Originally posted by: mbuf
Heat generation and dissipation rates determine CPU's temperature. Right after power-off, energy generation stops but heat generation does not (though its "rate" decreases sharply), while the heat dissipation rate decreases because the fan stops. If the heat generated around power-off cannot go off as fast as before, CPU's temperature might increase. That's just my guessing, but I don't know if the temperature will actually 'shoot up'...
Well, unless the laws of thermodynamics have changed radically since I got my engineering degree, what you're saying is contradictory. Heat generation and dissipation DO determine CPU temperature - it reaches a "steady state" where energy (lets only concern ourselves with heat here, not electricity) into the system (CPU and HSF) equals energy out. When energy in is greater than energy out, the temperature rises proportional to the energy difference and the thermal mass of the system. Vise-versa for energy out less than energy in. One thing to note - when electrical energy stops in the system, the production of heat stops. First law of thermodynamics - energy can neither be created or destroyed. It was being converted from electricity to heat. No electricity, no heat produced. The CPU will be hotter than the heatsink - the cpu converts the energy to heat and the heatsink transfers it to the air. When the computer turns off, the cpu is still the hottest component so it will transfer heat to the cooler heatsink until they are the same temperature (actually the cpu will be slightly hotter due to thermal resistance in the CPU/HSF interface) and will remain at the same temperature as they cool down to room temperature. The mass of the cpu (even if we include the heatspreader and/or ceramic/organic package) is still far less than the heatsink, especially if you have a big copper heatsink. The heatsink may
slightly increase in temperature due to heat transfered from the cpu, but it won't go above the CPU temperature. Because of the loss of airflow due to the fan not being on, the RATE of heat transfer from the heatsink to the air will be less than before, but it will still greater than the zero heat generation of the CPU, therefore the system is no longer in steady state and is losing energy - cooling down. The air temperature may rise slightly in the case, but no more than a degree or two, even if there isn't good ventilation in the case. Free Convection (engineering term for "hot air rises") will move some air through a normal case, providing some cooling, along with normal radiative and conductive cooling. You've got a (maybe) 1oz processor and 30+ pounds of computer (and maybe a pound of air), it won't heat up a measurable amount from residual cpu heat.
If you want a more common analogy of what's happening, think of it this way. You have a cup of water in a pan on the stove. The stove is low enough that it's hot, but not boiling. The water will reach a steady state (lets ignore water loss due to evaporation for now) where the heat put in from the stove equals heat radiated out from the pan. Now, turn off the stove. Does the water suddenly boil? Not in my kitchen. Heat is going out of the system, not in. The stove and water is like the CPU. Now pour the cup of water in a gallon of room temperature water. It will warm it slightly, but nowhere near the original temperature of the water in the pan. This is like your HSF when everything is turned off. It will warm slightly from the CPU, but because it has so much more mass, the temperature increase won't be much and the overall temperature WILL be less than what the water in the pan was. Eventually it will go to room temperature.
Don't even get me started on the myth of "aluminum gets rid of heat better than copper"....