Thanks for the hint!
I ran two tests:
- 100% CPU load (Prime95) for 30 minutes, no freezing, core temperatures around 83-84C
- 100% CPU load (Prime95) + ~70% GPU load (OCCT) for 30 minutes, no freezing, core temperatures around 83-84C
I suppose these temperatures are quite high, but still normal with NUCs.
Next (after a while) I ran the AV system scan twice:
- at first run the system freezed after 3 minutes (didn't measure temperatures)
- at second run it freezed at 4 minutes (core temperatures 79-80C)
There seems to be two sensors on the SSD:
- in idle state temperatures were 31C and 32C
- during AV system scan attempt they were 37C and 51C
Low enough in my opinion.
But here comes the interesting part: having read that there has been issues with hyperthreading previously, I disabled it in BIOS. Now the AV system scans ran through without freezing!
When I enabled HT again, AV scans and preview image creation in Lightroom lead to system freeze again.
So this problem has something to do with hyper-threading!
I would now very much like someone from Intel to comment this matter.
Hi, I mentioned the problem some time ago:
We simply aren't sending the NUCs to Intel and just deaktivated the planned AV task and wait until there is a) a BIOS fix from Intel or b) a fix from AV-Manufacturer I don't think there will be either one.
Why we do not send them back?
Since not every unit have that problem (all absolutly with the same hardware) the finding of every single unit out of ~100 NUCs with this problem take much time and nerves.
And even with that information we would have the option to...
- sending every single unit one by one to intel and wait until it is back - which is no option I think
- sending all at once - which is is not possible because that way the personnel is unable to work
If it is really no software/firmware-bug and "just" a hardware-bug then I have to say, that the NUCs have a massive failure spreading.