Schroedinger

Interesting results:

I underclocked to 1 volt and 1.2 Ghz last night:

No change in idle temps (maybe a degree relative to stock). Coretemp still shows 52-53 C idle temp and HWMonitor and Speedfan agree on Tcase 26-28 C. This is telling me that I have a near-25-degree gradient core to IHS at idle. Seems unlikely even with my heatsink on. I'd venture to guess that Realtemp is more accurate for E8400, but it will require further testing.

__________________
"The internet is text for people who can't read." -Andrew Eldritch

Burn

Identical results on an E6600, full blast...

Attached Thumbnails

 

__________________
My CNC Mill Project
We cannot change the cards we are dealt, just how we play the hand.
--Randy Pausch
Current Best Benchmark

GigaByte

Those are your accurate temps. E6600 is one of the most solid CPUs right now when it comes to Tj max.

__________________

tr8rjohnk

What an awesome thread but now my brain hurts, I've lost a couple of brain cells and I'm more confused than ever.

__________________
[Project] Goliath
Entertainment Center/Server/HTPC

"The reward of a thing well done is to have it done" -Ralph Waldo Emerson

"Life is what happens while you're making plans for something else." -John Lennon

"A foolish consistency is the hobgoblin of little minds." -Ralph Waldo Emerson

KarmaKiller

Well, after reading through the thread I had to test for myself. Both Coretemp and RealTemp show the same temps for my B2 stepping E6320. TJMax= 85C°

__________________

skunksmash

so my temps are wrong....?? if M0 is 100c tjmax

Attached Thumbnails

   

__________________

ATI83

Look like the CoreTemp is more accurate for my Q6600.
My CoreTemp,OCCT and HWmonitor read the same for me.
The RealTemp reading 5c lower than those 3.
The Asus AI Suite and speedfan are useless for reading TEMP.

__________________

20170 on 3DMark06
http://img110.imageshack.us/img110/9...ddddrq7.th.jpg
P11621 on 3DMark Vantage
http://img138.imageshack.us/img138/6...4018lx4.th.jpg

mothow

so whats better?

__________________

opt33

If more of you guys had flukes to measure the cpu casing, we could answer that question better. The problem is, intel has stated that TCC or "tjmax" may vary within the same stepping. So more data = better answers.

While it is difficult to prove a 5C difference with flukes, a 15C difference should be easy.

When DTS=0 at undervolted, underclocked state with no heatsink, if casing temp is measured with fluke at 95-100C, you would prove that your cpu had tjmax of at least 95-100C and irrefutably prove that tjmax is not 85C for your cpu.

Or, when DTS=0, if casing temp is measured at 83-85C, it would show that it is not possible to be 100C.

temp difference core/case = TDP x theta(core/case), at undervolted, underclocked state ~3-5W, theta is .3 to .4, so gradient is small, ie 1-3C or less at idle uc/uv.

zman

How is this possible? If you would use the theta .3-.4 figure for a QX9775 at load (TDP=150W) the difference from core to case would be 45 to 60 degrees! So with the specified Tcase max of 63C the cores would be at 108-123C, which seems quite high for sustained operation (keeping Tcase temp at or slightly below the intel thermal spec).

Or are you instead arguing that the core-Tcase theta value is highly non-linear? At least the thermal conductivity for copper is quite linear in the 0-100C range which would apply here. Of course I can not know which materials Intel is using between the cpu die and the IHS but I seriously doubt that the thermal conductivity is as bad as .3-.4, or as nonlinear as this value at idle would require.

So where does this .3-.4 value come? Intel documents? Someone removed an IHS and measured it?

Personally I have seen a value of .38 quoted for the core-case theta bu unfortunately I have no idea were that came from either.