Has any concensus been reached about the accuracy of "CoreTemp?"

[Frackal]

Just curious, the temps to me do seem a little misreported on the high side

[nikhsub1]

Just curious, the temps to me do seem a little misreported on the high side

LOL, and you base this on what? Everyone, yes EVERYONE refuses to believe actual temps. Coretemp is taping into the thermal diode in the cores of each CPU. Notice how quick temps jump from idle to load? I bet they are very close to reality unlike Asus probe or speedfan.

[lawrywild]

I think they are probably pretty accurate, just that we are all used to seeing lower temps on other cpus/ motherboards because of external sensors.

Although the AMD X2 on-die sensors report much lower temps.. and conroes are cooler.. hmm I don't know really...

[virtualrain]

BTW, Speedfan 4.29 also supports the on-die sensor and reports it as "core". Comparing the two temps in my experience, the old "CPU temp" from probe seems slightly lower at idle and slightly higher at load... in other words, it seems to have a wider range. In addition, as someone else pointed out, the core diode temp does increase much more quickly when a load is applied.

I wish someone could explain how the traditional CPU temp. (i.e. from Probe) was actually measured/calculated.

[nikhsub1]

BTW, Speedfan 4.29 also supports the on-die sensor and reports it as "core". Comparing the two temps in my experience, the old "CPU temp" from probe seems slightly lower at idle and slightly higher at load... in other words, it seems to have a wider range. In addition, as someone else pointed out, the core diode temp does increase much more quickly when a load is applied.

I wish someone could explain how the traditional CPU temp. (i.e. from Probe) was actually measured/calculated.

Uhh, speedfan does NOT report my core temps at all... cpu temp is same as asus probe... I have no 'core' in speedfan and I'm running 4.29

[ShiningArcanine]

My Core 2 Duo is currently reporting an operating temperature of 36 degrees Celsius, and I believe it.

I think that the thermal sensor in the Core 2 Duo is probably much closer to a hotspot than it was in the Pentium 4, hence why people think that it is inaccurate, but after using a XP-90 without a fan, and getting 76 degrees Celsius idle and then adding a fan to it and getting 36 degrees Celsius idle, I believe the readings.

[xgman]

Intel Thermal Anayzer is another worth trying

http://rapidshare.de/files/30387105/TAT.exe.html

seems similar temps to coretemp. even a deg or two higher. Has a max heat test.

oh, and press "stop monitoring" before closing the program.

[Frackal]

LOL, and you base this on what? Everyone, yes EVERYONE refuses to believe actual temps. Coretemp is taping into the thermal diode in the cores of each CPU. Notice how quick temps jump from idle to load? I bet they are very close to reality unlike Asus probe or speedfan.

Yeah, well how do you know it's being reported accurately, given your sarcasm ridden confidence.

[dinos22]

BTW, Speedfan 4.29 also supports the on-die sensor and reports it as "core".

ONLY ON AMD CPUs

Not yet on Intel

[aggybong]

Well, mine hits 70c which is quite high It could be my cooler, but I thought these chips were supposed to output less heat than others? I never broke 50c on my old A64 3200+ @ 2.5ghz :~(

[Fred_Pohl]

LOL, and you base this on what? Everyone, yes EVERYONE refuses to believe actual temps. Coretemp is taping into the thermal diode in the cores of each CPU. Notice how quick temps jump from idle to load? I bet they are very close to reality unlike Asus probe or speedfan.

All Intel cpu temp monitoring has been read from internal thermal diodes in the cpu since the P3 (P2?). The bios reads the core temp from the cpu diode and Asus Probe, MM, Speedfan, etc. read the temp from bios. When I start Orthos, Speedfan core temp jumps from 35C to 49C in a second.

Aggybong: If you could read the temp of your A64 with Coretemp it would probably read 65C.

[aggybong]

Yea, that's true. I need to find a good cooler

[Fred_Pohl]

Yea, that's true. I need to find a good cooler

I'm pretty pleased with the Ninja+. Quiet and I'm seeing mid 30s idle and low 50s under heavy dual-core load.

[aggybong]

I was looking at that but I can't find it in stock I was considering the Infinity, but I don't know if it would fit.

Then there's stuff like the Ultra-120, SI-128, and Big Typhoon. So many options!

[Fred_Pohl]

Yeah, there are a lot of good options. As much as I like the way a tower like the Ninja exhausts the heated cpu air directly to the rear case exhaust fan(s), I also like the way horizontal coolers like the Big Typhoon help to cool the mobo, ram and power circuitry.

[aggybong]

Yea, that's pretty much what I'm debating That and I'm not sure how powerful a fan to get for the fanless ones, since the Typhoon and Scythe stuff already include one.

[cooleye]

My asus prob2 agrees very well with core temp (at most 2c difference) , also agrees well with the intel program...

[mascarilha]

yes , i also think Coretemp its the best way to see temps :


with Opteron 165 = 32c >>>>>>>>> X1800XT = 35c

With Intel 920 D >>>>>>>>>> X1800XT = 44c

with Conroe E6400 >>>>>>>>> X1800XT = 42c


CPU + VGA >> watercooled .


so the only reason i find to X1800XT its + 7c hot with Conroe than with opty 165 its because the water from CPU to VGA came Hot (+7c) in Conroe that with Opteron 165 , so :


If i have 32c in opty 165 the conroe should be +7c hot = 32c + 7c = 39c


My temps E6400 :

Speedfan > 26c

Bios >> 26c

Core Temp >> Core1 = 41c || Core2 = 40c

Everest >> Core1 = 41c || Core2 = 40c


so CoreTemp and everest seams to be more accurate with my previous temps calculations (32c +7c = 39c)



and other thing , now that the days here are colder my E6400 last night with speedfan reads 15c what i think its impossible ( Core Temp read 30c):









and today :









sory for my bad english


regards

[LowRun]

Good job there mascarilha, interesting observations there. It's gonna take a lot of this to convince people that their chips simply run hot.

[nikhsub1]

Yeah, well how do you know it's being reported accurately, given your sarcasm ridden confidence.

Because the INTEL Thermal Analysis tool agrees with coretemp If Intel can't measure the temps of their own chips...

[lawrywild]

Because the INTEL Thermal Analysis tool agrees with coretemp If Intel can't measure the temps of their own chips...

Hehe, because they read the same sensors.. doesn't mean they're both right. TAT is old..

[fornowagain]

snip

So one day the mb and coretemp agree and the next there's a 15c delta?

[fornowagain]

Hehe, because they read the same sensors.. doesn't mean they're both right. TAT is old..

2002? And its for mobile cpu's, on mine it can't see the cpu internal command interrupts.

[couppi]

All motherboards are equipped with dreamometers... get over it.

CoreTemp is probably the most accurate temp reader there is. Who cares if it's 60 or 70c, as long as it's stable?

[Absolute_0]

Even if the temps are accurate, which i still doubt because i'm seeing a real lack of consistency with the results, the temps are out of the context we're used to.

If for years, we've only seen what the thermale diode gives us, and NOT the "true core temp" and now we see the "true core temp" then it's really two different readings, and these temps aren't comparable to anything we've had before. It's hard to swallow, coming from an AMD dual core Opteron where i got 39c load with watercooling to a C2D and get 70c load on watercooling, until i think HEY maybe the Coretemp of the Opteron was like 60c.

But what makes CoreTemp more important than the diode temps we've been using for years...