CPU sudden increase in overall temperatures

TwilightWolfi · Jul 16, 2016

CPU: i7-6600K, 3.5 GHz, OC 4.7GHz (stable, minor temp increase under full load compared to no OC, ~8C increase comparatively)
Cooler: Water-cooled, Corsair H100i
GPU: MSI NVIDIA GTX 970 Gaming Twin Frozr 4GB (refurbished, no current OC)

All components are only a month old.

I built my PC only a month ago, and when it started, idle temperatures were approximately 27/28 degrees Celsius. This persisted throughout the month until, without warning today, while running Overwatch, my computer gave me a Win10 bluescreen due to a CPU hardware error. I allowed it to restart my PC, confused, especially on top of having seen my GPU jump from the regular 63/64 to a solid 71 during play (no CPU temp display unfortunately). Now that my PC had restarted, I checked Corsair Link 4 to see my CPU temps and they had jumped up quite a bit to an idle 34/35, sometimes even temporarily jumping to over 40. On occasion, they even show 60 for a brief moment, and I'm worried for the state of my PC. I don't see any Task Manager abnormalities (about 5-6% usage constant), nor does Corsair Link show any consistent load increase, except on occasion (wavy between 5-15%, but no obvious effect on CPU temps). My cooler, when my computer crashed, was incredibly hot, even though Overwatch only ever took it to approx. 60-65C every other time I've played.

It's really worrying.

Thanks,
Bryn

EDIT: I've just seen a massive increase in temperature to approximately 55C and a constant 25% load on the CPU, and it's apparently something called "Service Host: Unistack Service Group (4)". It seems to keep kicking my temperatures up to 45-50C, and using up 5-6% of my CPU, along with many other programs suddenly using approx 1-2% for no obvious reason. Load is still 25%, and that's about the CPU usage in Task managed too. Unistack is always at the top, though, followed by Corsair Link. It's very disconcerting. The CPU also keeps jumping to its maximum OC of 4.7GHz as mentioned above, and the load is wavering up to 40% at times, consistent 30% often. I think it's getting worse.

Supahos · Jul 16, 2016

Disable automatic app updates in the store options

http://superuser.com/questions/1019522/windows-10-unistack-service-group-hogging-cpu-to-40-or-more

MeanMachine41 · Jul 16, 2016

If your temps are increasing as you describe, then your TIM needs to be checked or the pump on your H100i is not functioning correctly. What RPM is it spinning at in Corsair link.?

Even disabling the auto updates should not cause those temperature spikes.

Use HWMonitor to monitor and evaluate system.

Supahos · Jul 17, 2016

His idle temps have gone up because at 25-30% load it's not idle needs to kill the process hogs to fix the issue

MeanMachine41 · Jul 17, 2016

Supahos :

Oh yea, You seem so sure.? An increase to spike a 60c temp, even for a brief moment should not occur. I have an OCed system and multitask and I could run a MS update and heaps of other tasks and that never happens with my H110.
Using 5-6% of his CPU and kicking up to 45-50C is not normal and just how that relates to a process hog that's jacking his temperatures up. Please enlighten me with fact or at least a decent explanation.

Supahos · Jul 17, 2016

If it was 4-5% you'd be right but it's enough to get him out of power saving mode and go to his max oc, this will spike temps drastically above idle.

TwilightWolfi · Jul 17, 2016

MeanMachine41 :

It's running at a constant 1860 rpm, with the radiator fan at 1500 rpm.

MeanMachine41 · Jul 17, 2016

TwilightWolfi :

Ok well mine operates at around 860-1500 rpm at full load. My rad fans are adjustable.

I recommend you conduct a stress test on your system to determine what's going on.
Download HWMonitor and AIDA64 Trial version to your desktop and set them up side by side.

Run the tests for CPU and FPU for 10mins whilst monitoring your temperatures.Take a screen shot after 10mins and stop the test if temps reach 80C. You can also run the GPU test separately.

If you need to have the results analyzed then use IMIGUR to host your files and obtain the BB url to link to your post.

If temps are getting to 80C during the test and within 10 mins, You are going to have to replace your TIM.

Please report back.

TwilightWolfi · Jul 17, 2016

Temps quickly reached between 75-80 within seconds, and was at a solid 77-78 for package and quite a few cores regularly broke 80C. I cut the test after about 30 seconds.

MeanMachine41 · Jul 17, 2016

Well that's a conclusive result. The cooling pump is trying its hardest and can't keep up.
Don't try to run your system anymore as you will damage the CPU. You are going to have to strip it down and replace the TIM.
Do you know how to do this.?
Also check the hold down brackets as I suspect something has come loose.

TwilightWolfi · Jul 17, 2016

MeanMachine41 :

I'm fairly confident with it, though I don't know what I should use to clean the CPU and heatsink. Something like isopropyl alcohol (if I have it)? Is there any better, more common alternative so I don't have to go out and buy some?

MeanMachine41 · Jul 17, 2016

90% or higher ISOPRO and don't use rubbing alcohol. Don't use anything your not sure of.
I also updated my last post.

TwilightWolfi · Jul 17, 2016

MeanMachine41 :

I've cleaned both and reapplied the TIM (I apparently put on too much last time, so that might have been a contributing factor), and temperatures have dropped, though I just had an unprecedented spike to 50% and down to 4% on load, so I don't know why that happened. Stress testing again to check.

Cores temp suddenly unstable, but no stress test yet, thinking that the TIM wasn't the issue. I think it might be the massive CPU load program, though I don't know what is causing these huge spikes.

At first, for about 2 minutes, the temp was approximately 69, 71C. The temps slowly crept up until it was hovering at 75-79C. It broke 80, and I stopped again. It took longer, but it still got there.

The temperatures are once again stuck at high 30s. I don't think the water cooler is actually cooling hard enough, because it doesn't seem to be able to cool the CPU down to natural idle temps after heavy load. I also cleaned the air filters because they were very dusty, but that seems to have only delayed the temperature increase. I think it might be my pump, but I don't know why.

Supahos · Jul 17, 2016

If that process hogs is still there your cpu is never going to idle which is what I said 10 posts ago... Did you ever look at my link and disable the things in that post?

TwilightWolfi · Jul 17, 2016

Supahos :

_I don't know what the process is that keeps jumping my load to 50%+_. It's no longer that. I even stated that I didn't know what the process was.

Additionally, after turning my radiator and case fans on full, and my pump, the cooler slowly dragged it to 28. The processor jumps to high temps under high load, but the real problem is that it doesn't cool down enough after high load. I also stated that. Please read the posts rather than paraphrasing "I was right all along". It's extremely unhelpful.

MeanMachine41 · Jul 17, 2016

TwilightWolfi :

What made you conclude you applied too much TIM last time.?
Was the TIM squeezing outside the edge of the processor.? If so did you inspect for any contamination to the socket as that can be a major problem if not cleaned properly. And did you inspect for any bent pins in the socket using a magnifier.?

If you have run the system at temps over its thermal limit, then damage to the CPU may have occurred.
If you feel this is the case then RMA the chip. Intel are good in this respect.

TwilightWolfi · Jul 18, 2016

MeanMachine41 :

The TIM on the heatsink was over the edge of the CPU, but I did check and there isn't any in the socket or the motherboard. I made sure to use less, and it improved my idle temperatures, but it's the heatsink's capacity to dissipate heat during and after heavy load that worries me. It seems to be at a high idle temperature or high overall temp during or after heavy load, only cooling down slowly instead of as fast as it used to. It might be the H100i, but I'm not entirely sure. Could the CPU have been damaged in some way that it reduces its capacity to dissipate its heat effectively? If not, I might just dial down the overclock to 4.4GHz or something similar, as I don't think I need the full 4.7GHz. If I can reduce the required voltage, it might be able to cool down faster and avoid heating up so much during spikes. I'll also try and keep an eye on the CPU load and compare it to the processes when it reaches an unusual load.

I also don't feel that I've damaged the CPU or pins in any way, as it still runs just as well as it did on 4.7 GHz OC before.

MeanMachine41 · Jul 18, 2016

I guess you'll never know if the Cooler is efficient enough without swapping it out. However just dialing down your OC is not going to help without reducing CORE voltage as its from the voltage where the heat is generated. Do you happen to know what core voltage is set at.?

I don't have an i7-6600k however OCing to 4.7 is quite a decent OC for that chip and maybe you should have gone for a H110 or H115 for the 4.7GH.

TwilightWolfi · Jul 18, 2016

MeanMachine41 :

I did mean dialing down the core voltage, yeah, not just the overclock.
Let me just check the max core voltage.

It was on automatic voltage offset, but was apparently up to 1.21V. However, the voltage regularly reached about 1.34-1.35V from what I saw on CPU-Z.

MeanMachine41 · Jul 18, 2016

TwilightWolfi :

Make it 1.3V and 4.5GHz and test that. Then try 4.6GH at same voltage and test again. I wouldn't go above 1.4V even tho Intel state max is 1.52V.

TwilightWolfi · Jul 18, 2016

MeanMachine41 :

I offset the voltage to 1.296V, but the voltage leapt to 1.44V when I opened up Chrome. I shut down quickly. Do you suggest a manual and consistent 1.3V? It seems a bit high to go straight for a manual 1.3 under all loads, and the variance in voltage is minimal in manual-set voltages.

MeanMachine41 · Jul 18, 2016

Rather than a TLDR TwilightWolfi. There is an excellent article re Adaptive offset or manual.
Go here: http://rog.asus.com/forum/showthread.php?2162-Overclocking-Using-Offset-Mode-for-CPU-Core-Voltage
This will explain better than I can.

I like Adaptive Voltage method. This method is much safer, and consumes less power than a fixed core voltage.

TwilightWolfi · Jul 18, 2016

MeanMachine41 :

[strike]I've seen that, but the reference images are all missing. They seem rather necessary to understand exactly what the guide is detailing.[/strike]

I read it properly now and actually understand what to do. Thanks I'll update with the temps, although I think I may have found the actual source of the problem.

CPU sudden increase in overall temperatures

Reputable

Expert

Judicious

Expert

Judicious

Expert

Reputable

Judicious

Reputable

Judicious

Reputable

Judicious

Reputable

Expert

Reputable

Judicious

Reputable

Judicious

Reputable

Judicious

Reputable

Judicious

Reputable

Share this page