False CPU Temp readings (running too hot?) i7 6800K in Core Temp and HWMonitor

[R3jected]

Hello  guys,

 

I build a workstation for my friend today (i7 6800K cooled by a Dark Rock 3 in Fractal R4) and finished by running some Prime95 benchmark to test CPU temps.

The temps started to climb until they reached arround 72°C (running under 20 minutes) which is way higher than I expected because of the massive cooler.

As I stoped the benchmark the temps falled down from 68 to 40 in about 5 seconds which irritated me.

I made a reboot ran the bench again letting it rise to over 65° quickly shut it down and entered the bios as fast as possible and look at the uefi temps which happened to be at ~45°

What do I trust? HWMonitor + CoreTemp or UEFI readings or my instinct hoping I am fine.

 

Waiting on your advice.
Best regards.

[zMeul]

use the software from the mobo manufacturer and compare it with your other software

[Unexas.]

There could be an issue with the installation of the cooler. Make sure it is making good contact with the CPU, you put thermal paste on and make sure you don't have the coolers fan and a case fan blowing at each other. AKA -><-

[AresKrieger]

72 in prime 95 is actually really good and core temps drop fairly quickly when the load is taken away so the temp are probably no to far off, if you want a more accurate test to how hot you cpu will run while maxed out use intel xtu or aida 64 (trial) since prime goes way overboard often enough.

[SpaceGhostC2C]

1 minute ago, R3jected said:

Hello  guys,

 

I build a workstation for my friend today (i7 6800K cooled by a Dark Rock 3 in Fractal R4) and finished by running some Prime95 benchmark to test CPU temps.

The temps started to climb until they reached arround 72°C (running under 20 minutes) which is way higher than I expected because of the massive cooler.

As I stoped the benchmark the temps falled down from 68 to 40 in about 5 seconds which irritated me.

I made a reboot ran the bench again letting it rise to over 65° quickly shut it down and entered the bios as fast as possible and look at the uefi temps which happened to be at ~45°

What do I trust? HWMonitor + CoreTemp or UEFI readings or my instinct hoping I am fine.

 

Waiting on your advice.
Best regards.

I think Prime95 below 70C is an unrealistic expectaton...

 

Also, the cooler got you back to 40 in 5 seconds, and that irritates you? I'd say good job by the cooler (although maybe 5 seconds is too long, I haven't timed those things.

 

Given how fast temperatures go down according to your report, I don't think there is much of a disagreement between BIOS temps and HWmonitor. Bear in mind that motherboards report the socket temperature, while HWmonitos will report both the socket temp as measured by the motherboard, and the core temps measured directly by the CPU (just to make sure you compare the same temp in both cases).

[lh99]

i myself have a 5820k (so the older model from you) and i have similar temps like u under a synthetic cpu benchmark, also using a (technically) 2 tower cooler. So I think everything should be all right by you.

[R3jected]

Thanks guys, this forum is my favourite place to go when I have questions!
I am / was sceptical about the 5 second recovery cooldown, so I will try a different benchmark if I get another try at his pc. Maybe I fucked up with the thermal paste.
If you got another experience with similar or different temps, I would be glad if you could share them.
I will update you on my case if I meet him again.

[paddy-stone]

BTW using the UEFI to gauge temps isn't great, there's no way a CPU would retain it's heat through a restart to gauge the temps, and it's normal for the CPU to cool quickly when the load is taken away... it's actually the OS that does the job of lowering the voltage and frequency... this is half the reason why if you have a bad OC you sometimes have to reset the CMOS as the BIOS/UEFI will run full pelt on start up and cause it to crash.

I may not have explained that very well, but basically use AIDA64 is better for stress testing and coretemp etc is usually getting the same readings so should be OK.