset in BIOS to lower the clock, but since your current Vcore settings and overclocking seem stable during transcoding, it’s not necessary to change them just for AVX support.
As Phaaze88 noted, it’s essential to verify that your Vcore and power delivery remain consistent.
Download and open Hardware Info (HWiNFO) with P95 Small FFTs without AVX, then monitor the minimum and maximum Vcore values during steady-state tasks. This will reveal how much your Vcore fluctuates.
A 10 to 15 minute session with an air cooler provides sufficient time for the CPU, cooler, socket, motherboard, and VRMs to reach 95% thermal saturation—so exceeding 30 minutes isn’t necessary.
After that, run HWiNFO with "Sensors Only."
Once you understand your Vcore behavior in P95, repeat the transcoding test in HandBrake as you did initially when instability was detected. Again check the minimum and maximum Vcore values, then compare them with the fluctuating workloads from HandBrake versus those from P95’s steady state.
I performed extended tests on prime95 and encountered instability (BSOD). I adjusted the offsets and repeated until reaching an offset of +10 without BSOD.
The findings are:
Prime95
VCore - Min - 1.184V Max - 1.248V Avg - 1.192V
80 degrees
167W
Handbrake
VCore - Min - 1.216V Max - 1.360V Avg - 1.243V
77 degrees
160W
I believe my voltage regulation remains quite weak. Are there any additional steps I should consider?
Power usage and core temperatures behave as anticipated, but during the steady-state workload of P95 there are 64mv fluctuations in Vcore, suggesting inadequate voltage control. Please confirm the Prime95 Small FFTs no AVX test. Ensure HWiNFO is not launched until after starting P95, and capture the minimum and maximum Vcore readings before stopping P95 so the data accurately reflects only the P95 conditions. The average Vcore value is not relevant here, so avoid logging it. Transcoding generates varying workloads, which should result in much higher Vcore swings (144mv) compared to P95 (64mv). HandBrake isn’t required—just rerun P95 if you wish.