[SOLVED] Insane temps with manual voltage

scherioyt

Reputable
Jun 17, 2018
41
0
4,530
Hello guys, i own an i7 8700k and i'd like to overclock it but i'm new to CPU overclocking, everytime i put a manual voltage from the BIOS (asrock z370 extreme4) like 1.2V, my temps bump to 95 degrees on tests like cinebench. My cooler is the dark rock pro 4 and i have pretty cool temps with the automatic voltage. Thanks in advance
 
Solution
cinebench passed well
Was that just one run, or did you do 'infinite loop'?
Infinite Loop: Open, click on File, then Preferences. Under Minimum Test Duration, enter some ridiculous number, like 9999999999999, and it'll default to the highest it offers.
Then run it for like an hour - although, the longer, the better.

In BIOS I set LLC on level 3 (1-5), is it ok in your opinion?
Should be. No one needs to use higher than the medium LLC settings unless using LN2, or the mobo VRMs are liquid cooled.

jasonm9933

Honorable
Mar 19, 2014
60
1
10,645
Cinebench is an AVX Workload. AVX programs can go way above the max temps for something like gaming. Easiest way to fix is to set an avx offset in your bios. If you google "How to set avx off set for X motherboard" It should pull it up!
 
Okay not too much headroom on AVX workloads and yes we can put a AVX offset but wanted to know how much headroom you had.

Try changing the multiplier to 47 leaving everything else in auto. Then run Prime95 version 26.6 and do a blend test monitor temps for 5 minutes and vcore max, report back.

Once done I will try some specific settings to test with vcore and LLC. Just worried about your temps right now. I had an 8700K but I delidded mine which dropped temps by 15 degrees but you should not need to with adequate cooling especially for a simple all core overclock at 4.8GHz...We will work up to that!
 
  • Like
Reactions: scherioyt

Karadjgne

Titan
Ambassador
Follow, but don't copy. Your cpu, motherboard, ram, VRM's etc are different from the video, even if they look identical. So tailor the exact voltages to what works for you and your pc, don't expect to get what the video gets. You may get better or worse results, no exactness is possible. It's not uncommon to see 1.4v on an 8700k at 4.9GHz.
 
  • Like
Reactions: Afro_ninja199

Afro_ninja199

Reputable
Aug 10, 2019
365
41
4,740
Follow, but don't copy. Your cpu, motherboard, ram, VRM's etc are different from the video, even if they look identical. So tailor the exact voltages to what works for you and your pc, don't expect to get what the video gets. You may get better or worse results, no exactness is possible. It's not uncommon to see 1.4v on an 8700k at 4.9GHz.


Thanks, I'll try to follow that guide

as karadigne said follow not copy, i follows the video but had to change a few small things now my 8700k is 5Ghz @ 1.28 temps sit around 60c while gaming
 

Phaaze88

Titan
Ambassador
as karadigne said follow not copy, i follows the video but had to change a few small things now my 8700k is 5Ghz @ 1.28 temps sit around 60c while gaming
What did you do to confirm stability of that overclock? I know they make improvements in refining the silicon over time, but that's a really low Vcore...
Coffee Lake All Core SSE FrequencyAll Core AVX2 FrequencyBIOS Vcore% Capable
8700K4.80GHz4.60GHz1.375V100%
8700K4.90GHz4.70GHz1.387VTop 99%
8700K5.00GHz4.80GHz1.400VTop 83%
8700K5.10GHz4.90GHz1.412VTop 49%
8700K5.20GHz5.00GHz1.425VTop 17%
8700K5.30GHz5.10GHz1.437VTop 4%
8086K5.00GHz4.80GHz1.400V100%
8086K5.10GHz4.90GHz1.412VTop 94%
8086K5.20GHz5.00GHz1.425VTop 65%
8086K5.30GHz5.10GHz1.437VTop 15%
That's Silicon Lottery's binning statistics for the 8700K. Even the 8086K, which is just a cherry picked 8700K, doesn't have that low of a Vcore recorded for 5.0ghz.
That's why I have to put into question what some people are actually doing for stability testing.
 

scherioyt

Reputable
Jun 17, 2018
41
0
4,530
I managed to get 4.7ghz all cores at 1.264V, i tested it with ibt with the very high preset and it passed. Should i test it with prime? And there's a thing that i had never seen before: hwmonitor and cpu-z are saying that my vcore is 1.264, but in the BIOS i set 1.275. Is there an explanation for that?
 

Phaaze88

Titan
Ambassador
Should i test it with prime?
Prime 95, Small FFT, AVXs Off, is more of a cooler thermal stability test, not a voltage stability one. The load it applies is heavy, but it doesn't bounce around like that of say, Cinebench R20 or Asus Realbench(V 2.56 latest) which are better suited for testing Vcore stability.

Is there an explanation for that?
Vdroop and Load Line Calibration.
Vroop is built into the cpu on the hardware level; you can't stop it. LLC was implemented to counteract it.
 
  • Like
Reactions: scherioyt

Phaaze88

Titan
Ambassador
cinebench passed well
Was that just one run, or did you do 'infinite loop'?
Infinite Loop: Open, click on File, then Preferences. Under Minimum Test Duration, enter some ridiculous number, like 9999999999999, and it'll default to the highest it offers.
Then run it for like an hour - although, the longer, the better.

In BIOS I set LLC on level 3 (1-5), is it ok in your opinion?
Should be. No one needs to use higher than the medium LLC settings unless using LN2, or the mobo VRMs are liquid cooled.
 
Solution

scherioyt

Reputable
Jun 17, 2018
41
0
4,530
Was that just one run, or did you do 'infinite loop'?
Infinite Loop: Open, click on File, then Preferences. Under Minimum Test Duration, enter some ridiculous number, like 9999999999999, and it'll default to the highest it offers.
Then run it for like an hour - although, the longer, the better.


Should be. No one needs to use higher than the medium LLC settings unless using LN2, or the mobo VRMs are liquid cooled.
It was just one run, i'm going to test it with the infinite loop, thanks for helping. I'm also thinking of delidding the cpu.