Reapply Therm Paste on i5 6600k for Fluctuating Temps?

[rabaker07]

My CPU (i5 6600k) has lately been jumping from around 28 C to 62 degrees and even higher, and then back down to high twenties and thirties, then back up to 50s, 60s... up and down up and down every other second. And I can hear the fan going up and down, off and on, etc. to keep up with the fluctuation.

This has had me a little worried, particularly because I thought it was lower over the last few months since I installed this new CPU and Cooler (Hyper 212), but maybe I just don't remember correctly. During heavy load/gaming, I see that the CPU is going all the way up to high 80s, even hitting 90. That has me a little concerned as well. It's only been around 6 months, but I've also had it overclocked to 4.6 on air (but it drops all the way down to 0.8 on idle). And that's not a crazy OC for this chip.

Should I reapply thermal paste?

 

[n0ns3ns3]

the thermal paste is not related to your "issue".
temperature fluctuation is a normal thing. it is proportional to the load on the CPU at that moment.
4.6GHz overclock is considerable one for this "entry level" cooler.
So your main and real concern/issue should be the 80-90C during gaming as it's not the highest CPU load.
In order to troubleshoot this, please answer following questions:
1. what is the CPU vcore voltage under load
2. what is your room themperature
3. what case are you using, with how many fans and how those fans are configured.

 

[rabaker07]

Thanks for the prompt reply.

1. vCore is 1.504 at load (high-demand gaming with CPU usage from 70-99%, not stress test), but the voltage is variable based on demand, so is CPU clock, from 0.8 to 4.6 GHz
2. Room temp is around 73 degrees F/ 23 degrees C
3. Thor V2 Case with positive pressure (mostly intake fans and a big hole in back of case, unfortunately)

- Front 1 x 230mm Fan intake
- Rear 1 x 140mm Fan exhaust
- Top 1 x 230mm Fan exhaust
- Side 1 x 230mm Fan intake

And I have an EVGA GTX 1080 with ATX 2.0 fans, which is below the CPU based on case design/mobo layout

 

[n0ns3ns3]

well, go to bios, set the vcore to adaptive = 1.25v and additional turbo voltage to 0 (it will be 0.001 or something).
if that does not work, gradually increase voltage up to 1.3v. do not go over it with this cooler. If the system is not stable with 1.3v, start lowering the clock. 1.5v is way beyond the "recommended" values and is the root cause to the high temps.
Even with liquid cooling it is not healthy to go over 1.35. and definitely not above 1.4v for 24/7.
once you lower the voltage, temps should go to normal values below 75C.

 

[rabaker07]

Maybe I should lower the OC to 4.2 or 4.3. I don't know how much 'performance' I'm even getting from the bump up to 4.6 and if it's generating all this unnecessary heat, it might be prudent to scale it back. At least until (or IF) I get a stronger cooling solution. Any recommendations?

 

[rabaker07]

well, go to bios, set the vcore to adaptive = 1.25v and additional turbo voltage to 0 (it will be 0.001 or something).
if that does not work, gradually increase voltage up to 1.3v. do not go over it with this cooler. If the system is not stable with 1.3v, start lowering the clock. 1.5v is way beyond the "recommended" values and is the root cause to the high temps.
Even with liquid cooling it is not healthy to go over 1.35. and definitely not above 1.4v for 24/7.
once you lower the voltage, temps should go to normal values below 75C.

Bummer, it wouldn't boot.

And when I set it to adaptive, the only value I was allowed to configure was the standard offset voltage, with a minimum of 0.001. The Turbo voltage was locked at 0.250, and the base voltage was at like 1.040.

I guess I'll have to go back to the drawing board, and start over, using more advanced settings and slowly bringing up the clock speed with repeated boots and stress tests. I have to get this voltage down, as you mentioned. No wonder it was so stable at 4.6 this whole time!! Meanwhile, I've been frying my chip!

Google search, here I come... I'll need to find a modern guide to O/Cing this CPU/Mobo combo. Asus Z170-AR, btw. And Skylake-S i5 6600k RO step with original clock at 3.5, oc'd to 4.6.

 

[n0ns3ns3]

you can do the search for the max overclock on manual voltage mode. once you found it, post again, I'll look into bios and tell you exact names of the fields to change to convert that manual to adaptive. eventually you want adaptive voltage so it will go down on low load/clocks. with manual it will stay at that value all the time.

 

[rabaker07]

Yes, so I dropped my clock down to 4.4 and had it at 1.34 volts Manual, which withstood a quick stress test in Aida64, so then I changed it back to Auto to see how high it would natively raise the voltage at 4.4. It was down around 1.28 during the next stress test, so I figured I'd leave it at that. The auto voltage was dropping to 0.8 and peaking at 1.28, so all good right? Not quite...

Now I keep getting reboots and a blue screen for "kernel_security_check_failure".

I'll keep messing with it, and hopefully can get that Adaptive config setup so that I can lock down the variable 'speedstep' low voltage, dropping to 0.7 volts at 0.8 clock, and still keep 4.6 stable at sub 1.4 volts on the top end.

 

[n0ns3ns3]

not with this cooler. don't go over 1.3v

 

[rabaker07]

Wow, thank you. Dropping my overclock from 4.6 down to 4.4, and setting the voltage to Auto has completely stabilized my system, temps, and volts. I also opened the top panels of the case to help with heat dissipation, but it's obviously the lower voltage hitting the CPU that has made the largest impact.

I'm now maxing-out at 1.296 volts and my temps are staying level in the mid 30s across normal use with highs in the very low 50s during peak gaming. No more fluctuations!! Awesome!

But I'm afraid that the BSODs are not truly over. Time will tell. I also set my DRAM to Auto which brought it down from 3000 MHz to 1066 and this might resolve the crashing. I'll dig in and give it some more stress testing and play a very demanding game like Ghost Recon: Wildlands on Ultra.... we'll see if 4.4 offers adequate performance compared to 4.6 (I'm sure it will)...

p.s. my ram is : Vengeance® LPX 16GB (2x8GB) DDR4 DRAM 3000MHz C15 Memory Kit - Black (CMK16GX4M2B3000C15)

 

[n0ns3ns3]

Bring the memory back to XMP profile. Though do not overclock it yet. I Was able to take my DDR3 1866 to 2800MHz.
if you are still having BSODs, it means the CPU can not run this clock at this voltage.
Unless you changed something else.

 

[urbancamper]

I have the same cpu and can overclock to 4.6 stable with a vcore of 1.425. i know this because i have done it. The temps never go above 68c at full load on my Corsair H100i v2. While I proved to myself it is possible, I prefer to keep my vcore under 1.4v. I currently run at 4.5 with a vcore of 1.385 at load. Also at load i will rarely see 63c.

1.504 vcore is way to high for that cpu.

 

[rabaker07]

. I currently run at 4.5 with a vcore of 1.385 at load. Also at load i will rarely see 63c.

How do you configure your bios for this O/C? How do you set the voltage? Adaptive? Manual? Offset? Auto?

What mobo? MSI Z270 ?

 

[rabaker07]

So now my CPU core clock is ranging from 0.8 to 4.4 GHz, at voltages from 0.74 to 1.28. And my temps range from 26 to 51 degrees C, and the crazy fluctuation appears to be completely gone!

But I had a few more BSODs this morning, so I'll prob have to tweak things further. My RAM is still down at 1066 without XMP, and I think I need to adjust my voltages to have a higher min and max. That could explain the crashing. But I'm still not clear on what BIOS settings to use for Adaptive or Offset voltage on my Asus Z170 mobo.

 

[urbancamper]

Manual with LLC at 4. There are 8 levels of LLC on my motherboard. At LLC 4 my vcore stays at 1.385. Never wavers.

As far as your ram goes it is not at 1033 It is at 2067. 1033x2(they remove the decimal point).

And yes the other guy is right, Use your XMP profile, and also get a better cooler, or that cpu won't last long. You may have already damaged it.

 

[rabaker07]

Okay, I've got it stable now. Back up to 4.6 GHz, peak voltage at 1.376v, down to 0.7v at idle. Figured out how to setup the Adaptive profile.

XMP is on, and RAM is back at 3000 MHz at 1.35v using the extreme profile. (but it still looks like it's at 1066 in Aida64 using 32:3 ratio based on 100MHz FSB.) Maybe CPUID or HWInfo will help me to verify this...

Otherwise, I'm golden, no rapid fluctuation, no BSODs (for 24 hours now) and cool as a cucumber. Peak temps at 58 degrees C.

Let's hope I didn't do much damage when it was hitting 1.5+v, but supposedly this chip is rated to allow volts that high...

 

[n0ns3ns3]

Glad you solved your overclocking task.
it's not really probable that you damaged the CPU silicon. It takes time and even with extreme voltages it needs to run at full load for long time.
Gaming is kinda "partial" load for the most time. One of the first signs for silicon degradation is inability to hold previous clocks at the same voltage.
IMHO, 1.376v is still way too much. I'd prefer 100MHz less but to stay around 1.3v. but if you feel comfortable with it, go for it.

 

[rabaker07]

IMHO, 1.376v is still way too much. I'd prefer 100MHz less but to stay around 1.3v. but if you feel comfortable with it, go for it.

So, you suggest a 4.5 GHz clock? And drop the voltage?

Current settings

 

[n0ns3ns3]

I'd go for the highest stable clock at 1.3v. that might be 4.4-4.6GHz in your case.