We can say for sure: ThrottleWatch can really detect when a CPU switches to the throttling mode. Of course we don't know how the utility detects it, but we want to note a large spread of the "CPU Throttling, %" quantity compared to the "Throttled Clock" quantity in RMClock. Besides, we are not sure about what is taken for 100% throttling in this program. But the graphs (obtained with the same measurement interval) demonstrate that these 100% are reached a tad earlier than the 46% of duty cycle in our utility. To all appearances, the 100% throttling there is just a CPU performance drop to 50%. Let's proceed with the experiment: power on the cooler...

"CPU Throttling, %" and "Throttled Clock" curves in both utilities are again practically the same (except for the spread of values). And the most interesting things are reserved for the end: let's see how ThrottleWatch response to other throttling forms? Unfortunately, we had already no opportunity to test TM2 and DBS in this utility. That's why we could only test ODCM on Prescott, which we set to 50% using the CPUMSR utility.