There are several key parts of a computer, one being the CPU or central processing unit. This is the general purpose chip that deals with calculations that the computers operating system throws at it. It coordinates all the actions that your PC does. For example, it tells your graphics card to display a 3D model.

A faster processor will result in a better user experience; faster calculations (Super Pi for example), quicker and more responsive OS.

In a game, the CPU deals with physics (although graphics cards and the Ageia PhysiX card have recently been taking this load), game code and other such information. Graphics, networking, sound, storage interface are all offloaded to the respective chips (graphics card, network card, sound card and motherboard chipset).

Your CPU sits underneath a heatsink or cooling block (for more adventurous cooling). This heatsink dissipates the heat energy that the processor inevitably radiates. When overclocking your CPU, more heat is given off and it’s important to match the correct cooling with the amount of heat being given off. For the most part, air cooling is more than sufficient. But when the MHZ rise, a more extreme method is needed, such as watercooling, a peltier or liquid nitrogen to mention a few.

In order to transfer heat efficiently, a small amount of heatpaste is needed between your CPU core and the heatsink or cooling block. This ensures that the surface area between the core and heatsink is as large as possible providing an easier channel for heat to pass.

Processor speed is measured in MHZ (the number of calculations that can be performed in a second) and of recent times, GHZ (1000MHZ) have become the norm. Previously the more MHZ your CPU packs, the faster your PC. But as the Pentium 4/Athlon battle proved, higher clock speeds didn’t necessarily yield better performance. Instead the CPU has become more and more specialized to its environment and processors now feature different instruction sets that programmers can use to speed up their code.

As faster chips become more expensive to make, manufactures (AMD and Intel) have opted to fit two or more cores per physical socket enabling true multicore processing (Intel released HyperThreading which enabled pseudo multicore). This enables more steamlined multitasking. However, most programs don’t yet support more than one core so often one core is used more than the other. More than one core doesn’t necessarily yield a linear performance increase (i.e. 1 core @ 1GHZ doesn’t equal 2 cores @ 0.5 GHZ). Unless the program thread is properly split (i.e. 50% going to one core, and 50% to the other) there will be no performance increase as only one core will be used. Some operating systems provide a basic thread splitter which helps to keep the load on each core similar.

The two main contenders in the desktop CPU battle are Intel and AMD. As they use different architectures, they also need different shaped sockets on your motherboard. Every successive processor range that is brought out normally yields a new socket to not only prevent back compatibility (stopping a user using a chip in a non-compatible motherboard) but introduce new features, such as LGA775 which changed the pins from CPU to motherboard so you’ll never bend the pins on your CPU again. Different sockets normally improve performance due to the new layout and number of pins.