Any electrical circuit has resistance, and electrical resistance is the principle upon which both CPUs and toasters are designed. Electrical semiconductors have the unusual trait of being able to switch between low and high resistance when electrical current is applied in a certain way, and those states represent the 1s and 0s of a logic circuit. Although the logic circuits of a CPU are not intended to create heat, those principles mean that we’re all running little tiny hotplates in our machines.
Data processing is the added benefit of designing your toaster as group of logic circuits, but how do we prevent the little chunks of glass upon which those circuits are etched from melting into a toasted lump? A chunk of metal designed to draw heat away from another object is called a sink, and that’s what’s at the core of CPU cooling.
Yet the name “heat sink” really doesn’t mean much, because a sink is simply something that absorbs heat. More than simple sinks, a large surface area of fins helps CPU coolers transfer heat to a greater volume of comparatively cool air. Those fins make a standard CPU heat sink a type of radiator, in spite of the nomenclature. And like most radiators, convection is a more significant cooling force than radiation, as the rising of warm air causes cool air from beneath to take its place.
Heat is proportional to things like clock frequency, circuit voltage, circuit complexity and the material upon which the circuit is engraved. While a few low-energy CPUs can be cooled using nothing more than a set of fins, most desktop users want more performance, which comes at the cost of more energy (heat) to dissipate.
When natural convection doesn’t replace warm air with cool air fast enough, forced convection is accomplished by adding a fan. The above image shows a rare all-copper cooler, which relies on the principle that copper transfers heat faster than aluminum. It also weighs and costs more than aluminum. Manufacturers often surround a copper pillar with aluminum fins in order to achieve better cooling-to-cost and cooling-to-weight ratios.
More fans and more surface area give a CPU cooler even greater capacity. Liquid cooling allows enormous radiators that wouldn’t fit over the motherboard to instead be mounted to a panel on the case. The CPU is topped with a component called a water block, which transfers heat to the liquid. A pump, seen mounted on the side of the radiator depicted above moves water (or coolant) through a series of channels on the radiator and water block.
While any of these solutions maximize contact with moving air, none would work well if the CPU were not making contact with the cooling apparatus. Thermal interface material fills the gaps between the top of the CPU package, eliminating the trapped air that would otherwise act as an insulator. Most CPU coolers ship with some type of thermal interface material, and many coolers come with that material factory-applied to the mating surface. Performance enthusiasts often seek out better-performing products to replace the cooler manufacturer’s solution, although the difference between various products designed for the task is usually quite small.
Compressor units are on the extreme end of the complexity curve, using refrigerant to produce CPU core temperatures far below ambient conditions. These generally use far more energy than the processor does, and can scale up to designs that compress and cool air in several stages to produce a liquid-nitrogen drip. Condensation around cold components becomes a big concern, so that even the simplest “refrigerator” designs are generally used just for exhibition purposes.
Of course the Bigger Is Better rule is limited by case size, but there are a few other factors to consider. Since this is a beginner’s article, we’ll cover the range of coolers that make up our Best CPU Cooling awards showcase. These include Big Air coolers over 6” tall, Slim designs that are under 3” tall, Mid-Sized coolers from three to six inches tall, and pre-filled liquid cooling kits.