I set up my NUC8i7HVK today. Before installing OS I went to BIOS, checking various settings. Under Cooling - section I noticed that core temperature was 68C which seemed quite high to me because the computer was (naturally) idle. After installing OS I used RealTemp to measure core temperatures. For some reason the temperatures fluctuate wildly even under minor load, which in turn causes the fans to ramp up like crazy. And I am not talking about difference of few degrees, but going from 60C to 87C all of sudden.
I tweaked the fan settings to more "aggressive", and at first it seemed that temps would no longer hit 100C (though they still went over 90C), but further testing proved my initial assumption wrong. Simply playing a game (which should not be very taxing for CPU) for few minutes was enough to get CPU temps to 100C again.
I am wondering, could it be that I have faulty cooling system or is the temperature sensor faulty? Anyone else experienced anything similar?
Running Intel processor diagnostic tool resulted the following:
Current Junction Temperature is 100C
Current Degrees Below Max 0C
Maximum Junction Temp Allowed 100C
Test Result - FAIL
Please check your Processor Thermal Solution
Temperature Monitor Stopped by User
Processor Name: Intel(R) Core(TM) i7-8809G CPU @ 3.10GHz
Processor Information: Family 6 Model 158 Stepping 9
Number of Physical Cores: 4
Number of Logical Cores: 8
Operating System: Microsoft Windows 10 Home 64-bit
Graphics Information: Radeon RX Vega M GH Graphics
--- IPDT64 - End Time: 02/06/2018 15.29.34
--- IPDT64 - Result: Fail
In excess of 100c for the junction caused the test to fail the ONLY three reasons I can imagine would be an overclocked system, faulty sensors, or the thermal paste was poorly applied between the heat pipe and CPU/GPU die.
I did not want to run RealTemp myself other than to see steady-state operating temperatures. I watched it for quite a while and turned on logging and there was <0.7% change sitting idle for an hour.
The normal operating temperature of the NUC8i7HVK with a 2% load should look something like this.
It would be helpful to post HWiNFO64 results to verify that your memory and cpu did not arrive overclocked. Your's should look very similar to below.
Make you that you select GPU # 1 from the drop down too.
There may be another confirmation test, but I can't think of one.
Perhaps someone with much better knowledge of these NUC engineering specs and tools can chime in.
All of the testing I've seen on this NUC, the only time temps reach that high is in overclocking scenarios.
100 degrees C for normal stuff, like MS Word, non-gaming stuff, is not right.
So I don't think you guys have analyzed all possibilities and you may have prematurely come to your conclusions. Let me ask this: As the temperatures are rising, are the blowers responding? At the point where IPDT failed, were the blowers running at 100% duty cycle (i.e. full speed)? If they aren't, then the problem could (also) be the configuration of the fan speed control algorithms or the application of this configuration being done in/by the BIOS.
In both the 7th and 8th generation processors, Intel made an (IMHO) incredibly bad decision in choosing to use TIM (thermal interface material; slang: heat sink paste), instead of solder, to connect the die to the heat spreader (what people call the processor lid). The fact that this TIM is high-quality stuff really doesn't matter; it is simply not as good as solder. While Intel argues to the contrary, it has been shown over and over (check the Internet) that this can cause heat transfers to be slower and can result in higher and more-volatile temperatures. This means you see seemingly innocuous operations causing temperatures to rise to very high levels and temperatures changing drastically from one second to the next.
Hope this helps,
I'm in the same boat. I too fail the Intel processor diagnostics.
I purchased the NUC8i7HVK a week ago in Australia. Since that time, I have seen three NUC8i7HVK NUCs exhibit the same problems. I returned the first NUC because of the issues I was having. The tech I returned it to tried out the other unit in stock. He had exactly the same issues I describe later on. So I assumed it was a bad batch, and waited for the RMA to come back from Intel. The returned RMA NUC I received is showing exactly the same signs. I am ready to ask for a refund at this point unless someone at Intel can assist.
EVERY single one of the NUC I've had hits 100 degrees Celsius on the mildest of CPU tasks. Running a browser benchmark in Windows 10 (v1803) that only maxes out one CPU, the blower hits 2200RPM and the CPU hits 100C. This happens only after 2-3 seconds on benchmark. All three units did the same thing. I'd estimate that ambient temperature during the tests was never higher than 22C. Only one core was maxed out - yet the cooling system already peaking after 2-3 seconds. It cannot even handle one core maxing out. We tried a more intensive test - and it too resulted in the same behavior. And the cooler is most certainly NOT quiet. I'd have to have my music up fairly loud for it to drown out the blower noise.
Loading Windows from a cold start gets the blower over 2K RPM - before the password prompt even comes up. Just in case it was my Windows install, I loaded Ubuntu 18.04 and I'm getting dmesg warnings that the CPU is being throttled because it hits max temp. This happens while Ubuntu is still loading. I'm not actually getting to the point of doing anything useful before the CPU throttles.
So it seems that EVERY single NUC I've tried hits 100C and CPU throttles at the drop of a pin. I've tried updating the BIOS firmware to the latest as well. That didn't make much of a difference. Tweaking the cooling profiles seems to make little difference. The Intel processor diagnotics fail after a few seconds unless I lock the blower at max.
I'm at a loss to explain what is going on. Single faulty unit, I can understand. But three units? The reviews painted a good picture of the NUC. Anyone from Intel care to chime in? Is there a bad batch or something?
I am sorry to hear that you have the same temperature problem, though I am glad that I am not the only one who suffers from it. Your experiences with the replacement units are very discouraging however. Maybe I should simply ask refund and save myself a lot of time and trouble. If the chances are, as it seems, that the replacement unit is no better, then there is no point going through all the hassle.
It's a pity really. After reading the glowing reviews I had high hopes for this NUC, but it turned out to be a major disappointment. With all the money I've sunk into this disaster I could've bought a relatively high-end laptop. Furthermore, if I send the NUC back, I'll have two SSD drives and SO-DIMM memory sticks I cannot currently use. I just "love" wasting money.
Have you guys tried doing a custom fan speed control configuration? I have, in general, found their Cool/Balanced/Quiet configurations to be awful. You can do a much better job of it yourself.
Could you try an additonal process for measuring the thermal performance and spec of the system?
A good example would be measuring temperatures with HWinfo64 and running prime95 with smallFFT workset and furmark at the same time. I know this seems counter-intuitive but the phase change thermal material between the CPU and heatsink may need a quick reconditioning and given that the cpu and gpu share the same heatsink this would at the very least be a good point of data for sure.
The idea is to run the prime95 and furmark for about 5min total and then shut down both workloads.
The reason this might work or help point out an issue with your units is that the Processor Diagnostic tool doesn't run a hot enough workload for a long enough amount of time to cause the TIM to change phase and flow, if you run the above workload for 5min maximum and then give the system time to come back down to normal temp and run the processor diagnostic tool it should pass if this is specifically a TIM issue.
Keep in mind this won't need to be done every time you reload the system, should be a 1 time thing for reflowing the TIM.
I did as you suggested, except I kept Prime95 (smallFFT) and Furmark running longer than 5 minutes. No errors etc. CPU cores run very hot, but only Core # 3 seems to exhibit thermal throttling. Interestingly GPU seems to run pretty cool at stable 77C.
After 20+ minutes I shut down Prime95 and Furmark and ran the Intel processor diagnostic and it passed. Temps went pretty high again and Core # 3 throttled at one point, but no failure.
I tried the same thing. I must say that it made a huge difference. Thanks @Rwashington for the suggestion.
After running prime95/furmark for 20 minutes, I could pass the Intel diagnostics. The blower fans still spin - but are nowhere near as obnoxious. They are sitting around 1.8K instead of the 2.2K previously. Not great - but tolerable. I had the same temps for the Radeon as Ruuti of 76C when running Furmark.
The CPU stopped hitting 100C when running Prime95. Now running prime95 gets the CPU to peak at approx. 87C. Booting up Windows no longer gets the blower fans to spin up to max. The fans take longer to spin up - and I haven't heard them hit max. Overall the NUC's thermals operate completely different after this process.
So having said that, what should I expect to see in terms of CPU temps? Is 86C at 100% OK for the NUC? Could it get better if I cleaned and re-applied some thermal grease?