Funny temps

[T1MT1M]

Im going to start overclocking and i went to see what temp i had to start with. i downloaded TAT, Speedfan and coretemp and they all tell me around 55 degrees C on idle and 70 degrees C on load. But when i restart and go into bios it tells me 38 - 40 degrees, any ideas what is going on?

I have E6400 with stock cooling.
a 965P-DS3P
dont think it matters but, the blow dryer that is the 4870.

[NickN]

whats so funny about that?

[Slotback]

That's strange.... my Q6600 is 40 in Coretemp, and 40 in BIOS. I find it hard to believe a 15 degree delta between the two exists. :-/ Um, you said your Case temp was 50 degrees, and CPU temp was 40 degrees. Perhaps the BIOS temperatures are inversed? Try enabling and disabling PECI in BIOS, and also with CPU thermal monistor 2 (TM2).

http://www.tomshardware.com/forum/22174 ... ture-guide

[T1MT1M]

and something i dont get also is that i live in a tropical climate and its almost always 30 degrees in my room, that and an 80 degree video card makes it always 50 degrees just in the case. so what you are saying is that when i look at bios i see the case temp at 50 degrees and the cpu temp at 40 degrees i am seeing the case temp of the core which is 40 degrees. so why is the case temp of the core lower than the case temp itself?

[NickN]

That's strange.... my Q6600 is 40 in Coretemp, and 40 in BIOS. I find it hard to believe a 15 degree delta between the two exists. :-/ Um, you said your Case temp was 50 degrees, and CPU temp was 40 degrees. Perhaps the BIOS temperatures are inversed? Try enabling and disabling PECI in BIOS, and also with CPU thermal monistor 2 (TM2).

http://www.tomshardware.com/forum/22174 ... ture-guide

Its based onthe CPU and the BIOS and what is being read

that is why you can not trust what you see all the time

Q9 processors have a bug where the 2nd and 3rd cores read 10c different than cores 0-1

That Toms writeup is generic at best

Intel does not release the data on the procs and unfortunately there can be differences in which you can not ASSUME anything

there is a tCASE, tJUNCTION and tJUNCTION MAX spec

tCASE is typically 10-13c lower than tJunction

Thats one of the reasons why REALTEMP can read 2 different temps or be set to read the tCASE instead of tJ

If reading tCASE it will shift the tJunction delta down by a defined amount to compensate


not all processors are read correctly be it a BIOS, a hardware or other issue.

[NickN]

and something i dont get also is that i live in a tropical climate and its almost always 30 degrees in my room, that and an 80 degree video card makes it always 50 degrees just in the case. so what you are saying is that when i look at bios i see the case temp at 50 degrees and the cpu temp at 40 degrees i am seeing the case temp of the core which is 40 degrees. so why is the case temp of the core lower than the case temp itself?

No thats not what I am saying

The CPU TEMP in the BIOS may be reading the wrong sensor. It may not be reading the diode temp and instead the CPU CORE CASE not TOWER case temp. I dont know the board or the BIOS or the processor

I would trust coretemp over any other reading on the system and as VB mentioned, under the CPU settings in the BIOS make sure PECI is enabled if it is listed. It may not be listed becauase not all procs support it

[NickN]

Lets try this again


there is an ISA buss (usually Winbond ISA on register $290) where the sensor information comes from and there is a Intel buss (usually on register $0) where the internal diode data comes from

The motherboard, BIOS or software in use can mix those up and what happens is software like Asus Probe, Speedfan, etc may read the wrong temp. Instead of the diode they are reading an offset.

Therefore you must be careful in trusting anything till you confirm what you are reading is correct

Dual core, quad core and different models/stepping can have different specs and exhibit issues in reading the right values. Somtimes it requires calibration which is a real PITA however usually Coretemp will read the diode correctly

[T1MT1M]

I get what you mean now because



^^^ from everest.

so it is 55 degrees at idle

[NickN]

[quote]I get what you mean now because



^^^ from everest.

so it is 55 degrees at idle

[NickN]

In your case I do think you are running hot

http://processorfinder.intel.com/detail ... Spec=SL9T9


See the thermal spec?

61.4

[T1MT1M]

[edit]took out stupid remark [/edit]

It dosent go above 75 degrees over the full 10 mins, i know that isnt much time to check but meh.

So this pretty much means im gunna have to buy an after market cooler before i start OCing dosent it.

[NickN]

Go back up and read... you should not be over 68c on that proc full load

your limit is 68... 70 is absolute top end

and PRIME wont full load that system

Use OCCT. Prime is old news now for load testing



And yes, if you do not have a a good aftermarket heatsink you are not going to be clocking safely

The stock heatsinks are very limited for clocking

[T1MT1M]

sorry we posted at same time

Tj. max = 100

[NickN]

sorry we posted at same time

Tj. max = 100

I really question that one when the intel spec calls for 61.4

I need to shoot a question to my Intel contact to get the skinny on that one


if it is 100 then 75-78c is the max but that really sees strange for a E64 dual core proc. If the silicone is from the same batch as the Q67 then I can see that as a possibility. Usually Intel posts the Thermal Spec which runs about 5-8c lower than tJunction so it does seem strange they would apply a 61.4c max on that proc

100c tJMax is typically Intel spec'd @ 71-73c

[NickN]

OK

just got off the phone

Yes it is 100c... and the thermal spec they posted on that proc @ 61.4c is tCASE MAX which means you have a delta of about 13-15c on that to tJunction

so with the E6400 your absolute max diode temp under a full OCCT load test is 75-78c

FSX will never run that processor that hot, ever

The delta is 13-15c less which is why the different temp you are reading from the Intel 0 to the 290 buss on the chipset


This is what I am talking about ,,, this stuff has to be confirmed cause Intel did some crazy things with some of these procs. They are not all the same in their limits or their calibration