I installed and ran AIDA64 with logging turned on. I've attached the data for the last hour before the crash.
The temperature data seem fine. So do all the other logged parameters.
Two interesting things to note
- When the machine crashes, the display is gone, but the NUC unit is still on (blue light on) and I can hear the fan. I cannot shutdown the unit by keeping the power button pressed (nothing happens). The only way to shut it down is by cutting the main power.
- Once the machine crashes and the power is manually cut, then pressing the power button does nothing for a few hours after the crash (no blue light - nothing). However after a few hours, pressing the power button simply starts the machine up. If this was purely an over-heating issue, then the machine should've started up after 10 minutes or so after the crash (once everything cooled down).
Any ideas what I should look at next?
sensor-log-last-hour.csv.zip 125.9 K
On the contrary, the temperature data shows that you have a serious cooling problem. Here is a plot of the CPU temperature data in your attachment:
As you can see, the system spent a significant amount of time well above normal temperature levels and even reached the throttling point on a number of occasion. You need to resolve this problem immediately (if it's not already too late).
Here are my recommended settings:
These settings ensure that the fan is on full when processor temperatures rise above 85c or motherboard temperatures rise above 75c.
Try these settings and let's see if they help...
Thanks Scott. I've made the changes and have run the system for about 3 hours now.
The Motherboard CPU is around 60 degrees Celsius, while the CPU is around 75 to 80 degrees celcius. From my online research, it looks like these numbers are still too hot (or are they fine).
Is there anything else I can do about this? Apply thermal paste, add/change the cooling fan?
I have no experience with any of this, so any advice you can give me would be helpful.
In the roughest of terms (because it can vary from one individual processor to another), if processor temperatures are not going higher than ~85, then everything is ok. I suggest you run a test to verify this, however. Run something that places a significant load on the processor - like Prime95 - for a solid hour, watching the temperatures as it runs. If the temperature stays at or below ~85, then you are good to go. If it goes above this level - say, into the 90's or higher - then you will know that there is something further to worry about.