6700k insane load temperatures

[Cherry Dude]

Hi there

So just finished my new build and already in trouble.

• 6700k 4 GHz
• Cryorig C7
• Asus Z170i PRO Gaming
• 32 GB G.Skill Ripjaws V 3000
• Asus GTX 1080 Strix A
• Samsung EVO 850
• Corsair SF450

I have fine temps and no trouble, but then when I fire up a game... (Screenshot from after shutting down game, but see max)



100 C! Now you may call me a prude, but that's just a little too much for me.
Bear in mind this is with absolute stock default settings, no XMP or other fiddeling.

Any idea what is going on? Cooler spins, used pea-sized application of paste, have not forgotten to remove plastic cover. Only thing I can think of is remove and reapply cooler and paste, but the temperatures just seems too excessive for this to be the cause.

 

[Shintai]

That voltage is rather high and not stock. And I have a feeling your motherboard is auto overclocking.

Try see if you can find something called Multi Core Enhancement and disable it.

Your cooler seems to be able to handle 100W TDP. So stock clocks and stock voltage should work out fine. However are you sure your cooler runs at speed 100% when under that load? That means 2500rpm or so.

CRYORIG | C7

 

[Cherry Dude]

You have a point regarding auto overclock, let me just check that out.
The cooler runs at 100% (indeed 2500RPM) at 70C.

 

[Cherry Dude]

Well you definiatly seem to be on to something Shinati.
Unfortunatly I have come to the conclusion that I am now completely ignorant in all things BIOS. Last time i fiddled around with settings was perhaps 5 generations ago, so it all seems more or less to me now.
I did however try a few settings, whivh gave me approximatly 10 C reduction, I however feel that they might not be quite right.

All suggestions for settings are most welcome, I have tried taking pictures of the menu, but it might not be the best. At this point I'm more interessted in no overclocking in favor of acceptable temperatures.

 

[Shintai]

10C is quite a step. Was that from disabling the multicore enhancement?

 

[Hurin]

Make sure BIOS is newest on motherboard. Because just trying for 4.2GHz (even across all cores simultaneously) shouldn't cause that sorta heat in a game.

Keep in mind that the VID listed in your screenshot is essentially what the CPU is asking for. . . but the actual vcore being provided by the motherboard according to its own calculations is listed elsewhere (you need to expand out the motherboard listing in HWMonitor).

Finally, unless that CPU is just a dud, it already seems to be asking for more vcore (via VID) than it should at those clocks. But, first things first. . . I would check out the contact of the cooler. Personally, I stopped depending on the coolers to spread their own paste a while back and have only had good results with a method others may scoff at. . . a finger wrapped (carefully so that it contacts nothing else) in saranwrap. Used to spread the paste ever so thinly across the CPU heatspreader.

If you lift the CPU cooler off and find gaps in the paste on it, or there is too much paste, there's your culprit.

I don't have any experience with that cooler. I'd wonder if it's up to the task except that the temps you're seeing are what I'd expect under a stress test. . . not just a game. So I doubt its the cooler being inadequate.

 

[Cherry Dude]

10C is quite a step. Was that from disabling the multicore enhancement?

Apparently I was able to do even better. From 15 minutes ARMA 3:



This is with Core voltage set to 1.168v
Multicore enhancement disabled
RAM at stock speed and voltage (2113)

 

[Cherry Dude]

Keep in mind that the VID listed in your screenshot is essentially what the CPU is asking for. . . but the actual vcore being provided by the motherboard according to its own calculations is listed elsewhere (you need to expand out the motherboard listing in HWMonitor).

I was wondering why there was discrepancies with the VID, maybe you can make more sense of it/this ;-)

But, first things first. . . I would check out the contact of the cooler. Personally, I stopped depending on the coolers to spread their own paste a while back and have only had good results with a method others may scoff at. . . a finger wrapped (carefully so that it contacts nothing else) in saranwrap. Used to spread the paste ever so thinly across the CPU heatspreader.

If you lift the CPU cooler off and find gaps in the paste on it, or there is too much paste, there's your culprit.

Does thermal paste really have that much influence? Scary. Will check it later just to eliminate all possible culprits.

 

[Hurin]

If I'm reading that correctly, your vcore is at an astonishing (and terrifying) 2.336v. safe voltage for a skylake is in the 1.4v to 1.5v range. I had a 6700k die on me when I came back to find the screen frozen and showing a vcore of "only" 1.984v.

I don't want to freak you out. Maybe that reading is an aberration and not accurate. But. . . I would cease everything at this point and my very next step would be to look for a BIOS update for the motherboard. Which, ideally, will either fix whatever is causing the aberrant vcore readings, or fix the actual aberrant vcore that the motherboard is sending.

Personally, I'm thinking it's an aberrant reading because otherwise your CPU would probably already be toast.

 

[Cherry Dude]

Completely agree, I have just sat for the last few minutes thinking something does not quite add up...
Updating BIOS and will return

 

[Hurin]

Does thermal paste really have that much influence? Scary. Will check it later just to eliminate all possible culprits.

I think the thickness and getting it perfect is overstated (some will claim that a single human skin cell will cause issues, etc.). But how the paste appears when you pull the cooler off can give you some sense of how good the contact is between your cooler and the CPU. The paste is just there to seal any gaps (microscopic and otherwise) so that heat can transfer from the CPU to the metal of the heatsink. Without it, you have a bit of insulating air between the heatsink and CPU, and that's bad. But again, things don't have to be "perfect" for the cooler to work optimally. Just a very thin layer and complete contact are all that is needed.

But, please see my post above this one. Your vcore readings are insanely high. Worryingly high. I hate it when someone on a forum tells me something like that. But. . . I wouldn't do *anything* else before confirming that you are on the latest BIOS, or that your motherboard is known to report aberrant vcore in HWMonitor. . . or you have otherwise determined that the vcore reading is not accurate. Because if it is accurate, I'm surprised your CPU isn't toast already. =(

--H

 

[Cherry Dude]

So a quick check after BIOS update, still same ludicrus motherboard readings. Will have to contact ASUS but must be a misreading, otherwise the apartment should have at least burned down...

 

[exlink]

Those are insane temps indeed. My 6600k that is at 4.5GHz and 1.325v has gotten to 70°C max while running Prime95 and typically stays around 60°C during any other task/gaming. I personally wouldn't want my CPU going above 80-85°C at the absolute max.

 

[Hurin]

So a quick check after BIOS update, still same ludicrus motherboard readings. Will have to contact ASUS but must be a misreading, otherwise the apartment should have at least burned down...

What version of HWMonitor?

I'd say try other monitoring programs. . . but even that can be misleading. For example, the ASUS software I tried was actually reporting VID as VCore, which would have given you a false sense of security in this case. Even if the reading you're currently getting from HWMonitor is possibly giving you a false sense of alarm. =)

One other thing, monitoring software can conflict. So make sure you're only running one at a time. And that none are running behind the scenes.

 

[Cherry Dude]

I'm running ver. 1.30.0
What other programs would show motherboard temperature and VCore? Having a look around, but the ones I finde do not display it.

 

[Hurin]

I'm running ver. 1.30.0
What other programs would show motherboard temperature and VCore? Having a look around, but the ones I finde do not display it.

Longshot, but I was running v1.29 when I was still OCing and stress testing. Maybe it's the new version causing trouble? Again, I know, that's really grasping.

CPU-Z will report vcore. But I'm pretty sure it uses the same tech as HWMonitor.

AIDA64 reports vcore, but I haven't confirmed it's not just VID.

 

[Hurin]

Hmmm, we have different motherboards, but it's probably worth noting that I updated my HWMonitor to 1.3 and am not getting the aberrant vcore reading. Wish I had better news for you.

 

[Cherry Dude]

Right... new results are in:

The previous three iterations of HWMonitor is unable to locate motherboard i guess, since they do not display any info about it. (ver. 1.29, 1.28, 1.27)

HWiNFO32 v5.38-3000:


AIDA64 Extreme version 5.75.3900


Asus AI Suite 3 - Dual Intelligent Processors 5 (I guess)

 

[Cherry Dude]

Hmmm, we have different motherboards, but it's probably worth noting that I updated my HWMonitor to 1.3 and am not getting the aberrant vcore reading. Wish I had better news for you.

Awww... but thanks Hurin.

 

[Cherry Dude]

Oh i forgot the most important: BIOS own readings are completely normal, right voltage and motherboard at breezy 38 C...

 

[Shintai]

After you disabled MCE, it seems its all fine. Just keep it as it is.

Remember to enable XMP, so your memory runs at 3000Mhz again.

 

[Hurin]

Oh i forgot the most important: BIOS own readings are completely normal, right voltage and motherboard at breezy 38 C...

It's hard to generate a load in the BIOS though.

Run AIDA64's stress test. It should report temps and voltages as it does so. If those are okay, I'd feel better about the vcore in HWMonitor. The fact that this newest HWMonitor is the first that is able to detect the motherboard sensors is probably a good sign that this latest one that does detect them isn't doing so in an accurate fashion.

In my experience, Asus "Dual Processor blah blah blah" actually reports VID and not vcore.

Keep in mind, having recently fried a Skylake, I'm a bit gunshy. My thread about it all is around here somewhere. May have some interesting points of comparison to your own situation.

But if temps are now tolerable, benchmarks show proper performance level, and you're willing to assume that vcore reports are not accurate, I think you're probably good to go.

 

[Cherry Dude]

Well I have now been fiddeling around for too long and temps are still a little off.
Looking at real life usses, half an hour gaming and CPU at 89C...
VCore is as low as i could get it while still stable, running at 4 GHz. (should i use offset or adaptive instead of manual?) My case is a Ncase M1 which might not give the best airlflow. Any thoughts, defect components, to powerfull CPU for setup?

 

[Shintai]

Well 89C as such isn't an issue. But looking at the power drawn for the CPU. Airflow/cooler seems to be the next step. Because you are drawing 70-80W tops. Not even 91 or 95W.

 

[Cherry Dude]

Sooo buckle up, cause this is a doozy for your kids.

I have been so caught up thinking it might be defective components, that I have completely neglected to test one important and super obvious element...
Name calling is allowed, I used retard myself.
...
It's the case...

Tested open setup = everything fine. 76 C for an hour CPU intensive gaming, not over 80 C stress testing.
Still testing but it seems to be a combination of the panels (So far: No panels = 67 C, only top on = 67 C, only sides on 72), odd since they seem fine with large meshed areas for air. Oh well more testing will hopefully reveal it.

Thank you all very much for your assistance, I almost feel bad for discovering it to be such a trivial cause, and not at least that I did not think of it ;-)

 

[Shintai]

Good thing you got it (semi) solved!

For the case, make sure you check how the case you got should be used, if you are not using it out of the box. May be positive pressure, negative pressure etc. Its a small science of its own.