6700K OC temps

[GEHORC]

Hi Guys.

 

I'm kinda sorry to start yet another 6700K OC thread, but reading other posts and the OC-rank-table got me even more confused.

 

My 6700K is currently overclocked to 4.6 GHz with the Voltage set to 1.28V in the BIOS.

Just to be clear: The OC is NOT Prime stable.

Side note: The voltage readings from CPU-Z and HWMonitor are completely off.

For cooling i have the Corsair H115i with Arctic Silver V TIM.

 

With this OC my CPU instantly skyrockets to 80°C when starting prime small FFT, while idle is around 30°C.

However the Corsair Link reading of the AIO fluid temp is around 36-42°C.

So the delta T of ~40°C must be anywhere between the DIE and the Pump/Waterblock.

It this particularly bad? Is this normal? Can i blame this to Intel's bad TIM between DIE and IHS? Should i just go ahead an delid the CPU?

My main problem with this is, that my ASUS Maximus VIII Hero instantly ramps up all fans to 100% when the CPU hits 80°C, which is quite annoying. (no this is not configured that way in Q-Fan (BIOS))

And of course i want to increase voltage to get it stable or even OC further.

 

Screenshots:

Spoiler

Thank you for your help.

[Lays]

Are you by any chance using some sort of adaptive voltage, or do you have the load-line calibration set to high?

[Samppa221]

Those temps are odd. I have 4,7GHz with 1.4V and my max temps are 80c after 1h of stress testing. Idle under 30c

[GEHORC]

9 minutes ago, Lays said:

Are you by any chance using some sort of adaptive voltage, or do you have the load-line calibration set to high?

TBH i don't know. I just set the core voltage to manual override an d 1.28V. (screenshot below and added to original post)

Spoiler

 

[Lays]

Just now, GEHORC said:

TBH i don't know. I just set the core voltage to manual override an d 1.28V. (screenshot below and added to original post)

  Hide contents

 

 

Can you look for LLC or Load-Line-Calibration setting?

[racer243l]

Strange temps. I have 4,6Ghz @1.344v and get 77°C stress testing with a H80i GT as exhaust (worst CPU cooler I ever had). In CPU-Z in your scrrenshots it shows a voltage 1.344v. So some kind of adaptive voltage must go on here

[GEHORC]

11 minutes ago, Lays said:

Can you look for LLC or Load-Line-Calibration setting?

So LLC is set to auto in the bios and can be set from 1 (maxiumum v-drop) to 8 (maximum v-overshoot).

Man that photo turned out crappy...

Spoiler

 

[Lays]

1 minute ago, GEHORC said:

So LLC is set to auto in the bios and can be set from 1 (maxiumum v-drop) to 8 (maximum v-overshoot).

Man that photo turned out crappy...

  Hide contents

 

 

Try setting it to like 5 or 6, then retry in windows and see what it does

[GEHORC]

25 minutes ago, Lays said:

Try setting it to like 5 or 6, then retry in windows and see what it does

Step 1:

Set LLC to 5: Prime instantly crashed Windows.

Step 2:

Set Voltage to 1.30V: Still crashed Windows. (Nearly bricked it. Had to restart several times do reach the login screen.)

Step 3:

Set Voltage to 1.32V: Didn't crash Windows anymore but core 4 (worker 7) failed after just one minute. Voltage hit 1.429V according to HWMonitor.

 

After Step 3 i was back at 135W TDP and the fans ramped up to 100% again with the CPU reaching 76°C.

I guess, if i raise the voltage further to reach the 140W TDP i had before, it will also hit 80°C again.

[Djxinator]

1 hour ago, GEHORC said:

Step 1:

Set LLC to 5: Prime instantly crashed Windows.

Step 2:

Set Voltage to 1.30V: Still crashed Windows. (Nearly bricked it. Had to restart several times do reach the login screen.)

Step 3:

Set Voltage to 1.32V: Didn't crash Windows anymore but core 4 (worker 7) failed after just one minute. Voltage hit 1.429V according to HWMonitor.

 

After Step 3 i was back at 135W TDP and the fans ramped up to 100% again with the CPU reaching 76°C.

I guess, if i raise the voltage further to reach the 140W TDP i had before, it will also hit 80°C again.

Unless you're using adaptive voltage your CPU shouldn't be jumping 0.109v under Prime95

 

Theres no way LLC creates that level of Voltage jump.

 

Go back into your BIOS, reset all of your settings and start from scratch. Try a 100mhz bump on stock volts and go from there. 

 

Every time you fail a test or crash, bump the voltage up 0.01v and try again.

 

If you're prime stable for 30 mins, bump the multiplier up and repeat the process.

 

EDIT: Doing the above actually SAVES TIME. I learned this the hard way when I had to reset the CMOS after a week of tuning because of a failed overclock.

[GEHORC]

I found another setting:

>>CPU SVID Support<<
Disable this item to stop the CPU from communicating with the external voltage regulator.

So i disabled this setting.

As a side effect i don't get voltage nor power readings directly from the CPU anymore (In HWMonitor).

But according to the motherboard reading the maximum voltage was 1.344V after I manually set the voltage to 1.34V (still @4.6GHz).

With these settings the CPU successfully completed the 12k and 8k iterations of prime (yeah i know that's only 10 minutes) .

 

But this leads me to the original problem.

CPU temps are now hovering around 80°C again with a water temp of <40°C.

The voltage is set much higher now, probably it cranked itself to the same level before.

So 1.34V and 4.6GHz with 80°C on a H115i. Is this normal now? I still consider it too hot.

[Imakuni]

3 hours ago, Samppa221 said:

Those temps are odd. I have 4,7GHz with 1.4V and my max temps are 80c after 1h of stress testing. Idle under 30c

3 hours ago, racer243l said:

Strange temps. I have 4,6Ghz @1.344v and get 77°C stress testing with a H80i GT as exhaust (worst CPU cooler I ever had). In CPU-Z in your scrrenshots it shows a voltage 1.344v. So some kind of adaptive voltage must go on here

Download this and fire the Small FTT test, which I what OP used. Let's see if temps still get as cool as you claim.

 

51 minutes ago, Djxinator said:

Unless you're using adaptive voltage your CPU shouldn't be jumping 0.109v under Prime95

 

Theres no way LLC creates that level of Voltage jump.

Actually, yes it can. Mine shoots up from a 1.24 undervolt (at min lvl) to 1.35 super overvolt (at max), for example.

[GEHORC]

Bump.

Still not sure, if the temps are ok for the new voltage at the given frequency.