The Graphics Processing Unit (GPU) or just graphics card is the component which processes and outputs the feed on the monitor. Most CPU’s have integrated graphics, but expansion graphic cards are very popular since their performance is a lot better. GPU is the component which has the most performance in the system, it bypasses the CPU’s performance easily.
If you don’t need to run any GPU demanding programs, you probably won’t need any expansion graphic card. However, if you play new games or run GPU heavy programs, you need to get one to be able to play the games. In that case, you will also need a decent monitor for gaming.
Examples of GPU heavy tasks:
- Battlefield 4
- High-resolution videos and games
When you are selecting yourself a GPU, the outputs you have depends on the outputs the GPU has. Remember to check the compatibility between your monitor and graphics card before buying. The most common cable types are usually supported by both the monitor and GPU, but if you have an older monitor, you might run into issues with connecting the two.
Another thing to keep in mind is how many screens, what maximum resolution and what refresher rate the GPU supports. These things won’t be an issue for you if the computer and GPU are only for basic use. Then you don’t probably use more than two screens that are full HD and you don’t need higher than 60Hz refresh rate. However, if you are playing a lot of games and you want to have a high frame rate, and you have a monitor that supports 144Hz already, you will probably want to invest on some high-end GPU. If you decide to go for the 144Hz, make sure your all the components support it, GPU, monitor, and cables.
Make sure your graphics card fills your needs since they are expensive. Selecting the right GPU for your build may be difficult since there are so many things to take into consideration. This is why I go through the terms and any other important things you need to know about graphics cards before buying one.
The two largest companies producing graphics cards are NVIDIA and AMD. GeForce series is manufactured by NVIDIA and Radeon series of graphics cards is manufactured by AMD. Both of the series are very popular, but there are some differences between the series that may confuse and mislead people who are new to graphics cards. The differences are mostly in the naming of the graphics cards. The GPU’s of the same manufacturer is easy to compare if you have the knowledge. You can’t really compare AMD and NVIDIA graphic cards without real-world benchmarks since the technology used is different.
AMD Radeon naming scheme
- Prefix means the main tier of the GPU, for example, R5 and R7 are basic tier graphic cards and R9 and RX (R10) are higher performance tiers.
- The mobility of the GPU is marked with the letter M and it means that the particular graphics card is for laptops or tablet devices.
- Generation is simply the age of the GPU, the higher the generation is, the newer the graphics card is.
- Performance means the performance group the graphics card belongs to. A higher number is better.
- Revision separates two graphics cards that share the same performance, some are assigned the number 0 and some are assigned the number 5.
- The suffix X designates that the particular GPU has even higher performance than the normal version of the GPU.
Let’s take R9 390X as an example. It is a generation 3, high-performance GPU with extra power compared to the normal version of the GPU.
NVIDIA GeForce naming scheme
- Prefixes used are GT and GTX. Graphics cards that are basic class are marked with GT and high- and enthusiast-class graphics cards are marked with GTX.
- Generation means the same as in AMD’s graphics cards. A higher number means newer GPU.
- The performance level of the GPU, a higher number means more performance.
- Revision works the same way as in AMD’s graphics cards. GPU’s that share the same performance are separated by a number 0 and 5.
- Suffix gives information about the graphics cards special properties in a similar way as AMD’s mobility and X. Ti means that the GPU is faster, and M means that the GPU is for laptops and other smaller devices.
Let’s take GTX 980 Ti as an example. It is a high/enthusiast level, 9th generation GPU. It is a high-performance graphics card with even extra power added to it. Note that 9th isn’t the newest generation and therefore this generation’s graphics cards can’t perform as well as the newer ones. This GPU is the top class of 9th generation graphics cards.
The ads selling graphics cards have a few different terms mentioned on them. Graphics cards are almost impossible to compare if you don’t know what the terms mean. On top of knowing the naming scheme, it helps a lot to understand the terms of the ads, so you are able to really compare the products. Below are explanations for the most common terms used in the ads.
Meaning of clock rates, OC and DUAL
The clock rate is the GPU’s processing speed measured in MHz. Modern GPU’s also have their boost clock rate mentioned as well. This the speed the graphics card is able to boost when under a heavy workload.
Some graphics cards are also sold as OC models. OC simply means that they have been overclocked at the factory for maximum performance. If the GPU is said to be DUAL, it just means that it has two fans. The number of fans is not something you should be too worried about unless you are going to overclock some low-end GPU.
CUDA (Compute Unified Device Architecture) cores are the computing cores in GPU. The more cores there are to do the work, the more work it can get done. The number of CUDA cores is naturally higher in the high-end GPU’s.
Stream processors are basically the same as NVIDIA’s CUDA cores. Keep in mind that the technology is different, and you can’t compare the amount of CUDA cores to the number of Stream processors. The best way to compare different graphics cards from different manufacturers is real-world benchmark results.
Just like your processor has the RAM as its working memory, the GPU has its own working memory as well. It is used to store the data which is processed and then displayed on your monitor. The more memory your GPU has, the faster it can work. In graphics heavy games (or in other programs) this will affect the smoothness of the game.
In graphics cards, the memory is GDDR (Graphics DDR SDRAM). This means that the memory is specifically designed for graphics cards. It is fairly similar to DDR memory, but they do have differences between them. DDR is used as RAM in computers. The newest GDDR memory is at the moment GDDR5. If you are planning to play games with high graphics settings, you want to get a graphics card with at least 4 GB memory, my suggestion is 6 GB or more for graphics-heavy games.
Memory clock rate is different from the GPU’s clock rate. Memory clock rate means how fast the memory of the graphics card is. The higher the clock rate (MHz), the faster the memory simply is.
Memory bit interface determines how fast the actual processor in the graphics card (GPU) and memory are able to talk to each other. Higher bit interface designates faster communication between the two. However, it isn’t quite that simple. The memory type affects the bandwidth as well, GDDR5 with lower bit interface is faster than GDDR4 with higher bit interface.
Graphics cards are usually connected to screens with either HDMI, DisplayPort or DVI-D, depending on your needs. Make sure that the cable is capable of the data transfer speeds you need, and that your screen has the same connectors.
Many modern graphics cards are so big that they require you to remove two expansion slots from the back of your computer. You only need to connect the GPU to one PCIe slot, but you will be unable to install anything in the PCIe slot next to the GPU if it is a big model. Keep this in mind if you are planning to have multiple graphics cards or additional expansion cards.
It is important to have a good PSU, since it powers your whole system, including the GPU. Some graphics card advertisements mention what are the minimum requirements for the power supply unit for that particular GPU. This helps you to select the PSU since now you already know what it needs to be capable of.
Another thing related to power supply units is the power connectors the GPU has and needs. Usually, the power cables the graphics cards have are either 6-pin, 8-pin, 6+8-pin or 8+8-pin cables. The power cables you will need depends on the graphics card you have. Higher performance graphic cards require more power. The power cables you have depends on the power supply you have, make sure you will have enough and the right type of power cables to power your GPU and all the other components.
Depending on your use of the computer, you might want to make sure your graphics card supports some technologies. Most modern GPU’s usually support them, but it is better to be sure before buying.
Direct X and Open GL are competing technologies that provide a common interface for different needs, for example, 3D-graphics. We won’t dive deeper into these technologies since they are more related to programming than hardware. All we need to know when buying computer components is what Direct X and Open GL versions are used and supported, and basically, if you are buying a new graphics card, you don’t need to worry about these.
Multiple GPU setups
You are able to use more than one GPU in your system. It improves the performance of your computer but also the power consumption and heat dissipation. In the end, the gained benefit isn’t something you wouldn’t normally need if you can afford one good graphics card. Using two graphics cards can however almost double the number of pixels produced.
The image below is another example of multiple GPU setups, this one is used to mine the cryptocurrency called Etherium.
AMD Crossfire X
Crossfire X is AMD’s technology to support multiple graphics cards in one computer. With Crossfire X you are able to use two, three or four graphics cards in your system. The graphics cards need to have the same GPU architecture, otherwise, it won’t work. AMD’s Crossfire X technology doesn’t need any extra connectors between the GPU’s anymore. Crossfire X is cheaper for the manufacturers to implement, so it is commonly supported, and you can use GPU’s from different manufacturers as long as the GPU architecture is the same.
Just like Crossfire X, SLI (Scalable Link Interface) is a technology which makes it possible to use two, three or four graphics cards in one computer. Indifference to AMD’s technology, the graphics cards connected with SLI must have the same GPU architecture and amount of memory. If those requirements are filled, you can use GPU’s from different manufacturers.
Another difference to AMD’s solution is that SLI needs a physical device between the graphics cards, unlike Crossfire X. This makes the SLI cost more for the consumer but also for the manufacturer’s.
When overclocking graphics cards, you need to overclock the GPU and the GPU memory. As I have stated in other posts, I do not suggest overclocking if you are building your very first computer. However, try to focus understanding every component as well as you just can. After you have basic knowledge and understanding of the components, and in this case about the graphics cards, you can learn about overclocking. Nowadays manufacturers offer also overclocked models of their graphics cards, they aren’t a bad choice either.
If you are going to overclock your graphics card, the graphics card must be overclockable. Also, keep in mind that the card will need a more power and cooling than before. Make sure that you are able to offer good cooling and enough power for your graphics card and your other components as well. If you can’t cool the card enough, it isn’t a very good idea to overclock at all, also if your power supply unit can’t provide enough power, you aren’t able to overclock.
If your components are overclockable, you can provide good cooling and enough power, you can proceed with overclocking your graphics card. Overclocking your graphics card gives your system some nice extra kick and with a stable overclock you will get the most benefit out of your graphics card you can.