i7-6700k Overclocking voltage

[Airdragonz]

Hey all, so I was trying to oc my 6700k on my Asus z270 mobo. In bios, the core voltage is set at 1.39v in adaptive mode and a level 7 LLC. Once I booted, CPU-z showed voltages fluctuating between 0.74 volts and 1.434 volts. From my understanding, the "safe" 24/7 voltage for this chip is around 1.4v. Should I be concerned about a too high voltage even though it does not constantly hold at 1.434 volts? 

[Tabs]

42 minutes ago, Airdragonz said:

Hey all, so I was trying to oc my 6700k on my Asus z270 mobo. In bios, the core voltage is set at 1.39v in adaptive mode and a level 7 LLC. Once I booted, CPU-z showed voltages fluctuating between 0.74 volts and 1.434 volts. From my understanding, the "safe" 24/7 voltage for this chip is around 1.4v. Should I be concerned about a too high voltage even though it does not constantly hold at 1.434 volts? 

1.4 is fine if your cooling is good. I'm running at 4.6ghz, 1.29v with LLC on "Extreme" on my GA-Z170-XP-SLI. Under load, it gets around 1.34v.

 

So, being worried about the voltage is good, but it's not the voltage itself you need to be worried about, it's the spikes caused by having LLC enabled that can do damage.

 

The issue to be aware of is that vdroop is technically a feature, not a "bug", if that makes sense, so LLC in bios is potentially a very dangerous setting. vdroop is there to prevent inrush voltage when load drops from being super excessive. It takes a non-zero amount of time for the voltage regulators to update once load (and therefore current) changes, so when the cpu goes from full load to no load, the previous voltage (say 1.4@50A for around 70w of power) will shoot up super high since the VRM is still trying to deliver 1.4@50A. It takes time from the load dropping and the voltage being correctly lowered again, which causes a very sharp spike of high voltage.

 

Too many of these spikes and your cpu stops working as effectively or can outright die. It accelerates electromigration. 

 

If you're happy with your temperatures and know the risks, you're fine to stay where you are.

 

Can I ask what temperatures you get currently and what cooling you use?

Edited July 18, 2018 by Tabs
Formatting

[Airdragonz]

52 minutes ago, Tabs said:

Can I ask what temperatures you get currently and what cooling you use?

Thanks for your input. My current temps are 75C under load and I am using a custom liquid cooled loop

[Tabs]

2 minutes ago, Airdragonz said:

Thanks for your input. My current temps are 75C under load and I am using a custom liquid cooled loop

That should be absolutely fine then mate. Keep an eye on your stability, perhaps re-run a stability test every 6 months or so, since if your current OC starts to not be stable anymore it could indicate some underlying damage to your processor.

 

In reality though, it's a good board and you're set up very well with temps like that, so it should last a very long time.

[Marinatall_Ironside]

I overclocked my brother's i7-6700K to 4.6 GHz using 1.39V VCore (4.7 was my target, but alas it's not stable, even with 1.425V VCore). CPU is liquid-cooled via a custom loop, and package temperature rose to 78º, but I saw it consistently run in the low-70's for extended duration in AIDA64, according to HWiNFO64.

 

The board he has is not exactly the best (half-decent fan control, and pretty anemic VRMs); it's the MSI Z170A GAMING PRO CARBON. According to Play3r, it uses a 5+2+1 Phase configuration (5 for VCore, 2 for iGPU, and 1 for IMC/PCIe) with 2 Hi-Sides and 2 Lo-Sides per phase, the one phases halves the Hi and Lo-side IC count to 2 for that one phase. Obviously I prefer Asus boards, since from my experience, they're excellent on component selection, easy to work with regarding BIOS, and have a solid lineup of features. For this CPU, I would have liked to have worked with the Asus ROG MAXIMUS VIII HERO, which means I could have squeezed out that extra 100 MHz, or lowered the VCore voltage a little further to drive down both thermals and power consumption, because the VCore voltage determines those two metrics.

 

I don't have my own thermocouples and thermocouple reader setup to tape a K-type thermocouple to one of the phases, so I can't give you a scientific measurement of the temperature. But from a non-scientific standpoint, 9 out of 10 times I touch the top heatsink I say it's not going to give me second-degree burns if I left my fingers there for 1 minute.