Graphics card comparison

We've got everything you need to know to find the right graphics card for your needs and budget.

We break down the complexities of modern graphics cards to help you find the one that's right for you.

Need help picking out the perfect graphics card?

Whether you're an avid PC gamer, an up-and-coming 3D animator or a keen crypto miner, you're going to need a capable graphics card to power your passion.

Occupying the PCI Express slot in most modern computers, the humble graphics card is often the most expensive component of any high-end PC. This is hardly surprising when you consider what it's responsible for. Every pixel on your computer screen is controlled by the graphics card, from the icons on your desktop to the shrapnel raining down in a match of Call of Duty. In fact, without a graphics card of one form or another, a computer is incapable of displaying any sort of visual interface, rendering it practically useless for the vast majority of regular users.

While recent years have seen a rise in integrated graphics solutions as an alternative to traditional graphics cards, this approach has significant limitations. With an integrated graphics solution, the graphics processing unit (GPU) is built into a computer's motherboard often in close proximity to the central processing unit (CPU) – the chip that serves as the brains of a PC. This takes up far less space than a dedicated graphics card and is usually a lot cheaper, but the compact size means less power and typically requires the CPU and GPU to share resources like system memory – a further knock to performance that dedicated graphics cards don't have to worry about.

For laptops and PCs used primarily for work, study or browsing the Internet, integrated graphics are often sufficient. But for any serious gamer or digital artist, a dedicated graphics card is an absolute must.

Back to top

How do graphics cards work?

There are three main components to a graphics card: the GPU, the on-board memory and the display ports. The GPU is the heart of the operation, a tremendously powerful chip that performs the complex calculations required for constructing and displaying visual elements on a computer's screen. As data comes in for processing from other components of the PC, it is stored in the graphics card's on-board memory, which in most modern graphics cards consists of 1GB or more of GDDR RAM. Once this data has been processed, it is transmitted through one or more of the card's display ports and shows up on the connected screen(s).

While there's more to the process from a technical perspective, this is essentially what graphics cards spend most of their time doing. To ensure smooth, flicker-free output on your screen, they need to deliver a new frame – a static image just marginally different from the one before it – many times a second so as to create the illusion of motion. If you've ever drawn your own flip book, it's the same general concept.

Back to top

Nvidia vs AMD

For decades now, there have been just two key players in the graphics card space: Nvidia and AMD. Each has its own range of cards it updates with new models as technology and manufacturing techniques improve, with Nvidia currently focusing on its GeForce line and AMD putting its weight behind its Radeon branding.

Unlike the split between Intel and AMD in the CPU market, though, both GeForce and Radeon graphics cards are built on roughly the same technology base. This means that where Intel and AMD CPUs require different motherboards with specific CPU sockets, modern Nvidia and AMD graphics cards will work in any motherboard with a PCI Express slot. This makes it possible to install either a GeForce or Radeon card in the same computer, with no need to swap out other components.

What are your main options?


Modern graphics cards fall under one of two umbrellas: Nvidia's GeForce line or AMD's Radeon line. Both lines feature a range of models spanning budget, mid-tier and high-end offerings, as you can see below.

Nvidia GeForce

Nvidia's approach to graphic card design is all about power. Each generation of its GeForce line is driven by a desire to push the bounds of technology, and as such it typically begins each generation by launching a high-end model before gradually rolling out a variety of more budget-friendly options.

The most common GeForce line is currently the GTX 10 series consisting of the cards listed below. Nvidia has also recently revealed its new RTX 20 series graphics slated for release later in 2018.


GTX 1050

Base clock speed
1354MHz
Memory
2GB GDDR5 @ 7Gbps
Memory bus
128-bit
Price range
$150 - $250
the-witcher-3-small

GTX 1050 Ti

Base clock speed
1290MHz
Memory
4GB GDDR5 @ 7Gbps
Memory bus
128-bit
Price range
$250 - $300
the-witcher-3-small

GTX 1060

Base clock speed
1506MHz
Memory
3GB/6GB GDDR5 @ 9Gbps
Memory bus
192-bit
Price range
$380 - $550
the-witcher-3-small

GTX 1070

Base clock speed
1607MHz
Memory
8GB GDDR5 @ 8Gbps
Memory bus
256-bit
Price range
$750 - $900
the-witcher-3-small

GTX 1070 Ti

Base clock speed
1607MHz
Memory
8GB GDDR5 @ 8Gbps
Memory bus
256-bit
Price range
$900 - $1100
the-witcher-3-small

GTX 1080

Base clock speed
1607MHz
Memory
8GB GDDR5X @ 10Gbps
Memory bus
256-bit
Price range
$950 - $1200
the-witcher-3-small

GTX 1080 Ti

Base clock speed
1582MHz
Memory
11GB GDDR5X @ 11Gbps
Memory bus
352-bit
Price range
$1300 - $1500
the-witcher-3-small

AMD Radeon

Where Nvidia prefers the power play, AMD has set its sights on delivering the most bang for your buck with its mid-tier and budget-level Radeon RX graphics cards. While they might not match the raw power of the high-end GeForce GTX models, these cards are more than capable of handling the demands of modern games at a price that won't leave you eating packet ramen for the foreseeable future.

AMD's latest Radeon line is split between the standard RX 500 series and the high-performance RX Vega series.


Radeon RX 550

Base clock speed
1100MHz
Memory
2GB/4GB GDDR5 @ 7Gbps
Memory bus
128-bit
Price range
$150 - $200
the-witcher-3-small

Radeon RX 560

Base clock speed
1090MHz
Memory
4GB GDDR5 @ 7Gbps
Memory bus
128-bit
Price range
$200 - $300
the-witcher-3-small

Radeon RX 570

Base clock speed
1168MHz
Memory
4GB/8GB GDDR5 @ 7Gbps
Memory bus
256-bit
Price range
$430 - $530
the-witcher-3-small

Radeon RX 580

Base clock speed
1257MHz
Memory
8GB GDDR5 @ 8Gbps
Memory bus
256-bit
Price range
$570 - $650
the-witcher-3-small

Radeon RX VEGA 56

Base clock speed
1156MHz
Memory
8GB HBM2 @ 1.6Gbps
Memory bus
2048-bit
Price range
$999 - $1199
the-witcher-3-small

Radeon RX VEGA 64

Base clock speed
1247MHz
Memory
8GB HBM2 @ 1.9Gbps
Memory bus
2048-bit
Price range
$1149 - $1399
the-witcher-3-small
Back to top

What are the differences between branded versions of the same graphics card?

One of the trickier aspects of shopping for a graphics card comes from the fact that Nvidia and AMD aren't the only companies manufacturing cards. In fact, you're more likely to spot a Gigabyte-branded GTX 1060 or an XFX-branded Radeon RX 580 than you are a stock model from either of the two GPU designers.

This is because Nvidia and AMD provide third-party hardware manufacturers with the reference designs for their cards, which companies like Gigabyte and XFX then use as the basis for their own custom versions of those cards. This arrangement saves Nvidia and AMD money on production costs while still earning them revenue through licensing fees from those third-party manufacturers.

Key third-party manufacturers include:

  • Gigabyte
  • Asus
  • EVGA
  • MSI
  • XFX
  • PNY
  • ZOTAC
  • Sapphire
  • PowerColor
  • Inno3D
  • GALAX
  • Colorful
  • Leadtek

Because branded graphics cards are all based on the same reference design, they don't typically stray too far from the original specifications. Most of the time, the differences will be fairly minor and focused predominantly on slight improvements to efficiency.

Common differences between branded graphics cards include:

  • An overclocked GPU
  • Additional, more efficient cooling
  • Quieter fans
  • Smaller form factor
  • Extra display ports
  • Aesthetic elements like RGB lighting
Back to top

Graphics card terminology demystified

RAM

Random access memory. The high-speed, short-term memory built onto a graphics card that stores data while it's being processed by the GPU. Also referred to as VRAM. Typically, more memory means a graphics card can handle higher-resolution textures and load in new assets faster, minimising the stutters and hitches that occur when it has to pull in new data from the hard drive.

Memory bus

While the amount of RAM a graphics card has is often one of the first buying considerations, the memory bus is just as critical in determining performance. To process data, a graphics card must move that data from its RAM to its GPU along its memory bus. The wider the bus, the more data can be moved at one time, and in turn the faster the GPU can get on with its job. Memory bus width is measured in bits and goes hand in hand with the memory clock speed.

HBM2

High-bandwidth memory. An alternative to the standard GDDR specification adopted by GeForce 10 and Radeon RX 500 cards, HBM2 takes the small DRAM modules that store data and stacks them on top of each other, allowing for a much wider memory bus and therefore higher bandwidth than GDDR offers. Because of this, HBM2 can lower its memory clock speed while still achieving comparable performance, reducing the overall power consumption of the graphics card.

CUDA

An Nvidia technology, CUDA is an application programming interface (API) that enables general-purpose software to take advantage of the parallel processing capabilities of a GPU. This extends a graphics card's utility beyond video games and other graphical applications and into data-intensive fields where multiple large information sets must be manipulated at the same time.

Overclocking

A graphics card's GPU has a clock speed that essentially dictates how fast it can process individual instructions. The higher the clock speed, the higher the framerates you're likely to see in-game. When a GPU's clock speed is increased from the original rate set by Nvidia or AMD, that process is known as overclocking. While overclocking can deliver performance beyond a card's stock specifications, it also generates additional heat and can lead to software and even hardware failure if the cooling solution is not beefed up to keep the GPU cool.

SLI (Nvidia)/Crossfire (AMD)

Technologies that allow for the use of multiple graphics cards in a single computer. By splitting the graphical workload among two or more graphics cards in parallel, the task can be completed faster than with a single card alone. Once completed, output from the linked graphics cards is combined into a single frame and sent out to the monitor as per normal.

Back to top

What are graphics card drivers?

As far as graphics card technology has come, it's not quite plug-and-play yet. Before you can put your flashy GTX 1080 through its paces, you'll need to install the appropriate graphics card drivers so that your operating system can communicate with and control this new piece of technology you've just slotted in. Without the correct drivers, a modern operating system like Windows can only make use of a graphics card's most basic display functions, so downloading the latest drivers should be one of the first things you do after installing a new card.

It's also important to keep your graphics drivers up-to-date on an ongoing basis, as new driver versions often squash bugs and performance issues as well as improve compatibility with the latest games.

Which graphics card is best for you?

Deciding on the "best" graphics card is always going to be subjective. Your budget, your PC specs, what you plan on using it for and many other factors play a significant role in determining which model will deliver the biggest bang for your buck.

That said, it is possible to categorise the main graphics card models based on their general gaming performance, reducing the number of cards you'll need to compare. These categories are:


Budget graphics cards (~1080p Low-Medium 60fps)

These cards are designed for gamers happy to dial down the graphics settings of the latest 3D games or stick with the less-demanding 2D fare coming out of the indie space.

Name Product Brand Model Name Memory Provider in CTA
MSI
RX 560
4GB
ASUS
RX 550
2GB
ASUS
RX 560
2GB
ASUS
RX 550
4GB
XFX
RX 560
4GB

Compare up to 4 providers


Entry-level graphics cards (~1080p Ultra 60fps)

For gaming at 1080p, these cards will hit the coveted 60fps threshold at max settings in all but the most graphically intensive games.

Name Product Brand Model Name Memory Provider in CTA
GALAX
GTX 1050 Ti
4GB
MSI
GTX 1050 Ti
4GB
MSI
GTX 1050 Ti
4GB
ZOTAC
GTX 1050
2GB
EVGA
GTX 1050 Ti
4GB