Coretemp readings seem to be lower than expected?

thesuperguy

Honorable
Apr 19, 2013
348
0
10,810
I recently built a new rig with the FX-6300 and for the past few days, I've been plagued with the crappy feeling that my CPU temps have been high. Today I decided to get a second opinion and downloaded coretemp which tells me just the supposed CPU temp instead of a list of unnamed numbers which I have to guess from. The thing is, coretemp seems to be quite inconsistent with other programs which have shown idle temps of high 30s and above (temperature determined by finding the highest temp when using prime95). Coretemp is currently showing a temperature constantly fluctuating from 12 degrees (I doubt it) to mid 20s. It also doesn't display the temp for all 6 cores and rather just a single temp for "CPU #0". What is going on?

Edit: actually it seems like it consistently idles at 12 degrees and when simply browsing the internet, has a current max of 31 degrees.
 

thesuperguy

Honorable
Apr 19, 2013
348
0
10,810
Thing is, under load, other programs will shoot to over 70 degrees while coretemp ever so slowly climbs up. I'm also pretty sure 12 degrees is under my ambient temperature. Under 100 percent load coretemp is idling at 33 degrees... Speedfan and HWmonitor are both showing 64 and climbing.
 

RazerZ

Judicious
Ambassador


Over 70 degrees celcius is pretty dangerous... What's your CPU cooler?

 

RazerZ

Judicious
Ambassador


Hmm, that fan is decent. Either there's some installation issue, or your case is preventing air flow, or you overclocking too high.

 
Ah!

If you have a Radeon GPU, snag AMD Overdrive, for another set of info. Both AMD Overdrive and Core Temp snatch their thermal readings from the on-die CPU Core sensors. Other temperature reading software will sometimes show the CPU Socket temperatures, which are great for measuring idle temperatures, but inaccurate at load temperatures. Ideally, a program will display both the on-die CPU Core temperatures and the CPU Socket temperatures, so you can get the most information about your rig. The on-die CPU Core temperatures are innaccurate at idle temperatures (often showing below-ambient) but are accurate at load temperatures; this is because the readings are essentially fed through a type of weighted equation to display those accurate numbers for load temperatures.

As for Core Temp displaying only one set of temperatures, that is normal for AMD-based processors. I can't remember the exact reason why. I think I remember someone saying it is because they employ only one sensor? That, or the read-out displays only the average of all the sensors, for whatever reason?
 

thesuperguy

Honorable
Apr 19, 2013
348
0
10,810


I still think that a temperature of barely over 30 degrees at full load (which is what coretemp and AMD overdrive both show) is unlikely. Ambient is currently at the low 20s and full load only about 10 more degrees? I see people saying 50 degrees is a great temperature for full load... Not to mention my full load temperature on coretemp/overdrive is lower than my idle on speedfan/hwmonitor. It does seem however like AMD overdrive takes all temps from sensors and beside the 32ish degrees per core under load, it also shows 47 for tmpin1, 65 for tmpin2 which is what I assumed was the CPU temp for speedfan and hwmonitor (it is a cool day), and 33 for tmpin3 (same as GPU so probably GPU).

 
If you can figure out and match up which temperature reading is coming from which probe on each of your temperature monitoring programs, you'll be that much closer to your answer.

I agree, slightly over 30C does sound a bit low, but it isn't outside the realm of possibility. Check out the review from Frostytech on the Hyper 212 EVO: http://www.frostytech.com/articleview.cfm?articleid=2655&page=5

For the simulated 85W Intel Processor, the result was 9.3C; for the simulated 125W AMD processor they got 12.8C. Your processor is rated for around 95W, so getting roughly 10C above ambient sounds right - according to these results.

EDIT: 50C is a good temperature to have - but at what ambient temperature, and from which reading? General discussion, axioms, and advice don't always carry specifics.
 

thesuperguy

Honorable
Apr 19, 2013
348
0
10,810
Yes but my main concern is why the load temperature on what would be the actual core I assume is lower than the idle on the socket which you said is a more reliable gauge for idle temps. Then again, I'm still not 100 percent sure which each tmpin temperature represents. I kinda just assumed the hottest one under load was going to be my CPU.

Edit: overdrive seems to be tripping out right now. It is reporting temperatures of over 250 degrees on tmpin 2 through 4 :p
 
Without complete objective knowledge with the data at hand, anything else is conjecture.

You could have faulty temperature sensors, or, since you have only one case fan for ventilation, your CPU Socket could be reading higher temperatures since there is inadequate airflow.
 

thesuperguy

Honorable
Apr 19, 2013
348
0
10,810
This is unusual. Turns out the highest temperature (tmpin1) seems to be my vrm chips... At least the temperature quickly decreases as I blow directly onto them and slowly decreases as I attempt to blow onto the CPU. But what that has me wondering is what the other 2 are. Tpmin0 at full load stays at 45. Could this be the CPU? It didn't really decrease when I was blowing onto the CPU area. There is also tmpin2 which at full load stays at 32. I had always assumed that it was my GPU temperature but perhaps not? It doesn't decrease when I blow onto the GPU area but the GPU specific temperature does. Shouldn't one of these temperatures be my nb/sb?