It refers to how low you want to allow your CPU to go. There was a fellow, who posted about how it was not good to run these halt-on-idle programs, because they subjected the CPU to internal mechanical stresses by cycling them over a wide temperature range frequently. His post did not get much attention, probably because most of us OC'ers are in a constant battle with high temps. you rarely, if ever, see a post from someone worried that his cooling solution is running his CPU too cold. ;D
That fellow was more right than wrong. If you're running an underpowered peltier, for example, and at idle your temps drop to minus values, only to jump up to the 40s under load, you are, indeed doing a disservice to your CPU. Ideally, we would use adaptive cooling that held the CPU at a specified temperature all the time, but we don't. You have to pay good money for that kind of process control. We see an occasional HSF with a variable speed fan, but that factor does not sell them nearly as well as how well they do at the peak of the heat load.
So, most monitoring programs allow you to be alerted if you CPU drops below a certain temperature as well as climbing above a certain temperature. Some day, we'll do something with that information.
Hoot
