Curiously enough I did something a couple of days ago that may answer your question and help to dispel fears for others.
I was getting tired of hearing the Delta 7200 RPM fan in my system, so I chose to put the stock coolermaster fan back on. I took the delta off while the system was still going because I was first intending to put the other fan right back on. Somewhere in between I decided to see how well the heatsink could deal with head all on its own.
The results were interesting to say the least.
Here is the build on my system:
Enlight Mid-tower case
Enermax 350w
EPoX 8KHA+
Duron 600 @ 858 (143 * 6)
Coolermaster DP5 6H-11 lapped down with 1500 girt
Arctic Silver 3 themal paste
256 MB Crucial PC2100
Asus V7100 pure
I have the WPCREDIT tweak that halts the CPU when idle, installed. The idle temp that I got with the delta was reported by mobo monitor as 19 celcius. When I ran idle with just the heat sink, the temp never went past 24 celcius.
My desire for dangerous things did not stop there. What I did next will shed a lot of light on this whole temperature comparison thing.
I started up an opengl test called particle fury. It runs a fairly intensive opengl loop that runs in a window so I can see the temp rise in another. I still have the fan off during this. It took a good 15 minutes but the temp eventually rose to 48 celcius and crashed the PC.
Now according to the AMD tech specs, the K7 series are supposed to run as hot as 90 C. Because ALL motherboards that are currently on the only use an under the CPU sensor, the reading that they are taking is nowheres close to accurate. What I did really illustrates that point. A sensible conclusion that can be made is that the actual core temperature for the processor was 90 C when it crashed. Some 40 C higher than what was displayed.
As an earlier post had indicated, the ASUS boards report temperatures "funny" (my words). There is absolutly no way to ever compare even identical model motherboards' temperature results as there are far too many opportunities for difference between the setups. Actual core voltage, air circulation, sensor placement, BIOS calibration, etc... can all have an impact on the final result.
Here are the things to remember:
- Never try to compare external sensors
- If you are running stable then you are fine. To be sure, run 3dmark for 24 hours straight. If you don't crash you are fine.
- If you are crazzy like me, try running with no fan to find out what the breaking point for your setup and monitoring software is. My CPU is still running like it should. The trick is that the second that you lock up, you shut the PC right off. The storage temperature is higher than the running temperature and so the CPU can easily deal with the excess temperature when it is off.
I hope this helps you.
Thanks,
Jason.