Core Temp prograM

vollman1

Distinguished
Aug 8, 2011
1,060
0
19,360
Asus p8p67 ws revolution
i5 2500k
Xigmatek Gaia sd1283
4Gx2 G.SKILL F3-12800CL9D-8GBXL
VGA POWERCOLOR AX6950 2GBD5-2DH R

I have been running stress tests on my new build using Prime95.

I am monitoring temps using:

Real Temp 3.6
CPUID HW Monitor
Asus Probe II
SpeedFan 4.44
Core Temp 1.0


All of the monitoring programs show temps of 54-57c except the Core Temp 1.0 program. It shows temps spiking to 98c for all 4 cores and the monitoring shut down because of this. But, the computer never shut down.

I am not exactly sure if this is a problem or not.

Since all other monitors show good temps, is Core Temp 1.0 just having a problem? Or, is Core Temp 1.0 identifying a problem that the others aren't registering?

I don't want to damage my CPU.
 
Solution
RealTemp 3.69.1
http://www.mediafire.com/?4uixpjtezznuzkd

This version supports Sandy Bridge CPUs.

The Intel sensors count down as the temperature goes up. To determine the temperature, software does this calculation:

Reported Temperature = TJMax - Sensor reading

If there is a Core Temp glitch when it reads the sensor, it might assume a sensor reading of zero so the reported temperature will be TJMax. If all other programs disagree then obviously Core Temp is wrong. When a heatsink is attached, the core temperature can not instantly shoot up to the maximum temperature.

Another way to prove this is to look at the Thermal Status area in RealTemp. There is a separate bit in the CPU that keeps track of these sensors. If the...

gero9mo

Distinguished
Aug 28, 2011
1
0
18,510



Hi there Vollman1,

I am having the exact same problem. Temps spike to 98c but the computer is stable and all other temp monitors show a stable CPU. I made a post here: http://www.overclockers.com/forums/showthread.php?p=6945566&posted=1#post6945566 , adressing the issue and even linking to this post. Lets hope the developers adress this problem soon because all signs show that this is a Core Temp RC1 bug...
 

unclewebb

Guest
Sep 11, 2007
247
1
18,865
RealTemp 3.69.1
http://www.mediafire.com/?4uixpjtezznuzkd

This version supports Sandy Bridge CPUs.

The Intel sensors count down as the temperature goes up. To determine the temperature, software does this calculation:

Reported Temperature = TJMax - Sensor reading

If there is a Core Temp glitch when it reads the sensor, it might assume a sensor reading of zero so the reported temperature will be TJMax. If all other programs disagree then obviously Core Temp is wrong. When a heatsink is attached, the core temperature can not instantly shoot up to the maximum temperature.

Another way to prove this is to look at the Thermal Status area in RealTemp. There is a separate bit in the CPU that keeps track of these sensors. If the sensor ever counts down to zero, that triggers thermal throttling and this bit in the processor will be immediately set. It doesn't matter whether the CPU throttled for 5 minutes, 5 seconds or 5 milliseconds. There will be a record of what happened within the CPU.

If RealTemp shows OK, that proves that the CPU did not thermal throttle. If it shows LOG then that means the CPU logged at least one thermal throttling episode on that core. You don't even need to have RealTemp running when thermal throttling happens. You can start RealTemp up after a thermal throttling incident and it will read the throttling flag bit directly from the CPU and report if there has been a problem since you powered on.
 
Solution