The PNY GeForce RTX 4070 XLR8 OC is a fantastic improvement on an already fantastic product, I just wish it was a little cheaper.
Starting at $679 (about £560 / AU$1,020), the RTX 4070 XLR8 OC is a third-party version of our current top pick for best graphics card of 2023, so it's not surprising that the RTX 4070 XLR8 brings fantastic 1440p gaming performance to the table. It does manage to do so with a lot more flair than the Founders Edition card from Nvidia, however, and this is coming from someone who is generally anti-RGB.
Where the PNY card succeeds in its design is that, while bold, it's fairly subdued but allows for just enough customization to let you blend this card in fairly cleanly into just about any build.
(Image credit: Future / John Loeffler)
Assuming the card fits. Lightweight though it is at just 1.87 lbs (about 0.84kg), it is a good bit longer than the RTX 4070 Founders Edition, 305.1mm to the Founders Edition's 242mm. So if you've got a tight case, make sure you have the clearance for this one before you try to cram it into place.
As a dual-slot card, it's going to take up some room, but it feels like it takes up less space than the RTX 4070 Founders Edition despite being a couple of millimeters thicker. A huge part of that is the decision by PNY to go with the 8-pin power connector rather than Nvidia's 12VHPWR cable, the latter of which requires an cumbersome adapter if you don't have an ATX 3.0 power supply with the new power cable leads.
(Image credit: Future / John Loeffler)
I really can't emphasize how much easier working with a single 8-pin cable is versus having to try and finagle two 8-pin cables into an adapter cable and then into a graphics card.
While not the absolute worst — which would be the Nvidia GeForce RTX 4090 and it's 4-to-1 8-pin adapter cable — even two 8-pins plugged into an adapter can wreak havoc on your cable management, especially if you've got a fairly tight fit in your PC case. The 8-pin power connector alone makes this card worth buying over the Founders Edition card.
Yes, I would pay an $80 premium just to not have to use a 12VPWR cable.
(Image credit: Future / John Loeffler)
Fortunately, you do get some additional features to make the RTX 4070 XLR8 even more appealing than just it's power input. Though the card comes with the same base and boost clocks as the Founders Edition (unlike the PNY GeForce RTX 4080 XLR8 OC, which shipped with higher boost clock speeds), the PNY RTX 4070 XLR8 OC starts at stock speeds but PNY's VelocityX software allows you to overclock the card to a degree.
Just be sure that the card you're buying is the overclockable model, as the non-OC and OC cards look identical other than their price, and price can be a moving target sometimes.
(Image credit: Future / John Loeffler)
You also get a triple-fan cooling solution for the RTX 4070 XLR8 OC, which is 50% more fans than the Founders Edition, though I didn't notice too much of a difference in terms of temperature. But it's there, with the PNY RTX 4070 XLR8 staying about 4-5°C cooler under load, but with the RTX 4070's max temperature only hitting about 68°C, neither card was ever really at risk for throttling.
In terms of performance, at its stock clock speeds, the PNY GeForce RTX 4070 XLR8 OC has its performance land within 1% to 2% on either side of the Founders Edition's performance, which is what I expected. If you want to do a deeper dive into this card's performance, definitely check out my Nvidia GeForce RTX 4070 review for all my relevant testing data.
(Image credit: Future / John Loeffler)
Overclocking the card gets you somewhat better performance, but given that it's strictly an 8-pin card (which means you can only really push 200W safely), I wouldn't push the overclocking too hard for too long unless you want to burn your card out much faster.
All told, the premium price is a tougher pill to swallow than it should be, but given that one of the RTX 4070's better selling points was its price, it's hard to not wrinkle your nose a bit.
Still, if the RTX 4070 is what you're after and you want some extra bells, whistles, and a touch of RGB, you can't go wrong with the PNY GeForce RTX 4070 XLR8 OC.
PNY GeForce RTX 4070 XLR8: Price & availability
(Image credit: Future / John Loeffler)
How much does it cost? $679 (about £560 / AU$1,020)
When is it available? Available now
Where can you get it? Available in the US, UK, and Australia
The PNY GeForce RTX 4070 XLR8 OC is available now in the US, UK, and Australia with an MSRP of $679 (about £560 / AU$1,020). This price puts it a good bit higher than the RTX 4070 Founders Edition, but it's not entirely out of step with third-party graphics card pricing.
And while the price isn't great, this is more on Nvidia's head than PNY's since Nvidia is the one who significantly raised the price on the RTX 4070 over the RTX 3070. At the moment, AMD hasn't released the RX 7700 XT or RX 7800, which would be the natural competitors for this card, so it's uncertain how this card will stack up against AMD's lower-premium class of GPUs until those are released in the next few months, presumeably.
PNY GeForce RTX 4070 XLR8: Specs
(Image credit: Future / John Loeffler)
Should you buy the PNY GeForce RTX 4070 XLR8?
(Image credit: Future / John Loeffler)
Buy it if...
You want the RTX 4070 with some extras
The performance of the RTX 4070 XLR8 OC is on par or slightly better than the reference card.
You want to have customizeable RGB
The Nvidia RTX 4070 Founders Edition is pretty staid in terms of appearance, so the XLR8 version definitely lets you liven things up a bit.
Don't buy it if...
You want a cheap graphics card
The RTX 4070 might be one of the best graphics cards on the market, but it definitely isn't one of the best cheap graphics cards.
You have a very tight case
The PNY RTX 4070 XLR8 OC is a pretty long graphics card, so make sure you have clearance for it.
PNY GeForce RTX 4070 XLR8 OC: Also consider
How I tested the PNY GeForce RTX 4070 XLR8 OC
I spend about three weeks with the PNY GeForce RTX 4070 XLR8 OC
I used it to play games, produce and edit creative content, and more
I used our standard battery of benchamarking tools to test it
I used the PNY GeForce RTX 4070 XLR8 OC as the graphics card in my main PC at home for about three weeks.
During that time, I played PC games like Cyberpunk 2077, Star Wars Jedi: Survivor, and more. I also used Adobe Creative Suite extensively to edit photos and video.
I also ran my standard battery of benchmark tests to assess how much the PNY GeForce RTX 4070 XLR8 OC deviated from the baseline RTX 4070, if at all.
We pride ourselves on our independence and our rigorous review-testing process, offering up long-term attention to the products we review and making sure our reviews are updated and maintained - regardless of when a device was released, if you can still buy it, it's on our radar.
The Nvidia GeForce RTX 4060 Ti is without a doubt one of the most anticipated graphics card launches of this generation, and now that it's here, it should be an easy win for Nvidia over archrival AMD. I wish that were the case.
That's not to say that the RTX 4060 Ti doesn't hold up well against AMD's midrange offerings at the moment, it absolutely does, and there's no question that the features this card brings to the table are impressive, especially DLSS 3, which is the first time a properly midrange GPU (under $500/£500/AU$750) is seeing this feature.
It goes without saying that Nvidia is leaning into DLSS 3 as its biggest selling point, and as I'll get into later, it definitely delivers significantly better performance than the RTX 4060 Ti should be capable of given its various specs — even factoring in the expanded cache which widens up the memory bandwidth of the card despite still having just 8GB GDDR6 VRAM to work with.
The decision to go with 8GB VRAM for this card — a 16GB VRAM variant is going to be released in July 2023 for an MSRP of $499 (about £400/AU$750) — is probably the only thing that kept the price on this card under $400. With an MSRP of $399 (about £320/AU$600), the Nvidia Founders Edition RTX 4060 Ti 8GB is the same price as the Nvidia GeForce RTX 3060 Ti it is replacing, and generally, it offers a pretty good value for that money, with some caveats.
In terms of native, non-DLSS performance, there isn't a whole lot of improvement over the previous generation, which is definitely going to disappoint some, if not many. Given the kinds of performance advances we've seen with higher-end cards, we were hoping to see that extend down into the heart of the midrange, but it seems those benefits generally stop at the Nvidia GeForce RTX 4070.
Instead, you have a card that relies very heavily on DLSS for carry its performance over the line, and where it works, it is generally phenomenal, offering a real, playable 1440p gaming experience, and even brushing up against some decent 4K performance with the right settings.
This is something AMD has really struggled to match with its FSR, and so Nvidia really has a chance to score a major blow against AMD, but as we'll see, the best AMD graphics card in the last generation's midrange, the AMD Radeon RX 6750 XT, actually outperforms the RTX 4060 Ti in non-ray tracing workloads, including gaming, so this does not bode well for Nvidia once AMD releases its current-gen midrange cards.
This is somewhat exacerbated by the fact that the RTX 4060 Ti's ability to use its new features is fairly limited, and while features like DLSS 3 with Frame Generation are available on the best PC games like Cyberpunk 2077 and Returnal, as of the launch of the RTX 4060 Ti, there are only 50-ish games that support DLSS 3.
This list will surely grow over time, but you certainly won't get this kind of support on games that may be just recent enough to push the RTX 4060 Ti in terms of its performance, while being just old enough that you'll never see a DLSS 3 patch for it.
I can say that if you're coming from an RTX 2060 Super or older, then this card is absolutely going to blow your mind. It's effectively the RTX 3060 Ti's NG+, so if you missed what I consider to be the best graphics card of the last generation, you'll get all that and more with the RTX 4060 Ti. If you're coming from an Nvidia Ampere card though — especially from greater than the RTX 3060 Ti — chances are you are going to find this is really a downgrade with some neat features to soften the blow.
Nvidia GeForce RTX 4060 Ti: Price & availability
(Image credit: Future / John Loeffler)
How much is it? MSRP listed at $399 (about £240, AU$600)
When is it out? It is available starting May 24, 2023
Where can you get it? You can buy it in the US, UK, and Australia
The Nvidia GeForce RTX 4060 Ti 8GB is available starting May 24, 2023, with an MSRP of $399 (about £240, AU$600). This is the same launch price of the Nvidia RTX 3060 Ti that this card is replacing, so we're glad to see that Nvidia didn't increase the price on this variant with this generation.
This also puts it on par with the AMD Radeon RX 6750 XT, which initially launched for $549 (about £440/AU$825), but which you can find under $400 right now, even without discounts, at major retailers. AMD hasn't released an RX 7700 XT yet, which would be this card's more natural competition, so comparing the RX 6750 XT and the RTX 4060 Ti isn't really fair, but it's all we have for now until AMD launches its RTX 4060 Ti challenger.
Nvidia GeForce RTX 4060 Ti: Features and chipset
(Image credit: Future / John Loeffler)
Only uses one 8-pin...but still requires a 16-pin converter?
3rd-generation ray tracing and 4th-generation tensor cores
288.0 GB/s memory bandwidth, but 32MB L2 cache boosts effective bandwidth to 554.0 GB/s (according to Nvidia)
The Nvidia RTX 4060 Ti is the first properly midrange Nvidia Lovelace graphics card, and so it is built on TSMC's 5nm process, with about 22.9 billion transistors across 34 streaming multiprocessors (SM), which come with 128 shader cores (CUDA), 4 fourth-generation tensor cores, and 1 third-generation ray tracing core per SM.
The clock speed is a solid 2,310MHz base clock, which is about a 64% improvement over the RTX 3060 Ti's 1,410MHz, with a boost clock of 2,535MHz, or about 52% faster than the RTX 3060 Ti's 1,665MHz.
The biggest difference between the two cards is the memory bus. The Nvidia RTX 4060 Ti uses a 128-bit bus for 8GB GDDR6 VRAM, while the RTX 3060 Ti uses a 256-bit bus for the same amount of VRAM. The RTX 4060 Ti has a faster memory clock (2,250MHz), which combined with the expanded L2 cache (32MB), the RTX 4060 Ti has a slightly faster effective memory speed of 18 Gbps to the RTX 3060 Ti's 15 Gbps, while having a much faster effective memory bandwidth.
Still, this really does smack of over-engineering. The move to a 128-bit bus doesn't seem necessary, and given what we've seen of other Lovelace cards, like the Nvidia GeForce RTX 4070, I definitely think Nvidia could have stuck with a higher bus width and it wouldn't be catching nearly the grief it is getting over this.
What's more, even though the performance of the RTX 4060 Ti is better than the RTX 3060 Ti, I really think that had this card had the same bus width as the RTX 3060 Ti, this card would absolutely anything that approached it in the midrange. As it stands, the RTX 4060 Ti is great, but fails to score the knockout it really needed.
It is worth mentioning though that this card also uses significantly less power than the RTX 3060 Ti. That card had a TGP of 200W, while the RTX 4060 Ti 8GB comes in at 160W, which is a 20% improvement in efficiency. This is great for keeping your build under 600W, and it's a move in the right direction for everyone and deserves to be praised.
Nvidia GeForce RTX 4060 Ti: Design
(Image credit: Future / John Loeffler)
The Nvidia RTX 4060 Ti Founders Edition keeps the same black heatsink with chrome trim as the other Founders Edition cards this generation, and — unfortunately — it also sticks with the 12VHPWR 16-pin power connector. Fortunately, you only need to plug a single 8-pin into it, so it is at least somewhat easier to manage in a case.
Also easier to manage is the size of the card. Using the same dual-fan design as previous Founders Edition cards, the RTX 4060 Ti pretty much shrinks these down a bit. While it's still a dual-slot card, it comes in at just under 9.5 inches long and 4.5 inches tall, making it the kind of card that will easily fit in your case.
There's not much flash here, but that's a given with the Founders Edition, so if you're looking for visual bells-and-whistles like RGB or super-cooling features like a triple fan design, you're going to want to look at any of the third-party cards that release alongside this one for those.
Nvidia GeForce RTX 4060 Ti: Performance
(Image credit: Future / John Loeffler)
DLSS 3 is the real draw here
Improved ray tracing performance
Baseline performance not much improved over the RTX 3060 Ti
When it comes to performance, the Nvidia GeForce RTX 4060 Ti really leans on DLSS 3 for most of its major performance gains, and while this can be substantial, some are going to feel somewhat disappointed.
Test system specs
This is the system we used to test the Nvidia GeForce RTX 4060 Ti:
This is largely because even with the introduction of some pretty advanced tech, there aren't a lot of games out right now that can really leverage the best features of this card. While three of the games I use as benchmarks — F1 2022, Cyberpunk 2077, and Returnal — all feature frame generation, these are also three of the latest games out there from major studios that have the time and staffing to implement DLSS 3 with Frame Generation in their games.
There is a DLSS 3 plug-in coming for Unreal Engine, which should definitely expand the number of games that feature the tech, that's still going to be a ways off at this point before that starts to trickle down to the gamers who will actually be using this card.
I'll get more into DLSS 3 and Frame Generation in a bit, but a quick glance over the underlying architecture for the RTX 4060 Ti tells something of a story here, as shown by synthetic benchmarks using 3DMark and Passmark 3D.
Synthetic Benchmarks
As you can see, the RTX 4060 Ti beats out the RTX 3060 Ti, but only just barely, getting about 11% better performance than the card it's replacing. This is, okay, I guess, but hardly the generational leaps that previous Lovelace cards have been making.
For example, the RTX 4070 offers a roughly 21% jump over the RTX 3070 on these same synthetic benchmarks. In fact, this puts the RTX 4060 Ti just ahead of the RX 6750 XT, and ultimately just behind the RTX 3070 in terms of raw performance.
As a gaming card, the performance outlook is better, but not by a whole lot overall.
Gaming Benchmarks
On games with heavy effects-based visuals like Metro: Exodus and Cyberpunk 2077 where the advanced architecture of the RTX 4060 Ti can be leveraged, it does edge out the competition, sometimes. The RX 6750 XT still manages a slightly better fps on Returnal at 1080p, on average, when not using ray tracing or upscaling tech, for example.
The RTX 4060 Ti also gets crushed in CS:GO at 1080p, relatively speaking, which I chalk up entirely to pushing textures through the smaller memory bus of the RTX 4060 Ti. The 192-bit bus on the RX 6750 XT's 12GB GDDR6 VRAM and the 256-bit bus on the RTX 3060 Ti's 8GB GDDR6 really show up in cases like this.
Things do start to turn in the RTX 4060 Ti's favor once you start fiddling with ray tracing. The third-generation ray tracing cores on the RTX 4060 Ti are definitely much more capable than the RTX 3060 Ti's and especially more than the RX 6750 XT's, which are second-generation and first-generation cores, respectively.
The RTX 4060 Ti is the first midrange card I've tested that is going to give you playable native ray-traced gaming at 1080p consistently on max settings, though it will struggle to get to 60 fps on more demanding titles like Cyberpunk 2077.
But lets be honest, nobody is playing any of these games with native resolution ray tracing, you're going to be using an upscaler (and if you aren't then you really need to start).
Here, the performance of Nvidia's DLSS really shines over AMD's FSR, even without frame generation. In both Cyberpunk 2077 and Returnal, the RTX 4060 Ti can get you over 120 fps on average when using the DLSS Ultra Performance preset, and if you want things to look their best, you can easily get well north of 60 fps on average with every setting maxed out, even ray tracing.
Now, one of the things that the wider memory bus on the RTX 3060 Ti gave that card was a faster throughput when gaming at 1440p. Now, not every game is going to run great at 1440p, but for a lot of them, you're going to be able to get a very playable frame rate.
The RTX 4060 Ti improves over the RTX 3060 Ti here, but not nearly as much as it should, and on games like F1 2022 and CS:GO where that memory bandwidth difference is going to show, well, it shows up here at 1440p, too.
Of course, once you turn on ray tracing, most games are going to turn into a slide show, but unsurprisingly, the RTX 4060 Ti manages to score a win here on every ray-traced game I tested.
That said, you are really pushing it here on these settings, and you're better off using upscaling if you're going to go for 1440p, especially with settings turned up.
The biggest win for the RTX 4060 Ti here is with Cyberpunk 2077, where it manages 67% better performance at max quality settings than the RX 6750 XT, but maddeningly, it's only about 13% better than the RTX 3060 Ti on the quality preset. On ultra performance, the RTX 4060 Ti is about 52% better than the RX 6750 XT, but again, only 13% better than the RTX 3060 Ti.
When it comes to Returnal, the RX 6750 XT is essentially tied with the RTX 4060 Ti on the quality preset for FSR 2.1 and DLSS, respectively. Bump this up to ultra performance, and the RTX 4060 Ti does better, beating out the RX 6750 XT by about 22% and the RTX 3060 Ti by about 17%.
I imagine the RTX 4060 Ti will perform more or less the same across most games that still rely on DLSS 2.0, which number more than 200. For those games that really leverage DLSS 3 with Frame Generation though, it really is another story entirely.
With Frame Generation, you can get about a 40-60% performance improvement on games that support it. This isn't nothing, since this can even get you playing Cyberpunk 2077 at a playable framerate at 4K on ultra performance. The RTX 3060 Ti and RX 6750 XT really don't have any answer to this, and so they are going to lag behind considerably on any games that have DLSS 3 with Frame Generation.
Does Frame Generation increase latency on some titles, along with other issues? Sure. Will it matter to gamers who get to play Cyberpunk 2077, Returnal, and other titles that play like they were RTX 3080 Ti's? Probably not.
Will any of this matter to anyone who doesn't play those games? Obviously not. And that is ultimately the issue with this card. For what it does well, it has no peer at this price, but if you already have an RTX 3060 Ti, then there is really very little reason to upgrade to this card. Hell, if you have an RX 6750 XT, you might feel like you're better off just waiting to see what AMD has in store for the RX 7700 XT, and I would not blame you in the slightest.
This isn't a whiff by Team Green by any means, but there's no getting around the fact that the performance of the Nvidia GeForce RTX 4060 Ti absolutely leaves a massive opening for AMD to exploit in the coming months with the RX 7700 XT, or even the RX 7650 XT.
Should you buy the Nvidia GeForce RTX 4060 Ti?
(Image credit: Future / John Loeffler)
Buy it if...
Don’t buy it if…
Also consider
How I tested the Nvidia GeForce RTX 4060 Ti
I spend several days with the RTX 4060 Ti running benchmarks, playing games, and generally measuring its performance against competing cards.
I paid special attention to its DLSS 3 Frame Generation technology, since this is one of the card's biggest selling points, and played several games at length with the tech turned on.
Having covered and tested many graphics cards in my career, I know how a graphics card perform at this level.
The PNY GeForce RTX 4080 XLR8 has a lot of things going for it, but one of those is not that it's something other than an RTX 4080.
In my Nvidia GeForce RTX 4080 review, I found that the Founders Edition card had excellent performance, and the same is true with PNY's XLR8 OC version. In fact, its performance is slightly better, on average, than the base reference model thanks to this card being overclocked and having some third-party extras that the Nvidia reference card lacks.
While the underlying GPU architecture and memory between the two are the same — so you get the same number of SMs, RT cores, Tensor cores, and the like, as well as 16GB GDDR6X VRAM — you do get faster clock speeds with the XLR8 OC card. The base clock speed is the same, but the boost clock is about 2% faster, while the memory clock speed is about 3% faster.
(Image credit: Future / John Loeffler)
Toss in some RGB lighting (if that's your thing) and a triple-fan cooling solution with a larger heat sink, and you have a card that will throttle less, have customization options that many PC gamers will love, and everywhere I've seen it, you can get it for less than Nvidia's frankly outrageous $1,199 (about £960, AU$1,740) MSRP.
I've seen the PNY RTX 4080 XLR8 OC retailing for $1,189 (about £950 / AU$1,725) on PNY's website, and found it selling for even less on most retail sites I checked.
This isn't the biggest savings, but you do get some extras here that the reference card lacks and sales make this card a lot more accessible than the RTX 4080 was at launch. If you are going to pick up the PNY RTX 4080 XLR8 OC though, make sure you are actually buying the OC version of the card since PNY also makes a non-OC XLR8 version for slightly less, and they are otherwise identical and easily confused online.
On the design front, the RTX 4080 XLR8 OC looks much more like a gamers' RTX 4080 than Nvidia's reference design, so if you're looking to outfit your gaming PC with a new GPU and you want it to match the look of your other addressable RGB components, this will do that much better than Nvidia's reference card will.
The RTX 4080 XLR8 OC sticks with the same 3-to-1 8-pin-to-12VHPWR cable adapter that its other cards have gone with since Nvidia Ampere launched in 2020, so if you don't have an ATX 3.0 PSU with a dedicated 12VHPWR cable, cable management with this card will be challenging, though less so than with the laughably tentacled RTX 4090.
Cable management will also be a concern here as the RTX 4080 XLR8 OC is larger than the RTX 4080 reference card, which is already a monster GPU. At 13-inches long, this card is just over 7% longer than the reference card, and about 17% thicker, taking up 3.5 slots. It has the same display output as the base RTX 4080, so three DisplayPort 1.4a ports and an HDMI 2.1 port, with no USB Type-C output.
(Image credit: Future / John Loeffler)
If you want to know more about the kind of performance the RTX 4080 is capable of, I would refer you back to the original RTX 4080 review, since the effective difference isn't enough to thoroughly rehash here.
On that note, given the effectively identical performance of the base Nvidia RTX 4080 and the PNY RTX 4080 XLR8 OC, the PNY RTX 4080 XLR8 OC suffers from the same problem as the base reference card, which is that you can get the best AMD graphics card on the market for cheaper.
In my AMD Radeon RX 7900 XTX review, I found that the 7900 XTX effectively outperformed the RTX 4080 (with the exception of some ray-tracing workloads and overall creative performance), and did so while costing roughly 16% less than Nvidia's MSRP, making it the best graphics card for gamers overall.
(Image credit: Future / John Loeffler)
You can get the PNY RTX 4080 XLR8 OC for less than Nvidia's MSRP, but it's still going to be much more expensive than AMD's flagship GPU, so unless you need to do rendering work in Lumion or Blender, the RX 7900 XTX is going to be a much better bet for just about everyone, and those who must have an Nvidia RTX card, the RTX 4070 Ti offers a far better value and creatives should be going to the RTX 4090 anyway.
All things considered, the problems with the PNY RTX 4080 XLR8 OC are hard to get around, but only because it's an RTX 4080, and there isn't much PNY can do about that. That said, if you absolutely cannot spend the money to get the RTX 4090 and you need an RTX card, rather than the RX 7900 XTX, then the PNY RTX 4080 XLR8 OC at least offers you a better proposition than Nvidia's base model, and it does so while mitigating the price somewhat.
So, if it's gotta be an RTX 4080 for whatever reason, do yourself a favor and make it the PNY GeForce RTX 4080 XLR8 OC. At least that way you'll get something extra thrown in to help choke down the price you're going to have to pay.
PNY GeForce RTX 4080 XLR8: Specs
While the base specs of the RTX 4080 Founders Edition and the PNY RTX 4080 XLR8 OC are essentially the same, there are a few differences worth noting beyond the differences in aesthetic design.
PNY GeForce RTX 4080 XLR8: Price & availability
(Image credit: Future / John Loeffler)
How much does it cost? $1,189 (about £950 / AU$1,725)
When is it available? Available now
Where can you get it? Available in the US, UK, and Australia
The PNY GeForce RTX 4080 XLR8 is available now for $1,189 (about £950 / AU$1,725), which is slightly less than Nvidia's official MSRP, but not by a whole lot. There are still plenty of sales you can find on this card to help defray the cost even further, but it's still selling for way too high when the AMD Radeon RX 7900 XTX offers better performance when it comes to gaming and costs about 16% less.
The RTX 4090 is much more expensive, but considering how much you are splurging on a card like this, there is no reason to not just spend the money to get an RTX 4090, which is much more expensive but is the best graphics card on the market by a mile, especially for creative workloads like architectural design and 3D modeling.
Should you buy the PNY GeForce RTX 4080 XLR8?
(Image credit: Future / John Loeffler)
Buy it if...
You can't afford the RTX 4090 If the RTX 4090 is out of reach (an understandable concern, for sure), then this is the next best Nvidia graphics card in terms of raw performance.
You want an overclocked card The Nvidia RTX 4080 Founders Edition can't be overclocked, but you have some wiggle room with the PNY RTX 4080 XLR8 OC to boost some performance with some settings tweaks.
Don't buy it if...
You can afford the RTX 4090 I really can't stress enough that the RTX 4090 should be the graphics card to buy if you can afford to get it. It's the best there is, and there's no reason to settle for second best at this price point.
PNY GeForce RTX 4080 XLR8: Also consider
How I tested the PNY GeForce RTX 4080 XLR8
I spent several weeks with the PNY GeForce RTX 4080 XLR8
I used the PNY GeForce RTX 4080 XLR8 for gaming and content creation
I tested it with our standard battery of benchmarks
In addition to running the standard battery of tests on this card, I also used it as my main GPU for work and gaming for several weeks. This included creative workflows like Photoshop and Lumion, as well as gaming on the most demanding titles like Cyberpunk 2077.
I have been reviewing computer hardware for several years now, and have an extensive computer science background to make sense of the data I'm collecting.
We pride ourselves on our independence and our rigorous review-testing process, offering up long-term attention to the products we review and making sure our reviews are updated and maintained - regardless of when a device was released, if you can still buy it, it's on our radar.
The Nvidia GeForce RTX 4070 is here at long last, and for gamers who've been starved for an upgrade, go ahead and pick this one up. It can do just about everything.
It's hard to follow up the RTX 3070, one of the best graphics cards of all time, and in our Nvidia GeForce RTX 3070 review, we praised that card for being an outstanding performer at 1080p and 1440p — which is where the overwhelming number of PC gamers game at — while also being a much more affordable option over the other two launch cards for Nvidia's Ampere lineup. We especially noted how the RTX 3070 offered comparable performance to the RTX 2080 Ti for half the price.
Everything we said about the RTX 3070 applies just as easily to the RTX 4070, only now it doesn't just dabble in 4K; it can competently game at every resolution, making it a graphics card that everybody can fall in love with without spending a fortune.
A lot has changed since the RTX 3070 launched towards the end of 2020, and unfortunately, not everything changed for the better. Things are more expensive pretty much everywhere you look, and the Nvidia RTX 4070 isn't immune. At $599 (about £510 / AU$870), the RTX 4070 is fully 20% more expensive than the RTX 3070 was at launch.
(Image credit: Future / John Loeffler)
I'm not happy about this at all, and you shouldn't be either, but all you have to do is look at the scores the RTX 4070 puts up on the board and you'll be as hard pressed as I am to dock it any points for this. It consistently puts out RTX 3080-level performance more or less across the board and even manages to bloddy the nose of the Nvidia GeForce RTX 3080 Ti, and while the RTX 3080 beats out the RTX 4070 at native 4K, turn on DLSS and the RTX 3080 simply gets blown out.
On the other side of the aisle, the AMD Radeon RX 7900 XT is Team Red's nearest real competition, and it struggles to justify itself in the presence of the RTX 4070. While the RX 7900 XT solidly outperforms the 4070, it's also 50% more expensive, and the benefits of the RX 7900 XT get quickly drowned out by the power of DLSS, especially in titles with DLSS 3.
Moreover, the RTX 4070 makes for a pretty competent creator GPU, offering indie developers and artists who don't have the funding to get themselves an Nvidia GeForce RTX 4090 a handy option for getting some work done within a more limited budget. It's not going to power a major movie studio or anything, but if you're dabbling in 3D modeling or video editing, this card is great compromise between price and performance.
Finally, wrap this all into a package that feels like a downright normal graphics card from ye olden days, back before you needed to include support brackets and balast to keep your gaming PC from tipping over, and you end up with a graphics card that can easily power some of the best gaming PCs that can actually fit into your PC case and your budget.
This graphics card has its issues, which is inevitable, but given what's on offer here, it's easy enough to look past its shortcomings and enjoy some truly outstanding performance at at a reasonable enough price.
Third-party cards retail prices will match or exceed Nvidia's MSRP
The Nvidia GeForce RTX 4070 is available starting April 13, 2023, with an MSRP of $599 (about £510 / AU$870). Third-party partners will have their own versions of the RTX 4070 that will vary in price, but they will always have a matching or higher regular retail price than the Nvidia GeForce RTX 4070 Founders Edition.
Notably, the RTX 4070 is getting a 20% price increase over the card it's replacing, the RTX 3070, which had a launch price of $499 in the US (about £425 / AU$725). While we'd have loved to see the price stay the same gen-over-gen, this should come as no surprise to anyone who has been watching GPU price inflation recently.
Meanwhile, we haven't seen AMD's direct RTX 4070 competitor yet, the AMD Radeon RX 7800 XT, but the AMD Radeon RX 7900 XT is the closest AMD has this generation with an $899 / £799 (around AU$1,350) MSRP, putting it 50% more expensive than the RTX 4070.
This card is also the same price as the Nvidia GeForce RTX 3070 Ti, for what it's worth, and considering that the RTX 4070 punches well above the 3070 Ti's performance, you do at least get a better sense of value out of this card than anything from the last generation.
Price score: 4 / 5
Nvidia GeForce RTX 4070 review: Features & chipset
(Image credit: Future / John Loeffler)
DLSS 3 with full Frame Generation
Third-gen Ray Tracing Cores and fourth-gen Tensor Cores
Lower TGP than RTX 3070
The Nvidia RTX 4070 doesn't change too much on paper over its last-gen predecessor, featuring the same number of streaming multiprocessors, therefore the same number of CUDA cores (5,888), ray-tracing cores (46), and tensor cores (184).
It does bump up its memory to the faster GDDR6X and adds an additional 50% VRAM for a total of 12GB. With a 192-bit bus and a memory clock of 1,313MHz, the RTX 4070 has an effective memory speed of 21 Gbps, equal to that of the Nvidia RTX 4070 Ti, for a memory bandwidth of 504.2 GB/s.
It has a lower base and boost frequency than the 4070 Ti, clocking in at 1,920MHz base and 2,475MHz boost (compared to 2,310MHz base and 2,610MHz boost for the 4070 Ti), but this is a substantial bump up from the 1,500MHz base and 1,725MHz boost frequency of the RTX 3070.
This is owing to the 5nm TSMC process used to fab the AD104 GPU, compared to the Samsung 8nm process for the RTX 3070's GA104. Those faster clocks also power next-gen ray tracing and tensor cores, so even though there are the same number of cores in both the RTX 4070 and the RTX 3070, the RTX 4070's are both much faster and more sophisticated.
Also factor in Nvidia Lovelace's DLSS 3 with Frame Generation capacity, something that Nvidia Ampere and Turing cards don't have access to, and what looks like two very similar cards on paper turns out to be anything but in practice.
Finally, thanks to the 5nm process, Nvidia is able to squeeze more performance out of less power, so the TGP for the RTX 4070 is just 200W, making it a fantastic card for a lower-power, sub-600W build.
Features & chipset: 5 / 5
Nvidia GeForce RTX 4070 review: Design
(Image credit: Future / John Loeffler)
Same size as the RTX 3070
16-pin power connector
Same design as RTX 4090 and RTX 4080
With the RTX 4070 Founders Edition, Nvidia finally delivers a next-gen graphics card that can actually fit in your case without requiring a construction winch to hold it in place.
OK, the previous cards weren't that bad, and even at the reduced form factor and weight, you'll still want to toss a GPU bracket into your case for good measure (there's no harm in protecting your investment, after all).
Image 1 of 6
(Image credit: Future / John Loeffler)
Image 2 of 6
(Image credit: Future / John Loeffler)
Image 3 of 6
(Image credit: Future / John Loeffler)
Image 4 of 6
(Image credit: Future / John Loeffler)
Image 5 of 6
(Image credit: Future / John Loeffler)
Image 6 of 6
(Image credit: Future / John Loeffler)
But holding the RTX 4070 in my hand, this is the first card of this generation that doesn't feel like a piece of machinery. Even the more modestly-sized AMD Radeon RX 7900 XTX and RX 7900 XT feel substantial, while the RTX 4070 feels like an old school GeForce graphics card from a couple years back.
The RTX 4070 Founders Edition keeps the same fan design as the RTX 4090 and RTX 4080 that preceeded it (a fan on the front and back), but it shrinks everything down to a dual-slot card about two-thirds the size of those monsters. The RTX 4070 also features the same outputs as previous RTX Lovelace cards (so no USB-C out), and a 16-pin power connector with an included adapter for two 8-pin leads to power the card.
With a TGP of 200W, Nvidia could theoretically have just gone with a single 8-pin connector, but Team Green seems absolutely committed to the 12VHPWR cable, it seems. I'll never stop complaining about this, but it is what it is. If you have an ATX 3.0 power supply, you won't need to worry about that, but the rest of us will have to deal with additional cable management.
Design score: 4.5 / 5
Nvidia GeForce RTX 4070 review: Performance
(Image credit: Future / John Loeffler)
Phenomenal gaming performance
Can easily push 60 fps in 4K gaming with DLSS
RTX 3080 performance at 60% of the power
Right out the gate, let's just say that the Nvidia RTX 4070 is the best 1440p graphics card on the market right now, and it's likely to remain at the top of that list for a good long while.
Its performance prowess isn't limited to just 1440p, mind you, and when I get into the gaming performance, you'll see that its 4K gaming potential is exciting (with caveats), but for starters, we can dig into its synthetic performance in tests like 3DMark to see how the fundamentals stack up.
General Performance
As you can see, the RTX 4070 outperforms the RTX 3070 by about 21% overall, while underperforming the RTX 3080 by about 1.37%, which is close enough to effectively tie the last-gen 4K powerhouse, and underperforms the RTX 3080 Ti by about 6%. Considering that the RTX 3080 Ti's MSRP is nearly twice that of the RTX 4070, this is an astounding result.
The RTX 4070 does lag behind the RTX 4070 Ti and the RX 7900 XT by quite a bit, averaging about 22% worse performance than the RX 7900 XT and about 13.5% worse performance than the RTX 4070 Ti. These current-gen cards also have substantially better hardware, so this isn't unexpected.
Creative Performance
When it comes to creative performance, well, we have a more limited dataset to work with since Blender Benchmark 3.5.0 decided it only wanted to test half the cards I tried to run it on (including failing to run on the RTX 4070), so we'll have to come back to that one at a later date once the benchmark is updated.
In the meantime, the tests I was able to run really showcased how well the RTX 4070 can handle creative workloads. On Adobe Premiere and Adobe Photoshop, the RTX 4070 performed noticeably better than the RTX 3080 across both apps and fell in very close behind the RTX 4070 Ti for an overall second place finish.
In lieu of Blender's Benchmark, V-Ray 5 is a fairly good stand-in, as well as an excellent benchmark in its own right. Here, the RX 7900 XT wouldn't run, since it doesn't use CUDA or Nvidia's RTX, but we can see the RTX 4070 coming in a respectable runner up to the RTX 4070 Ti.
One of my recent favorite workloads, Lumion 12.5, renders an architectural design into either a short movie clip at 1080p or 4K at 60 fps, making it one of the best benchmarks for creatives to see how a graphics card handles production level workloads rather than synthetic tests.
It requires the same kind of hardware as many of the best PC games in order to light a scene, create realistic water effects, and reproduce foliage on trees, and it's the kind of real-world benchmark that tells more about the card than a simple number devoid of context.
Considering that it can take a five-second, 60 fps movie clip an hour to render at production quality, I switched things up a bit and rather than calculate frames per second, like I do with Handbrake's encoding test, I use frames per hour to give a sense of how long a movie clip you can produce if you leave the clip to render overnight (a common practice).
In the case of the RTX 4070, it rendered a five-second movie clip at 60 fps at draft (1-star) quality 13% faster than the RTX 3080, about 7% faster than the RTX 3080 Ti, and nearly 23% faster than the RX 7900 XT.
It lagged behind the RTX 4070 Ti, though, by about 8%, a deficit that grew wider at 1080p production (4-star) quality, where the RTX 4070 rendered the movie 25% slower than the 4070 Ti and 6.78% slower than the RX 7900 XT.
For Handbrake, the RTX 4070 manages to pull out its first clean win on the creative side, though not by a whole lot. Still, 170 frames per second encoding from 4K to 1080p is not bad at all.
Overall then, the RTX 4070 puts in a solid creative performance, besting the RTX 3080, the RX 7900 XT, the RTX 3070 Ti, and the RTX 3070, while barely losing out to the RTX 3080 Ti.
Gaming Performance
(Image credit: Future / John Loeffler)
As good of a creative card as the RTX 4070 is, in its bones, this is a gamers' graphics card, so gaming performance is definitely where I spent most of my time testing the RTX 4070. I want to note that the included figures here are a representative sample of my testing, and that not all test results are shown.
When it comes to gaming performance, the RTX 4070 offers some of the best you're going to get at this price, though there are some stipulations to bring up right out the gate.
First, broadly speaking, this card can game at 4K on most games not called Cyberpunk 2077 or Metro: Exodus using max settings natively, so long as you keep things within reasonable limits. Or, really, one limit: keep ray tracing turned off.
Overall, the RTX 4070 gets about 58 fps on average at 4K when not ray tracing, with a floor of 45 fps at 4K, which is eminently playable. Turn ray tracing to the max and your get an average fps of 34 with a floor of 25, which is just better than a slideshow.
The RTX 3080 doesn't fare too much better on this metric, managing 40 fps on average with a floor of 29 fps at max settings with ray tacing turned on, while the RTX 3080 Ti averages about 36 fps and a floor of 19 fps. This does put the RTX 4070 just behind the 3080 Ti in terms of average fps and with a higher fps floor than the 3080 Ti.
If you're dead set on ray tracing, the RTX 4070 can certainly deliver, thanks to DLSS, which can bump those numbers back up to 79 fps on average with a floor of 55 fps. Compare that to the RTX 3080's 80 fps average with a 58 fps floor in our tests and the RTX 4070 can definitely go toe to toe with the RTX 3080 when ray tracing on max settings if DLSS is on.
In addition, the RTX 4070 gets about 10% less fps on average than the RTX 3080 Ti at 4K with ray tracing and DLSS on, (79 fps to the 3080 Ti's 88 fps), and a roughly 14% lower fps floor than the RTX 3080 Ti (55 fps to the 3080 Ti's 64 fps).
Overall, the RTX 4070 manages an average 57 fps at 4K, with a floor of 41 fps, across all the settings I tested. This is about 28% lower than the RTX 4070 Ti (79 fps average, overall), about 10% lower than the RTX 3080 (63 fps average, overall), the RX 7900 XT (64 fps average, overall), and the RTX 3080 Ti (64 fps average, overall).
These numbers skew a bit against the RTX 4070, since the RTX 4070 Ti, RX 7900 XT, RTX 3080, and RTX 3080 Ti all handle native 4K gaming much better, but so few people play at native 4K anymore that is a fairly meaningless advantage.
Meanwhile, the RTX 4070 actually beats the RX 7900 XT by about 20% when using DLSS (versus the RX 7900 XT's FSR) at 4K with max settings and ray tracing; 79 fps on average to 66 fps on average, respectively. It also manages to strike a dead heat with the RTX 3080 (80 fps average) and come just 10% short of the RTX 3080 Ti's average RT performance at 4K with ray tracing.
It's important to note as well that these don't factor in DLSS 3 Frame Generation, to make it a fair comparison.
As for the RTX 3070, the RTX 4070 manages about 39% better average 4K performance, with a 53% higher fps floor (57 fps average with a 43 fps floor for the RTX 4070 compared to the RTX 3070's 41 fps average and 28 fps floor).
When it comes to 1440p gaming, the RTX 4070 is on much more solid footing, even if some of the bigger cards definitely perform better in absolute terms. The RTX 4070 underperforms the RTX 3080 by about 8% in non-ray-traced, non-upscaled 1440p gaming, on average (105 fps to the RTX 3080's 115 fps), though they both have a very similar floor around 80-85 fps.
Meanwhile, the RTX 4070 falls about 12% short of the RTX 3080 Ti's 119 average fps at non-ray-traced, non-DLSS 1440p.
Both the RTX 4070 Ti and RX 7900 XT kinda clobber the RTX 4070 with roughly 25-29% better performance at non-ray-traced, non-upscaled 1440p gaming, and this carries over into gaming with ray tracing settings maxed out, though the RTX 4070 is still getting north of 60 fps on average (67 fps, to be precise), with a relatively decent floor of 51 fps.
The real kicker though is when we turn on DLSS, at which point the RTX 4070 beats out everything but the RTX 4070 Ti and RTX 3080 Ti, including the RX 7900 XT, which it outperforms by about 29% on average (125 fps to 97 fps), with a much higher floor of 88 fps to the RX 7900 XT's 60 fps, a nearly 49% advantage.
The RTX 4070 also beats the RTX 3080 here too, with about 5% better performance on average and a 7.5% higher fps floor on average than the RTX 3080. Incredibly, the RTX 4070 is just 3% slower than the RTX 3080 Ti when both are using DLSS at 1440p with max ray tracing.
As for the RTX 3070, the RTX 4070 gets about 35% better performance at 1440p with ray tracing and DLSS 2.0 than the card it replaces (125 fps to 93 fps), with a nearly 53% higher fps floor on average (87 fps to the 3070's 57 fps), meaning that where the RTX 3070 is setting the 1440p standard, the RTX 4070 is blowing well past it into territory the RTX 3070 simply cannot go.
The story is pretty much the same at 1080p, with there being essentially no difference between the RTX 4070, the RTX 3080, the RTX 3080 Ti, and the RX 7900 XT, with the RTX 3070 languishing about 30% behind and the RTX 4070 Ti off on its own out ahead of everyone else.
There has been a lot of talk about the RTX 4070 ahead of its launch as benchmarks have leaked and people have looked at numbers out of context and downplayed the performance of the RTX 4070 based on one or two tests. They've even pointed to the price increase to say that this card is a disappointment.
Granted, I'm not thrilled about the 20% price increase either, but there's no getting around the fact that you're getting a graphics card here with just 200W TGP that's putting up numbers to rival the RTX 3080 Ti. And I haven't even touched on the new features packed into Lovelace that you can't get with the last-gen Nvidia graphics cards.
The numbers are what they are, and the RTX 4070's performance is simply outstanding across every resolution in all the ways that matter.
Performance score: 5 / 5
Should you buy the Nvidia GeForce RTX 4070 ?
(Image credit: Future / John Loeffler)
Buy it if...
You want next-gen performance for less than $600
The Nvidia RTX 4070 offers performance on par with the RTX 3080 and even the RTX 3080 Ti for a good deal less.
You don't want a massive GPU Graphics cards are starting to resemble transformers nowadays (both the autobot and power plant variety), so it's nice to get a graphics card that's just normal-sized.
You want next-gen features like DLSS 3
Nvidia's hardware is often on the bleeding edge of the industry, but things like DLSS 3 and Nvidia Reflex are Nvidia's not-so-secret force multiplier here.
Don't buy it if...
You can get an RTX 3080 cheap
Generally, the RTX 4070 is going to outperform the 3080, but if you don't care about the advanced features and can grab the 3080 in a bargain bin, you could save some money.
You're looking for Nvidia's next budget card The RTX 4070 is a lot cheaper than the rest of the current-gen graphics card lineups from Nvidia and AMD, but at $600, it's still too expensive to truly be a "budget" GPU.
Nvidia GeForce RTX 4070 review: Also consider
If our Nvidia GeForce RTX 4070 review has you considering other options, here are two more graphics cards to consider...
How I tested the Nvidia GeForce RTX 4070 Ti
(Image credit: Future / John Loeffler)
I spent about 50 hours with the RTX 4070 in total
Besides general benchmarking, I used the card for everyday gaming and creative work
My test bench specs
Here is the systems I used to test the Nvidia GeForce RTX 4070:
When I test a graphics card, I start by making sure that all tests are performed on the same test bench setup to isolate GPU performance. I then run it through a series of synthetic benchmarking tools like 3DMark as well as in-game benchmarks in the most recent PC games I can access like Cyberpunk 2077 and F1 2022.
I run everything on the maximum settings possible without upscaling tech, and I run all tests at the resolution a reader is most likely to use a given card at. In the case of the RTX 4070, this meant testing at 1080p, 1440p, and 2160p.
I also make sure to install the latest relevant drivers and rerun tests on any competing graphics card that I might have already reviewed and tested, like the RTX 4070 Ti, RX 7900 XT, and RTX 3080 to make sure that I have the most current scores to account for any driver updates. All of these scores are recorded and compared against the card's predecessor, its most direct rival, and the card directly above and below it in the product stack, if those cards are available.
I then average these scores to come to a final overall score and divide that by the card's MSRP to see how much performance every dollar or pound spent actually gets you to find how much value the card actually brings to the table.
Finally, I actually use the card in my own personal computer for several days, playing games, using apps like Adobe Photoshop or Adobe Illustrator, and watching for any anomalies, crashes, glitches, or visual disruptions that may occur during my time with the card. Having extensively covered and tested many graphics cards over the years, I know what a graphics card should do and how it should perform, and can readily identify when something is not performing up to expectations and when it exceeds them.
The Nvidia GeForce RTX 4070 Ti came onto the scene needing to score a real win if Team GReen had any hope of reigning in a resurgent AMD, and this is exactly the right graphics card to do that.
The RTX 4070 Ti isn't the best graphics card Nvidia's ever put out, and its launch has been somewhat overshadowed by the major stumbles Nvidia has made since Jensen Huang first announced the Nvidia Lovelace launch lineup back in September — which was really more of a brief aside during a presentation overwhelmingly devoted to getting us to care about the Omniverse, but I digress.
And, lest we forget, the RTX 4070 Ti is identical in substance to the "unlaunched" RTX 4080 12GB that was initially announced to the confusion of many, and it's not surprising that there is no Founders Edition for this card, since you really can't scratch a 7 out from an 8 in RTX 4080.
What's more, the RTX 4080 that we did get is too expensive to really recommend, so it's disappointing that the RTX 4070 Ti wasn't the card to carry the 4080 brand into the next generation. It is without question the best Nvidia graphics card you can buy right now (by value) from this new generation of GPUs, and it represents a major leap forward for everyday, mainstream PC gaming. It's not without its flaws, but on balance, it's the Nvidia GPU that anyone looking to upgrade with Team Green ought to be buying unless they have a couple of thousand dollars to burn.
With an MSRP of $799 / £799 / AU$1,469, it's cheaper than the cheapest RDNA 3 GPU, the AMD Radeon RX 7900 XT, and half the price of the Nvidia GeForce RTX 4090. And while both of those cards outperform the RTX 4070 Ti in raw performance terms, there are a number of value adds for the RTX 4070 Ti that collectively make it worth major consideration regardless of its limitations.
The Nvidia RTX 4070 Ti isn't an undisputed winner in the lower-premium GPU class, but this is where Nvidia really needed to shore up its flank after AMD crushed it in my AMD Radeon RX 7900 XTX review and to that end it is exactly what Nvidia needed right now.
Nvidia GeForce RTX 4070 Ti review: Price & availability
(Image credit: Future)
RTX 4070 Ti MSRP is the lowest of all the latest next-gen graphics cards
Some third-party cards can even be bought at MSRP
Availability is generally pretty good
The Nvidia RTX 4070 Ti, available now in the US, UK, and Australia, comes in at the lowest MSRP of any of the next-gen cards to hit the market in recent months, and it's good to see that Nvidia took a lot of the criticism about price inflation directed toward it after the Lovelace announcement to heart.
The RTX 4070 Ti, with an MSRP of $799 / £799 / AU$1,469, is $100 / £100 cheaper than the RX 7900 XT (though it is actually AU$60 more expensive in Australia), and even though there is no Nvidia Founders Edition card guaranteed to sell at MSRP, even some third-party cards can be found very close to or even matching Nvidia's MSRP.
That said, this is also much more expensive than the Nvidia GeForce RTX 3070 Ti, which had a launch MSRP of $599 / £529 / AU$959. That makes the RTX 4070 Ti $200, £270, and AU$510 more expensive than the card it is technically replacing, so we can't go praising Nvidia too for relaunching the RTX 4080 12GB (originally slated for an $899 / £849 / AU$1,659 launch price) at a somewhat lower price point.
And, of course, prices for other third-party cards may end up being substantially higher, especially for OC versions that squeeze a few hundred MHz more out of the GPU's clock speeds or special edition cards with premium design or water cooling.
Still, there's no getting around the fact that this is the cheapest next-gen card we're going to have for a while, so the RTX 4070 Ti is going to score some major points here by default.
Price score: 4 / 5
Nvidia GeForce RTX 4070 Ti review: Features & chipset
(Image credit: Future)
Third-gen RT cores
DLSS 3 with full frame generation
Lower TDP
The RTX 4070 Ti is built on Nvidia's new Lovelace GPU architecture, which features a significantly smaller TSMC process than the last-gen Nvidia Ampere architecture. At 4nm, as opposed to Ampere's 8nm process from Samsung, we're getting significantly faster clock speeds as well as more energy efficiency with this generation. The RTX 4070 Ti has just over twice as many transistors as the RTX 3070 Ti while packing them into a GPU die about 75% of the size of the RTX 3070 Ti's silicon, and you can see it in the RTX 4070 Ti's slightly lower TDP (285W to 290W for the RTX 3070 Ti).
The RTX 4070 Ti we reviewed, the Asus Tuf GeForce RTX 4070 Ti Gaming OC 12GB, also features a higher boost clock (but not that much more than Nvidia's reference specs) but both the reference clocks and the actual clocks on our Asus card are nearly a full 1,000 MHz faster than those of the RTX 3070 Ti, so this card is seriously fast.
As I said before, this is essentially the unlaunched RTX 4080 12GB, and so it has the same specs as that unlaunched card did, including 7,680 CUDA cores, 60 ray tracing cores, and 240 Tensor cores for doing all those tricky machine learning calculations needed to power the new DLSS 3 with full frame generation, which is as big a deal today as DLSS 2.0 was when it launched with Nvidia Ampere and made fast 4K gaming a reality for gamers around the world.
On the memory side, there is 12GB GDDR6X VRAM, the same as in the unlaunched RTX 4080, as is both the memory clock (1,313MHz) and 192-bit memory bus (for a total memory bandwidth of 504.2GB/s). If you're worried that 12GB might be a bit too low for a 4K graphics card, you don't need to worry about that with this card. There is more than enough to power a high-refresh 4K display, which is something that the RTX 3070 Ti could only do on the most restrictive of settings.
That memory though is too little to really power 8K content at fast speeds, and even if you could get more than 60 fps at 8K in any given game, the RTX 4070 Ti is locked in at 60Hz for 8K video thanks to its lack of DisplayPort 2.1 output, so you'll never be able to game faster than 60 fps at 8K.
The RX 7900 XT, on the other hand, has both the memory and the output capacity to go as high as 165Hz at 8K, though it would only really be able to natively manage more than 60 fps on an 8K display with very low lift games like esports titles and the like.
Still, it's possible for the RX 7900 XT, and it's really not with the RTX 4070 Ti, which is a shame. 8K gaming isn't really here yet beyond a couple of titles like Spider-man: Miles Morales, but with higher refresh rate 8K displays set to hit the market in the next year or two, the RTX 4070 Ti feels less future-proof than a card this expensive should be.
Features & chipset: 4 / 5
Nvidia GeForce RTX 4070 Ti review: Design
(Image credit: Future)
No Founders Edition
Designs will vary, but none of them will be small
USB Type-C output
The Nvidia RTX 4070 Ti we reviewed is actually a third-party card, since there is no Nvidia RTX 4070 Ti Founders Edition, but there are definitely some things that I can generalize about the design of the RTX 4070 Ti. Namely, that this is going to be a honking-big card no matter who you buy it from.
Image 1 of 3
(Image credit: Future)
Image 2 of 3
(Image credit: Future)
Image 3 of 3
(Image credit: Future)
Side by side with the RTX 4080, the RTX 4070 Ti takes up just as much space and will be just as challenging to wedge into all but the largest of tower cases. It's a triple-slot card, so a normal ATX motherboard will leave little room for anything else to slot in next to it.
Then there's the matter of its 16-pin connector. Given the lower power requirements for the RTX 4070 Ti, the card only requires three 8-pin connectors to be plugged into the included adapter (as opposed to four 8-pins for the RTX 4090), but the adapter is still going to be clunky to deal with, so unless you have a new ATX 3.0 PSU that comes with a 16-pin connector cable, your cable management skills are going to really be put to the test.
In terms of output, there is no DisplayPort 2.1, as I mentioned, but there's also no USB Type-C output either, something that would make a lot of sense on this card since it has very strong creative workload performance, so a lot of creatives on a budget would be tempted to give this card a look, but since the best USB-C monitors are also very popular among the creative professional crowd, they'll have to use an adapter, and no one likes having to use those if they can help it.
There is also an included support bracket for the Asus card we received for testing, and I imagine that a lot of other manufacturers will be including them as well. This card weighs a good bit, so torque forces are not going to be kind to it (or any other RTX 4070 Ti) in the medium-to-long term, so make sure you use one if you aren't running the card upright.
Finally, as for Asus's Tuf Gaming design, the open metal shroud exposes more of the heat sink while a triple-fan array will keep air moving through it. The cage-like shroud does look cool, and there is some RGB along the top of the card as well if you're into that.
Design score: 3.5 / 5
Nvidia GeForce RTX 4070 Ti review: Performance
(Image credit: Future)
4K gaming on a 1440p GPU
Great "budget" graphics card for creative pros
Competent ray tracing at 4K, especially with DLSS
As we move on to the RTX 4070 Ti's performance, the long and short of it is what it's been for two generations now: the RTX 4070 Ti is better at most creative workloads and ray tracing while AMD pulls ahead in rasterization, especially in gaming, though it is worth noting that the RX 7900 XT doesn't fall as far behind the RTX 4070 Ti in ray tracing as it would have a generation ago.
Image 1 of 6
(Image credit: Future / Infogram)
Image 2 of 6
(Image credit: Future / Infogram)
Image 3 of 6
(Image credit: Future / Infogram)
Image 4 of 6
(Image credit: Future / Infogram)
Image 5 of 6
(Image credit: Future / Infogram)
Image 6 of 6
(Image credit: Future / Infogram)
Nvidia's edge in ray tracing performance can be seen in our 3DMark Speedway and Port Royal tests, both of which are ray tracing-heavy benchmarks. The RTX 4070 Ti pulls out a fairly close win here, edging out the RX 7900 XT by a few hundred points in each.
Once we move onto Timespy and Firestrike though, both at 1440p and 4K, AMD's rasterization advantage really pulls ahead of Nvidia's RTX 4070 Ti with the RX 7900 XT blowing out the RTX 4070 Ti by a few thousand points at times.
Image 1 of 5
(Image credit: Future / InfoGram)
Image 2 of 5
(Image credit: Future / InfoGram)
Image 3 of 5
(Image credit: Future / InfoGram)
Image 4 of 5
(Image credit: Future / InfoGram)
Image 5 of 5
(Image credit: Future / InfoGram)
Whatever gloating the RX 7900 XT might be doing gets quickly shut down in creative tests, especially in our Blender Benchmark tests. Admittedly, this isn't a fair fight, since Blender Cycles is highly optimized for the CUDA instruction set. Even the last-gen RTX 3070 Ti pushes AMD's RX 7900 XT aside in Blender Benchmark, and the RTX 4070 Ti fully laps its Team Red rival and then some.
This extends to Adobe Premiere as well, where the RTX 4070 Ti outperforms the RX 7900 XT by about 17.5%. If it's any consolation for the RX 7900 XT, it's that it edges out the RTX 4070 Ti in Photoshop, which is the living definition of a rasterization workload, so this shouldn't be surprising. Still, the RTX 4070 Ti manages to only lose by about 1.5%, so it's enough to call it a wash.
The key takeaway for me from these creative benchmark results though is that the RTX 4070 Ti is quite adept at creative work normally reserved for graphics cards twice as expensive, so any creatives out there looking for a more "budget" GPU option for their workstation actually have one now.
Image 1 of 13
(Image credit: Future / InfoGram)
Image 2 of 13
(Image credit: Future / InfoGram)
Image 3 of 13
(Image credit: Future / InfoGram)
Image 4 of 13
(Image credit: Future / InfoGram)
Image 5 of 13
(Image credit: Future / InfoGram)
Image 6 of 13
(Image credit: Future / InfoGram)
Image 7 of 13
(Image credit: Future / InfoGram)
Image 8 of 13
(Image credit: Future / InfoGram)
Image 9 of 13
(Image credit: Future / InfoGram)
Image 10 of 13
(Image credit: Future / InfoGram)
Image 11 of 13
(Image credit: Future / InfoGram)
Image 12 of 13
(Image credit: Future / InfoGram)
Image 13 of 13
(Image credit: Future / InfoGram)
In gaming performance, we again see the RTX 4070 Ti besting the RX 7900 XT when it comes to ray tracing while often slipping into second when ray tracing isn't a factor. Ultimately, though, the difference here isn't too significant to go with one over the other on this basis alone.
All games need rasterization performance while only a few games even implement ray tracing, so the RTX 4070 Ti's third-gen ray tracing cores are only really a factor when it comes to premium ray tracing experiences like Cyberpunk 2077 or Hitman 3, and even then, it's still at the point where you need to rely on DLSS 3 for a good frame rate.
It also needs to be said that, technically, the RTX 4070 Ti is a 1440p graphics card. It's not supposed to really perform this well at 4K, as evidenced by the RTX 3070 Ti's rather pitiful showing in a number of 4K game benchmarks. Heck, the RTX 3070 Ti could barely get through the Hitman 3 Dubai benchmark at 4K without ray tracing and totally craps the bed when you turn ray tracing on to the point of crashing to the desktop.
The fact that the RTX 4070 Ti is competitive at 4K is a huge win for the RTX 4070 Ti here, especially given that the RX 7900 XT has a much more built-in hardware advantage at 4K owing to 66% more VRAM and 58.66% more memory bandwidth for pushing through 4K textures.
AMD might have the edge in our gaming tests, but the raw number doesn't tell the whole picture. In our formal benchmark tests, we don't use DLSS or FSR to improve frame rates algorithmically, since updates can make or break their functionality, and this happens too often for any numbers gathered while using upscaling to really be valid after even a single update. As such, it's more important to get a baseline figure that can't change much over time when comparing hardware, and the non-assisted hardware rendering taking place in the PCIe slot can rarely, if ever, change.
That said, no one plays games nowadays without some form of upscaling. Even if you've got a GTX 1060, you can still use FSR and you undoubtedly will. With the RTX 4070 Ti, DLSS 2.0 is already phenomenal, but DLSS 3 with frame generation takes DLSS 2.0's performance gains and pretty much doubles it. In practice, with DLSS 3 you will almost always get more fps in-game than you will with the RX 7900 XT running FSR 2.2, so even though the RX 7900 XT barely pulls ahead on gaming performance, your actual experience of gaming on the RTX 4070 Ti probably won't reflect that at all.
Image 1 of 4
(Image credit: Future / InfoGram)
Image 2 of 4
(Image credit: Future / InfoGram)
Image 3 of 4
(Image credit: Future / InfoGram)
Image 4 of 4
(Image credit: Future / InfoGram)
In the end, the RTX 4070 Ti comes in very close behind the RX 7900 XT in overall performance, owing mostly to the RX 7900 XT's strong absolute performance in synthetic benchmarks. Normalize all of that mathematically, though, and the RTX 4070 Ti is going to be the better bet here even without DLSS 3. Throw that in on top of everything else, and the RTX 4070 Ti simply walks away with it in terms of performance against both its predecessor and its direct rival from AMD.
Performance score: 4.5 / 5
Should you buy the Nvidia GeForce RTX 4070 Ti?
Buy it if...
You want the cheapest next-gen card available At MSRP, the RTX 4070 Ti is the cheapest next-gen GPU on the market right now.
You want the best bang for your buck
The RTX 4070 Ti has the best performance for price of any of the newest graphics cards from Nvidia and AMD, so your money will go farther with the 4070 Ti than with anything else from this generation.
You want next-gen features like DLSS 3
Nvidia's hardware is often on the bleeding edge of the industry, but things like DLSS 3 and Nvidia Reflex are Nvidia's not-so-secret force multiplier here.
Don't buy it if...
You plan on doing a lot of 8K gaming With just 12GB VRAM and no DisplayPort 2.1, 8K gaming with modern AAA titles on the RTX 4070 Ti is going to be a challenge.
You're on a very tight budget While the RTX 4070 Ti is the cheapest graphics card on the market right now, it won't be the cheapest for long, as more affordable cards from both Nvidia and AMD are set to drop in the next few months.
Nvidia GeForce RTX 4070 Ti review: Also consider
If our Nvidia GeForce RTX 4070 Ti review has you considering other options, here are two more graphics cards to consider...
How I tested the Nvidia GeForce RTX 4070 Ti
(Image credit: Future)
I spent about 30 hours with the RTX 4070 Ti in total
Besides general benchmarking, I used the card in my everyday gaming and creative work
In addition to standard benchmarks, I played games for several days with a framerate monitor active and recorded the real-world average
When I test a graphics card, I start by making sure that all tests are performed on the same test bench setup to isolate GPU performance. I then run it through a series of synthetic benchmarking tools like 3DMark as well as in-game benchmarks in the most recent PC games I can access like Cyberpunk 2077 and F1 2022. I run everything on the maximum settings possible without upscaling tech, and I run all tests at the resolution a reader is most likely to use a given card at. In the case of the RTX 4070 Ti, this means nothing less than 2,560 x 1440p, with 3,840 x 2,160p wherever possible.
I also make sure to install the latest relevant drivers and rerun tests on any competing graphics card that I might have already reviewed and tested, like the RTX 4080 and the RX 7900 XT, to make sure that I have the most current scores to account for any driver updates. All of these scores are recorded and compared against the card's predecessor, its most direct rival, and the card directly above and below it in the product stack, if those cards are available. I then average these scores to come to a final overall score and divide that by the card's MSRP to see how much performance every dollar or pound spent actually gets you to find how much value the card actually brings to the table.
Finally, I actually use the card in my own personal computer for several days, playing games, using apps like Adobe Photoshop or Adobe Illustrator, and watching for any anomalies, crashes, glitches, or visual disruptions that may occur during my time with the card. Having extensively covered and tested many graphics cards over the years, I know what a graphics card should do and how it should perform, and can readily identify when something is not performing up to expectations and when it exceeds them.
