Nvidia RTX 4070 vs 3080: two of the best GPUs around go toe to toe

The Nvidia GeForce RTX 4070 has finally arrived, and we're pumped - at long last, a next-gen graphics card that doesn't require you to negotiate a bank loan before you buy it!

We're exaggerating, of course; its big brother the RTX 4070 Ti is already more affordable and offers better performance-per-dollar than the more powerful cards in Nvidia's 'Lovelace' RTX 4000 lineup, but it's great to see an actual midrange GPU landing. At $599, this is the card most PC gamers are realistically going to want - but is it the best choice?

We've already compared the new GPU to its direct precursor, the RTX 3070, and concluded that while it's the superior card, the lower price of the 3070 means that it's still a very strong value proposition versus its successor. But how does the RTX 4080 stack up against a more expensive GPU from Nvidia's previous generation?

Enter the RTX 3080: the graphics card that brought 4K gaming to the masses back in the locked-down days of 2021. Mass-scale scalping and crypto mining might've meant that not as many gamers got their hands on the 3080 as they'd have liked - but in this post-crypto-crash world, GPUs are more accessible. Why not pick up a last-gen card instead of spending hours hunting for a newly-launched next-gen model?

It's a valid question, and Nvidia's RTX 3000 series remains relevant to this day; so we're going to break down the nitty-gritty details and help you work out what the best graphics card for you is. Let's get down to business.

Nvidia RTX 4070 vs RTX 3080: Price

On paper, the RTX 4070 should win this round cleanly. After all, its $599 / £599 / around AU$900 MSRP sits it squarely beneath the RTX 3080, which retailed for $699 / £649 / about AU$950.

You'll notice there's no regional currency adjustment this time around though, so the RTX 4070 is effectively more expensive in most regions outside the US, which is a shame for non-Americans. It's also generationally more expensive than its predecessor, which hurts its value proposition somewhat; the RTX 3070 retailed for $499 / £469 / AU$809 back in 2021.

Of course, the RTX 4070 is likely to be scalped and price-gouged to all hell in the period immediately after its launch (that sucks), but we won't factor that in here - it's not really Nvidia's fault, after all. The pricing will hopefully stabilize somewhat, and retailers are likely to run 'lotteries' for retail-price cards, as they did with the RTX 4090 launch.

On the other side, it's still hard to find an RTX 3080 at its original retail price; sure, you can pick up this $699 model on Newegg (opens in new tab), but it's a Biostar card - not one of the mainstream GPU brands we'd typically recommend. RTX 3080 cards from more trusted manufacturers such as Asus and Gigabyte will run you upwards of $900 right now.

This is doubly strange given that the similarly-priced RTX 4070 Ti is now available (as well as the far less impressive RTX 4080, the 3080's direct descendant) has been available for a little while now. It's possible the 3080's pricing will drop a bit in the wake of the 4070's arrival, but nobody can truly predict that.

Ultimately, GPU pricing remains at the mercy of many external factors; the current explosion in AI research thanks to everyone's favorite chatbot ChatGPT is threatening to lead to another GPU shortage, and there's not much Nvidia (or PC gamers as a whole) can do about that. We'll give the win to the 4070 this round, but only narrowly - as you'll see further down, it offers slightly more bang for your buck than the 3080, so it deserves the win here.

• Winner: Nvidia RTX 4070

Nvidia RTX 4070 vs RTX 3080: Design

Good lord, a next-gen GPU that isn't the size of a paving slab! We're joking, of course, but the RTX 4000 GPUs released so far have certainly been pretty darn chonky; the flagship RTX 4090 is bloody massive.

As such, it's gratifying to see Nvidia dial things back a bit for the RTX 4070, with a more sensible two-slot design that matches the RTX 3080. It's actually even smaller than the 3080, being a bit shorter thanks to a more compact heat sink behind the front casing.

In terms of aesthetics, these two GPUs are practically identical - note that we're specifically comparing Nvidia's own Founders Edition (FE) models of the cards, not any third-party iterations, all of which will have their own designs and will appeal differently to different users.

Unlike the RTX 3070 FE, which featured two open-face fans on the front side, the new RTX 4070 FE mirrors the split-fan design of the RTX 3080 FE (and the other currently-available RTX 4000 cards), where one fan is situated on the reverse side and hidden by a panel bearing the card's designation.

We maintain that Nvidia's new FE designs (starting with the RTX 3000 series) are perhaps the best Team Green has ever produced, replete with their sleek metal housing and clean curved lines for cooling. Both cards look fantastic.

Unfortunately, both also feature non-standard power connectors. The traditional 8-pin power connector (the best GPU power connection standard out there) was abandoned for both of these cards, with the RTX 3080 using a larger 12-pin port while the 4070 uses the highly contentious 16-pin 12VHPWR connector.

This 16-pin slot requires an adaptor to be used with older ATX power supplies (you may need one for the 3080's 12-pin one too), and was the subject of 'cablegate', a thankfully short-lived crisis that saw some RTX 4090 cards overheating and melting the power connector due to a thermal variance issue in high-powered GPUs. Thankfully, that shouldn't be an issue here.

• Winner: Tie

Nvidia RTX 4070 vs RTX 3080: Performance

Damn, this round is going to be a hard-fought contest. An older - but ostensibly more powerful - GPU, pitted against the newer, sleeker, but more affordable challenger? It's the battle of the year.

And it's a seriously close battle, too. Looking first at our suite of creative benchmarks, the RTX 4070 gets off to a strong start, just about managing to pip the 3080 to the post across almost every test. The only benchmark that showed significantly better performance was the Lumion 12.5 4K render test, where the 4070's newer hardware saw it soar to victory.

Overall, creative performance is extremely close, with the 4070 securing an edge of just over two percent on average. Considering the lower MSRP, that's impressive.

Moving on to synthetic tests, and it's a similarly close fight. The RTX 3080 takes the overall lead here, but by an even smaller margin - just 1.37%. (Bear in mind that the averages provided at the bottom incorporate the results of tests not shown here - check out the full Nvidia RTX 4070 review if you want all the numbers.)

Interestingly, the biggest gulf in performance was found in the 3DMark Firestrike Ultra test, which tests graphical and physics performance at 4K resolution. Could this be indicative of how our gaming results will come out...?

Let's not get ahead of ourselves. We'll take a look at 1440p gaming first - everyone knows both of these cards will dominate at 1080p, so higher resolutions are where the real battle lies.

The 3080's hardware might be older, but it's simply a higher-spec graphics card than the 4070; it wins in most areas here, with the biggest difference seen in Cyberpunk 2077 with ray tracing turned off.

Once we're working with ray-traced graphics, though, the RTX 4070 bares its teeth again. It closes the gap significantly in Cyberpunk, and actually manages to overtake the 3080 by 5.4% in Returnal too - proving that Nvidia's third-gen RT tech has made some serious steps forward.

Shifting over to 4K, and the gap widens - as predicted by our synthetic benchmark results. At this resolution, the RTX 3080 outperforms the new card on the block by almost ten percent on average.

However, once again the 4070 demonstrates the power Lovelace architecture brings to the table in ray-traced workloads, matching the 3080 in Cyberpunk and absolutely thumping it in Returnal with a differential of more than 15%.

Of course, neither of these cards are really going to be used for 4K gaming without employing Nvidia's nifty DLSS upscaling tech. In terms of raw performance, the RTX 3080 has the edge - but's a close race, and it's missing the next-gen tech that the 4070 offers.

DLSS 3 is amazing. Seriously; the full-frame-generation it packs means that the 4070 was able to basically wreck the 3080 in most supported games compared to the previous-gen DLSS 2.0. No, we still don't know why Nvidia chose to drop the '.0' this time around.

In technical terms, then, the 3080 is the more powerful GPU, but the 4070's slick new technology gives it the real-world advantage. Indeed, using DLSS 3 with Cyberpunk 2077's Psycho RT preset allowed for 60+fps at 4K, which is frankly bonkers.

To wrap things up here, we've got one last statistic locked and loaded: the RTX 4070 offers about fifteen percent better performance per dollar than the RTX 3080. Boom, mic drop, bell ring, the fight's over. The 3080 is tapping out.

• Winner: Nvidia RTX 4070

Which one should you buy?

The arrival of the RTX 4070 might just herald the end of the 3080's previous reign in the midrange 4K gaming space. Frankly, we can't think of any reason to recommend it over the 4070 - once you factor in the pricing and the awesome power of DLSS 3, it's a no-brainer.

Neither of these are what we could call cheap graphics cards, but the 4070 is currently the most affordable next-gen GPU on the market - with nothing to show from AMD in the midrange space right now. $599 might still be too steep an entry price for some, but there's no denying that this GPU gives you a ton of bang for your buck.

We will drop one small caveat in here: if you're planning to game at 1440p or below, you really might want to consider opting for the RTX 3070 or RTX 3060 Ti instead - at least, until the expected RTX 4060 arrives. With DLSS 2.0, those cards can more than handle gaming at 1440p, and won't cost you as much as the 4070.