Need Help! e8400 huge variation in core temps!

[Scrooge2]

Hi! I just bought a new e8400 to replace my old e6320.

However when I popped in the cpu, my mobo bios temp read it at 49C! This is weird because I had ran my e6320 (1.86ghz, 65nm) at 3ghz @ 1.287V. With my Thermaltake big typhoon, it was always at 25C idle and 35C load.

The e8400 is native at 3ghz and at 45nm uses 1.18v should run a heck of a lot cooler! What's even stranger is that Core Temp .98 reads core #1 as 43C and core #2 at 27C. And Real Temp 2.5 reads core1 as 32C and core2 at 17C.

What the hell? Why is core 1 and 2 so drastically different in temps? I'm quite sure my HSF is attached right!

 

[Scrooge2]

help?

 

[Razorfist]

I had heard that e8400 and some temperature readers were not playing nice together, including Core Temp.

I suggest trying some other temp readers and see whats up, Ill bet its not really at 47C.

 

[kpo6969]

Originally posted by: Scrooge2
help?

http://www.xtremesystems.org/f...howthread.php?t=179044

 

[rge]

The bios is reading the cpu temp (diode between the cores), and that diode is dependent on bios calibration. First try to update your bios to the latest. If that does not help, write your mobo maker and tell them to get in gear and release a bios that supports E8400 with an accurate cpu temp calibration.

Regarding core temps...they are independent of bios calibration but...Coretemp uses 105 tjmax which is too high, and thus gives temps 10C too high. I would read the thread that kpo6969 linked to above...post 916 will explain why tjmax 105 used by coretemp, everest, HWM is incorrect.

Check you load temps, if under load the temps even out, you're fine. Idle temps are nonlinear and inaccurate, yours are just more noticable, as core 1 is reading high and core 2 is reading low, instead of both reading high or low where it is less apparent.

If load temps even out at high range, you can use realtemp with guide to calibrate your inaccurate idle temps. Both of my sensors are reading high like your core 1, but using fluke to case, at UC, undervolted state, I was able using realtemp to make accurate temps from idle to full load (full load needs no calibration, around 60C temps are accurate on mine.

 

[Caminetto]

Originally posted by: rge

Regarding core temps...they are independent of bios calibration but...Coretemp uses 105 tjmax which is too high, and thus gives temps 10C too high. I would read the thread that kpo6969 linked to above...post 916 will explain why tjmax 105 used by coretemp, everest, HWM is incorrect.

Yes Core Temp is 10 degrees off for this chip.
You can use Real Temp or just edit the Core Temp settings.ini file from TjMaxOffset=0 to TjMaxOffset=-10