--Conroe max temp debate--

I have read many conflicting results on temp monitoring. My general understanding was that to be 100% safe keep ur cpu under 55c and 1.5v (for air). But, using which type of temp monitoring? Here is my example.

I am assuming both of my readings are correct here because it would make sense for speedfan to read lower then a direct core reading.

Speedfan:
Mobo - idle39c/load41c
Cpu - idle25c/load46c
Hdd - 22c
For comparison the mobo and cpu temps are reported the same in the bios (1305)

Coretemp:
Cpu - idle40c/load58c (both cores are kept around the same temps cause of Orthos priming, so I made it one reading).

So whats the deal here? Assuming they are both correct, which one should be under 55c? And if its speedfan, what should be my max coretemp because the Tcoremax is rated at 85c.

P.S. I have a stacker with a 1 120mm back exaust (having it intake make help my si-120 though...), 1 120mm front instake, 1 80mm top exaust, 1 makeshift 120mm blowing down on my ram/chipset, and 1 120mm mounted on my si-120.

go with what your motherboard says. MY motherboard says my e6400 runs at 56C on full load in dual prime95 but core temp says something like 70C or something absurd. If the CPU shrivles up and dies, I'll get asus to spring for a new one for not giving me correct temperatures. I'm sure thousands of people would get asus to spring for a new one in this way, but I'm also sure that the motherboard is reporting the correct temperature.

Why debate with absolute strangers... when the Specificationsare available for all to read.

Thats one of the great things about Intel. Their documentation.
None of this brokeee ingleesh from competitors.

Look go to the Intel web site. Look up processors.... Core 2 duo
Then look up documentation.

What you want is a pdf sorta like E66xx series family DATASHEETS.

There one can find EXACTLY how it is supposed to work.

Granted Chapter 5 is a tough read... It is written for engineers.
But if one was half way decent in High School, they should be able to
muddle through and get the drift.

One can ask a hundred people online and get a hundred different opinions.

What you need to do is go to the source.

God only knows if the temps are right and who cares. All that matters is that your clocks are stable. As long as I'm stable and I can touch the heatsink without being burnt, I don't worry.

Quote:

---

Originally Posted by chron

go with what your motherboard says. MY motherboard says my e6400 runs at 56C on full load in dual prime95 but core temp says something like 70C or something absurd. If the CPU shrivles up and dies, I'll get asus to spring for a new one for not giving me correct temperatures. I'm sure thousands of people would get asus to spring for a new one in this way, but I'm also sure that the motherboard is reporting the correct temperature.

---

lol i am 99.999999% sure that the motherboard is reporting wrong temperature. i highly recommands you follows core temp results.

Quote:

---

Originally Posted by Bail_w

lol i am 99.999999% sure that the motherboard is reporting wrong temperature. i highly recommands you follows core temp results.

---

Then you are 99.999999% wrong. Both the mobo and Core Temp read temp sensors located on the C2D die and both are accurate. The mobo temp is read from a conventional thermal diode that approximates Tcase while Core Temp, iTAT, etc. read separate digital thermal sensors (DTS) on each core that are completely unique to C2D and not comparable to any other CPU temps.

If the thermal diode temps are under 60C max and the DTS (Core Temp, etc.) readings are under 70C max, you have nothing to worry about. Thermal protection kicks in at 85C based on the DTS temp.

Why does every program I use for temp monitoring give me the exact same results then? (TAT, CoreTemp, Asus Probe, Speedfan)

I am slo confused as to what "safe" temps are for 24/7 usage... im at 50C idle and 62C load no matter what program I use... it seems high to me, but I guess its not all that bad.

Quote:

---

Originally Posted by Rmc[AZ]

Why does every program I use for temp monitoring give me the exact same results then? (TAT, CoreTemp, Asus Probe, Speedfan)

I am slo confused as to what "safe" temps are for 24/7 usage... im at 50C idle and 62C load no matter what program I use... it seems high to me, but I guess its not all that bad.

---

I suppose I shouldn't assume that every mobo reports the same diode temps as the DS3. It seems that most C2D mobos report diode temps that are ~10C lower than the DTS temps but not all do. Since Intel QST doesn't even spin the fans up to 100% until 66C (DTS), 62C load seems very reasonable and safe.

Believe the temps, especially the digital temps.
If they were broken, do you think both core digital readers would break, and that they would both read over?
Read the E6XX family processor datasheets from Intel Chapter 5.

I have two heat sinks for the E6600 and both were concave.
I smeared AS6 on the ceramic top of an old 120MHz Pentium and rubbed the base of my stock heatsink trying to transfer as much AS5 as I could.
When I pulled it away to examine, only a pencil thin ring was transfered to the heatsink. I sanded the heatsink with 400 grit and then 600 grit.
Retested to see how much would now transfer. Almost the entire HS now got the paste transferred. This dropped my temps by 10C degrees.

My AC freezer 7 pro Heatsink was worse. If two out of two Heatsinks were bad for me.... shouldn't you do the same test if you are experiencing bad temps. Oh by the way the OS matters too... see my signature

So sick of this CPU temp argument, just load of the orthos and run it without air condition or above your average room temperature while running throttling monitor program. You'll get a clue how hot is hot ! That is the easy way.

Or the other way, a bit hassle though, just detach your cpu fan from the hsf and while holding it with your finger, move it closer or away to see how the cpu react on throttling temp and take a note when it starts to throttle, trust me, you'll surprise cause some of you might get above 100C which is way off the theoritical limit ! :D

Guys, after looking at those gazillion posts of "mine are accurate I trust it" vs "mine are sux cause its way off", it is easy to conclude that those temp sensor either in the cpu or mobo sensor are not "accurately" calibrated, your dumb program can't do anything to fix that.

C'mon guys, your mobo or cpu are a mass product commodity stuff and anyway they're not supposed to sell you a thermometer. Get use to it. :D

Attachment 51567
Image courtesy of www.overclockers.com

Or, just take a brave steps like above pic if you're still not satisfied, with an accurate "real" temp probe and stick it right in the middle of the IHS.

Fyi, this is the only "proper" method to measure CPU temp and Intel use this method too on their standard lab measurement, just search for the white paper on measuring CPU temp on their site, you'll get tons of it.


-

From Intel:

Intel core 2 extreme Processor x6800 and Intel core 2 duo desktop processor E6600 sequence

First of all Throttelling does not occur at just one temperature, it is a function of power being consumed by the processor at that instant.

Y = mX + b (Y = Throttle temp, x = watts currently being consumed)
(Thank you Mr Yaches, HS math teacher}

************************************************
For the Core 2 Exteme X6800 : Y = .23x + 43.2C
X or the Power, ranges from 0 to 75 watts
Thus throttling for this processor can occur anywhere from 43.2C to 60.4C
I don't know how to tell exactly how many watts the processor is consuming at an instant. But this much I know: IF THE E6800 IS AT OR BEYOND 60.4C IT IS THROTTLING, AND YOU ARE LOSING PERFORMANCE.

************************************************** *******
For the Core 2 Duos with 4MB L2 cache Y = .26x + 43.2C
X or the Power, ranges from 0 to 65 watts
Thus throttling for this processor can occur anywhere from 43.2C to 60.1C
I don't know how to tell exactly how many watts the processor is consuming at an instant. But this much I know: IF THE E6700 OR E6600 IS AT OR BEYOND 60.1C, IT IS THROTTLEING AND YOU ARE LOSING PERFORMANCE.

************************************************** *******
For the Core 2 Duos with 2MB L2 cache Y = .28x + 43.2

X or the Power, ranges from 0 to 65 watts
Thus throttling for this processor can occur anywhere from 43.2C to 61.4C
I don't know how to tell exactly how many watts the processor is consuming at an instant. But this much I know: If the E6300 OR E6400 is at or beyond 61.4C IT IS THROTTLING AND YOU ARE LOSING PERFORMANCE.

================================================== =====
NOW WAIT JUST ONE DANG SECOND !!!

YOU MEAN TO TELL ME MY CHIP CAN BE THROTTLED AT EVEN 50c ????

YES !

I know its not intuitive... but bear with me for one more second...

Consider two identical chips, one is at 59C currently consumming 65 watts
the other is at 50C consumming only 10 watts... Which one is in trouble?

Actually it is the cooler second chip that is in trouble, and the one that will be throttled.

The first chip is consuming almost max power and working its tail off, but it is staying within the defined limits of throttling. He is ok.

Its the second chip, the cooler one, is beyond his recommended temps, given the power he is consuming.

Analogy: Two people:
One is currently exercising (running a mile) and has a current heart rate of 90
the second is sitting and has a heart rate of 85
Who is more in trouble?

================================================== ======
Ok now lets define throttling:

Thermal Monitor : turns the clocks on and off rapidly
Will occur at Throttle point defined above,
but definitely by (60.1 to 61C depends of Processor type)


Thermal Monitor 2 : Lowers the voltage to the processor,
and reduces the multiplier
Will be applied after Thermal Monitor if Thermal Monitor
doesn't bring temps back below current throttle point.

Thermal Trip - the drastic action of shutting off the power to the processor
after it reaches about 25C higher than the throttle point.

================================================== =======

I don't know of any up to date software that actually shows Throttling for the New C2D and C2 extreme processors.

Some people say they monitor it in TAT, but I think they are misinterpretting the Software induced throttling feature of that package. The processor not only throttles itself, but allows the software to command it to.

If I am wrong about the TAT, someone please correct me,
I would like to see a screen shot...if you could.
{I don't like to heat my 600$ investment up just to prove this point. :)

================================================== ====
Hope I'm correct on all the above...
One always loses something in simplification...
Its not simple...
But remember you do have the manual...
{written for a higher level than me... but I tried...

This really is getting out of hand. Yes, the DTS are more accurate than the thermal diode but the temps they report have nothing to do with the max Tcase temp of 60C. Additionally, since ONLY C2D has DTS, those temp readings are NOT COMPARABLE to any other existing cpu on the market. So if you like, you can use iTAT or CoreTemp DTS readings for C2D comparisons but you can't accurately compare them to any other cpu that uses a conventional thermal diode. For that, the best temp reading to use is the C2D's thermal diode reading. Does anyone actually believe that a 65nm 65W C2D runs hotter than a 90nm 125W FX62? Because that's what iTAT and CoreTemp suggest.

Further, C2D thermal throttling protection occurs at 85C based on the DTS readings and not before, unless you're allowing EIST and C1E to throttle your cpu for power savings and thermal reduction. There are plenty of programs available that monitor cpu clock throttling. CPU-Z, iTAT, Throttle Watch, etc. will all show you if your cpu is throttling. I can state with absolute certainty that my E6400 has never throttled down even at 70C (DTS) but I'm not going to turn off the fan or remove the heatsink to prove that it starts throttling at 85C. Others have already done it.

Quote:

---

Originally Posted by Fred_Pohl

This really is getting out of hand. Yes, the DTS are more accurate than the thermal diode but the temps they report have nothing to do with the max Tcase temp of 60C.

---

Fred some of what you say is true, but you are mistaken about a number of points. I can only address them one by one. I will address the others as I have more time.

See Intel Core 2 Duo Destop Processor Sequence Thermal and Mechanical Desing Guidelines.

Specifically Paragraph 1 Section 4.2.10

"The DTS is monitoring the same sensor that activates TCC."

So you see it is the digital thermal sensor (the one reported by core temp)
that one should be concerned with.

Quote:

---

Originally Posted by Fred_Pohl

Further, C2D thermal throttling protection occurs at 85C based on the DTS readings and not before, unless you're allowing EIST and C1E to throttle your cpu for power savings and thermal reduction.

---

This is just blatantly wrong, and it might lead others to ignore bad temps which need correcting.

Throttling can occur any time the die temp exceeds the thermal profile for the processor. This can occur at as little as 43.2C depending on power consumption, but since the processor cannot consume more than its max(65 watts for C2D, 75watts for Extreme), you are definitely gonna trip TCC if you are over 61.4(extreme)

Please see my earlier post.

What you refer to at 85C, is what is generally referred to as Thermal Trip, or the point where the processor takes the drastic action of shutting itself off.

EIST is not really triggered by temperature. But rather, is triggered when the Processor has no work to do. Instead of letting the processor spin its wheels at full speed while idle, it lowers the multiplier, and voltage setting,
and lets the processor cool faster. {don't confuse EIST with Thermal monitor 2 which also reduces voltage and multiplier}

My Infinity heatpipes are cold, yet coretemp shows 65C, even up to 70C at only 1.45V full load orthos (contact is good, don't blame that :p: )

And there is def no throttling and I am not losing performance since under single stage I get the same performance at that speed), at least 3dmark and SP :)

http://img140.imageshack.us/img140/4...0268fo5.th.jpg

Check that , I am running on the edge of things

Quote:

---

Originally Posted by GWillakers

Throttling can occur any time the die temp exceeds the thermal profile for the processor. This can occur at as little as 43.2C depending on power consumption, but since the processor cannot consume more than its max(65 watts for C2D, 75watts for Extreme), you are definitely gonna trip TCC if you are over 61.4(extreme)

---

The maximum wattage can easily exceed specified wattage with overclocking. Based on the square of any overvoltage and increased frequency.

e.g. an E6600 at 3.6Ghz
max wattage = 65*(3600/2400)*(1.5/1.325)² = 125W

With your interpretation, throttling can never occur as low as 43.2C, even an idling CPU uses more than 0W. Also the thermal profiles you refer to are the external case temps Tc, not on die.

What it does say is that the thermal monitor TCC modulation activates when the TDP is exceeded, i.e Tc >Tc-max

Quote:

---

Originally Posted by GWillakers

This is just blatantly wrong, and it might lead others to ignore bad temps which need correcting.

Throttling can occur any time the die temp exceeds the thermal profile for the processor. This can occur at as little as 43.2C depending on power consumption, but since the processor cannot consume more than its max(65 watts for C2D, 75watts for Extreme), you are definitely gonna trip TCC if you are over 61.4(extreme)

Please see my earlier post.

What you refer to at 85C, is what is generally referred to as Thermal Trip, or the point where the processor takes the drastic action of shutting itself off.

EIST is not really triggered by temperature. But rather, is triggered when the Processor has no work to do. Instead of letting the processor spin its wheels at full speed while idle, it lowers the multiplier, and voltage setting,
and lets the processor cool faster. {don't confuse EIST with Thermal monitor 2 which also reduces voltage and multiplier}

---

Excellent points. I was wrong about 85C being the thermal throttle point. In reviewing Intel's data sheets more thoroughly I found that 85C is the max Tstorage which has nothing to do with operating limits. That's what I get for believing other member's opinions without doing my own research.

Thermtrip causes a complete shutdown, as you said, and occurs at 20C above Tcase max which should be 80.4-81.4C depending upon C2D model. One point that I kept seeing repeated is that all of these thermal features need to be enabled to work except for Thermtrip. Since I believe that most C2D overclockers have these features disabled in bios, that might explain why no one seems to be seeing any throttling even at core temps above 61.4C.

I'm beginning to think that a lot of people might be running their C2Ds outside Intel's recommended thermal guidelines. The more I read the data sheets, the more it appears that ~61C (DTS) is the max recommended Tcase. Of course, with TM and TM2 disabled, the only protection left is Thermtrip at ~81C.

However, questions remain. For one, why would iQST reduce fan speeds at DTS temps below 66C if ~61C (DTS) is the max Tcase? When you disable C1E, EIST, TM2 and QST in bios, does this disable TM too? It seems to me that it would since without any means to reduce clock, multi and Vcore, how can TM work?

Quote:

---

Originally Posted by fornowagain

Also the thermal profiles you refer to are the external case temps Tc, not on die.

---

Ok... Hold on...
I'm just trying to piece this stuff together like all of you.

Since I have a few people to answer, let me take em one at a time..

1.3 Definition of Terms
Tc The case temperature of the processor, measured at the geometric center of the topside of the IHS

Yes some of the stuff you quoted, I did not properly phrase...
I will rephrase...

Quote:

---

Originally Posted by fornowagain

The maximum wattage can easily exceed specified wattage with overclocking. Based on the square of any overvoltage and increased frequency.

e.g. an E6600 at 3.6Ghz
max wattage = 65*(3600/2400)*(1.5/1.325)² = 125W

---

I grant you here is where I could have phrased things better.
I should not have said the processor couldn't consume more than the max speciified by Intel... obviously it could.

I am trying to constrain the analysis to situations where one is running within the guidelines Intel put down.. {i.e. running in spec, no O/C}

The point that I was trying to make is that if a processor is beyond 60.1 - 61.4C (depending on type) then it has passed the maximum temp on the defined Thermal profile that Intel has set forth, and the TCC will be activated.

http://users.erols.com/chare/elec.htm#intel

Here is the list for those who missed it. Also I noticed the other cores have better and higher thresholds but that also maybe something Intel wants people to beware and be scared cos its a new core and a highly overclockable one. It is the first year of this core and its the second 65nm chip , the pentium D had higher threshold than core 2 duo , weird or is that something intel wants us to believe because of how much power the chip is capable of.

I am going to find out , I am running at 83c on both cores with core temp , its been running for a hour and it hasn't stopped or throttled as yet and I have not disabled this feature in the bios either.

I have seen it run 70c without it even being overclocked on Prime , I think intel has got us all scared.

There have been a lot of overclocks round forums and we still havent heard anything about this but we have heard that the core 2 duo does run hot.

Also the HSF is dissipating heat really well , also noticed the laptop cpu's where they can withstand 100c and our desktops cant , I recon its lies . Can you imagine if instead of intel telling the public 61c they wrote 100c what would happen?

We would all be running them at 4.3ghz 24/7 .lol

Quote:

---

Originally Posted by GWillakers

The point that I was trying to make is that if a processor is beyond 60.1 - 61.4C (depending on type) then it has passed the maximum temp on the defined Thermal profile that Intel has set forth, and the TCC will be activated.

---

It's also important to note that TCC activated clock throttling, reduced multiplier and Vcore will only occur when TM and TM2 are enabled. Otherwise you'll have no TCC thermal protection until ThermTrip kicks in at ~81C.

As long as it stays out of the high 70s / low 80s, I can care less. I load 65c in TAT @ 3750mhz @ 1.52v w/ my Infinity. I never see it go over 60c because I never use 100% ;).

MY thought process on the subject has stayed the same... If asus is telling me my processor is at 40C idle 55C load, I'm going by that. Core temp tells me I idle at 58C load 70C. When I touch my heatsink it does not feel this hot. I don't know what else to do than to say F it and wait for my computer to splode.

Quote:

---

Originally Posted by EZClock

As long as it stays out of the high 70s / low 80s, I can care less. I load 65c in TAT @ 3750mhz @ 1.52v w/ my Infinity. I never see it go over 60c because I never use 100% ;).

---

Agree with you on that , same here

http://img81.imageshack.us/img81/7195/img0305rp9.jpg

Anyone of you are electronic enthusiast or hobbyist here ? or better electronic engineer ! Not apllicable if you don't undertand few electronic jargons below here.

Have you ever make a DIY electronic thermometer using plain silicon diode/transistor as the temp probe ? You will have to go through tedious steps to calibrate that sensor just to get a "good enough" accuracy. Worst, if you replace that sensor even the same type/manucfacturer/batch, you'll have to recalibrate again !

Same problem apply to Intel too, anyway the CPU is made from the same silicon as ordinary stupid diode/transistor, and it is way too costly to calibrate that on die sensor at each CPU since they need to toast it at constant known temperatures hot and cold to get an accurate calibration at "EACH" CPU, so it will never be accurate, period.

Also, those what you called cool feature "Digital Sensor" ..yuck.....they are just plain olde stupid current sensing method to guestimate the cpu temp.

Making an accurate resistor in silicon wafer is easy and just use current sensing method and feed it thru the on chip A/D and voila !!!! A digital sensor was born ! :D Again its just to guestimate power consumption and dissipation.

While making a consistent/same forward voltage on every piece of silicon to detect same temp changes will be never easy, it is the law of the nature, unless they're capable of putting silicon atom by atom "accurately' to make that piece of diode temp sensor.

Now, why those 85C magic number, cause they've decided even the sensor reading is off by 10-20C, it is still considered safe as long you don't overclock, overvolt or within the acceptable ambient temp range like your dad or mom's puter, of course with the proper HSF mounted.

Just read tons of electronic components datasheet out there, almost all of those silicon based electonic components are capable to withstand an operating temp at around 100C or even higher.

It doesn't matter what methodology/mechaism you folks are arguing, if the sensor is not accurate, its useless.

Worry and want a safe long life cpu ? Then stop OC-ing or overvolting, simple as that.

Cpus can live long even when oc and overvolting :)

Do you care if it lives 15 years instead of 30+? :p:

Well, sure we are not talking about running Conroe at 1.9V 24/7 fully loaded :D

Quote:

---

Originally Posted by Fred_Pohl

Then you are 99.999999% wrong. Both the mobo and Core Temp read temp sensors located on the C2D die and both are accurate.

---

Quote:

---

Originally Posted by chron

go with what your motherboard says.

---

:rolleyes: :rolleyes: :rolleyes:

Cooling in sig.

http://www.rose-hulman.edu/~sharpjm/...ore2%20035.jpg

Well, the sensors are reasonably accurate within normal operating conditions but the mobo doesn't necessarily interpret the sensor readings correctly. Case in point.

Quote:

---

Originally Posted by Fred_Pohl

Well, the sensors are reasonably accurate within normal operating conditions but the mobo doesn't necessarily interpret the sensor readings correctly. Case in point.

---

oRlY?:stick:

Quote:

---

Originally Posted by dr_sharp

oRlY?:stick:

---

Maybe you got a bad thermal diode. Who knows? Does that mean that none of them are reasonably accurate?

Quote:

---

Originally Posted by Fred_Pohl

Maybe you got a bad thermal diode. Who knows? Does that mean that none of them are reasonably accurate?

---

What is your definition of reasonably accurate?

Quote:

---

Originally Posted by dr_sharp

What is your definition of reasonably accurate?

---

At least as accurate as the thermal diode on any other cpu, much more accurate than mobo socket thermistors and within 5C +/- of actual Tcase.

I modified my e6300's IHS, by protecting it with the shipping cover, then rubbing it flat on very, very fine grade wet & dry sandpaper.
This put the IHS's sides on the same level as the middle and it had an unexpected benefit... it exposed a nice little copper spot directly above the core, right in the middle of the IHS...

The reason I did this is because, when I place a straight edge over IHS, it was dished in middle. it was the best part of 1mm higher on the outsides!!! Meaning no contact at the core (middle)!!

Then I lapped it with good quality car polish...

Voila!!!, 10oC cooler than previous!!! :D

Sorry this is a repost from earlyer today.

Onboard temperature monitoring is pointless, you are better off giving your dog 100 buscuits to eat, letting it eat for 3 minutes then counting the remaining biscuits and using this number as your load temp. Give the dog another minute of eating to get idle temp..
For greedy fat dogs/prescott use 200 biscuits, for small dogs/via chips use only 50.


IMO all the onboard monitors and motherboard readings are suspect. None are propperly calibrated and none of them can be trusted to give an accurate reading IMO.

If your system is running STABLE and your voltage is sensible then I wouldnt worry about your tmeps at all.
If they suddenly jump up for no apparent reason then check it out.
Otherwise ignore the readings.

Has ANYBODY killed one of these CPU's yet? Ive not seen many posts reporting dead ones.
Your all worrying about nothing IMO.

I went through this years aggo with the EPoX 8K3A+ when I spent ages trying to calibrate the sensors with many different XP chips. They were all different and general calibration was deemed impossible..
Diode reading was subsequenty removed from following epox products due to customer confsusion and loads of extra annoying support mail to answer.

im thinking Holst has a point

Im also having temp porblems with my p5b dlx and 6600, as soon as i raise voltage above normal i get throtteling under load and im sure my hsf can handle it so i turned of TM and even tested when computer is gonna shut of cause of the heat and it was at 111c (asus probe) but it was usualy throtteling at 85ish, so either there is a 25c difference between TM2 and shut down or the sensors are wrong

edit:

what i would really really like to know is: Is a Same CPU on different mobos that give out different readings going to throttle at same temperature or compesated, just as an example

one P5BD with 40 idle 70 load throttle at 85
and Ga DS3 (or whatever) that does 30 idle 60 load is it also going to throttle at 85 or is it going to throttle at 75

(cause i have seen some ppl that say that a same cpu was 10c "cooler" or hoter on different mobos)

I think the variations everyone is witnessing is more CPU side than Mobo side. I have a L626 and a L629 , here are some temps I reached with both chips. Core temp used for all testing

----------L626 / L629

Stock idle 35c / 40c
stock load 47c / 54c

Oced idle 42c / 52c ( this oc is based on the L626, 1.6 vcore @ 3.5 ghz / L629 set to match)
oced load 54c / 65c

At one point intime I put 1.7 vcore on the L626 to get 3.6 and it never loaded over 60c , but as you can see it wasnt the best ocer. Now my L629 uses 1.6 vcore for 3.82 but runs so much hotter according to Core temp. In closing I beleive Bing was on the mark with his explanation :)

it has to be on both sides, on the cpu and in the BIOS, in combination.

The electical resistance curve is a function of the silicon diode/thermister variability.

The formula to associate the resistance to the temperature is in the BIOS or software used to report the temperature.

There is no calibration made between the silicon resistance reading to the predicted temp from the formula so the variability is a function of both sources of error.

my Abit AW9D-Max reads my CPU temps roughtly 8-10C higher than my DFI Infinity did. Same CPU, same cooling, 3 remounts already to make sure...

Coretemp temperatures, or some other program? If it's Coretemp, then that's pretty big information.