the right temp?

[Maverick986]

for a while now i've just used MSI's CoreCenter for my CPU temps and assumed they were accurate.....however the other day i started using SpeedFan and found that it is giving me drastically different temps than CoreCenter....CoreCenter usually gives me a reading around 105-109degF and now SpeedFan is telling me its more around 120-126degF. which is right???

I'm running a AMD 4000+ on a MSI K8N mobo

 

[Peepaw]

You need to take any software temp monitoring program with a grain of salt.
None of them are dead on accurate.
I have as much as 20°c difference on diff programs.

 

[unclewebb]

SpeedFan 4.32 does an excellent job of reporting the core temperatures correctly. They're the only important temperatures to monitor anyhow so you can ignore the rest.

 

[Maverick986]

Sweet deal....thanks for the info

 

[pdawg17]

SpeedFan 4.32 does an excellent job of reporting the core temperatures correctly. They're the only important temperatures to monitor anyhow so you can ignore the rest.

I'd love to use Speedfan but I was told that Coretemp or S&M give the "real" temps of the cpu which unfortunately report around 15C higher than Speedfan...I hope I was told wrong...

 

[unclewebb]

CoreTemp 0.95 assumes a Tjunction of 100C for the revision L2 E4300, E6300 and E6400 which I believe is a mistake. There is no publicly available documentation from Intel to support that and evidence from other users does not support that either. All Core 2 Duo desktop processors seem to have a maximum junction temperature of 85C which results in it showing core temperatures that are exactly 15C too high.

CoreTemp 0.95 will let you read the on chip digital thermal sensors (DTS) directly. That is the only Intel documented 100% correct temperature information for a Core 2 Duo desktop processor.

When CoreTemp is set like in the above picture it will show you how much head room until your processor reaches the thermal throttling point.
50C until it reaches the maximum Tjunction temperature at idle is pretty normal.

SpeedFan 4.32 assumes the maximum Tjunction = 85C in its calculations, which I believe is correct, so it can be trusted when reporting core temperatures.

Reported Core Temperature = Maximum Tjunction of the Processor - Digital Thermal Sensor Reading

CoreTemp = TjMax - DTS

 

[pdawg17]

Actually for me and my E6400 Coretemp reads about the same as Speedfan but S&M reads 15deg higher...what is S&M reading then? And I've read that when testing an overclock (orthos, etc), max core temp should not get closer than 20-25deg from Tjunction (which would mean a max temp of 60-65deg)...do most people base this on Coretemp then or a prog like S&M?

 

[unclewebb]

what is S&M reading then?

Who knows?

I haven't investigated the readings from S&M but I'll probably have a look at it now. I have done a lot of testing with CoreTemp 0.95 and I'm happy with what it tells me.

In my own testing I've found that when my revision B2 E6400 is overclocked it needs approximately the following amount of head room to be 100% Orthos stable:

3200 MHz ~ 5C to TjMax
3400 MHz ~ 15C to TjMax
3600 MHz ~ 25C to TjMax

Regardless of core voltage or anything else, if my core temperature increases and my head room decreases by a degree or two from those numbers, my computer will usually automatically re-boot soon after ward.

You can reliably run a Core 2 Duo at a higher temperature the less you overclock it but if you're going for big MHz then you better invest in some decent cooling to maintain stability.