[SOLVED] Can a CPU be faulty on a thermal level?

[RobertDH]

Hi all,
I'm building a small lab server with a Supermicro X10SRH-CLN4F-O board and a Xeon E5-2650 v3 cpu. The heatsink is a Noctua NH-U12DXi4.
All settings are default and no overclocking is done.
Everything sems to work fine, the only thing is the cpu gets way too hot. I have tried re-installing the heatsink several times using different amounts of thermal paste, every time with the same result.
Once fully booted into ESXi with no vm's running the cpu temperature is about 55 degrees C, wich is allready a lot I think. Once I start a FreeNAS vm the temperature stays almost the same but when I copy a file to FreeNAS, the temp rises to above 80 degrees within a minute.
According to ESXi, cpu usage isn't even reaching 25% while copying files.
So my question is: Is it possible the cpu is faulty?
Or am I missing something here?

Any help would be appreciated!

Cheers,
Robert.

 

[SteveRX4]

the only thing is the cpu gets way too hot - how are you measuring the temp?

I don't think the cooler is compatible with your motherboard. https://noctua.at/en/products/cpu-cooler-workstation-server/nh-u12dx-i4/specification
Your CPU is LGA1151.

 

[helpstar]

Did you use thermal paste?

update the BIOS:
https://www.supermicro.com/support/resources/results.aspx

Your CPU is LGA1151.

No, the CPU and motherboard is socket 2011-3

https://ark.intel.com/content/www/u...-processor-e5-2650-v3-25m-cache-2-30-ghz.html

 

[RobertDH]

the only thing is the cpu gets way too hot - how are you measuring the temp?
Is your case well ventilated?
What speed is the cooler fan running at?

Hi, Thanks for helping!
Yes, at the moment the case is even open, fans are at 1500RPM, i'm reading the temperature from the mainboard sensors.

 

[RobertDH]

Did you use thermal paste?

Thanks for your reply! Yes I even tried different amounts of it.

 

[psilokamenos]

Try stress testing your cpu and let the heat rise, if you cpu reaches around 100-105c your system is probably gonna crash and reboot if that happens there is something wrong with your cpu and it will be best to replace it. If your cpu stays stable at around 80c and dont rise more while the usage is 90-100% in the test
then maybe the measurement from your mainboard sensors isn't working right and it does not compare to the real temperature of you cpu.

 

[helpstar]

the CPU power management might not work correctly. Intsall the latest chipset and APM/ACPI driver and set the power management profile to balanced

 

[RobertDH]

Try stress testing your cpu and let the heat rise, if you cpu reaches around 100-105c your system is probably gonna crash and reboot if that happens there is something wrong with your cpu and it will be best to replace it. If your cpu stays stable at around 80c and dont rise more while the usage is 90-100% in the test
then maybe the measurement from your mainboard sensors isn't working right and it does not compare to the real temperature of you cpu.

Yes i'll give that a try, but at the moment automatic throtling is kicking in at 88 degrees so first have to find out how to disable that.

 

[RobertDH]

..Come to think of it, the throttling happens by the cpu itself so apparently it is overheating and the IPMI temperature reading is probably right.
Do you guys agree?

 

[psilokamenos]

I think so, if your cpu is throttling then the temps are correct.
Well i never heard before in my life of a faulty cpu cooler but if you have another cooler that you can use give it a try
other than that i think you should find your cpu warranty.