Today, we’re going to look at the GPUs available at the time of writing (Q1 2019), and help you choose which is right for you. Whether it’s for your budget-friendly PC or a bang of the buck set for 4k Gaming, any number of cards could be considered the best graphics card for 2019. With that in mind, we’ve selected eight to fit every possible type of user we could think of.
Let’s hop into it.
Best Graphics Card For 2019
|Graphic Cards||Specifications||GPU Architecture||Price|
|GTX 1050 Ti||Clock Speed – 1354 MHz Base/ 1468 MHz Boost|
VRAM – 4GB GDDR5 RAM
|Nvidia Pascal||Check Price|
|RX 570||Clock Speed – 1168 MHz Base/ 1244 MHz Boost|
VRAM – 8GB GDDR5 RAM
|AMD Polaris||Check Price|
|RX 580 (8GB)||Clock Speed – 1257 MHz Base/ 1340 MHz Boost|
VRAM – 8GB GDDR5 RAM
|AMD Polaris||Check Price|
|RX Vega 64||Clock Speed – 1247 MHz Base/ 1546 MHz Boost|
VRAM – 8GB HBM2 RAM
|AMD Vega||Check Price|
|RTX 2070||Clock Speed – 1410 MHz Base/ 1620 MHz Boost|
VRAM – 8GB GDDR6
|Nvidia Turing||Check Price|
|GTX 1070 Ti||Clock Speed – 1607 MHz Base/ 1683 MHz Boost|
VRAM – 8GB GDDR5
|Nvidia Pascal||Check Price|
|RTX 2080||Clock Speed – 1515 MHz Base/ 1710 MHz Boost|
VRAM – 8GB GDDR6
|Nvidia Turing||Check Price|
|Titan RTX||Clock Speed – 1350 MHz Base/ 1770 MHz Boost|
VRAM – 24 GB GDDR6
|Nvidia Turing||Check Price|
And, if you want a detailed review of our top picks, here you go.
Table of Contents:
- Best Budget Graphics Card – GTX 1050 Ti
- Best Graphics Card Under $200 – RX 570
- Best Value Graphics Card – RX 580 (8GB)
- Best AMD Graphics Card – RX Vega 64
- Best Nvidia Graphics Card – RTX 2070
- Best Graphics Card For VR and 1440p Gaming – GTX 1070 Ti
- Best Graphics Card For 4K Gaming – RTX 2080
- Best Overall Graphics Card – Titan RTX
- Breaking Down Graphics Card Tech and Terminology
- The Importance of GPU Benchmarks (and resources)
- Frequently Asked Questions (FAQ)
Best Budget Graphics Card – GTX 1050 Ti
The GTX 1050 Ti is the best budget GPU and has been since its release. A glance over the benchmarks will easily reveal why: this card is doing an excellent job at driving AAA games at 1080p and medium-to-high settings since it was released!
One thing that’s helped the 1050 Ti stay ahead of past cards, like the GTX 760 (and even the 960 in some cases!) is its VRAM allocation. Specifically, the 1050 Ti boasts 4GB in VRAM, whereas most previous generation cards targeted at 1080p only had 2GB in VRAM. This may seem like a lot for 1080p, and in most cases, it is…until you get to the latest releases.
When it comes to the latest releases, that extra VRAM really starts coming in handy. Perhaps due to the large VRAM pools on current-gen consoles (8GB GDDR5 on the PS4, for instance), texture resolution across the board has been pushed higher and higher. While the PS4 and Xbone weren’t particularly well-known for pushing higher resolutions, the extra VRAM helps accomplish that on PC too.
The end result is that the 1050 Ti ends up being an excellent card for modern games at 1080p and reasonable settings, and even 1440p with older titles. At the time of its release, the 1050 Ti didn’t quite demonstrate its value over a 760 or 960, but with the passage of time, we’ve seen how that has changed.
- The best entry-level GPU, perfectly suited for 1080p gaming at high settings
- Low price, strong value
- Can’t really go above 1080p
- Beaten in value by a few more expensive picks- extra saving recommended if you ever plan on playing above 1080p
The main catch with the 1050 Ti isn’t really a problem with the card itself, just what it can’t do. Despite how well 4GB VRAM turned out for a 1080p card, more intensive modern games have proven to need more than just a large memory pool to swim around in. When it comes to raw GPU clock performance and other concerns, the 1050 Ti doesn’t quite stack up against titles like Monster Hunter: World…at least not at max settings.
In other words, the 1050 Ti is still a GPU from 2016. The latest AAA releases are starting to leave it behind, making gaming at 1080p60 and max settings with this card not quite as viable as it once used to be. While this is still a pretty decent buy for your money, it’s not the best bang for your buck and it won’t last too much longer if you demand the highest graphics settings.
Best Graphics Card Under $200 – RX 570
The RX 570 is our pick for best GPU under $200, and for good reason.
First and foremost, let’s talk performance.
Unlike the GTX 1050 Ti, this card hasn’t fallen victim to the passage of time nearly as much. Namely, because it’s a little bit newer (early 2017), but also because it was built to be better from day 1.
The RX 570 is perhaps the best 1080p graphics card currently available on the market, capable of pulling 1080p60 at high-to-max settings in pretty much any scenario. It pretty handily beats the GTX 1050 Ti in every scenario despite not being that much more expensive, and in fact competes with the GTX 1060 3GB and 6GB, which start in the low $200s.
In addition to 1080p performance, the RX 570 also puts on an admirable showing as a 1440p card…as long as you opt for the 8GB version. With the 8GB version of the card, you’ll be able to play titles like the 2016 DOOM reboot at 1440p, Ultra settings, and around 80 FPS on average. Not bad at all for a card retailing under $200.
Ultimately, the RX 570 is really a stripped-down version of its bigger brother, the RX 580. It comes roughly in 2nd in terms of bang-for-buck, with the 580 itself in 1st and the 1050 Ti in 3rd.
- Pretty good value for its price
- Offers strong 1080p performance and entry-level 1440p/VR performance
- Has most of the RX 580’s performance, but doesn’t quite beat it in value
- 4GB version isn’t much cheaper- opt for 8GB, even if only playing in 1080p
- More fiercely competed with by the GTX 1060 variants- watch for their pricing before buying this one
That being said, let’s talk about where this card falters.
First and foremost, don’t bother with the 4GB version. Even if you’re only playing in 1080p, there is little, if any, appreciable price difference between the 4GB and 8GB versions of this card. You won’t save money by buying the version with crippled VRAM, and you’ll also miss out on better performance in 1440p or with high-res textures.
Next, let’s talk about its rivals: the two GTX 1060 cards. The RX 570 competes with both of these cards, especially in its 8GB configuration. In most cases, it beats them in price and almost always beats the 3GB in performance.
However, due to the recent launch of the Nvidia RTX cards, you will see many price drops and sales on the GTX 1060 cards. If you notice that the GTX 1060 6GB is at the same or lower price as the RX 570, you should probably grab that one instead. It won’t be a major performance improvement, but it will still be an improvement.
Best Value Graphics Card – RX 580 (8GB)
Breach $200 and higher and you start to see the AMD/Nvidia rivalry really heat up. The GTX 1060 3GB, 1060 6GB, RX 580 4GB, and RX 580 8GB, all exist in this price range. It’s…a lot to keep track of, to say the least.
However, we’re going to go ahead and simplify it for you.
Don’t bother with the 580 4GB or the 1060 3GB. Neither of these cards are worth spending $200 on, and the latter is beaten by the cheaper RX 570. The only choices worth considering are the 580 8GB and 1060 6GB…and not only is the 580 usually cheaper, it also performs better in most games.
In terms of 1080p gaming, congratulations: you aren’t going to have any problems, ever. The massive 8GB VRAM pool ensures that high resolutions and HD textures are not a problem whatsoever, and the core performance of the actual GPU is nothing to scoff at, either. The 570 already dominates the sub-$200 due to its excellent performance-per-dollar, and the 580 is anywhere from 10 to 20 percent better in games.
As long as you’re willing to cut down on a few settings (namely AA), this card is also an amazing value pick for 1440p gaming. If you want the best possible value- that is, performance-per-dollar- this is the card for you. Just don’t try to max out games at 4K and you should be fine.
- Stellar value, topping the chart in performance-per-dollar
- Decimates 1080p gaming- suitable for 1440p and VR as well
- High price
- Just suitable for 1440p and VR- won’t push these to their highest settings and fastest speeds
First and foremost- the 4GB version. Seriously, do not buy the 4GB version. Your 1080p performance will likely be just about the same, at least for now, but your 1440p and VR performance will plummet, and we doubt you want that.
Additionally, the 4GB version isn’t even a good value! Where the 570’s 4GB version would be at least $5 or $10 cheaper than the 8GB counterpart across the board, the 580’s 4GB versions usually cost the same or more than their 8GB counterparts!
Despite their names, the 1060 3GB and 6GB are actually different cards. The VRAM isn’t all that changes between them. There is a meaningful distinction between their prices and usage. This isn’t the case with the 570 and 580- the 4GB versions of these cards are just straight-up worse and don’t save you much money at all.
Best AMD Graphics Card – RX Vega 64
Until the release of the Radeon VII on February 7th, 2018 (comment below if this date has passed and we haven’t updated this article yet!), the RX Vega 64 is the best AMD GPU on the market.
The Vega 64 boasts a great many claims to fame.
First and foremost, let’s talk about what makes it most unique: the HBM2 memory inside the card.
The Vega 64 boasts 8GB of HBM2 VRAM. If you don’t understand what HBM2 is, we’ll explain it in more detail later in the article. For now, just understand that HBM means High Bandwidth Memory and that it outperforms the industry standard GDDR5 pretty significantly. That makes the Vega 64 particularly well-suited for handling high-quality assets and resolutions up to 4K with little issue.
As one may imagine, that HBM2 pays off in spades when it comes to gaming performance. 1080p is an afterthought on this card, if only because it’s so thoroughly decimated. 1440p and 4K are where this card really shines because these resolutions will actually put the cards through its paces. The latest AAA games should never be a problem with this card at 1440p, and many of them will be quite playable at 4K and high settings.
If you fancy yourself a proud member of Team Red and you want amazing high-resolution gaming performance, the RX Vega 64 is a pretty good card for you.
- Great performance in 1440p and VR games- better than the 1070 Ti, in fact!
- HBM2 memory results in faster overall speeds and better management of high resolutions and HD assets
- Somewhat poor value, especially compared to the RX Vega 56
- High heat output and power consumption
- Somewhat expensive
However, it is a bit questionable.
First and foremost, the Radeon VII looks to retail for just about the same price and outperform this one pretty significantly. If you’re willing to wait just a few months or so before upgrading your GPU, you probably should: you’ll spend the same amount and get more performance for your money.
Even without the Radeon VII looming in the distance, however, there’s also the value proposition. The 1070 Ti comes fairly close to this card in performance, and retails a lot cheaper most of the time. The RX Vega 56 is also very close to this card in performance, and also a lot cheaper. Value-wise, this isn’t the best pick you could make.
Aside from value and new GPUs on the horizon, there is also the issue of heat. This card sucks in a lot of power, and that means it exhausts a lot of heat, which can be bad if you don’t already have a very strong cooling setup in your PC.
While we still consider this card good enough to make this list, it is important that you keep in mind these caveats before buying. We have noticed that prices are starting to drop with the announcement of its successor, though, so maybe you’ll be able to get it for a better deal.
Best Nvidia Graphics Card – RTX 2070
The RTX 2070 is our pick for best Nvidia GPU. It isn’t the strongest Nvidia GPU- that’s later in this list, as our strongest overall GPU- but it’s one of the strongest GPUs on the market, and one of our favorites. However, we still should explain why it won out best over the obvious two contenders of RTX 2080 and RTX 2080 Ti.
In the case of the RTX 2080, we passed over it because right now, it’s essentially just a 1080 Ti that costs an extra $100. Or it would be, if Nvidia hadn’t cut down 1080 Ti shipments in order to force people buying new GPUs to move to this card sooner rather than later.
In the case of the RTX 2080 Ti, we’ll discuss that in a later entry.
Ultimately, though, we came down to the RTX 2070 for this entry. It outperforms the GTX 1080 by a good bit, but falls short of the 1080 Ti. This still makes it an excellent card for 4K and 1440p gaming, however, since even a basic GTX 1080 is still enough to tackle today’s hottest games. The value scaling isn’t quite perfect, sure, but we’d say it’s about here where buying a high-end GPU still makes sense.
If what you want is a stellar experience gaming at 4K or 1440p, congratuiations: this card will definitely deliver that. If what you want is RTX.
This card will deliver that, too.
At launch, the RTX line was fraught with a great many issues. Most especially, performance issues and a general lack of programs for RTX to take advantage of. With the release of Battlefield 5’s ray-tracing, however, the RTX 2070 reveals itself as the best way to play games at 1440p and Ultra settings with ray-tracing enabled.
This is a step up from the RTX 2060’s High/Ultra mix with ray-tracing enabled at 1080p. (Nvidia claims that RTX 2060 is enough for BF5 with ray-tracing and ultra settings at 1080p60, but Digital Foundry proves this not to be the case.)
- Stellar 4K, 1440p, and VR performance
- RTX features, where supported, are excellent
- Poor price-to-performance, high price overall
- RTX isn’t really worth paying extra for yet
Truth be told, the only problem there is with this card is that the performance-per-dollar isn’t all that great. Nvidia has passed on the cost of RTX to the consumer, and at least for now, that gambit hasn’t quite paid off. Most games aren’t supporting the technology yet, and even in the best case scenarios, it does still seem to come at a severe performance penalty. Severe enough that a card that would normally run a game maxed-out at 4K needs to drop down to 1440p.
If you’re willing to pay a bit extra and don’t plan on selling a kidney for your GPU, however, we believe the RTX 2070 is a good pick. It’s not as absurdly overpriced as the RTX 2080, it doesn’t have the manufacturing mishaps of the RTX 2080 Ti, and overall it just seems like a good card.
Best Graphics Card For VR and 1440p Gaming – GTX 1070 Ti
The best GPU for VR and 1440p gaming is the GTX 1070 Ti, hands down. It’s not the strongest GPU for these purposes, but it is the most sensibly-priced, at least, for these purposes. It’s still pretty expensive in general.
In terms of pure performance, the GTX 1070 Ti offers fairly superb performance in all but the most demanding games. Neither 1080p or 1440p should pose any challenge to this card, even if playing at high refresh rates (120+ Hz), which makes it particularly ideal for a G-Sync monitor setup.
Additionally, the performance-per-dollar actually does line up a bit here, at least for 1440p gaming in particular. Options like the 1080 (Ti) and RX Vega 64 are much more expensive but don’t provide a particularly big performance boost over this card. In fact, the 1070 Ti pretty handily poaches the GTX 1080 with some overclocking but that’s just how the GPU market is sometimes, we guess.
Worth noting is that if you’re reading this article after the release of the RTX 2060, you should buy that card instead if it isn’t much more expensive. The 2060 beats the 1070 Ti in most scenarios and even competes with the GTX 1080, all while retailing at an MSRP of $350- even cheaper than some 1070 Tis out right now. Plus, it adds RTX!
- Great value pricing compared to the competition in this price range
- Completely equipped for VR gaming, 1440p gaming, and even some 4K gaming
- While better value than contemporaries, still somewhat low in terms of price-to-performance
- No RTX
The biggest catch concerning this card is the lack of RTX, but that isn’t a very big catch. Despite the extensive marketing around Nvidia’s RTX technology, the truth is it’s going to be a very long time before that feature becomes truly mainstream in gaming. If it’s really that big a problem, just wait for the RTX 2060 to come along and grab that one instead- everything we said here applies just as well to that card.
The price, will high, is still relatively fair for what you’re getting, especially in comparison to higher-end offerings from both AMD and Nvidia.
Best Graphics Card For 4K Gaming – RTX 2080
In our humble opinion, the RTX 2080 is the best GPU for 4K gaming. There are a great many reasons for this, but we should address the elephant in the room: why not the 2080 Ti?
With the 2080 Ti, we had to look at the astonishingly poor performance-per-dollarand the fact that it’s defective. While there is a possibility that this was exclusive to early Founders Edition runs, it’s still a startling problem for a top-billed product that was sold in preorders and delayed shipments. RTX 2080 Ti is steadily improving as a buy, but we’re cautious to recommend it for now.
In the absence of the 2080 Ti as a viable recommendation, we come instead to the RTX 2080, which offers very comparable performance at a lower price and lower failure rate. This may seem like we’re unreasonably ragging on the 2080 Ti, but we’ve done enough research that we just don’t want to risk recommending a potentially bad card to our readers. It’s not often that a GPU launches with as many issues as the 2080 Ti, but when it does we take notice.
If what you’re looking for is top-of-the-line 4K, 1440p, VR, and even RTX performance, Congratulations. You’ve found it. No game on the market, and no game coming for at least the next few years, should give this card a worthy fight. As long as the rest of your system can keep up to it, you should be just fine with this one for the foreseeable future.
- Nvidia’s flagship GPU for 4K gaming
- Most of the performance of the 2080 Ti at a lower price and without the alarmingly high failure rate
- RTX features, where supported, enhance the gaming experience
- Poor value for the high price
- RTX technologies have yet to reach the mainstream
The main catch here is the price, which is fairly high. However, AMD hasn’t released any competing cards in this range yet, and until that happens, the prices are going to stay where they are.
The higher prices on RTX versus the GTX 10 Series are partially due to the new technology being introduced, and partially due to Nvidia’s market dominance in the $400+ GPU range. With any luck, releases from AMD later this year should shake things up on this list and give us either a better option for 4K gaming or make this one cheaper.
Additionally, as we’ve noted on all RTX cards thus far, RTX support has yet to meaningfully penetrate the mainstream. What this means for gamers is that only a few of the latest AAA games are going to support the tech you’re paying for today, and chances are that ray-tracing enabled GPUs are going be faster and cheaper a few years down the line.
Even with those downsides in mind, though, this is still a genuinely great card. It’s also easily your best option if you want to game at 4K without splurging both your kidneys and an eye for a Titan RTX.
Best Overall Graphics Card – Titan RTX
This is a difficult card to discuss, but the Titan RTX is undoubtedly the best overall graphics card on the market. It’s mainly gaming-oriented, but the Titan series has proven to be quite capable in pro workloads as well, effectively blurring the lines between consumer and pro GPUs.
If you’re a serious pro user, we recommend opting for one of the workstation GPUs we’ve placed above. If you’re just interested in playing games, you should probably go for one of the mortal-tier graphics cards listed above as well.
The Titan RTX isn’t.
- The best-performing graphics card on the market
- RTX architecture taken to its fullest- should be leagues better at handing RTX than any other Nvidia GPU
- Should be capable of just about anything, even some pro work
- Insanely expensive
- Poor value
It’s $2500 (over twice the price of an RTX 2080 Ti) for a roughly 20% performance increase.
That about sums it up.
Breaking Down Graphics Card Tech and Terminology
In this section, we’re going to break down graphics card tech and terminology that you may not be familiar with. Whether you’re a semi-savvy gamer or someone completely new to this whole thing, we hope this section helps demystify certain aspects of PC graphics technology and helps make buying a GPU easier.
GPU stands for “Graphics Processing Unit”, similar to how CPU stands for “Central Processing Unit”. GPU is usually used interchangeably with the term “graphics card” (describing the physical expansion card you buy for your PC), but there are a few cards out there that have multiple GPU chips inside them. These have become increasingly rare in recent years, however, due to lacking developer support and sometimes poor performance scaling.
VRAM refers to “Video RAM”, and RAM is Random Access Memory. Boiled down to the essentials, VRAM is the memory that your GPU uses for processing. VRAM usually matters most when playing in VR, at higher resolutions, or with large, HD textures. The more detailed the texture or overall picture, the more VRAM you’re probably going to need.
Here is a VRAM scale, for your convenience:
- 2 GB VRAM – Good for 1080p gaming at medium to high settings. Most common with GPUs from a few generations back, or modern budget GPUs. Definitely sufficient for 720p.
- 4 GB VRAM – Commonly seen in modern 1080p-focused GPUs. This will be able to more adequately handle high-resolution textures and resolutions like 1440p, but isn’t exactly optimal. VR starts becoming viable here.
- 6 GB VRAM – A sweet spot for 1080p and 1440p performance at high-to-max graphical settings. From here onward, VR has its VRAM requirements met.
- 8 GB VRAM – Great for 1440p, VR, and 4K. Most games won’t use all of this, but the most cutting-edge titles will.
- 12 GB and higher VRAM – Rare, but starting to penetrate the market. Practical gaming usage is questionable at best, but this will definitely be able to handle 4K and higher-res VR headsets.
GDDR5 and GDDR6
GDDR5 is the most common form of VRAM you’ll see in the GPU market today, and the minimum recommended spec. A few cards may use desktop RAM (like DDR3 or 4), but these will be far too slow for VRAM purposes and come with a severe performance penalty- never buy one of these.
In addition to GDDR5, there is also GDDR5X and GDDR6. We’ll list these standards below, in order of speed:
- GDDR5 – Standard graphics memory.
- GDDR5X – Slightly faster graphics memory.
- GDDR6 – Faster graphics memory, made to compete with HBM2.
HBM2, or High Bandwidth Memory, is a form of VRAM used by high-end GPUs from AMD and Nvidia. (AMD has also taken to using HBM2 in their high-end consumer chips, like Vega 56 and 64.) As the name implies, HBM2 has more bandwidth than its GDDR equivalents, and for this reason, is often considered an “improved” form of VRAM compared to other implementations.
And it is. For this reason, less HBM2 is needed to accomplish the same tasks as GDDR, but the same general rule of more = better still applies here.
Clock speed refers to the speed of the GPU core, measured in MHz. (Memory clock is the same, but refers instead to the VRAM inside the GPU.) Like with CPUs, this alone doesn’t necessarily indicate performance, especially if used to measure between disparate architectures. Only when used between different variations of the same GPU, or different GPUs sharing an architecture, does this become a performance metric.
That being said, the general rule of higher = better does apply here. If you’re able to push your GPU clock speed higher than it originally came, you will enjoy higher performance to match. We’ll discuss this in more detail in the “Factory Overclock” section below.
RT Cores, Stream Processors, Cuda Cores, etc
If you’re a consumer, this stuff is honestly kind of useless to you. The number of cores in a GPU does matter, but not as a consumer-facing performance metric. It’s even more difficult to quantify and compare than clock speed is.
If you want to understand how a GPU performs, stick to actual benchmarks and reviews. All this helps with is comparing performance between GPUs sharing the same architecture- otherwise, it is completely useless.
The reference design of a GPU refers to the standard design created by the original manufacturer. In this case, AMD or Nvidia. Reference designs create a basic version of a graphics card that other manufacturers, like MSI or EVGA, modify or build upon to create their own versions of a GPU.
In recent years, AMD has stopped selling reference designs. Nvidia has rebranded their reference designs as “Founders Edition” cards, which sometimes come at a price premium compared to third-party cards. Reference design cards are generally just fine, but may not have as much overclocking headroom or cooling performance as non-reference cards.
Additionally, many non-reference cards will ship with something called a factory overclock. We’ll explain that concept below.
Traditionally, overclocking refers to the practice of an end user overclocking their components past manufacturer’s recommended specs to get more performance. A factory overclock, on the other hand, refers to overclocking done in-factory, by the manufacturers themselves, to grant extra performance at no penalty to the card or end user.
The most common benefit of buying a non-reference GPU is a factory overclock. Cards with beefier cooling solutions tend to boast higher factory overclocks than competitors as well, due to increased thermal headroom. Choosing an individual GPU based on factory overclocks, especially as a user who doesn’t intend to do it on your own, is a great way to get more bang for your buck.
G-Sync and FreeSync
G-Sync and FreeSync are technologies intended to replace VSync.
VSync, for the uninitiated, refers to “Vertical Sync”- a rendering technique that prevents screen tearing when playing games. Screen tearing occurs when framerate (the number of frames rendered onscreen per second) doesn’t match with refresh rate (the number of times the screen refreshes per second).
VSync, however, comes at a penalty to overall performance and responsiveness, causing more intense FPS dips and input lag where performance falters.
G-Sync and FreeSync solve this problem by directly synchronizing the display to the GPU in question. The catch is that the display in question has to actually support the technology. G-Sync, in particular, used to be harder to achieve, since it required a built-in module on the monitor. As of January 2019, this is no longer the case.
With a -Sync technology in play, the resulting gameplay will feel much more smooth and responsive than it would otherwise. All modern AMD and Nvidia GPUs support the tech- you’ll just need to find a compatible gaming monitor to use it.
Nvidia Share and AMD ReLive
Nvidia Share and AMD ReLive are applications used to record and sometimes stream your gameplay. These come with the GPU’s driver suite and are typically the best way to record gameplay on PC, outside of a dedicated external capture card. In recent years, the headroom of these applications has lowered significantly, resulting in little performance penalty to speak of.
Another feature popularized by these apps, especially Share, is something called a “shadow recording”. When Share debuted, it was called Shadowplay, and its best feature was the ability to save a recording of the past few minutes of your gameplay, ensuring that you only captured the highlights of your games. This feature has carried on to Nvidia Share, and was later picked up by AMD ReLive as well.
The Importance of GPU Benchmarks (and resources)
In this section, we’ll explain everything you need to know about GPU benchmarks. This is very important to understand before buying a GPU, so don’t skip this one if you’re new!
What is a benchmark?
Outside of PC gaming, a benchmark is understood as a standard by which other things are measured against. Inside PC gaming, this is basically the same. But, there is no one universal benchmark used by all GPUs and GPU reviewers.
Instead, benchmarks come in two varieties:
- Synthetic – Synthetic benchmarks use applications like 3DMark to measure synthetic performance of a GPU. In theory, this will correspond to real-world performance, but this isn’t always the case. For more precise measures of performance, you want to use real-world benchmarks.
- Real-world – Real-world benchmarks use actual games and applications to measure performance. These are the most useful benchmarks, especially if you can find one that corresponds to the game you’re playing and the GPU you’re buying.
When buying a GPU, make sure to take a look at real-world benchmarks corresponding to the games that you’re playing.
Framerate (FPS) and frame-time
Framerate is a measure of performance, specifically individually-rendered frames (pictures) of a game rendered per second. Frame-time is a similar but separate metric, used to measure the latency between individual frames (in milliseconds). There is much debate as to which should be used, but both can be helpful metrics.
With framerate, however, you’ll also want to look at lows. 1% and .01% lows will show you the worst-case performance scenarios, which are important to take into consideration (especially if they dip well below your monitor’s refresh rate or 60 FPS).
Where you can find benchmarks
For real-world benchmarks, your best bet is individual GPU reviews and Google Fu. Running a search like “GTX 760 PUBG” for instance, will show you many YouTube videos of that game being run with that graphics card, which is excellent when you’re looking for specific real-world benchmarking scenarios.
Frequently Asked Questions (FAQ)
What is a bottleneck?
A bottleneck is when one component of your system is restricted by another component in your system, sometimes multiple. Most common in a gaming PC build is a GPU being bottlenecked by a weak CPU, but the opposite is pretty common, too. It’s always important to buy a balanced PC build for this reason, since you want all of your components to perform at their best.
Is Nvidia better than AMD, or vice versa?
Short answer, No.
Long answer, depends on what you want and how much you’re spending.
A common perception is that one manufacturer over the other has better drivers, software, and etc. This generally isn’t the case, at least not across-the-board. Some games, especially those explicitly developed around Nvidia or AMD tech, may perform a bit worse on the opposite company’s hardware. There are a few particularly egregious examples of this, but in the past five years, this has thankfully become much less common.
Where one company does become better than the other is raw performance in a given price range. At the time of writing, AMD is doing an excellent job dominating the low-end and mid-range GPU market, while Nvidia is competing much better in the high-end GPU range. If you want the absolute best GPU available and don’t care how much money you spend on it, Nvidia is typically the way to go.
Is the graphics card the most important component in a gaming PC?
With some games, however, this won’t always be the case. Some titles may favor CPU power over GPU power, especially titles like CS:GO and Team Fortress 2, which are based on the Source engine. Be sure to keep in mind what components your games favor when budgeting for your gaming PC.
Even if the GPU is the most important, though, make sure you don’t skimp out on your CPU. If your CPU is too weak, it’ll bottleneck your GPU’s performance.
Should I use SLI, NVLink, or CrossFire?
SLI, NVLink, and CrossFire are all technologies used in multi-GPU setups.
On paper, this sounds like an awesome idea. However, any performance gained from a setup like this won’t be additive. That is, slapping two GTX 760s together won’t grant you twice the performance. It’s usually much less than that…and if the game doesn’t support it at all, you’ll just end up getting single-GPU performance instead. In those cases, that means you’ve pretty much wasted your money.
Simply put…no. At least, not starting out. Slapping in a second GPU as a future upgrade can be a decent way to improve performance, especially if you snag the second one used and for a low price. If you’re buying all your parts today,aough, just buy a stronger GPU from the start and save yourself the headache.
Additionally, SLI and CrossFire are essentially the same thing, just differing based on the manufacturer. NVLink is Nvidia-exclusive and actually a good bit better than SLI, though.
Should I overclock my GPU? Is it safe?
Overclocking your GPU will enhance your gaming performance, but usually not by massive margins. We’re talking 5 or 10 FPS at best in most cases. In exchange for this overclocking, however, you’ll also deal with more system instability, especially if you haven’t properly stress-tested your overclocks.
In terms of safety, your GPU should be fine with an overclock. If it can’t handle the overclock you’ve given it, the app will crash or your PC will shut down. At that point, you can just turn down the OC and try again, or stop entirely. To get a stable overclock, you’ll need to be patient and ready to run multiple failed stress tests before you find your best one.
If you aren’t ready for all the research and work that overclocking a GPU entails, though it’s best to stay away.