9700k overvolting and high temps - How do I stop it?

Associate
Joined
14 Jan 2013
Posts
77
Evening all,

My 9700k has started overvolting lately,
Despite having adaptive voltage set at 0.060, it is overvolting way too much on occasion.

it has only started doing it lately, notably when I play total war three kingdoms, and only on the load in screen for battles for a couple of seconds once the laoding bar reaches around 90%.

I have also noticed the fans ramping up quite a bit on boot up, but there is no way to get a temp reading that I know of on boot up, as it is again only for a brief period.

OC is 5.1ghz, cooled on a new alphacool eisabear 360,
c states enabled.
Intel speed shift enabled.
Adaptive voltage at 0.060
Bios f9a
Gigabyte pro wifi z390 mobo
32gb 4000 mhz teamgroup ram.
Zotac amp 1070 gpu.
WD Black 1tb NVME.

What is going on here?

CPUID HWMonitor is what I am getting my readings from.
CPUVcore is getting up to 1.4- 1.5V and temps can peak at 90c sometimes.
Doesnt seem healthy to me hence why I am asking here.

Thanks in advance folks.
 
Last edited:
OP updated with mobo and BIOS,
I havent actually thought about using a fixed voltage as I'd rather have an adaptive setting to lower power draw when idling etc...

Vcore is it 1.200 v normal
 
Last edited:
I have put load line calibration on power saving and cpu vcore calibration on low, set adaptive voltage at 0.100 over 1.200 normal, tjmax at 81c.
Still getting voltage readings at around 1.350v on loading into battles in tw3k.
So LLC in power saving seems to have dropped a bit of voltage.

Will be ordering some crynaught thermal paste this weekend and see if that helps at all as well.
Thanks for the help, sorry for the late reply but, real life and all that ****.
 
It could be AVX instructions in certain games/apps which boost voltage when Adaptive mode is enabled, which is why I suggested testing Fixed voltage to determine if it's that.

Thank you, will test and report back.

EDIT:..... Won't even boot on manual with 1.4v.... I may need to go back to the drawing board here...
 
Last edited:
Back to adaptive voltage:


5.1 ghz
Intel speed shift - enabled
C states - enabled
AC/DC LLC - power saving
CPU vcore LLC - Low
AVX offset 2
Adaptive voltage 1.200 normal + 0.040 adaptive

Boots fine, a quick benchmark in tw 3k fine, HOWEVER:
Prime95 stress test... the workers started stopping saying 1 error no warnings.

Max voltage hit 1.308 volts / Max temp on package 85.c, core 3,4,5,7 hit 100c!!!
but due to the avx being on 2 now I'm not sure how this affected the benchmarks.

Sorry everyone I'm not the best at overclocking as I only understand the basics...
and gigabyte BIOS is... difficult to understand sometimes.
Why the hell it wouldnt boot at manual voltage I dont know, was I missing something?
 
Last edited:
Will try in the next couple of days and if not will try on sunday, should i reset everything to default and just use adaptive with no modifier first then?
Also to note, I installed new BIOS as I noticed F10 is out now, and simply put in an overclock of 5.1 with everything else on auto and it booted fine, forgot to test anything then though because it was just a test after installing the new bios.
 
Sounds good. Just have to keep in mind that Adaptive will boost voltage more than Fixed, with AVX, and account for it when setting voltages. At least that's what seems to be the case going by the fact the Vcore did not shoot up as much last time on Fixed. I'd keep things slightly over 1.3v max (not in BIOS but what you see after running Prime with AVX or Cinebench R20 that has AVX). If you can keep temps 80C or under with Cinebench R20 that's a better indicator than Prime of what'll be good for 24/7 usage.

So would you reccomend using avx with adaptive?
 
Ok new settings seem to be a success:

5.1 ghz
Intel speed shift - enabled
C states - enabled
AC/DC LLC - power saving
CPU vcore LLC - Low
AVX offset 3
Adaptive voltage 1.200 normal / -0.005 adaptive

6CyqGnD.png

Thoughts @Danny75 ?
AVX seems to be very heavy handed though as I am not seeing 5.1ghz very often, even in tw 3k, haven't had time to test any other games though.
Are there any other AVX settings in the gigabyte bios I may have missed?
 
Last edited:
Try MCE (Multi Core Enhancement) or whatever the setting is to set "all cores" to 5.1 (other than when AVX running lowers them to 4.8). I think the lack of MCE right now could be the explanation for not seeing 5.1 more often.

Also, you're still hitting 100C on a 360 AIO, with 1.2 Vcore... that's weird and I got much better results on a 9700K at 1.224 Vcore with an air cooler (Dark Rock Pro 4). Recommend using HWiNFO64 to double check the readings, and also try out Cinebench R20 to see max temps with that.

Thanks, I have tried Hwinfo but never liked it before, will give it a go on sunday as I am away for a few days from tomorrow, also will be ordering some new thermal paste.
 
@Danny75

zBlKlsS.png

Nt14cVF.png

Fresh Bios, no adaptive, everything on auto except 5.1ghz overclock and 32gb memory at 18 18 18 38 4000mhz... still high temps and almost instant themral throttling even after re-applying themral grizzly.

I am at a loss...
 
I did suggest you set a Vcore of 1.2v (for example) rather then leave absolutely everything on Auto. Did you try that? On Auto your voltage is spiking to around 1.37-1.39v which would easily explain the high temps.

To be clear, set 1.2v* Vcore, and Adaptive, and test.

* You may need more than 1.2v to maintain those clocks.

At least all cores now seem to be hitting 5.1? Which was an issue last time? Can you confirm that?

Hi,

Tried a bunch of stuff, managed to get fixed working @ 1.325 for 5.1ghz, 1.275 for 5.0ghz, and also adaptive @ 0.025, however adaptive seems to run cooler and c-states run more often, but maybe I'm just imagining that one.
LLC seems to be the key, using turbo to minimize vdroop and prevent any big voltage spikes.
Cinebench is fine to run, temps get to around 80c max. however with prime, still getting massive temp spikes, and also workers stopping and instant thermal throttling once it really kicks in, even with avx on.
AVX is still being weird though, it seems to have my cpu permenently running at 4.8ghz when doing nothing in windows, with the occasional jump to 5.1 here and there.

With MCE on/off doesn't make a difference to how avx is working, and my google fu isn't turning up any definitive answers on it.

As for the temps issue....
I may just not have a good chip I guess, although why I have only just noticed this now I don't know. Has been running fine since christmas until recently.


EDIT: This my explain the AVX issue https://www.reddit.com/r/overclocki...out_avx_offset_with_gigabyte_z390_ud/eautlus/
 
The above is using blend of all options in prime 95.

5ghz oc, adaptive voltage with +0.000
LLC power saving.
C states on.
AVX off.

No failures this time at least. Maybe I was just pushing it too much at 5.1ghz.
Those temps are crazy though....
 
Using CPU-z what is the voltage going to the CPU when it is being stressed by Prime and running at 5ghz.?
Is that voltage stable or does it fluctuate.?

With temps like that at 5Ghz then it is not going to keep at that speed without throttling, but you know that.

What do you mean by AVX off, it is either auto or a numeric value .?

The voltage fluctuates slightly, but will usually hang around the 1.23 - 1.27 mark:

Note this is using the best setup I have found so far for stability when using prime 95 as a stability test.

Internal AC/DC LLC : power saving
CPU Vcore LLC : low
Adaptive voltage : 1.200 normal / +0.020 adaptive.
C states all enabled.
Intel speed shift enabled.
Multi core enhancement : auto

Avx seems to have no impact, as soon as prime95 kicks in heavily the CPU begins to thermal throttle.

I have also noticed the temps fluctuate quite a bit when just in windows ranging from 33c - 60c.

I have an old asetek aio that i am going to put back in. As I really cant see why I am getting these results with those voltages!

Any other combination of settings I have tried are either not stable or introduce way more voltage than I am comfortable with.

Edit: AVX is either auto or a numeric value.
 
I have read of a number of people who simply avoid using Prime with their 9xxx CPU's. I don't bother with my 9900k.
My dynamic Vid is set to -0.45v and the LLC at low. That seems to offer 1.296 steady at 5Ghz for my 9900k.
I leave AVX to auto, otherwise it generally gimps the CPU.

If you have that low voltage under load your temps are higher than I would have expected. I'm surprised that 0.20 added to your adaptive equates to 1.23 - 1.27v, thought that it would be higher than that.

Have you used CPU-z like I asked, it is said to report voltages for the CPU quite accurately.

Yes I have used CPU-Z.
I will test again and confirm tonight.
 
Use and leave open CPU-z whilst running Intel Burn Test v2. Let that loop whilst you watch for voltage fluctuations within CPU-z, cross check that with the values you are reading withing HW Info.

Have you a link for the DL for that?
I have heard that it can really stress things to breaking point... which I dont want to do.

Currently trying realbench, as it has been suggested as more of a "real world usage" benchmark, and all seems fine, if somewhat warm.

Will be getting the old asetek 360 aio hooked up tomorrow to see if that solves the heat issue.

The only thing I can think of after this is delliding and applying the heatsink straight onto the die?

Ed79BcO.png
1aac401e-f918-43b2-80c4-e305e3e3ae4e
 
Last edited:
That is pretty toasty at 4.8ghz. But then again look at the max voltage, around 1.34v.does it need that at 4.8ghz..?

That is with an avx of 2 on the 5ghz overclock.
I used your setting of -0.045 adaptive.
Low LLC.
Uncore at 4.7.
C states on or off doesn't matter.
Voltage is weird as hell, will try reducing it further tonight.

What could be wrong here?
Bad chip?
Bios being wonky?
 
Warm!!! Havent measured but I'm guessing it's around 20 - 22c in this weather at the moment.

I've also tried turbo myself but it shoots voltage up way to high from my testing so far.

Fixed voltage may be the way to go but it gets even warmer with that on, but with less fluctuation in voltage.

EDIT: AVX offset is only on to limit thermal throttling in benchmarks. I will turn it off once I've figured out what's going on.
 
But why the AVX offset..?

ok read your reason from your edit.

I prefer to have that set to 0 and then work from there. It should not be thermally throttling, mine used to, if you get the voltages and the cooling sorted.

I've read somewhere that vcore LLC at normal is the most stable so I'll try that tonight.
 
Update:

I have finally found a fixed voltage & LLC combination that works:

1.295 & LLC high

Highest vcore has been 1.284 so far.

Prime has package temps around 90c after half an hour of stress testing, no cores have throttled and no errors.

So it seems that adaptive has been the cause all along, temps and volts are much steadier with this setting. Been a nightmare to find this combination of settings as no matter what I did with fixed I was getting bsod or crashing in p95.

I have also however changed the aio to another older 360 asetek, but before I tested with fixed, I was getting similar results on adaptive with it. Will test for another hour, and refit the alphacool aio tomorrow again and see if temps are similar to my current using fixed core.
 
Back
Top Bottom