If you’re thinking of buying a new graphics card, it pays to understand all of that jargon
One of the most high profile upgrades in the PC market has to be the graphics card. If you’re into games it’s an essential purchase, and one that requires inevitable upgrades as the technology and games become more and more advanced. Other users like graphic designers and animators also need good visual hardware, albeit often specialised, and not the same breed as most consumer models.
Despite it’s popularity and mainstream saturation, PC graphics technology is far from straightforward if you’re not an experienced user. For first time buyers, or users who have little technical knowledge, buying a graphics card can be a very tricky affair, as the tech is drenched in complex terminology and jargon, not to mention differences between manufacturers, and software that requires different specifications. This all makes it hard to make a well-informed decision, and a lot of buyers are left with only sales people or guesswork to go on. This doesn’t have to be the case, though.
As confusing as graphics card terms can be, and as wide as the market is, it’s actually not that difficult to understand the important elements. The basic jargon is easily explained, and for the more complex and deep technical specifications, it’s often not essential to understand it to get a decent card.
We’re going to look as this jargon and help you understand what it all means in order to give you a better understanding of the technology so you can make the right decision when shopping for a new model. So, without further delay, let’s jump right into those confusing terms.
Jargon Buster
Anti-Aliasing / Anisotropic Filtering
This is a graphical technology designed to make games look better by smoothing out jagged edges on models and other assets. Low resolutions are the most problematic, so the issue isn’t a noticeable for users running HD displays, but AA can usually make any game look better, at the overall cost of performance.
Anisotropic Filtering is similar, and deals with the smoothing of textures at extreme angles, reducing the amount of fuzzy, jagged edges. Again, this comes at a performance cost.
Lower-end cards are often unable to handle these technologies without major performance issues, and these are often the first detail settings to be sacrificed.
Architecture
The architecture of a GPU is the base technology it’s built on, and this technology is improved time and time again, bringing with it more and more capable GPUs. New generations of architecture bring the big, new versions that happen infrequently. Most of the new chips we see are revisions and updates of existing architectures, bridging the gap between old and new architecture generations.
Core clock
This is the measurement in Mhz and GHz of the GPU’s core clock frequency, or how fast it can perform operations. The speed of a GPU depends on many aspects, such as quality, architecture, and the amount of stress placed on it.
Generally, you want a card with a higher core clock speed, but this specification alone does not guarantee a good performance, and it’s greatly affected by other elements. Some cards with higher clock speeds, for example, may not perform as well as cards with a lower core clock rating and a better memory clock speed.
CrossFire / SLI
Depending on the motherboard and the graphics card, systems can now host multiple graphics cards at once, increasing the graphical power of the system. Two or more cards can be run with obvious benefits. This uses either CrossFire or SLI tech, with CrossFire being used by AMD/ATI cards, and SLI used by Nvidia.
CUDA cores / Stream Processors
CUDA core and Stream Processors are parallel processing cores featured on the graphics card that increase the speed and power of the GPU. Nvidia uses CUDA cores, whilst AMD/ATI uses Stream processors. Both are different at an architectural level, but they accomplish the same thing.
It’s important to note that the two technologies are not directly comparable, though, and the numbers of cores don’t directly tally. For example, an AMD card may have more cores than an Nvidia card on paper, but this doesn’t mean the AMD card will be a better performer.
Die / Die Size
You may hear the words die and die size used when people talk about both CPUs and GPUs. This is basically the term for the piece of silicon used to produce the actual processor. The die size is simply the actual physical size of the silicon die used to make the processor.
Over the years, manufacturing advanced has reduced the size of the die needed for processors. This has allowed processors to be smaller and run cooler whilst still being powerful and using less power.
FPS
This stands for Frames Per Second, and is a simple, easy way to measure the performance of a display. The higher the FPS, the smoother and faster the graphics appear. 30 and 60FPS are common standards in gaming, especially with consoles, and PC users can achieve higher depending on hardware and the games being run.
GPU
The Graphics Processing Unit is the brain of the graphics card, and is to the display what the CPU is to your entire PC. This is where most of the work is done when it comes to outputting graphics, and it’s the most important part of any graphics card.
Integrated Graphics
Most modern motherboards feature a built-in GPU, meaning no separate graphics card is needed. However, this GPU is usually less powerful than a dedicated graphics card, and few are all that suitable for today’s games.
Integrated cards are mainly useful for more casual PC users, or those that simply don’t need too much graphical power. If you’re going to need your PC to amp up the display power, though, including image work and design, a graphics card is a better choice. Serious gamers should usually avoid on-board video at all costs.
Low Profile
Low-profile cards are smaller in physical form than full-sized models and designed for use in smaller-sized systems and for smaller motherboards. Low-profile cards are therefore often less powerful than full-sized models as they simply can’t accommodate the required cooling or memory of larger boards. This means low-profile cards are not usually the best choice for high-end gaming.
Memory Bandwidth
Arguably the most important technical specification of a graphics card, the memory bandwidth measures how effectively your card’s GPU can utilise the on-board RAM. The wider the bandwidth, the better the card’s performance. Higher bandwidths reduce data bottlenecks, and this will ensure games and other apps run more smoothly.
Memory clock
Like the core clock speed of a GPU, memory clock speed is the measurement of the speed of the video RAM. The faster the core clock, the faster the overall performance of the graphics card.
As graphics is an area of computing that’s very memory intensive, this is a very important specification of any GPU, and a good amount of RAM is paramount.
Memory Interface
The memory interface of a card is the measurement, in bits, of the pipeline the video memory utilises. Typical measurements include 256-bit, 384-bit, and so on. This higher the number, the wider the interface, which delivers better memory performance.
Nvidia G-Sync, AMD FreeSync, V-Sync
The company-specific successor to V-Sync; G-Sync and FreeSync technologies are used to try and eliminate screen tearing or frame misalignment caused by issues with a monitor’s refresh rate. Nvidia’s and AMD’s solutions strive to attain the same goal, but work differently, although they do both enable a monitor to change its refresh rate dynamically when needed in order to combat screen tearing.
V-Sync is the older, still common form of this technology, but unlike the new versions, it limits a game to running at a specific frame rate in order to prevent screen tearing. This can not only limit how well a game can run, but is also prone to causing stuttering or other negative performance issues.
PCI, AGP
Short for Peripheral Component Interconnect, the PCI ports on a motherboard are usually white and are now used mainly for expansion cards such as audio boards, USB extension cards, video capture cards, and so on. Before the advent of AGP and PCI’s successor, PCIe, PCI was also used for graphics cards, although AGP was more popular, and offered the best performance.
PCIe
Available in various revisions, such as 2.0 and 3.0, PCIe (Peripheral Component Interconnect Extended) is the faster and more powerful incarnation of PCI, and is used for graphics cards. PCIe boasts faster data transfer rates and a higher bandwidth.
The interface replaced AGP, and soundly bettered it in every way. PCIe features multiple lanes, denoted by the multiplier (x4, x8, x16, and so on). These lanes increase the bandwidth of the slot, so the more, the better. Most graphics cards these days utilise PCIe x16, and most motherboards feature this kind of PCIe, as well as secondary, slower slots.
PCI Lanes
PCI lanes are simply the routes data takes to and from the PCI or PCIe card. Each lane is able to carry one bit per cycle, so the more lanes a card and motherboard support, the faster the communication and the better the performance, with a x16 lane handling 16 bits per cycle.
Refresh Rate
The refresh rate of a monitor is how many times per second an image is refreshed. The higher the refresh rate, the smoother and clearer the image will be. For many years the standard monitor refresh rate was 60Hz. Today this has risen, and displays can handle higher refresh rates, such as 120Hz. This accommodates graphics cards that can output more than 60 frames per second, allowing users and developers to make them most out of current hardware.
Resolution
Resolution is the size of the display height and width in pixels, such as 1024x768 and 1900x1080. The higher the resolution, the crisper and more defined the image. This is why HD displays are so popular, and why 4K displays look so impressive.
All monitors come with what’s called a native resolution, which is the setting a monitor is optimised for. Running at this resolution, usually the highest a monitor will support, is optimal and gives you the best possible display for the monitor in question.
Revision
Many graphics cards are produced in revisions, meaning updates to an existing chipset. This usually means a new card may simply be an updated and enhanced version of an existing, older model, and not a totally new card, or range of cards.
Revisions also apply to the underlying technology or interfaces, such as PCIe 3.0.
Shaders
There are a few different primary shaders used in graphical applications, including pixel shaders, vertex shaders, and geometry shaders. These handle various different techniques, with pixel being the least demanding, vertex being more impressive, and geometry being them most complex.
Stock / Reference
Reference graphics cards are those produced and released by the original manufacturer, such as Nvidia or AMD. They’re the card exactly as originally developed with no changes and original coolers and features.
Stock boards are graphics card produced by other companies who license the technology from the manufacturer, also called card partners by some. Vendors like MSI, Gigabyte, EVGA, and Asus produce stock boards, which modify the original design of a reference model.
TDP
Thermal Design Power is the measurement of how much heat a GPU generates when it’s placed under strain. The higher the TDP, the more heat is generated and represents a rough estimate of how much power the device uses. TDP is measured in watts, and high-scoring cards will require larger coolers to offset the greater level of heat produced, and will often use more power.
Texture Fillrate / Pixel Fillrate
This is the measurement of the speed a GPU can render textures and effectively draw what’s on screen. A higher texture fill rate means faster rendering and a far better performance. You’ll want to find a card that has a higher fillrate if you want the best performance from your GPU. The specification of Pixel Fillrate is the same, except it deals with the number of pixels a GPU can draw on screen per second.
VGA / DVI / HDMI / DisplayPort
VGA (Video Graphics Array), DVI (Digital Visual Interface), HDMI (High Definition Multimedia Interface), and DisplayPort are all different connection methods used to connect your graphics card to a display.
VGA and DVI are the older options here, with HDMI now being considered the best by a good deal of users. DisplayPort isn’t as popular as the other three, but is still used by a good deal, and is the best way to ensure support for 4K displays at 60Hz. HDMI 2.0, however, is also able to handle 4K at 60Hz, whilst older HDMI will struggle.
HDMI is now hugely popular thanks in part to the support for higher resolution devices, and its ability to carry both video and audio. It’s also used in most non-PC displays, whereas older methods aren’t always present. This opens up the connectivity possibilities above and beyond standard monitors.
VRAM
This is Video RAM, and is the memory mounted on the graphics card for the sole use of the GPU. It’s often labelled as GDDR followed by a number, with the higher numbers being the latest, and therefore usually faster and more powerful RAM.
3D API
3D Application Programming Interfaces like DirectX and OpenGL are used to deliver the actual graphical interpretations on screen, and are what facilitates all of the impressive shaders, lighting, mapping, and other effects. To get the best performance in gaming your card will need to support the latest version of these, otherwise some new graphical effects may not be available to you. You may not even be able to run some games at all.
Levelling
Very common terms used in the graphics card market are entry-level, mid-range, high-end, and other similar phrases. As with other industries, these simply denote a video card’s market and budget range, with entry-level being the cheaper, lower-end card, and high-end often representing the cutting edge but expensive options.
Entry level cards in general are either non-gaming models designed for every day simple use, or gaming cards that are cheap, but decent enough to play games at the minimum level, often with a need to reduce visual quality.
Mid-range cards bridge the gap between low and high end and offer an affordable, all-round solution that can handle all the basics and deliver decent gaming performance with far less compromise in visual quality. They won’t break benchmark or visual quality records, though.
High-end cards are the enthusiast models that may cost a lot of money, but deliver the best possible graphical performance at the highest detail settings. Going this far, many uses even opt to run two or more of these cards in CrossFire or SLI, further increasing their PC’s visual muscle. If you’re serious about gaming, want the best experience, and can afford it, this is where you’re going to have to look.