Has any genus of graphics card been as dramatically storied as the GeForce XX80s? The RTX 3080 was a thing of beauty, only to be tarnished by the worst wheeler-dealing spree (and crypto mining misappropriation) in PC component history. Then the RTX 4080 rocked up with its laughable £1269 / $1199 price tag, a miscalculation so severe that the RTX 4080 Super looked good – despite hardly being any faster – simply for not repeating it. For the new RTX 5080’s sake, you almost want it to be boring.
It isn’t. But then, neither is it a blood-boiler like the RTX 4080, nor a largely aspirational show-off piece like the RTX 5090. By maintaining the 4080 Super’s course correction on price while tooling up on compelling DLSS 4 improvements, the RTX 5080 is an agreeable GPU from the off. Particularly, if you’ve got the 4K monitor to take full advantage of it.
It’s certainly a more sensible buy than the RTX 5090 for high-rez gamesplaying. A Founders Edition, like the one I’m testing here, will set you back £979 / $999, or roughly half the price of the next (and final) RTX 50 model in the pile. Plus, while it has the same 16GB of 256-bit VRAM as the 4080 and 4080 Super, that’s still enough for all the AAA industry’s biggest resource-eaters, and it’s been upgraded to GDDR7 for a stretch of extra bandwidth. Get the Founders Edition and you’ll also bag yourself the same lovely metallic design and effective dual-fan cooler that the RTX 5090 FE has, running even colder here thanks to the RTX 4080’s lower power needs.
It’s still a thirsty creature, to be sure: Nvidia reckon you’ll need at least an 850W PSU to run it, and max usage is rated at a toasty 360W. Unlike on the RTX 5090 FE, however, I didn’t record the RTX 5080 FE actually hitting its limit, with it peaking at 322W while running games. Its GPU temperature high of 64°c was a good nine degrees cooler the 5090 as well. See? Sensible. Ish.
Nvidia GeForce RTX 5080 review: 4K benchmarks
The biggest and scariest of mainstream monitor resolutions doesn’t faze the RTX 5080 a jot, even in the more recent tough runners. The RTX 5090 is faster still, though the RTX 5080 is always able to match a good majority of its sibling’s frames-per-second output for half the cash. It’s also a juicy upgrade over the RTX 30 series’ elite – encouraging, if you’re upgrading from an older card – and it even comes somewhat close to the RTX 4090, which at the time of writing is still going for upwards of £1300 in the secondhand market.
Keen bar chart fans may recall that the RTX 4080 did beat the RTX 3090 and RTX 3090 Ti outright, so it’s a bit of a backwards shuffle for the RTX 5080 to not completely outpace the previous generation’s top dog in turn. I’m not sure that’s a particularly huge deal, outside of maintaining conventions, though I do wish the RTX 5080 gave more of a speed boost to rasterised, native-rez games over the RTX 4080 Super.
To look at this, we must – temporarily but with apologies – return to the old RPS test rig and its Core i5-11600K, which is the only one I have RTX 4080 Super results for. Cyberpunk 2077 sees a decent bump, rising from 51fps (with Ultra settings) on the 4080 Super to 71fps on the 5080, though the newest GPU’s 81fps in F1 22 (on Ultra High) is an uptick of just 12fps. Metro Exodus barely improved either, averaging 99fps on the 4080 Super and 106fps on the 5080.
What none of these RTX 40 series cards have, however, is a full suite of DLSS 4 framerate-massaging tools. While one of my personal favourites among these – the new Transformer model option for upscaling, which enhances overall image quality in exchange for a smaller performance boost – is available to older RTX models, the headlining multi Frame Generation (MFG) is exclusive to the RTX 50 series. And this actually can nudge the RTX 5080 past the RTX 4090, at least in terms of purely visible smoothness.
If you’re unfamiliar, MFG takes the AI frame generation of DLSS 3 and can throw up anywhere between one and three generated frames for every one of the frames that your PC actually renders. This adds some extra input lag, and it won’t necessarily feel smoother (as the generated frames don’t take into account your mouse movements or keyboard presses), but the visual impact can look very fine indeed. On the RTX 5080, MFG also opens up the possibility of above-60fps framerates at while combining 4K with some of the most demanding ray tracing and path tracing effects in the business. In all three games I tried, applying the 4x frame gen option let it saunter past the RTX 4090, whose DLSS 3 version maxes out at the equivalent of MFG’s 2x setting.
Outside of benchmark nerdery, everything looks and feels playable, though Alan Wake II did start pushing 100ms of latency with 4x MFG on – enough for aiming to get a little sluggish. I’d probably drop DLSS upscaling from Quality to Balanced, in that case, to bump up that base of 37 non-AI frames per second. That would improve visual smoothness even more while touching up responsivity. Otherwise, though, DLSS 4 makes a persuasive argument in the RTX 5080’s favour, and with 75 compatible games confirmed, it’s getting a far more widely-supported start to life than rival frame gen tech like FSR 3 and XeSS 2 did.
Nvidia GeForce RTX 5080 review: 1440p benchmarks
As ever, anything 4K can do, 1440p can do faster, and there are good reasons to choose the RTX 5080 for this resolution if you’ve redirected all your monitor upgrade budget into a GPU fund. CPU bottlenecking snips the tall-poppyness of the RTX 5090, allowing this cheaper card to keep pace more effectively, while also narrowing the gap against RTX 4090 and downright styling on the RTX 3090.
That said, it’s worth waiting for the RTX 5070 and RTX 5070 Ti at this stage. The RTX 5080 is undoubtedly a barnstormer at 1440p but unlike with 4K, there is sometimes a feeling that you’re paying for more power and features than you can use. 4x MFG, for instance, isn’t as useful because the framerates you can get without it are already sky-high: take Cyberpunk 2077 with Psycho ray tracing, which the RTX 5080 could run at 106fps with just Quality-level DLSS upscaling. Most monitors wouldn’t even be able to display all the extra AI frames, even if your eyes could perceive the difference between such speeds – and diminishing returns very much do come into effect.
Then again, there is a case for 2x frame gen in some games. F1 24 on Ultra High rose from 116fps with only Quality DLSS to 148fps with frame gen, and that actually did make for a noticeable (if minor) smoothness improvement. The RTX 5080 clearly isn’t a bad GPU for Quad HD – it’s just not the best resolution for spreading its DLSS 4 wings, hence why it’ll be intriguing to see whether the more affordable RTX 5070 duo can make for more suitable matches.
Or AMD’s Radeon RX 9070, maybe. That will have its own souped-up DLSS rival in FSR 4, though it sounds like this will be a simpler 2x frame generator in the DLSS 3 vein. With the faster and more flexible MFG, DLSS 4’s pin-sharp Transformer model option, and the GeForce line’s historic advantages with ray and path tracing, it’s hard to see how RDNA 4 GPUs like this will be able to stand up to the RTX 50 series on features.
For 4K, anyway, the RTX 5080 is the one to beat. It’s cooler, more efficient, and (not, uh, counting any potential stock shortages) drastically more attainable than the RTX 5090, while still having enough muscle for slick, max-quality performance. Is it as lovable as that poor, reseller-abused RTX 3080? Not so much – a bigger jump up from the RTX 4080 Super would have been appreciated. But it’s better deal overall than both the Super and the original RTX 4080, especially once you start exploiting its more advanced tech. Tech that can help it leave the RTX 4090 behind, let alone those 4080s.
This review is based on a retail unit provided by the manufacturer.