A Heat "Sink" as its commonly, but perhaps incorrectly called, has two jobs:
- evacuate heat directly from the CPU die as quickly as possibly
- to allow evacuated heat to dissapate without reintroducing the heat into the system you are attempting to cool.
This second part is why I don't like the term "sink". it implies that you can keep pouring heat "down the drain", without ever filling the sink.
Traditionally the metal "fins" of a heat spreader provided maximum surface area in contact with the surrounding air, so that heat would move from the fins to the air and be blown out by the system exhaust fan as quickly as possible. If the fins were not able to get rid of their own heat fast enough, a fan could be attached to the spreader to increase the amount of air that comes into contact with the spreader surface per time interval, presumably increasing the transfer of heat from the fin to the air.
Some of the key concepts are:
- Heat (energy) will always move from the highest energy conductor to the lowest energy conductor available. When you put an ice cube in a glass of water, the heat moves from the water into the ice cube, thus reducing the heat in the water.
- You can never cool an object by moving its heat to an object that is already hotter than it is. This generally means that you can never cool a CPU to less than room temp without refrigerant.
- Materials with a lower specific heat will both heat faster and cool faster than materials with a high specific heat.
So in a traditional scenario, the heat spreader has a lower specific heat than the CPU die, so the heat flows into it quickly. The low specific heat also allows the heat to exchange into air, but the rate varies based on the amount of heat produced with any given time interval.
The flaw in this approach, is, what happens when the CPU is producing so much heat/second that the heat spreader cannot evacuate it fast enough, causing the spreader to heat up to the point where heat will no longer flow into it (or will only take heat so slowly, in too little quantity) such that the CPU's heat doesn't go anywhere.
Thats where it becomes important to seperate the concerns of immediate evacuation of heat from the die, and more gradual evacuation of the heat from the heat spreader. if we can keep taking heat from the CPU quickly, then it really doesn't matter to us much how quickly the fins disappate it into air.
Heat pipes are specifically designed to evacuate heat from the CPU as quickly as possible, and move it further away from the die than was traditional, which makes the whole system more effective when the CPU is running hot, and might have swamped a traditional spreader.
The material heat pipes are composed of (as well as the material in direct contact with the CPU die) must have a very low specific heat and conductive resistance to be effective. The mass of conductive material has an impact on this equation, such that the more pipes of lower mass individually will perform better than fewer pipes that are aggregately of the same mass. So, from that perspective, all else being equal (and it rarely is), the more pipes the greater the capacity to evacuate in any given time interval.
So, that was way more answer than you wanted, and represents only a minimal knowledge of thermodynamics, but hopefully its of some use to you.