Well, I haven't read all about these things yet, but from what I've read so far, both Core Temp and Real Temp use the DTS, which is what the internal automatic Intel/BIOS fan speed regulator uses as well.
A very interesting document (from Sep 2010) on the DTS sensor from Intel:
http://edc.intel.com/Link.aspx?id=2612Intel actually review both Real Temp and Core Temp. Apparently you can't trust any of them. One very interesting aspect is that the displayed temperature is close to the truth only when you are rather clsoe to max allowed temp. This might explain why ET6 produces quite unrealistic values (3degC below room temp for me!). Designers of Real Temp at least seem to be aware of these weaknesses, so I guess of the two, I'd trust Real temp quite a bit more. And they do say calibration of Real Temp is necessary. However, they calibrated their own stuff using an IR thermometer, which to me seems optimistic. If I believed an IR thermometer would show a correct reading of the core temp I would buy one for €50. You have to take into account thermal resistance from junction to point of measurement. But maybe they did that, I havent read all about the tests. They among other things present a table of expected diff between room temp and reported temp when idle, saying e g +10degC for Intel stock cooler. This seems quite useful and believable to me. If I get a believable value when idle, and the sensor is more trustwothy at higher temps, and the relations are linear (or 2-piece linear), I think the reported values actually might be close to the truth.
http://www.techpowerup.com/realtemp/docs.phpThe values in BIOS and in Real Temp and Core Temp all are around 29degC for me, and ET6 says 18.
Hopefully, for my purposes, I can set the automatic fan control so that it keeps the fans very quiet as long as the temperature is low, meaning it might let the fan speed be a bit low at low temps and increase the speed rather suddenly when it's really needed. Might decrease CPU life slightly but I want fans to be quiet during low usage. Whatever mistake developers of ET6 made, it's likely to be linear, and for that I can compensate with the parameters at hand.
Would be nice if Gigabyte posted at least a hint on these problems. But maybe they did, can't say I browsed all info.