The 10 Best Graphics Cards
This wiki has been updated 24 times since it was first published in March of 2016. The graphics card is one of the most important components in a PC and can make or break the entertainment or work experience. Available in a huge range of prices, there are models of varying video capabilities to suit the needs of everyday users, gaming enthusiasts, and professional designers and engineers. We've selected some of the most powerful GPUs as well as the most affordable. When users buy our independently chosen editorial recommendations, we may earn commissions to help fund the Wiki. If you'd like to contribute your own research to the Wiki, please get started by reviewing this introductory video.
October 16, 2020:
The release of Nvidia's RTX 3000 series was a welcome announcement, but the GPU leader was immediately criticized when anxious consumers realized that there would be massive shortage of these powerful GPUs extending well into 2021. As such, it's not exactly the best time to buy a new graphics card, but there are some useful products worth highlighting.
The two that we really want to mention are the Gigabyte Radeon RX 5700 XT and XFX RX 5500 XT, which are capable of decent 1440p and 1080p performance, respectively. Their main positives, however, are their cost-effectiveness and, quite frankly, their availability.
We've also upgraded our references to a handful of last-generation Nvidia cards, such as the EVGA GeForce RTX 2070 Super, PNY GeForce GTX 1660 Super, and Gigabyte GeForce RTX 2080 Super. These "Super" variants are specifically engineered to endure faster clock speeds and have higher memory bandwidth, so they consistently outperform their vanilla counterparts.
With all that said, we hope to revisit this category as well as the VR GPU field in a few or several months, when a consistent supply of RTX 3000 series cards, and hopefully AMD's new "Big Navi" heavyweight GPUs, are readily available in various outlets.
July 23, 2020:
We're currently awaiting the announcement and subsequent release of both Nvidia and AMD's new lines of graphics cards. When we have more information on the potential Nvidia GTX/RTX 3000 series as well as AMD's "Big Navi," we'll update this and some similar Wikis with the most up-to-date information. Check back around mid-October and there will hopefully be further developments.
April 24, 2019:
It was surely an exciting ride, but thankfully, we have left behind the dark days when even the average 3D gaming enthusiast was nearly forced to sell body parts just to buy a passable GPU. That's not to say that high performance is suddenly inexpensive -- or that those older cards have dropped in price with the release of new technology, because they haven't -- but we're finally seeing a wide range of hardware that doesn't elicit an "oof" as soon as you see the sticker. So we've compiled some of the most lauded variants of the most capable chipsets around.
To that end, it shouldn't be a surprise that the top spots go to Nvidia's brand new line of RTX cards. the 2060 is just about the best overall value out of everything on the market right now, and its mid-range price won't hit most users too hard. If you're looking for an upgrade, it's hard to go wrong with the 2070, and all but the most demanding gamers will be satisfied with it. On the other hand, the 1660 is ideal for those who stick to low-intensity games, and it's priced as such. The 1660 Ti is an excellent video card as well, and does a killer job at pumping out 1080p titles smoothly, though it occupies a somewhat middling spot: for just a bit more money, you could spring for the noticeably more powerful 2060. And if you want buttery-smooth 1440p gaming on even the newest games, the 2080 is a great choice, though it is considerably costly.
I could go on about how impressive the 2080 Ti is, but the fact is that it costs an absolutely ridiculous amount of money. If you're willing to make that investment, the EVGA model we've highlighted is consistently viewed as one of the best, though all of EVGA's closely related variants are of great quality. I'd also like to mention that, according to vendors, RMA of the 2080 Ti has finally sunk to the same rate as those of other cards, which means that the memory issues that the card experienced at launch appear to have been fixed. Do be aware, however, that the 2080 Ti requires a power supply that is not only powerful, but can also support multi-rail configurations using two independent 8-pin power cables, rather than a split adapter. In other words, it takes a considerable amount of tech know-how to ensure that your installation is appropriate and safe.
Nvidia's 1650 is the most recent chipset to come out, and it will be followed by the Ti version soon, but know that the base 1650 isn't really ideal for gaming. It's outperformed by multiple of the AMD models we've covered in this list. It is a great option for everyday PC usage though.
Speaking of AMD, they do have some great offerings. The 570 is about as cheap as a video card gets nowadays, and it performs surprisingly well. The 590 is one of the most recent releases, and likely the last Polaris-based model, and it's more efficient than its predecessors. Finally, the Radeon VII is so named because of its 7nm transistors, and it performs similarly to the RTX 2080, plus it's priced about the same as well. Its unmatched 16GB of high-bandwidth memory make it able to hold multiple buffered frames at high resolutions, which is great for 4k performance.
The Four Most Important Things to Know about Graphics Cards
The higher the clock speed, the faster your graphics card can process the data stored by the VRAM.
Every time you upgrade one piece of hardware in your beast of a desktop PC, you always have to consider whether or not the rest of the hardware can handle it. A motherboard without PCI Express 3.0 is not going to support a current generation graphics card and a current generation graphics card is not going to accept anything less than a 400-watt power supply.
Knowing that the most important aspect of a high-end graphics card is whether or not it will support the maximum settings on the most recent games, the fact still remains that you cannot take advantage of such a card if you don't have the power to run it.
Thus, the first thing you need to know about your new graphics card is whether or not you have the wattage to get so much as the fans up and running. Does it require a 750-watt power supply? Can you even fit a 750-watt power supply, which is going to be twice the dimensions of a 250-watt supply, in your current case? If not, chances are your motherboard is too small, you may not have enough dynamic random-access memory (DRAM), and your processor may be behind the times.
In the event you already have enough power to run your card of choice, there are three other things worth taking into consideration: video RAM (VRAM), memory interface width, and clock speed.
The more VRAM your card has, the higher the resolution at which your card can render graphics in real-time. The greater the memory interface width, the quicker your card can render those graphics. The higher the clock speed, the faster your graphics card can process the data stored by the VRAM.
Thus, these three things can be prioritized in order: the more data you store, the more data you have to work with, and the more data you have to work with, the more inclined you are to maintain the highest possible speed.
Understanding the Law of Diminishing Returns
While shopping around for a new graphics card, we often come across the term diminishing returns--a term that applies, in our case, to both the prices of graphics cards and the number of polygons used in computer graphics during the video game development process.
As computer-generated imagery (CGI) artists double the number of polygons they use while generating a particular image, from 2,000 polygons to 4,000 to 8,000 and so on, the difference in quality between subsequent generations becomes less and less noticeable to the untrained eye. Thus, the difference between the original Xbox and the Xbox 360 is much more drastic than the difference between the Xbox 360 and the Xbox One, despite the fact the first and second consoles were released three years apart while the second and third were released eight years apart.
By comparison, more recent advancements in computer graphics technology have resulted in marginally better graphics than prior advancements made between, say, the original 16-bit Nintendo and the 32-bit Super Nintendo. As a result, what we end up with are price-based diminishing returns that help us to determine where to draw the line on cost while shopping around for graphics cards that best suit our needs.
On the one hand, a graphics card capable of displaying 4,000 polygons is going to be drastically better than a card only capable of displaying 2,000 polygons, but the price will be marginally different. On the other hand, a graphics card capable of displaying 2,000,000 polygons is going to be drastically more expensive than a card only capable of displaying 1,000,000 polygons, but the image quality will only be marginally different, and may even be unnoticeable to customers that don't fancy themselves videophiles.
When Graphics Cards Are Too Big for Their Britches
One of the main issues with purchasing a current graphics card, an issue consumers never seemed to have in the past, is the ever-increasing size of the cards. Due to overclocking and the need for upwards of three large fans, thin cards that once fit snugly in PCI slots at right angles perfect enough to balance marbles on, are now fat cards that sag at obtuse angles to the point their immense weight has been known to tear PCI slots straight out of the motherboard if not given additional support; support not supplied by the manufacturers of either the over-sized graphics cards or the computer cases that houses them.
Alas, computer hardware manufacturers have never been ones to cooperate.
What consumers end up with is the need to use household items that would otherwise never be seen inside a computer case performing feats of great strength that motherboards themselves are not designed to adequately perform. Bundles of twigs wrapped in duct tape should not be considered obligatory while building a high-end gaming PC.
Alas, computer hardware manufacturers have never been ones to cooperate. An NVidia graphics card manufactured by EVGA should be paired with a motherboard manufactured by EVGA, otherwise you run the risk, however so slight, that the drivers for each will start bickering just as you're about to beat the final boss of Dark Souls III or finish rendering an animation for your latest youTube video.
In the end, it's up to you. Can your current motherboard support the weight of your new graphics card? Do you have the engineering skills to prop it up with sticks or toys? Will the new card fit inside your current case?
Perhaps on top of recommending one brand over the other, the gurus of the PC master race should recommend carrying a ruler and a spring scale whenever it comes time to do some shopping.
Statistics and Editorial Log