View Single Post
  #9 (permalink)  
Old October 15, 2007, 01:52 PM
Lithotech's Avatar
Lithotech Lithotech is offline
Join Date: Apr 2007
Location: North Vancouver Rain Belt
Posts: 409

Originally Posted by burebista View Post
I guess you're aware that those bold temps are entirely off. Unless you're out of home at -5C ambient.

Will need cooler than -5C ambient to get those temps!

Yes I am aware. I don't pretend to understand how AMD monitors and displays the internal diode temperatures. I can only say that I've run into these crazy readings before with AM2 chips. I've even seen negative numbers.

Those temperatures are straight out of Everest Ultimate. They are verified with CoreTemp 0.95.4, which shows exactly the same numbers. CBId states that the thermal offset is -21C, but I fail to see how this can be applied to the temps that I get, and still have reasonable numbers.

So I go by the main external CPU temp for a rough idea of how hot she actually is, and use the borked diode temps to measure how much temp has risen from idle readings, disregarding the actual number itself. Sort of like Thermal Delta:

The formula is Thermal Rise = CPU Temperature - Ambient Temperature.
Except I don't do any calculations with Ambient temps, just how much it rises from idle:

Rise = CPU Diode Temperature Load - CPU Diode Temperature idle
Using this, I can see that the diode temps rise almost exactly as much as the main CPU external temperature. So while the numbers may be unrealistic, the rise is usually accurate.

I left it folding@home last night, 3ghz and 1.4v vcore. Current temps are:

CPU 47
Core 1 34
Core 2 22

Compared to idle temps:

CPU 36
Core 1 20
Core 2 9

An 11C rise on main cpu temp, 14C on Core 1, and 13C on Core 2. Reasonable for sure? Now, just add 13C and 25C to the core temps to bring them in line. Allowing a couple more degrees fudge factor, I therefore never want to see the core temps past 37C to 40C for Core 1 and 25C to 28C on Core 2.

Compared to idle and load temps at stock speed and 1.3vcore:

CPU 40
Core 1 26
Core 2 13

CPU 33
Core 1 18
Core 2 4

Under load, we see a 7C rise in the main cpu temp, not bad considering the speed and vcore boost. On core diode temps we see an 8C and 9C rise respectively; again, not bad.

At idle, we see only 3C rise in the main cpu temp, 2C rise for Core 1, and 5C rise in Core 2. There seems to be too much rise in Core 2 (compared to other results), which tells me I should double check my numbers and monitor it closely for a while to be sure things are OK.

My socket 939 X2-3800 doesn't show negative numbers, but I still have a 10 degree difference between the two cores. It still drives me nutz, and I have yet to see an X2 dual core cpu that didn't have a fairly large difference between the two cores, and between the main cpu temp.

eVGA 680i SLI (A1) P31
Intel Q6700 Kentsfield | Scythe Ninja-B | Antec Tri-Cool 80cfm
4GB (4x1024MB) Corsair Dominator 8500
boot: 4x WD Caviar SE16 250gb SATA2 Raid-0
backup: Seagate 500gb 32mb SATA2
eye: eVGA 9800GTX+ SC 512mb 770/2.25ghz
candy: Samsung 940B-HAS | X-Fi XMusic | Pioneer DVR-212D SATA
juice: PC Power & Cooling 750Quad | Cooler Master HAF932
steering: Logitech G15 & G5 | E3D Pro/Momo Wheel | X-530's
Reply With Quote