Build:
i9-10850k
EK-AIO 360 Elite Rgb (the brand new one that uses a push-pull config on a 360 rad - i believe its aluminum, 30mm thick, with 18fpi)
- set up as intake
MSI MEG z490 Unify
- 16+1 power stages (90A rating)
G Skill Trident Z Neo - 64gb, 3600mhz, CL16 (pretty sure it's b-dye)
- (XMP is enabled on all tests, plan to OC later once I get the cpu to where i want it)
2tb Samsung 970 evo plus
1000W EVGA Supernova
GeForce gtx 1080 ti (yeah I know it doesn't quite fit in with the rest of the build lol, its a relic from the previous one, planning to replace this in the spring time with a 3080/3090)
Built this computer a couple of weeks ago and finally got the AIO in the mail yesterday; put it together last night and was finally ready to start overclocking it. Not knowing a whole lot about what I was doing, I decided to give Intel Extreme Tuning Utility a try with some numbers that I knew worked in a youtube video I saw with identical everything, except with a 10900k .
[test#1] 5.1ghz on all cores, 4.7ghz cache, with 1.4v vcore with Intel Extreme Tuning Utility.
- LinpackExtreme - 10gb. Ran it 15 times and it was stable. It was running hot, as expected, but it was stable. 95c max with each taking about 62 seconds to complete.
- Prime95 with 32K FFT's (what I'd seen buildzoid run alot of tests on) - Ran this for 20 mins and I got the same results as Linpack. About 95c max.
- Monitored it with cpu-z and hwinfo. What was alarming to me about using Intel extreme tuning was that that it was letting the vcore go well above 1.4v, sometimes over 1.5v
- I ran cyberpunk for 2 hours and it was sitting around 70c, hitting a peak of 85c.
I was quite pleased with the results but I didn't like how high the vcore was going and I wanted to push it further. Given the results that I'd gotten, I came up with my goal for this overclocking venture:
1) Figure out how to do it all manually through the bios and figure out the voltage issue
2) Hopefully try to push the cpu to 5.2 ghz. I want to get it as close to a 10900k as I can in terms of performance.
3) Once I figure out what the limit of this bin is, drop it down to a solid 4.9/5.0ghz, figure out how to undervolt it as much as possible and hope that would be enough to get the thermals below 90c on stress tests for daily usage.
So, I set out and watched a bunch of videos online and tried some settings in the bios.
[test#2] 5.1ghz on all cores, 4.7ghz cache, 1.4vcore, mode 4 llc (first one i've tried, before this, when I was using extreme tuning, this was set to whatever it as at default - auto, i think), left the rest of the voltage settings on auto, foolishly set avx offset to 0. This test failed. CPU temps spiked to 100c and the system crashed
. I figured it was because I changed the avx from auto to 0 but I was wrong about that.
[test#3] Changed avx offset from 0 to -3, temps spiked to 100c, it didn't crash immediately but I didn't want to wait to find out if it would, and i ended the stress test and reset.
[test#4] Set avx offset from -3 back to auto, lowered vcore to 1.37, lowered the frequency to 5.0ghz on all cores, 4.6ghz on cache and and tried again. I can get Linpack to pass but barely, it will still hit 100c on 2 of the cores and that's just not gonna be stable.
Andd that's as far as I've gotten. It's idling around 33c with these settings. I can't figure out what the difference is between what Intel's extreme tuning utility was doing for me and what I did. And that's the main reason I came here. Why can I get 5.1ghz to work on extreme tuning utility and can't even get a stable 5.0ghz stable manually in bios?? The biggest difference I can notice is that the voltage isn't spiking high like it was before but that wouldn't make sense to me; less voltage should equate to less wattage and lower thermals, right? The only other difference I can spot is that Linpack did complete significantly faster in test#4 than in #1 - it took 54 seconds to complete as opposed to 62 seconds in test#1 - so, I'm clearly getting more performance for the increased thermals but I have no clue why the lower clock speed in bios is performing better and creating so much heat.
tl;dr, to summarize what I'm asking:
I'm hoping you guys can help shed some light on the differences between extreme tuning utility's settings and setting them manually, why I can get a stable overclock with 1 and not the other. Or rather, what did I do wrong? As well as maybe some help in achieving the 3 goals listed in the 3rd paragraph. Thank you all for your time and sorry this thing turned out so long, I just like to try to be thorough.
i9-10850k
EK-AIO 360 Elite Rgb (the brand new one that uses a push-pull config on a 360 rad - i believe its aluminum, 30mm thick, with 18fpi)
- set up as intake
MSI MEG z490 Unify
- 16+1 power stages (90A rating)
G Skill Trident Z Neo - 64gb, 3600mhz, CL16 (pretty sure it's b-dye)
- (XMP is enabled on all tests, plan to OC later once I get the cpu to where i want it)
2tb Samsung 970 evo plus
1000W EVGA Supernova
GeForce gtx 1080 ti (yeah I know it doesn't quite fit in with the rest of the build lol, its a relic from the previous one, planning to replace this in the spring time with a 3080/3090)
Built this computer a couple of weeks ago and finally got the AIO in the mail yesterday; put it together last night and was finally ready to start overclocking it. Not knowing a whole lot about what I was doing, I decided to give Intel Extreme Tuning Utility a try with some numbers that I knew worked in a youtube video I saw with identical everything, except with a 10900k .
[test#1] 5.1ghz on all cores, 4.7ghz cache, with 1.4v vcore with Intel Extreme Tuning Utility.
- LinpackExtreme - 10gb. Ran it 15 times and it was stable. It was running hot, as expected, but it was stable. 95c max with each taking about 62 seconds to complete.
- Prime95 with 32K FFT's (what I'd seen buildzoid run alot of tests on) - Ran this for 20 mins and I got the same results as Linpack. About 95c max.
- Monitored it with cpu-z and hwinfo. What was alarming to me about using Intel extreme tuning was that that it was letting the vcore go well above 1.4v, sometimes over 1.5v
- I ran cyberpunk for 2 hours and it was sitting around 70c, hitting a peak of 85c.
I was quite pleased with the results but I didn't like how high the vcore was going and I wanted to push it further. Given the results that I'd gotten, I came up with my goal for this overclocking venture:
1) Figure out how to do it all manually through the bios and figure out the voltage issue
2) Hopefully try to push the cpu to 5.2 ghz. I want to get it as close to a 10900k as I can in terms of performance.
3) Once I figure out what the limit of this bin is, drop it down to a solid 4.9/5.0ghz, figure out how to undervolt it as much as possible and hope that would be enough to get the thermals below 90c on stress tests for daily usage.
So, I set out and watched a bunch of videos online and tried some settings in the bios.
[test#2] 5.1ghz on all cores, 4.7ghz cache, 1.4vcore, mode 4 llc (first one i've tried, before this, when I was using extreme tuning, this was set to whatever it as at default - auto, i think), left the rest of the voltage settings on auto, foolishly set avx offset to 0. This test failed. CPU temps spiked to 100c and the system crashed
. I figured it was because I changed the avx from auto to 0 but I was wrong about that. [test#3] Changed avx offset from 0 to -3, temps spiked to 100c, it didn't crash immediately but I didn't want to wait to find out if it would, and i ended the stress test and reset.
[test#4] Set avx offset from -3 back to auto, lowered vcore to 1.37, lowered the frequency to 5.0ghz on all cores, 4.6ghz on cache and and tried again. I can get Linpack to pass but barely, it will still hit 100c on 2 of the cores and that's just not gonna be stable.
Andd that's as far as I've gotten. It's idling around 33c with these settings. I can't figure out what the difference is between what Intel's extreme tuning utility was doing for me and what I did. And that's the main reason I came here. Why can I get 5.1ghz to work on extreme tuning utility and can't even get a stable 5.0ghz stable manually in bios?? The biggest difference I can notice is that the voltage isn't spiking high like it was before but that wouldn't make sense to me; less voltage should equate to less wattage and lower thermals, right? The only other difference I can spot is that Linpack did complete significantly faster in test#4 than in #1 - it took 54 seconds to complete as opposed to 62 seconds in test#1 - so, I'm clearly getting more performance for the increased thermals but I have no clue why the lower clock speed in bios is performing better and creating so much heat.
tl;dr, to summarize what I'm asking:
I'm hoping you guys can help shed some light on the differences between extreme tuning utility's settings and setting them manually, why I can get a stable overclock with 1 and not the other. Or rather, what did I do wrong? As well as maybe some help in achieving the 3 goals listed in the 3rd paragraph. Thank you all for your time and sorry this thing turned out so long, I just like to try to be thorough.
Last edited:
