Updating a PC or laptop’s graphics card can be a cost-effective upgrade. A graphics card upgrade allows your PC to play games smoothly at higher resolutions, with more special effects, but there are plenty of options and potential pitfalls to consider.

A powerful graphics card is an essential component in any high-performance desktop or even laptop PC. A separate card is required in order to handle the complex 3D maths and geometry that go on behind the scenes to add realism to games and HD content. AMD and nVidia release a new generation of their graphics cards every year or so, with refined hardware and more transistors squeezed in, and here’s our guide to selecting the best for your requirements.

All PC’s and laptops will come with some form of a graphics processor, whether it be one built into the CPU itself (which is becoming more common) or a basic one on the motherboard. For general day-to-day tasks such as surfing the internet of office work these are generally fine. But what should you look for when considering games or photo/video editing or wanting to watch high definition content such as blue-ray films.


GTX 1080 made by Evga

The GPU:

The key feature of any graphics card is the graphics processor, more commonly referred to as the GPU. There are two names you need to know in graphics processors: ATI and nVidia. Much like the Intel and AMD duopoly in PC processors, these two manufacturers produce the chips around which others build graphics cards.

Put simply, the GPU determines the quality of images displayed, as well as their refresh speed. At the time of writing the nVidia GTX1080Ti is the highest-performing card, while the GTX 1060 and Radeon Rx480 represent excellent value.

Generally speaking, the more powerful the GPU the better – gamers may have to switch off some detail settings with a cheaper card, for instance.

Stream processors are a significant feature of today’s cards, able to work as either vertex or pixel shaders. Older graphics chips used dedicated vertex shaders and pixel shaders, but if a game used more of one than the other, the card wasn’t used to its full potential. When an object has been ‘created’ by the shaders, a texture or image will be applied to it. Texture units work with the shaders to apply textures as quickly as possible.

This is a resource-intensive process, so the more texture units, the better the fill rate will be. We calculate the fill rate by multiplying the number of texture units with the core clock speed of the card. The higher the resulting number of texels (texture elements), the more textures the card can fetch. Raster output units (ROPs) are also important since these turn the information from vertex and pixel shaders to complete pixels to create the final image.


The next most important factor is dedicated video memory, or RAM. Even those on a budget should be able to get a minimum of 2GB of DDR memory in a standalone graphics card. For high-end cards, look for 1,280MB or even more – these days we’re seeing cards with 8GB coming on a regular basis.

Most decent graphics chips now come with GDDR5 RAM. This is more sophisticated than GDDR3 and will in effect quadruple the clock speed, whereas GDDR3 merely doubles it.

Another important factor is the memory interface (or bus), which is determined by the chip. This governs how much data can be sent through at once. A 256bit interface can let through twice as much data as a 128bit interface. The memory can work at higher speeds, as determined by the clock speed. The cards with the best combination of memory interface size and clock speed should produce the best performance.

A decent measure of this is the memory bandwidth, which is the memory interface divided by eight, multiplied by the effective memory clock speed. If the memory is GDDR5, you get the effective speed by quadrupling the basic memory clock; for GDDR3 you double it. Measured in gigabytes per second (GBps), the higher the memory bandwidth, the better the card will be able to handle textures and other features.



DirectX 11.0 support is required for advanced, modern games; cheaper cards are likely to lack the firepower to do justice to DirectX 11.0 games, however.DirectX is present in every game and usually will require version 9+ in any current game. If your looking at a new graphics card always make sure it supports the current version, older versions will automatically be supported as they are backwards compatible.

Currently if you are looking at games such as Battlefield 3, Call of Duty 4 or Crysis which all use DirectX 11 you would  expect to pay around £2-300 for a card capable of running these games at a mid to high level of detail. That is assuming the rest of your PC system is up to spec.

Also if you are thinking about multiple monitors then the requirements for both the graphcs card itself and the PSU to run it (and in some cases multiple graphics cards) quickly adds up.

For example if running 4k monitors for a current game you would be looking along the lines of  an nvidia GTX1080 or Radeon RX580 graphics card as a starting point, possibly even two of them and this would require a substantial PSU of around 850watts or more. However the end results are quite something.

Multi monitor setup


One thing that is often overlooked but could be crucial, is the practical consideration of fitting a graphics card. You need to make sure it will physically fit – not all cards fit in all cases – and that you have sufficient connections from the PSU (power supply unit).


The cheapest cards require no additional power, but most cards require one six-pin connector and some will need two. The PSU should be rated high enough to handle the demands of the graphics card. Typically a 550watt PSU is the starting point.


If you are thinking about a new graphics card then please feel free to call and get in store prices on 0131 6629955.


Author: Alexander Wright

Posted in Custom PC, Gaming and tagged , , .

Leave a Reply

Your email address will not be published. Required fields are marked *