Graphics cards use the same basic components as processors, and they're even easier to overclock than their socketed stablemates.
For starters, GPU overclocking doesn't require fiddling in the BIOS. It can be done in software, with some options available in AMD and Nvidia's drivers, and other third-party tools that can improve speeds by clicking a button.
Graphics overclocking involves tweaking the core and memory clocks, and sometimes the boost limits - so cards can dynamically overclock to even higher speeds.
The GPU equivalent of upping the voltage involves raising the card's power limit, which is usually as simple as raising a slider.
It's easy to overclock a graphics card but, as with desktop processors, it is a risky business. If damage is caused to the card because of overclocking, it can void its warranty - as well as break the card.
Slots and display connections
A graphics card interfaces with the rest of a PC via a PCI Express slot, which can cope with the kind of demanding data transfers required by graphics cards. Motherboards have PCI Express sockets in a variety of sizes, and most graphics cards use x16 slots - the largest. These fall into two standards: PCI Express 2.0 and 3.0. The latter is newer and has ample bandwidth for even the most powerful cards, where 2.0 might struggle.
Also check whether a PCI Express slot is able to use its full bandwidth. Most motherboards have one PCI Express x16 slot that runs in x16 mode, but secondary slots are often restricted to just x8 or x4 modes - so they're only able to use a half or a quarter of their potential bandwidth - due to chipset restrictions.
Graphics card installation isn't as simple as pushing the card into a slot - the card needs to have the right connections, and the case needs to be large enough.
Powerful cards generate more heat, which means larger and more sophisticated cooling systems are required to keep the core chilled.
Larger heatsinks are sometimes too long to fit inside smaller cases, and double-width coolers can also block motherboard slots below the card.
That's not the only physical consideration. More powerful cards require more electricity, with some top chips needing two eight-pin power plugs to function. Most power supplies come with these connections, but it's worth checking in advance.
Every graphics card has a variety of outputs that connect to monitors. The most common connections found on modern cards are HDMI, DVI and DisplayPort - the latter is the newest, and is useful for connecting extremely high-resolution displays.
It's worth checking which outputs a card has before purchase to make sure they match your monitor - or so you can order an adaptor.
AMD and Nvidia decoded
AMD and Nvidia are the big players in graphics, with both firms battling for GPU supremacy. Both companies use different naming schemes. Nvidia's cards are all prefixed with 'GeForce', and top-end cards also use 'GTX'; low-end products only use 'GT'. Beyond that, the bigger the number the better - cards such as the GTX 750 and 760 are mid-range products, with the GTX 770 and 780 at the high end. Cards with the 'Ti' suffix are slightly more powerful versions of their origin card.
AMD's Radeon naming scheme has recently changed. Now its cards are divided into three groups: R5 cards are entry-level products; R7-branded parts sit in the mid-range; and high-end chips are denoted by the R9 prefix. Beyond that, it's a similar story to Nvidia - the bigger the number, the better the card, and the higher the price. Cards with an X attached to their names are more powerful.
The companies tend to follow each other closely when it comes to price and performance - it's handy to look for overclocked cards and game bundles to help differentiate between the two when buying a new card.