CPU IDLE:  CPU Temperature
 CPU-IDLE: The frequency and the voltage of the processor are the two most important factors that controls the temperature of the CPU.
CpuIdle uses a HLT instruction to stop the CPU to work and let it use less electric power: less power dissipation, less temperature. But the different CPUs have different instructions, and each of them (K6 PMMX PII 6x86) have different reactions when you use CpuIdle.

The Results.txt file of CpuIdle gives this results:      (in degree centigrad)
 
CPU and it speed: K6 166 at 166 K6 166 at 200 K6 166 at 208 K6 200 at 200 K6 200 at 210
Normal Temp 33 45 53 37 45
CpuIdle Temp 26 (-21.21%) 34 (-24.44%) 35 (-33.96%) 25 (-32.43%) 32 (-28.89%)
 
K6 200 at 225 K6 200 at 262 K6 233 (3 results) K6 233 at 250 (3r) 6x86 P166+ 6x86 P200+
52 64 70 / 51 / 35-40 49 / 56 / 70 52 33
39 (-25%) 43 (-32.8%) 52 / 32 / 29 (-27.88%) 38 / 32 / 49 (-31.77%) 41 (-21.15%) 19 (-42.42%)
 
P5 166 at 166 P200 MMX at 200 PII 233 at 233 PII 300 at 300
62 64 ?? 98*
53 (-14.52%) 54 (-15.625%) 15 *(lower than without) 80* (-18.37%)
* in degree Farenheit

This results show that a iP200MMX runs hotter than a K6-200, but it is not true.
The text in results.txt explains tha the tests had been made with different environments.
Don't worry about the 15-80 degrees of the PII too.

My Opinion :

I understand that if you have an AMD K6 you should use CpuIdle (or DosIdle) because you can have a decrease superior to 20% and even 30% in the half of the cases. If you want to overclock your AMD CPU it is a good choice to download CpuIdle.
 
You can download CpuIdle:  DISCLAIMER  

THIS SOFTWARE AND THE ACCOMPANYING FILES ARE DISTRIBUTED "AS IS" AND WITHOUT  
ANY WARRANTIES WHETHER EXPRESSED OR IMPLIED. NO REPONSIBILITIES FOR POSSIBLE  
DAMAGES OR EVEN FUNCTIONALITY CAN BE TAKEN. THE USER MUST ASSUME THE ENTIRE  
RISK OF USING THIS PROGRAM. ALL TRADEMARKS ARE PROPERTY OF THEIR RESPECTIVE  
OWNERS. 

PLEASE:  READ also the other LEGAL TERMS before using CpuIdle (in the files contained by cpuidle252.zip).
The cyrix results are strange !
21.15% and 42.42%,  this two numbers are very different. But it must be right.
I think you must try CpuIdle to see the real improvement in cooling, 20% is an important value.

All Intel tests show approximatively the same results (about 15%) and is due to the low temperature of Intel products even without CpuIdle. And Intel products don't allow all of the CpuIdle features.
As you can see in my P200MMX test it is also an improvement of the FPU under Dos (7.00). (but not with K6 200)