RealTemp & Speedfan II (Huge difference between them)

[Nikolayevich2010]

Hi guys; hope you can help me; I just read this thread;

http://forums.techpowerup.com/showthread.php?t=108326


and I'm in a quite similar situation:

I've tested the temperature with: speedfan; realtemp and CoreTemp.

The result: Speedfan and coretemp with very similar lecture; but quite different to RealTemp lecture with a difference ≈(15°C to 20°C at max)

I'm running windows 7. 32bits with intel core 2 duo;

Best seen here http://img188.imageshack.us/img188/5448/picxu.png


Who should I trust ?!?! I'm afraid the actual temp might be RealTemp's since it shows a max of nearly 70°C with a huge fan underneath the PC! If that's the actual temp, the thing may just fry when I turn the fan off...

Thanks very much for any help.. (I pretty much of a noob regarding to procesors and stuff; the only thing I know (cos I just read it here) is that the Max TJ is the max temp the procesor can deal with before shutting off)

 

[ehume]

You need to make sure you get Speedfan 4.40. The developer fixed his problems, and now his core temps are right in line with Real Temp.

Before that it was not something you could rely on.

 

[somebody]

If Intel's specification of Tjmax 100C for the T5250 is correct then Real Temp would be right. You appear to be using the Gulftown version of RT but it's probably only aesthetic.

Tjmax is the temperature the cpu will start throttling, if enabled. It will probably shutdown some ~20C above that. Since this mechanism is only supposed to be a secondary safety feature for the cpu your ACPI should take care of things before then.

 

[Nikolayevich2010]

Hi there; thanks very much for the reply;

A few moments before the snapshot I had uninstalled SpeedFan 4.40 cos it had some weird big sudden jumps (about 10°C);

so I intalled SpeedFan 4.38 wich in vista didn't do that;

but after a while I realised it does the same big jumps...

Anyway If ehume say the developer solved some of those stuff; I'm installing 4.40 and

posting new pics of comparision

Cheers guys

 

[Nikolayevich2010]

 

[unclewebb]

http://processorfinder.intel.com/details.aspx?sSpec=SLA9S

Here's the basic formula that all software uses:

Reported Temperature = TJMax - Digital Sensor Reading

All software is reading data from the exact same digital sensors. This sensor counts backwards so as your CPU gets hotter, it counts down instead of up. The digital sensor data is reported as Distance to TJMax in RealTemp. To convert that data into a meaningful temperature you need to know what core temperature your CPU will be at when it counts down to zero. This is called TJMax or the Intel site lists that as the Thermal Specification.

Luckily for your CPU, Intel says this number is 100C. If SpeedFan or other software assumes that this is 85C then hopefully by looking at that formula you will see that other software will report your core temperatures 15C cooler than what RealTemp is reporting.

Laptop CPUs can run very hot because many of them are poorly cooled. A heatsink that is not attached snugly with a poor application of thermal compound will result in high core temperatures. If you ever get bored you can pull your laptop apart and try to re-engineer your heatsink. Your core temperatures are presently fine. These CPUs won't explode even when operated in the 95C to 100C range for extended periods of time. Your lap will complain before your CPU does.

Stop worrying and enjoy your laptop.

RealTemp 3.50 beta
http://www.sendspace.com/file/u79b5p

 

[Nikolayevich2010]

unclewebb ! Thanks very much for the clear explanation !!!!

I'd have never guessed that actual temp was RealTemp's ;

BTW Is there a way to modify the TJMax in speedfan ?

Speedfan should show the actual temp instead of the TJMax distance don't you think?

I think i'd be easier for people to handle (despite it's a simple formula; by default I assumed it was the actual temp)

 

[burebista]

BTW Is there a way to modify the TJMax in speedfan ?

You can specify an offset for every core.

 

[Nikolayevich2010]

Just one final thought;

If SpeedFan temp. is actually the distance to tmax;

Why when the PC is offloaded; the temp is really low; like 20C and increases as the PC gets work.. ??


Shouldn't it be otherwise ?

 

[unclewebb]

Nikolayevich2010: You're not understanding this correctly. SpeedFan and RealTemp are both using the same formula and both reporting the temperature of your CPU. The difference is that SpeedFan is using the incorrect TJMax in its calculation. If it uses a TJMax of 85C when the actual TJMax is 100C then SpeedFan will always report exactly 15C less than RealTemp.

When the internal sensor is reading 40

RealTemp = 100 - 40 = 60C
SpeedFan = 85 - 40 = 45C

When the internal sensor is reading 25
RealTemp = 100 - 25 = 75C
SpeedFan = 85 - 25 = 60C

Now does it make sense?

The internal sensor that Intel uses is the equivalent of a reverse thermometer. As the core temperature gets hotter, this sensor counts down to zero. It's up to software to use the correct TJMax for that processor to convert the sensor data into a meaningful temperature number.

If you want SpeedFan to report the same temperature as RealTemp, you will have to go in and set an offset of 15 as burebista has shown. This will correct for the fact that SpeedFan didn't bother to follow the Intel documentation for your CPU.

 

[Nikolayevich2010]

unclewebb; Now I understand ! Thanks very much for the reply;

I've now set the offset values in both cores to 15°C in SpeedFan;

So; there's no diference in speedfan and RealTemp's lecture;

 

[Nikolayevich2010]

BTW; is it safe for it to be; between 50°C to 60°C ?

Yesterday; I did some tests; and stared running many heavy programs; and the thing reached above 80°C (without the external vent)!

 

[unclewebb]

Here's my E8400 running for 3 hours with the fan disconnected. It got a little hot and bounced off the Intel thermal throttle for 3 hours but never skipped a beat. Now that's what I call Prime stable.

This is my daily use CPU and I'm still using it today. I've been running it at 4 GHz and 1.40 volts for a long, long time without any problems.

When you are not stress testing your computer, the temperatures will be very reasonable and you shouldn't have any long term problems.