Umair
Our Verdict
At a price of $999, the Nvidia GeForce RTX 5080 is now easily the best sub-$1,000 graphics card. Multi Frame Gen offers an enormous boost to performance in games that support it, with a negligible impact on image quality, and while the RTX 5080 is only a little faster than the 4080 Super without DLSS 4, the fact that it costs the same price makes it a no brainer.
- Incredible performance with Multi Frame Gen
- Much cheaper than RTX 4080
- Fantastic card and cooler design
- Half the price of the RTX 5090
- Only slightly quicker than RTX 4080 Super without DLSS 4
- Multi Frame Gen needs wider game support
- Price is still expensive
- Not enough VRAM for some top game settings
With its slimline two-slot cooler, the Nvidia GeForce RTX 5080 Founders Edition might look practically identical to its two-grand sibling when you lift it out of its slender, recyclable cardboard box, but there’s a very different GPU under that cooler. You can basically think of the RTX 5080 as half a 5090 – you get around half the GPU power and half the VRAM. Perhaps more importantly, it’s also half the price. Despite all this, though, the RTX 5080 performs exceptionally well, and can still be regarded as a 4K gaming GPU.
This new Nvidia top-end sub-$1,000 GPU picks up from where the RTX 4080 Super left off, and that’s not a bad place to start. This GPU is already the best graphics card you can buy if you can’t stretch to over a grand, and the 5080 brings support for DLSS 4 and Multi Frame Gen with it, which uses AI to insert up to three frames between each pair of genuinely rendered ones, massively increasing the frame rate, with a negligible impact on image quality.
I’m also really glad to see that Nvidia has abandoned the ludicrous $1,199 price tag that held back the RTX 4080 when it was launched, and instead gone with the $999 tag of the later 4080 Super model. On the other hand, the RTX 5080 has a very similar spec to the 4080 and its Super variant in terms of basic numbers, which means it’s heavily relying on new features such as Multi Frame Gen to really give it a generational edge. Is this enough? I’ve run the RTX 5080 benchmarks on our test rig to find out.
Why you can trust our advice ✔ At PCGamesN, our experts spend hours testing hardware and reviewing games and VPNs. We share honest, unbiased opinions to help you buy the best. Find out how we test.
Specs
| RTX 5080 specs | |
| CUDA cores | 10,752 |
| RT cores | 84 (4th-gen) |
| Tensor cores | 336 (5th-gen) |
| Base clock | 2,300MHz |
| Boost clock | 2,620MHz |
| L2 cache | 64MB |
| ROPs | 128 |
| VRAM | 16GB 32Gbps GDDR7 |
| Memory interface | 256-bit |
| Memory bandwidth | 960GB/s |
| Interface | 16x PCIe 5.0 |
| Power connectors | 1 x 16-pin (3 x 8-pin with adapter) |
| Total graphics power (TGP) | 360W |
While the RTX 5080 still maintains the same 16GB frame buffer as its predecessor in terms of capacity, the new GPU uses fast GDDR7 VRAM to improve performance. Not only that, but it’s running at a super-fast speed of 32Gbps, which is even quicker than the 28Gbps memory supplied with the RTX 5090.
The RTX 5080 also partners this memory with a 256-bit wide interface, which is tight for a $999 GPU in 2025. As a point of comparison, the cheaper AMD Radeon RX 7900 XTX has 24GB of VRAM attached to a 384-bit interface. However, this AMD GPU only uses GDDR6 memory, and the huge speed of the RTX 5080’s GDDR7 memory results in a massive total memory bandwidth of 960GB/s. That’s not far off the 1.01TB/s bandwidth of the RTX 4090, and it means this GPU will be able to get data to and from its memory incredibly quickly.
Is 16GB enough? At the moment, yes, in the vast majority of most cases, but as I’ll come to later, there are already some games that demand more if you want to max them out. However, there are rumors that a 24GB RTX 5080 is also on the cards for the future, as the new Samsung 3GB GDDR7 memory chips can theoretically enable graphics card makers to drop 24GB of VRAM into the same PCB as a 16GB card, without any fundamental changes needing to be made to the GPU. For the moment, though, if you want the RTX 5080, you have to make do with 16GB of memory.
Another area where the RTX 5080 has a bonus over the 5090 is its GPU clock speed, with the smaller, less powerful GPU enabling Nvidia to crank more megahertz out of the 5080, while using the same cooler. The RTX 5080 can boost to up to 2,620MHz, compared to 2,407MHz on the 5090 – a 213MHz difference.
The downside is that the new $999 GPU has around half the number of internal processors as the RTX 5090. You get 10,752 CUDA cores in the RTX 5080, which the GPU uses for the rasterization tasks used by most games. This is barely any more than the 10,240 in the 4080 Super, or the 9,728 in the original 4080, so we’re not expecting the RTX 5080 to be much quicker than its predecessors in games without ray tracing or DLSS.
Those CUDA cores are spread across 84 of Nvidia’s streaming multiprocessor (SM) building blocks, each of which has a new 4th-gen RT core for ray tracing and four 5th-gen Tensor cores for AI work, such as DLSS and Nvidia’s new neural rendering features. This is less than half the number of SMs in the RTX 5090, but it still gives you 84 RT cores and 336 Tensor cores, which results in a supremely powerful GPU when it comes to frame rates.
The other key area of the spec is the power draw, which at 360W is a good 215W lower than the RTX 5090. That means you can happily use the RTX 5080 with an 850W PSU, or even a 750W PSU if you’re using a power-efficient CPU, such as the AMD Ryzen 7 9800X3D.
The RTX 5080 Founders Edition uses the same cooler as its larger sibling, and you can read all my detailed thoughts on it in our Nvidia GeForce RTX 5090 review. In short, it’s a fantastic piece of engineering, with the GPU mounted on a small PCB in the middle of the card, allowing room for the two fans on the front to pull cool air from the bottom of your case over the heatsink fins, and then expel it straight out the vents on the back of the card.
It’s quiet, good-looking, and it barely makes any noise. It also only takes up two slots in your PC case, which was a spectacular achievement on the RTX 5090, and still worthy of mention on the RTX 5080. After all, the RTX 4080 used the same massive cooler as the RTX 4090, and the RTX 5080 board partner cards are all much bigger.
I also really like the angled 16-pin power connector, which means you don’t have to turn your cable at a right angle and risk an insecure connection starting a fire. It also makes for neat cabling if you pass the 16-pin plug through a cable-routing hole in your case on the right of the card. You get a 3 x 8-pin adapter in the box for older PSUs that don’t have a 16-pin cable too.
Features
As there’s only a small step up in the basic spec, the main difference between the RTX 5080 and the 4080 cards is the former’s support for Multi Frame Gen, which is part of the new DLSS 4 suite. As I mentioned above, this is a clever way of using AI to insert extra frames between the ones rendered by the GPU, to smooth out the frame rate.
As you’ll see in the benchmarks below, it can have a transformative effect on frame rates, enabling you to max out all the ray tracing settings in your games, and still get fast performance. It’s this tech that enables the RTX 5070 to outperform the 4090 in games that support it.
Comparatively, the RTX 4000 cards only support standard frame generation, which just inserts one extra frame between each rendered pair, while the RTX 5000 cards can insert two or three extra frames, depending your settings. You can enable this feature in your game’s graphics settings, if it supports it, but Nvidia is also offering a way to force Multi Frame Gen support into games, as long as they already support standard frame gen, and as long as you’re using an RTX 5000-series GPU.
You can do this using the Nvidia App, where you’ll soon be able to find a section called DLSS Override in each game’s settings profile. From here, you can force the frame gen model from the standard one to 2x or 3x Multi Frame Gen. At the time of writing this review, I was able to test this feature in Marvel Rivals, and it worked amazingly well, more than doubling the frame rate between standard frame gen and Multi Frame Gen, while the action was still smooth.
Nvidia also says the RTX 5000 series is optimized for its new neural rendering features, such as RTX Skin and Neural Radiance cache. I recently tried Half-Life 2 RTX with neural rendering, and it makes a significant difference to the depth of shadows, as well as the translucency of skin, making for much more realistic graphics. Nvidia will be hoping that big games start to introduce these features soon, but they’re not used in any games right now.
How we test
To assess the performance of the RTX 5080, I’m running a number of benchmarks using real games, rather than synthetic benchmarks. Each test is run three times, recorded with Nvidia FrameView, and we report the mean average of the results, discarding any obvious anomalies. I report two figures for frame rates – firstly, the average, which gives you an idea of the general frame rate you will achieve. Secondly, I report the 1% low, which is an average of the lowest one percent of results recorded during the benchmark.
This is a more reliable indicator of performance than the minimum, as it removes outliers, such as moments where a Windows system event causes the game to stutter, which is unrelated to the performance of the GPU. The 1% low is what you can expect the actual minimum frame rate to be in these games.
GPU test system specs:
- CPU: AMD Ryzen 7 7800X3D
- CPU cooler: Corsair H100X RGB Elite
- Memory: 32GB 6,000MHz G.Skill TridentZ RGB, CL28
- Motherboard: MSI MAG X870E Carbon WiFi
- SSD: 2TB WD Black SN850X
- PSU: Corsair RM1000X Shift
Benchmarks
Cyberpunk 2077
I’m going to start with the game test that best shows off the potential of the RTX 5080, which is Cyberpunk 2077. This aging game still looks amazing, thanks to CD Projekt Red’s ability to keep adding the latest tech to it. I’ve tested the game with the latest build, which adds support for Nvidia’s new DLSS 4 transformer model for upscaling, as well as support for Multi Frame Gen.
The latter is absolutely the RTX 5080’s killer feature. Just look at the 1440p results in the graph above, and you can see that the RTX 5080 averages a massive 228fps, and that’s with the game fully maxed out with path tracing enabled. It looks incredible at these settings, particularly with Nvidia’s new transformer model handling resolution upscaling.
In this test, the only setup that’s faster is the RTX 5090 with Multi Frame Gen also enabled. Comparatively, the RTX 4090 with standard frame gen averages 148fps, while the RTX 4080 Super hits 115fps and the standard 4080 averages 113fps. Basically, in games that support Multi Frame Gen, the RTX 5080 can run at double the speed of the RTX 4080.
As a point of comparison, with no help at all from DLSS Super Resolution or frame gen, the RTX 5080 only averages 41fps in this test, which really goes to show how challenging it is, and how much work Nvidia’s AI tech is doing. Add Super Resolution with the transformer model at the quality setting, and the game averages 61fps, which is a good starting point before you start adding frame gen to the mix, and the game is smooth and responsive when you’re playing.
It can even handle the game at 4K using these settings, averaging 128fps with a 1% low of 98fps, which is again well in front of the 82fps from the RTX 4090 and 59fps from the 4080.
I’ve also included the results from the AMD Radeon RX 7900 XTX in these graphs for reference. As you can see, the AMD GPU really struggles with path tracing, even with both FSR frame gen and upscaling enabled. The AMD GPU averages just 36fps at 1440p and 17fps at 4K. If you want to play games with all the ray tracing eye candy enabled, the RTX 5080 is easily the best choice under $1,000.
As you would expect from the specs, though, if you don’t enable DLSS, there’s very little to separate the RTX 5080 from its predecessors. For example, I ran Cyberpunk 2077 at the Ultra ray tracing preset with no DLSS upscaling or frame gen enabled, and the RTX 5080 averaged 70fps at 1440, while the 4080 super hit 63fps and the standard 4080 runs at an average of 62fps – that’s a 12.9% performance increase, which shows that you really need DLSS and frame gen support to get the most out of this new GPU.
F1 24
Next up is F1 24, which doesn’t have support for Multi Frame Gen support yet, but does showcase the RTX 5080’s ray tracing performance at Ultra settings. The early driver I used for testing also didn’t allow me to enable DLSS Override in this game to force Multi Frame Gen either. However, even using standard frame gen gives this game a decent boost.
At 1440p, the RTX 5080 averages 133fps at Ultra settings without DLSS, which is already a good result, being 20.9% quicker than the RTX 4080. Add frame gen and Super Resolution to the mix, though, and the frame rate climbs to a massive 219fps, with a 1% low of 119fps – a great result for a game running at these settings, and I’m hoping it will be even faster again when I’m able to switch on Multi Frame Gen.
This was also a test that benefited the AMD Radeon RX 7900 XTX when we enabled FSR upscaling and frame gen to the mix, with the AMD GPU averaging a massive 270fps. That’s great for performance, but out of the two options I’d choose the slower RTX 5080, simply because the image quality is so much better.
At the moment, FSR still has a lot of problems with noisy shadow dithering, ghosting, and blurriness, as well as shimmering in fast motion. These artifacts are still present with DLSS, but are much less pronounced. Hopefully AMD will address a lot of this with FSR 4, but at the moment it’s important not to be led astray by that higher frame rate, as the RTX 5080 still provides a better game experience.
Similarly, at 4K, the average of 84fps is already pretty solid, being 21.7% faster than the RTX 4080, but enabling frame gen sees the frame rate rise to 146fps. It’s a shame I’m unable to test this game with Multi Frame Gen at the time of writing, but I’ll be going back to this benchmark with the RTX 5080. Basically, in this game, the RTX 5080 is already a decent amount quicker than the RTX 4080 cards, and frame gen makes it even faster. That’s a good result for this GPU, and it could get even better with Multi Frame Gen enabled.
Indiana Jones and the Great Circle
The latest Indy title is supremely demanding anyway, requiring ray tracing at all of its settings, and being absolutely punishing if you enable the full RT mode with path tracing. This test also saw the limits of this 16GB card, as running this game at 4K with the Supreme preset and full RT enabled maxes out the VRAM and causes performance to fall off a cliff, as you can see in the graph below.
No amount of frame gen can help if the GPU doesn’t have enough VRAM to render the basic frames in the first place, and this test already shows the limits of the 16GB memory allocation. That’s a shame, as the GPU itself looks as though it would have the power to run the game at these settings at a playable frame rate, if only it had the VRAM for it. Hopefully, there will indeed be a 24GB RTX 5080 card at some point to help out here, as the only other way to run the game at these settings is with a 24GB RTX 4090 or 32GB RTX 5090, and these cards cost a lot more money.
And, before you ask, no, a 24GB AMD Radeon RX 7900 XTX won’t work here. Not only are AMD GPUs terrible at path tracing, but they’re also unsupported by this game in the full RT mode. Fire up the game with an AMD GPU, and you won’t even see the options to enable path tracing.
On the plus side, you can at least max out this game at 1440p, where enabling frame gen and setting DLSS Super Resolution to the Quality setting enables the RTX 5080 to average 98fps with a 1% low of 75fps – that’s a good 21% performance increase over the RTX 4080. This game is also on the list of 64 Multi Frame Gen games that Nvidia has listed for the January 30 release date, so it should be able to go even faster with the new tech enabled.
Run this game at these settings without DLSS Super Resolution and frame gen, and the average frame rate drops right down to 41fps, which shows just how much of a challenge it is to run on today’s GPUs without AI tech. Add Super Resolution at the quality setting and the average goes up to 61fps which, as with Cyberpunk 2077, is a solid starting point before you add frame gen.
As with other games, though, if you don’t enable DLSS, the performance difference between this GPU and its predecessors shrinks in this game. The RTX 5080’s average of 90fps at 4K is still a fantastic result in this game at Ultra settings without DLSS, but it’s an 11.1% improvement, rather than a 21% improvement.
Similarly, running the game at the same settings at 1440p results in a 131fps average, which is just 7.4% quicker than the 4080. That’s still a great result, and it’s well in front of the 111fps from the Radeon RX 7900 XTX, but it’s clear that you really need to be running your games at high resolution with DLSS to get the most out of the RTX 5080.
Call of Duty Black Ops 6
Now, the not so good news. With barely any change to the CUDA cores in the RTX 5080 compared to its predecessors, you don’t get much of a performance boost in games that don’t support ray tracing, especially if you don’t enable DLSS.
Call of Duty Black Ops 6 is a case in point. Running this game at the Extreme preset showed only a tiny difference between the RTX 5080 and the 4080, climbing from an average of 89fps at 4K to 93fps – a difference of just 4.5%.
On the plus side, this game’s support for frame gen enables you to increase this to 128fps on the RTX 5080, but you can also enable this feature on the RTX 4080 cards, and even that only gives you a 37.6% performance increase. Once this game either supports Multi Frame Gen natively, or you can force it on with DLSS Override, however, the RTX 5080 should pull ahead, and I look forward to retesting this game when it’s possible.
This is another test where the Radeon RX 7900 XTX is ostensibly faster than the RTX 5080 when you enable FSR and frame gen, but again, the superior image quality of DLSS makes for a sharper image and a better overall gaming experience.
Multi Frame Gen may be a great tech, but this test really shows that you don’t get much of a boost in basic rendering performance by upgrading from the RTX 4080 to the 5080 if you can’t enable it. The results are even starker at 1440p, where the average frame rate only increases from 138fps to 142fps between the RTX 4080 and 5080 – a tiny improvement of just 2.9%. This test also marks the one rare occasion where the Radeon RX 7900 XTX is slightly quicker than the RTX 5080, though not by a huge margin.
Doom Eternal
Similarly to Call of Duty Black Ops 6, Doom Eternal doesn’t scale to new heights with the extra power of the RTX 5080 either. This aging game still runs like a dream on the new GPU, of course, but the performance benefit is negligible compared to the 4080 cards. At 4K without ray tracing, for example, the RTX 5080 averages 342fps, while the 4080 averages 325fps – a boost of just 5.2%. Add ray tracing to the mix, and the difference is still slim at just 6.25%.
We also saw this same limited benefit of the new architectures when comparing the RTX 5090 to the 4090 in this game, but the comparatively lower results of the 5080 and 4080 suggest this isn’t down to the game being CPU limited. Basically, older games don’t appear to respond particularly well to the benefits of the new Blackwell architecture, especially without DLSS. Really, you need a lot of ray tracing in the scene, such as path tracing, and Multi Frame Gen, to really reap the benefits of the RTX 5080.
Power draw
After the big shock of the RTX 5090 power draw, I was delighted to see that the RTX 5080’s power consumption is much tamer. With the RTX 5080 installed, our AMD Ryzen 7 7800X3D test rig drew 454W from the mains at peak load, while playing F1 24. That’s only 2W more than with the RTX 4080 Super fitted, and 32W more than with the RTX 4080.
Basically, the RTX 5080’s power draw is in the same league as the 4080 cards, and a long way off the 524W of the RTX 4090 and the massive 682W of the RTX 5090. A 750W PSU will be fine for this card if you’re using an eight-core CPU, and an 850W unit is recommended if you’re running a power-hungry chip such as the Intel Core i9 14900K. If you’re already using an RTX 4080 in your PC, and your PSU isn’t right up to its limit, you’ll be able to use the same PSU to power the RTX 5080.
Price
The Nvidia GeForce RTX 5090 price is $999, which is $200 lower than the original MSRP of the RTX 4080, and the same price as the more recent RTX 4080 Super. This is still a high price – comparatively, the RTX 3080 launched at just $649 – but with little competition in this area, Nvidia has control over pricing, and thankfully it’s learned a lesson from the RTX 4080 price.
$999 isn’t a bargain, but it’s a no-brainer over the RTX 4080 cards – it may only be a little faster in rasterization, but it handles advanced ray tracing well, it doesn’t cost any more money, and Multi Frame Gen makes a phenomenal difference to performance in games that support it.
Verdict
In many ways, it’s a shame that Nvidia hasn’t yet opened up DLSS Override for testing, as being able to enable Multi Frame Gen in Indiana Jones, Black Ops 6, and F1 24 would have had a big impact on frame rates, assuming it worked properly. As it is, though, Cyberpunk 2077 shows the potential for this tech, with the RTX 5080 being able to completely max out this demanding game at 4K and still average 128fps – a phenomenal result for a $999 card, and if you run the game at 1440p and it runs at an even smoother 228fps.
It remains to be seen how the other games in our test suite will benefit from the new tech, but if there’s one company that has the budget and resources needed to persuade game developers to implement its new features, it’s Nvidia. I’ll be retesting the RTX 5080 once Multi Frame Gen is available to benchmark in more titles, which may result in this review being revised, but even as it is, the RTX 5080 is easily the best sub-$999 gaming GPU.
It annihilates the competition in nearly every test, and while some games aren’t massively quicker on the RTX 5080 than they are on the 4080 cards, the fact that the 5080 has the same price gives it an immediate advantage. Whether you’ll actually be able to buy this card at MSRP, of course, is another question, and you’ll need to check our RTX 5080 stock page regularly if you want to buy one. Assuming the supply situation settles down from current predictions, though, and the RTX 5080 does indeed become available for $999, it’s the best card in its league.
The AMD Radeon RX 9070 XT isn’t expected to challenge it, and the RTX 5090 is twice the price. It’s a little disappointing that the basic rendering performance isn’t much better on the 5090 than the 4080 Super in some games, but that’s outweighed by the fact it has the same price and also has Multi Frame Gen under its belt. The latter needs more game support to make it worthwhile right now, but if it really takes off then RTX 5080 will be even better.
The elephant in the room, of course, is the stock situation. RTX 4080 and 4080 Super cards are impossible to find at MSRP at the moment, often going for at least $1,500, and if RTX 5080 supply is indeed low, then that $999 price is only nominal. My advice here is to wait it out for a few months until the stock situation settles, and you’ll then hopefully be able to pick up an RTX 5080 for a normal price. If you can genuinely get it for $999, then it’s a great GPU.
