3 CPU temps on mobo, which to believe?

[SlushpuppieT]

I've attached a screenshot of my hwinfo after running prime95 for ~30 seconds and then stopping that program and letting it sit for a minute or so.

Obviously 99-100 degrees is way too high, I have an H100i cooling system that I have checked again and again that it is mounted securely. Have not pulled off of the processor as I don't have any thermal paste and don't want to clear off the stock stuff. The processor itself has not been overclocked (Aside from factory stuff, normally 3.5, now 3.9 GHz).

As you can see on the screenshot the first CPU only ever reached 52C which I would be much more inclined to believe but I'm still worried about the other temps. I have been running this system for a year now without any trouble, no blue screens, no random shutdowns, so I don't know if this is a recent issue or not, just decided to check it a few days ago. Running RealTemp gives me the same temps as the DTS sensor for the most part, but realtemp will sometimes report 90 degree temps even if every other monitor is displaying ~30 degrees so I've stopped checking that. Also, the temps ran up VERY quickly, jumped from 30 to 50 to 90 in the span of 4 seconds, while the first one slowly crept up to 40, then 50, over the 30 second period that prime95 was running.

Any suggestions? Should I ignore the sensors that say my processor is currently boiling water?

 

[Martin]

Indeed hitting 100 C is too high. Those DTS values are the internal temperatures inside the CPU and are confirmed to be valid (in sync with the Package, IA Cores and CPU (PECI).
The other CPU temperatures are most probably measured in the socket by an external diode. So they are lower as expected.
You might also check the value of Thermal Throttling, which is expected to kick in if such temperatures are measured (for this protection mechanism the DTS values are of importance) and that will reduce the performance of the CPU until temperatures go down (which you should notice by a lower clock).
It's quite normal for temperatures to ramp very quickly when there's a high load executed.
I believe this problem is because that CPU integrates the voltage regulator (Fully Integrated Voltage Regulator = FIVR), which is the source of a significant rise in temperatures. This is quite common for the Haswell-family of CPUs and one might argue if this was a smart move by Intel.

 

[SlushpuppieT]

Thanks for the reply, I have a feeling it does have something to do with my cooling unit, I'll probably try to remount it at some point, I noticed that if I press down firmly on it, the temps only went up to about 48C (Maximum) after an hour skyrim, as opposed to the 60-70 I would get normally.

Looks like the thermal throttling only happens at 100C which is why the temps never go higher than that, and idling my processor voltage is about .8, running at 800 MHz. It went down to 22 momentarily but after a while of running it's back up at 34, still idling.

I'll probably try to remount the cooling unit soon and if that doesn't work just try a regular heatsink/fan to see if possibly my liquid cooler is just acting up.