i7 6700k @ 4.7 GHz 1.39v safe?

twerkking23

Commendable
Nov 26, 2016
3
0
1,510
PC SPECS:
Processor: i7 6700k
Motherboard: Asus Z170-A
GPU: GeForce GTX 1070 FE
RAM: 16GB Kingston HyperX Fury DDR4 @ 2133Mhz
CPU Cooler: Corsair H100i v2
PSU: 650W EVGA 80+ Gold

I'm new to overclocking, but currently am running stable @ 4.7 GHz with 1.39 volts.

Under 100% load my CPU is around 75 degrees Celsius, running idle my CPU is around 25 degrees Celsius.

My motherboard is an Asus Z170-A and I had heard after finding a stable OC, I should set it to adaptive mode from Manual.

I tried that and my voltage was still 1.39 volts when running idle?

I thought the purpose of adaptive mode was to only use the voltage I suggested under heavy load?

I even tried setting the offset to 0.001 and still the same outcome.

With all that being said would it be safe to just run it on manual @ 1.39 volts, or is there a way to make it so adaptive mode actually works? Or should I bring the OC down a bit?

Any input is appreciated! Thanks.
 
Hi twerkking23 :) You should go here for a decent OCing guide: http://www.overclock.net/t/1570313/skylake-overclocking-guide-with-statistics

You have to know that the CPU has a table of voltages pre-programmed into it (the VID table).

The VID of your CPU is the default VCORE your CPU needs in order to run at stock to run Intel’s standards for stability and programmed into the core

As multiplier changes with load on the CPU it asks for more or less voltage. this would be the default AUTO behavior.
Also, at load the CPU is programmed to protect itself from spikes in voltage and Vdroop you may have heard of.

There are a number of ways you can change what voltage the CPU receives.

Offset: Applies a positive or negative value you define to the whole range of voltages in the VID table...from idle to load

Adaptive: Applies a positive or negative offset to only the turbo multipliers

Manual: overrides VID table and feeds CPU constant voltage even if multiplier varies with load!

LLC: overrides and helps with Vdroop so when CPU is under load it does not drop the voltage as much as it wants to.

Offset as I have said applies a value to the whole VID table. Sometimes when setting a negative value idle voltage becomes too low even though the negative value is appropriate for load. In this case adaptive is a benefit since it only applies the offset at load and leaves the idle state alone.

The actual voltage you get in CPUz can often be different from what you set or what you think you are going to get when you set offset. This is because LLC will kick in at different levels if you leave it on auto....and the CPU does some complicated calculating not only based on offset values but also taking into account thermal conditions current power draw..etc etc. i.e. it takes some executive decisions of its own based on your settings...so exact calculations especially with positive offset values is often not possible. getting to your target voltage is often a bit of trial and error as you will find.

Hope this helps.
 

twerkking23

Commendable
Nov 26, 2016
3
0
1,510


Thanks for the reply meanmachine41!

I read the guide and it was a big help!

I followed it and got the same results, which is good, stable at 4.7 GHz @ 1.39v

I also fixed the problem with the adaptive setting, I just had to turn my PC's performance preset to balanced.

My CPU now runs at around 0.6v when idle and goes up to 1.392v under load.

Not sure why it goes up to 1.392 when I have it set to 1.39 in the BIOS but that's okay I think.

Average 70-75 degrees when under 100% load.

Do these temperatures and voltages seem safe for the current overclock of 4.7GHz?

Thanks!

 
Well if t's stable and Max temps under load are OK at MAX 75c then yes, that is a decent chip and a decent OC.
The Overclock stabilty depends too on the stress tester you use and the amount of time you run the test.
To consider it is 24/7 stable and safe you should use either AIDA64 running CPU and FPU for 30 mins.
Another realworld tester is Realbench and if you can run that test and complete the benchmark, you can consider your system stable.
Bring down your Core voltage in small increments till its unstable, then raise it back up to when it was last stable and you may gain some temperature drop.
 

twerkking23

Commendable
Nov 26, 2016
3
0
1,510


Thanks for the response!

I tried to bring down the voltage in small increments for 4.7GHz and didn't find stability until around 1.39v, I then tired to find stability @ 4.6 GHz which I managed to do with 1.325v.

For just 100 MHz a voltage jump of 0.075 volts seems like a lot to me, so I decided to stick with 4.6 GHz.

I stress tested 4.6 GHz @ 1.325v in realbench and AIDA64 for 4 hours each, both passed. I then brought the voltage down to 1.32v and stress tested on realbench for 30 minutes, which passed. Then, finally, I brought it down to 1.315v and stress tested on realbench and the application crashed. So I bumped the voltage back up to 1.32v and am currently running a stress test in AIDA64 for 4 hours again, if it passes, then I'll do the same on realbench.

Current average CPU temp under 100% load is 53c, and maximum temp was 67c while @ 4.6GHz / 1.32v, one core reached 73c but that's it.

Does everything seem okay for the 4.6GHz OC?
 

brunor

Prominent
Jan 11, 2018
9
0
510


Just want to point out for anyone reading this thread, that completing the RealBench benchmark does NOT guarantee overclock stability. From my testing at least, you must run the RealBench stress test for at least 30 minutes, AND the benchmark.