if the thermal diode's used on current 45nm are fubar'd and the only one that works correcrly is distance to TJ. then we should be using that, as changing the TJ be it 95c or 105 the temps reported change. Though if you notice the distance varition never doesn. IE if you changed coretemp to a different TJ settting Distance to tj stays the same, though die temp changes. try it with core temp and watch die temps change as you increase or decrease the TJ setting. then with the new tj setting in the INI file switch from temps to distance to TJ. you will see the distance to TJ nevers varies when you adjust the TJ setting in the ini file.
I must be missing something here, maybe on how "distance to tjmax" works but... if you adjust what the program thinks your TJMax is (like, from 105 to 95), then wouldn't the distance to change, since it now thinks this value is lower?
Or, is distance to TJMax an absolute reading that comes from the diode itself, which would make it independent of what the software thinks the TJMax really is...
Eureka! Its sort of coming to me a I write this... if I understand correctly, then:
diode does not report an actual temp reading, bu just a distance to TJMax. TJMax is something which is "programmed" inside the chip, and as of yet, the community at large dosn't know what this value actually is (hence the 105/95 guesswork). Programs like coretemp and realtemp just use their own "theoretical" TJMax (a user defined setting), to give you an "actual" temp. That is:
(user defined TJMax {may not be accurate}) - (TJMax {absolute reading}) = "actual" temp (may be inaccurate due to user defined TJMax)
The reason TJMax is "guesswork", "theoretical" and "user defined" is that Intel has not released this info to the public, and 95C was a number arrived at through empirical evidence from testing and comparison against 105C, and NOT necessarily absolutely correct.
Therefore, the distance to TJMax is the only absolute value coming straight from the diode and not calculated on "guesswork" values, so it is the only "across the board" value for us to compare temperatures.
If this is correct, then as Evilsizer says, we need to stop comparing temps which are based on "guesswork" (i.e. TJMax in programs) and start using DISTANCE FROM TJMax, which is the only value we have coming straight from the diode and thusly not then calculated on guesswork. *WHEW*
So... is what I'm saying here accurate?
Stellar results at beautiful voltage.
(it's a lot of pages to run through)
ya TJ MAX is the only constant...
