Sign in with
Sign up | Sign in
Your question

Core temperature vs CPU temperature

Last response: in CPUs
Share
a b à CPUs
June 22, 2009 5:06:37 PM

Hi there,
I've read a few articles on the difference between t-junction temperature and t-case temperature for cpus.

Now I have an OCed E6300, and I recorded the temperature using Everest when under 90+% load.

My "CPU temperature" stays around 35'C-40'C while my Core #1 and Core #2 temperature ranges from 55'C-60'C

Since the thermal specification for the E6300 is 65'C, am I cutting it close? Should I go by CPU temperature or core temperature?

thxs
a b à CPUs
June 23, 2009 2:02:03 AM

Also, it's strange that a lot of other hardware reader reads my CPU temp at 30'C, which Everest states is actually my motherboard temperature.

June 23, 2009 3:31:50 AM

Thermal spec. for Conroe (your E6300) is 95C. A guy on XS used an IR thermo probe to confirm that and is the value now used by both CoreTemp and RealTemp.
Tj comes from on-die DTS and "CPU temperature" usually comes from sensor on motherboard (underneath the CPU).
Related resources
a b à CPUs
June 23, 2009 4:06:33 AM

I see...so my actual temperature is around 35'C?

That means I can OC it quite a bit more. :D 
June 23, 2009 4:59:20 AM

With good aircooling an E6300 usually tops out @533x7=3731.
You'll need RAM that can do 533Mhz w/ 1:1 of course. Should be pretty easy with P45/35.
a b à CPUs
June 23, 2009 6:54:18 AM

wuzy said:
Thermal spec. for Conroe (your E6300) is 95C.

Not Thermal Spec, Tjunction Max ;) 
June 23, 2009 7:10:21 AM

randomizer said:
Not Thermal Spec, Tjunction Max ;) 

Right right....

I forgot to add sand in my pussy today. :p 

BSD, don't treat the values of 55-60C as an absolute. They just mean you've still got 40-35 units left till 95C, at which point the processor throttles.
June 24, 2009 2:32:56 AM

Of course we're talking about the older Conroe E6300.

Quote:
2.33GHz OCed E6300

266x7 -> 333x7, easy to tell.
Pentium E6300 has 266x10.5


I don't read into my temperature readings. If I'm running stock then I just need to know the processor isn't throttling. e.g. I let my E6300@stock in the HTPC/NAS run as high as 73C under load, it's passively cooled.
If I'm overclocking then the higher workable freq. automatically requires you to cool it to a temperature much lower than Tj max for it to be stable.

Temperature doesn't kill AMD/Intel processors now, it's voltage.
June 24, 2009 8:52:26 PM

Read his siggy again, it's implied.
Would would BSD overclock it to 2.33Ghz if it runs @2.8Ghz stock???

Reading comprehension and paying attention to details will do you good. ;) 
a b à CPUs
June 25, 2009 8:15:06 AM

Quote:
Are we talking about a 65nm Conroe E6300 or a 45nm Wolfdale E6300, big difference, hope you guys aren't giving him the wrong info.

I trust Coretemp after comparing it to several other temp programs.

Coretemp tells me the t-junc max is 85c, not 95c for my E6300.

Either way, your saying Core0 and core1 temps reported by coretemp is actually the tj. not the actual cpu temp? So If my temps read 65c thats not high, I can go all the way to 85c?

Well Core temp is probably still using the value that program developers pulled from Intel's Thermal Analysis Tool. This was all they could do without actually testing the chips, but since TAT was designed for mobile processors it just doesn't apply. Plus TAT hasn't been updated since 2006 so nobody considers it worth using for anything now.

The numbers Core Temp reports depend on your settings. By default it will report absolute temperatures (absolute is not equal to accurate), but can be changed to report the actual DTS output which is always the same regardless of which Tj Max you use.

Remember:

Absolute temps = Tj Max - DTS output.

If you want accurate temps then calibrate Real Temp or Speedfan. Trusting any program without calibration requires as much faith as selling an item on eBay and shipping it out before getting paid.
!