C2D E6700 hot? 3 diff temps from all apps?

[snadge]

I have just acquired a Core2Duo E6700 2.66Ghz from eBay, Iam using it in a Foxconn 45GM motherboard, Iam using an Arctic Freezer Pro cooler which has been great previously getting my 90nm Pentium 4 3Ghz down to 30-35'c idle

when I got it running with the E6700 (idle at 1.6Ghz and idle at 2.6Ghz) it seemed that the following temperatures are given...

1.6Ghz - 2.6Ghz

RealTemp = 49'c - 57'c
Core Temp = 43'c - 52'c
Speedfan = 39'c - 48'c

I have re-sit the CPU and cooler again and got same results...

which one do I trust?

is it possible the E6700 has been damaged from Overclocking?

EDIT: right now Iam using SpeedFa in tray and its reporting 42'c idle - when i leave in BIOS its 39'c-40'c flipping between the two, the fan is up from 1300 to 1500 as it kicks in at 35'c... seems speedfan is about right... the fan isnt speeding up so cant be 50'c - still, 42'c is a little high for my liking...

thanks

 

[newtekie1]

Idle temps don't matter, what are you load temps?

And take note that these chips tended to have temperature sensors that would stick at lower temperatures, making them extremely inaccurate for reading idle temps anyway.

 

[snadge]

Hi

thanks for speedy reply

I have not load tested yet... but... right now Iam using SpeedFan in tray and its reporting 42'c idle - when i leave it running in BIOS its 39'c-40'c flipping between the two, the fan is up from 1300 to 1500 as it kicks in at 35'c (this is when Im in BIOS)... seems speedfan is about right... the fan isnt speeding up so cant be 50'c - still, 42'c is a little high for my liking...

I have ordered some MX-4 paste as the stuff I used was what came with my dads cooler last year...crap stuff (although seemed to work ok on my 90nm Pentium 4 @ 3.06Ghz getting it too 35'c idle)

when I get the paste I will re-sit it and burn test it..

am i right in assuming that the sensors dont report actual tempratures but are some form of 'gauge' that the program has to read and determine tempratures from?

I always though RealTemp was the best but for me it seems to be way out?

thanks again

 

[unclewebb]

The way Intel's core temperature sensors work is that they only tell you how far away you are from the thermal throttling point. RealTemp reports this value as the Distance to TJ Max.

If a program knew the value of TJ Max then it would be simple to calculate the core temperature.

Core Temperature = TJ Max Value - Distance to TJ Max Reading from CPU

The TJ Max value is fully documented for the newer Core i processors but in the early days, Intel was never willing to document TJ Max for their Core 2 processors. Anything published was misleading to say the least.

RealTemp is the only program that I know of that assumes a TJ Max value of 90C for these early processors. If you have a look at that formula, that's why it reports the highest core temperature compared to other monitoring programs.

Most software originally assumed that the correct value was 85C but this was based on some Pentium 4 documentation that was completely irrelevant to the newer Core 2. The theory was that if a certain bit in the CPU was set then TJ Max was 100C and if that bit was not set then TJ Max was 85C. This is how the 85C TJ Max myth got started.

RealTemp decided to go with 90C based on some hands on testing with an IR thermometer of an early E6400 as well as an E6600. I am confident in that value for these CPUs but you will never see that number in any Intel documentation. Intel refers to TJ Max as TJ Target in later documentation. They admit that there is no such thing as an exact value. They might shoot for 85 or 90 but these sensors are not 100% accurate so the exact value is a mystery. I think they didn't want anyone to know that they were using temperature sensors that were not 100% accurate so they kept this quiet for as long as possible. These sensors were only designed to control thermal throttling and thermal shutdown and they are always good enough for that. Intel never thought that users would need to know about their exact core temperature as long as their CPU was working correctly.

If you think RealTemp is full of crap, go into the RealTemp Settings window and you can change this value to 85C just like Core Temp uses.

Core temperatures should always be higher than bios temperatures. These two temperatures are being measured at different locations on the CPU and there can be some significant temperature gradients on a CPU, especially when fully loaded.

Here is some of the "hands on" testing I did in the good old days.
I don't know of any other programmers that abused their personal CPUs like this in the name of science.

http://www.techpowerup.com/forums/showpost.php?p=962305&postcount=25