core temps vs MB diode temps

[pogiman]

Ive seen many people claiming 30C CPU temps with over 1.45vcore and 3.4GHz. Then i realize that these temps are not the actual CPU die itself. Speedfan reports 37C underload, while coretemp and Intels temp analyzer reports 52C. (for me) Im assuming the core temps would be more useful in your overclocks, but why do these people here keep posting the crappy MB diode temps? it confuses me when i try to search for a comparison. And then they go bragging about how low temps they got from a air cooler. Sorry i need to vent this, becsaue i kept seeing these low temps on air, and i thought water would be better, but in fact, its not as low as i hoped to be.

[Fred_Pohl]

The temps you see being reported are from a thermal diode located on the CPU die just as all Intel CPU temps have been read from an on-die thermal diode since the P2. Core Temp BETA and the Intel Thermal Analyzer are estimating temps from different thermal diodes located on each core. Which is more accurate is subject to debate but you certainly can't compare CT/ITA results to any other CPU in existence today because C2D is the only CPU with those extra thermal diodes located in hot spots on the cores. AMD and P4 use conventional thermal diodes just like the one that the bios reads from on C2D so those results are comparable and more meaningful with regard to Intel's thermal guidelines.

[Btrice]

For me, speedfan, core temp, DFI's tool, and the bios all report temps within 5 degrees of each other and they all seem to be relatively reasonable in terms of accuracy. They are a little higher than expected though, since I am using higher end air and my cpu heatsink is cool to the touch.

[Fred_Pohl]

For me, speedfan, core temp, DFI's tool, and the bios all report temps within 5 degrees of each other and they all seem to be relatively reasonable in terms of accuracy. They are a little higher than expected though, since I am using higher end air and my cpu heatsink is cool to the touch.

It could be that DFI interprets the diode reading a little higher than other mobos like my DS3, which is ~10C lower than Core Temp Beta at full load. Or maybe you're talking about idle temps, in which case a 5C spread is to be expected.

Different mobos have been reporting different temps from the same cpus for as long as there have been on-die thermal diodes. C2D might be a little different but not much, IMHO.

[Btrice]

It could be that DFI interprets the diode reading a little higher than other mobos like my DS3, which is ~10C lower than Core Temp Beta at full load. Or maybe you're talking about idle temps, in which case a 5C spread is to be expected.

Different mobos have been reporting different temps from the same cpus for as long as there have been on-die thermal diodes. C2D might be a little different but not much, IMHO.

The 5C spread was at load, idle they are all each within 1-2 degrees C of each other.

Unless we talk to the writers of the software and the engineers who designed the measurement points, I don't think we will ever quite solve this one.