cancel
Showing results for 
Search instead for 
Did you mean: 

Extremely high Vcore voltage (4 V) on Prime Z270-K with 7700K

dirtyred
Level 7
Hello!

Yesterday I was doing some OC stress testing my 7700K with Realbench. I'm using AVX offset since my cooling is not (yet, soon getting a better one) enough to handle high multipliers and voltages. Applied a multiplier of 47 and AVX offset 3 with an adaptive voltage of 1.25 V, no offset. In RealBench since it's using AVX the CPU was running on 4.4 GHz @ around 1.152 Vcore, occasionally jumping to 1.6 V. Since it wasn't stable I raised the LLC level from 3 to 4 (or from 4 to 5, can't remember exactly but definetly one of those). IA AD/DC load line is set to 0.01. Pretty much everything is set as Raja@ASUS wrote in the Kaby Lake overclocking guide. Now my Vcore was around 1.6 V with occasional spikes of 1.8 V but still unstable. Under non-AVX load the Vcore was pretty stable at 1.248 V.

Did not want to raise the Vcore above 1.25 V since above it my CPU is reaching 80-85 degrees under RealBench. So I decided to play around with the offset. Lowered the Vcore to 1.215 V and applied an offset voltage of 0.04 V resulting in Total adaptive mode CPU core voltage of 1.255 V. Rebooted, did 15 min RealBench stress test, passed it and the Vcore was 1.248 V stable.

Today I turn on the computer, do some easy stuff like browsing the internet, reading forums. Just out of curiosity I take a look at HWiNFO (with is autostarting at Windows startup and has a polling rate set to 250 ms) and I see Vcore current 0.112 V, min 0 V, max 4.064 V! :eek:

63845
63846

I took a screenshot and instantly rebooted to get into BIOS. Removed the offset voltage and now after 2 hours of monitoring my voltages seem to be perfect. Min 0.784 V, max 1.2 V while browsing.

I'm no expert but I don't think that 4 V of Vcore is in the safe zone yet alone it fits Intel's specifications. I don't know for how long and how many times did it spike to 4 V but at least once it did. It shouldn't have happened. I don't know what to think of: HWiNFO is inaccurate, offset mode is not working as it should, the VRM's are bad, or I'm doing something wrong.

Also with optimized defaults the Vcore is around 1.152 V under load in RealBench but it crashes in less then 10 minutes. Need to raise the Vcore to be stable even at stock frequency.

I'm not expert but this doesn't seem right.
767 Views
7 REPLIES 7

Zka17
Level 16
He-he, 4V would have smoked your CPU for sure...
I would say that something is interfering with sensor polling... usually that is the case when you have multiple softwares to check the same sensors...
Make sure that you are using only one monitoring software!

I am using only HWiNFO. SpeedFan is also running but all voltage monitors have been turned off, it monitors only CPU and GPU temperatures and fans speeds. I was guessing that 4 V would have fried the CPU but maybe it barely survived as it was only a short enough peak to get noticed by the sensors?

Also what could be the cause of RealBench crashing with completely stock settings and the CPU needing voltage increase? Is it more possible that the CPU is not requesting enough voltage or that the motherboard is not supplying enough?

jab383
Level 13
That CPU would be a toasted brick if it got 4.06 volts. Sensors glitch every now and then - give false values for one reading then get back to reality. You might try a monitoring program like Aida64 that includes a running graph you can check back through.

The need for higher than default volts to pass a stress test is just normal variation between CPU samples. 1.15 is quite low and your LLC setting would lower it farther in the test. A small (say to 1.17, but not to 1.2) increase in setting would be normal. High temperatures can cause instability. Waiting for the better cooling gear would be worthwhile.

Zka17
Level 16
Whenever you have installed multiple softwares polling the same data, conflicts (like aberrant reads) can happen... you don't have to run them, just have them installed... - if you know what are you doing in BIOS and the temps are fine, I wouldn't worry about it...

If RB is crashing then your system is unstable... which part of it? well, that's where all the fun starts... you will have to figure it out...

So I shouldn't worry about that scary 4 V and just ignore it as a glitch? Scared the hell out of me.

It's really odd that you have to increase voltage for stock CPU in order to pass a test. Should be rock solid. Maybe the QC requirements are pretty low at Intel these days.

I had couple of monitoring stuff at the beginning like CoreTemp, HWMonitor, CPU-Z, Aida64, AI Suite 3 and now even RB has temp monitoring but the only software that's allowed to startup is HWiNFO right now. All services of Asus have been disabled just to be sure. Maybe I'll do a fresh install and see if that solves the issues at stock speed.

Usually LuxMark crashes around the 5-8 minute mark running RB stress test. If it passes that, usually it passes 2 hours as well. It's odd because it's a GPU workload but increasing CPU voltage solves the instability. My GPU is stable even after 15-20 minutes of FurMark at 1920x1080 with 8x MSAA and overclocked to 2075/9400 MHz so I ruled it out but for RealBench I even reverted it to defaults.

When I'm using AVX offset then the voltage for that is too low. That's the reason I started applying offset voltage for while lowering the adaptive turbo. Will try it again with the offset and hopefully no blue smoke followed by the smell of Amps will come out of my system.

I never let my temps go above 80 degrees average or a spike once in a while to 85. My voltage limit is 1.25 V right now but will stress it when my Phanteks PH-TC14PE cooler arrives. Hopefully will bring temps down 15-20 degrees compared to my little Arctic Freezer i32.

Honestly I don't trust the sensors. I've read it at couple of places that the temp readings of the BIOS and of AI Suite are not from the sensor on the CPU but from somewhere near it on the MB and those temps are off by 13 degrees. Could be some issues with the voltage sensors as well.

Was a cheap motherboard (for Z270) so maybe I shouldn't complain but I'm still a bit disappointed. Maybe my expectations were a bit too high especially towards Intel as well.

Thank you for the informations!

giukj
Level 7
I used to have the same issue with Z370 TUF gaming plus, even on newest BIOS 0615.
it's because of Windows Power Plan. When set to "High Performance" plan then the issue is gone, but in "Balanced" plan then the issue comes back

AndreyBeard
Level 7

This is very old thread, but... I have this voltage now in BIOS of P5N32-e sli plus... Plus 3,3 V reads 5 V, 12 V reads 16 V. But I can enter into BIOS still.