Some newbie questions about 8700k OC and settings

[Grabob]

Hello friends,

Yesterday has arrived my new boy the 8700k and the Asrock Extreme 4 so i started to do some benchmarks and optimizing the settings. Keep in mind that my last big boy was the 4790k which i never even tried to go after 1.3V and 4.6GHz.

I noticed that the mobo (last bios) sets the automatic voltage of the core at 1.45V which i think it s insane infact i can t get on stress test without reaching 95+ on some cores, i decided then to set a fixed voltage of 1.2V and a 4.7 Ghz on ALL CORES and it seems pretty stable, i m running aida for about 2 hours and i get no more than 76°.

I read a lot of people complaining about the right voltage for this chip, a lot if them suggests using 1.36V but it seems pretty high to me. Am i missing something? Is it normal that the mobo auto-voltages the cpuat 1.4 V+? 

Are there any settings that are causing issues? 

 

I m sorry for these dumb questons but i read a lot of topics and i can t get any good (consistent) news fron them, i know that chips are not built in an equal way but i can t figure out if this is or not a good chip.

 

Thanks for the replies

[ZeouLs]

CPU's differ, everyone has a different sweet spot and some people get the golden lottery ticket cpu which runs by far better than the average. Maybe you were lucky and got one of those golden lottery tickets and that's the reaosn your chip is running so smooth with a lower voltage

[Fasauceome]

Lower is better if you're stable. No point in using more power for the same task.

[Grabob]

3 hours ago, ZeouLs said:

CPU's differ, everyone has a different sweet spot and some people get the golden lottery ticket cpu which runs by far better than the average. Maybe you were lucky and got one of those golden lottery tickets and that's the reaosn your chip is running so smooth with a lower voltage

 

2 hours ago, fasauceome said:

Lower is better if you're stable. No point in using more power for the same task.

Thanks for the replies, i will keep experimenting other settings and if anything usefull will be found i will update this post.

[Grabob]

Pardon me for the bad english .

Hello again, i haven't specified the cooler, i'm using a H80i GT and the Artic MX4 (i believe).

So i tested 4.8 on all cores using 1.25V because 1.2V was not enough and i get around 80° while running Cinebench, 

on idle i get like 36/37 on all cores except the second which is always a bit hotter then others.

The problem is that with Prime95 on the first test (the hotter one) the chip throttles down because temps get on the 95/100 °, i tried to reseat the cooler but temps improved for about 4 degrees on idle with no significant changes on load, the pump stays around 40 to 42 ° during stress tests, is there something wrong?

Maybe the TIM under the heatsink is not good enough (Thanks Intel, this is the second chip i found) but i don't want to delid until warranty expires.

Furthermore i have a question, CPU voltage and frequency stays at maximum during idle and i can't get to lower it, i changed the power plans option and set the minimum cpu load at 20% while maintaning the maximum at 100%, in the bios i have all intel's technologies ON and the LLC to level 3 (from 1 to 5 levels scale), any suggestions?
Thanks 

 

Edit:

Found out that for a realistic ish real world usage simulation is recommended to use Prime 26.6 because it does not use some AVX istructions (correct me if i'm wrong), anyways temps stabilize at 85° for most cores while core 2 stays at 93° 

[Fasauceome]

6 hours ago, Grabob said:

but i don't want to delid until warranty expires.

Your warranty is void after overclocking.

 

You would be throttling at 4.8 with a mere single rad, those sort of thermals would require a 240 rad at most. Gaming benchmarks will be more realistic, so the frequency should be fine in those workloads

[_Hustler_One_]

On 19/9/2018 at 8:28 PM, Grabob said:

Are there any settings that are causing issues? 

 

I m sorry for these dumb questons but i read a lot of topics and i can t get any good (consistent) news fron them, i know that chips are not built in an equal way but i can t figure out if this is or not a good chip.

 

Thanks for the replies

That was the Intel Turbo Boost in action. I also use i7 8700K, while I run it for the first time I also got the same experience, then I turn disable the Turbo and manually OC for 4.8GHz at 1.25v stable, and stress test it with Realbench for 30 mins was stable and temps was 79^oC max.

[Grabob]

On 20/9/2018 at 6:07 PM, fasauceome said:

Your warranty is void after overclocking.

 

You would be throttling at 4.8 with a mere single rad, those sort of thermals would require a 240 rad at most. Gaming benchmarks will be more realistic, so the frequency should be fine in those workloads

That's bad news, anyways i have to wait for the 1080 ti to arrive then i will see if the temps are acceptable.

My friend has a h100i and infact temps are lower but i don't understand why... okay long term benchmark should benefit from the way more dissipation area that the h100 offers but during idle when the pump is let's say at 34° there should not be any differences, the contact area of the pump is the same for both.

[Fasauceome]

3 minutes ago, Grabob said:

That's bad news, anyways i have to wait for the 1080 ti to arrive then i will see if the temps are acceptable.

My friend has a h100i and infact temps are lower but i don't understand why... okay long term benchmark should benefit from the way more dissipation area that the h100 offers but during idle when the pump is let's say at 34° there should not be any differences, the contact area of the pump is the same for both.

More water means more thermal mass, and more fins in the radiator means more surface area. Same reason larger air coolers cool better. Idle is a similar story.

[Grabob]

22 hours ago, _Hustler_One_ said:

That was the Intel Turbo Boost in action. I also use i7 8700K, while I run it for the first time I also got the same experience, then I turn disable the Turbo and manually OC for 4.8GHz at 1.25v stable, and stress test it with Realbench for 30 mins was stable and temps was 79^oC max.

Tried your suggestion but there is something i don't get, Boost technology allows you to basically overclock the chip infact if i disable it in the BIOS the freq. stays at stock (3.7Ghz) and the voltage lowers a lot (as expected), but if i manually set the core ratio to 4.8 it re-enables the boost by default giving me the same problems.

Anyways i found that using balanced plans in power option fixes the thing on idle but i still don't understand why selecting performance mode and lowering the minimum cpu usage to 5% does not archieve the same results...

[Grabob]

1 minute ago, fasauceome said:

More water means more thermal mass, and more fins in the radiator means more surface area. Same reason larger air coolers cool better. Idle is a similar story.

Oooh, i see, well i could upgrade the cooler in the future, don't quite like the idea of delidding.
Thanks sir for this information

 

[_Hustler_One_]

9 minutes ago, Grabob said:

Tried your suggestion but there is something i don't get, Boost technology allows you to basically overclock the chip infact if i disable it in the BIOS the freq. stays at stock (3.7Ghz) and the voltage lowers a lot (as expected), but if i manually set the core ratio to 4.8 it re-enables the boost by default giving me the same problems.

Anyways i found that using balanced plans in power option fixes the thing on idle but i still don't understand why selecting performance mode and lowering the minimum cpu usage to 5% does not archieve the same results...

Go to BIOS, make sure that you disable the Turbo after you manually OC, double check the status of Turbo Boost after you OC manually, if it automatically turns back on, turn that off again, so you'll have your own OC without being intervered by the Turbo Boost. That's how I did on my own.

[Grabob]

26 minutes ago, _Hustler_One_ said:

Go to BIOS, make sure that you disable the Turbo after you manually OC, double check the status of Turbo Boost after you OC manually, if it automatically turns back on, turn that off again, so you'll have your own OC without being intervered by the Turbo Boost. That's how I did on my own.

doesn't work for me:

If boost enabled i can change the clock with no problem

if boost disabled i can change it but after pressing enter boost reappears 

disabling it again set's clocks multipliers to AUTO 

changing them of my choice re enables boost ;v

 

Anyways i played around with c-states, deactivated C7 then in windows i have resetted the power plans default settings and set the maximum performance minum cpu power to 5% and now it works (???)

 

Dunno what to say, minimum voltage at 0.740V and 800 Mhz on idle, maximum on 1.3V and 4.8 even tho i set the bios @1.21V, LLC value to 2 which is the first non flat level it downvoltages to 1.26 on load.

 

Cinebench score 1577 temps from 76ç to 78° except core 2 which is always angry and ramps up to 82° 

[_Hustler_One_]

48 minutes ago, Grabob said:

doesn't work for me:

If boost enabled i can change the clock with no problem

if boost disabled i can change it but after pressing enter boost reappears 

disabling it again set's clocks multipliers to AUTO 

changing them of my choice re enables boost ;v

 

Anyways i played around with c-states, deactivated C7 then in windows i have resetted the power plans default settings and set the maximum performance minum cpu power to 5% and now it works (???)

 

Dunno what to say, minimum voltage at 0.740V and 800 Mhz on idle, maximum on 1.3V and 4.8 even tho i set the bios @1.21V, LLC value to 2 which is the first non flat level it downvoltages to 1.26 on load.

 

Cinebench score 1577 temps from 76ç to 78° except core 2 which is always angry and ramps up to 82° 

Idk how ASRock Intel Turbo Boost act, but my MSI motherboard Intel Turbo Boost can be enabled/disabled as you wish whenever you want whether you manually OC or not.

 

1.3v for 4.8GHz is too high. Actually i7 8700K can be set 4.8GHz at 1.22-1.27v, I guess.. while mine is locked at 1.25v for 4.8GHz

[Grabob]

1 hour ago, _Hustler_One_ said:

Idk how ASRock Intel Turbo Boost act, but my MSI motherboard Intel Turbo Boost can be enabled/disabled as you wish whenever you want whether you manually OC or not.

 

1.3v for 4.8GHz is too high. Actually i7 8700K can be set 4.8GHz at 1.22-1.27v, I guess.. while mine is locked at 1.25v for 4.8GHz

Wait, i just realized that i confused the VID with the Vcore.... 

VCore is @1.2V VID is 1.3V, my bad.

[Grabob]

Have some other questions..

This is what i get after 50 mins in AIDA

  

 

Anyways yersterday during a test aida gave me a couple of times HArdware Failure Detected and i kinda freaked out, i read that it may be caused by bad RAM stick so i ran memtest86 during the night but it reported 0 errors so i thought it could be lacking some voltage on the cpu, i then set it to 1.22V instead of  1.21V and the above graph is what i get, what do experts think?