It’s great being a tech reviewer. I get to play with shiny new hardware before Joe Bloggs gets his hands on it. There’s something warm and fuzzy about a GeForce RTX 5090 sitting on my desk… and it’s not incontinence, despite my advancing years. Yet even I need to take a step back and look at launches through the lens of people laying down their hard-earned cash. I’d hazard most readers won’t upgrade on a generational basis. Instead, they look on with envious eyes and twitching wallets when comparing their card’s performance to the latest and greatest.
This is why examining Nvidia GeForce RTX 5080 performance against immediate peers is technically useful but misses the point for millions of gamers. They upgrade when their present hardware no longer keeps up with demands, usually on a three-to-five-year cadence. Nowhere is this truer than for graphics card.
Looking back through recent history, Nvidia’s GeForce RTX 3080 was a great card from 2020. Launched at $699 to almost universal acclaim, I wondered at the time why it was so cheap. Even now, it’s a solid FHD and QHD performer, though the rise of ray tracing and DLSS magnifies what’s lacking compared to the best of today.
That’s why I’m comparing freshly minted RTX 5080 against RTX 3080. I fully expect the new Blackwell card to stomp over its nomenclature namesake, but is it worth upgrading from one to the other across a bevy of games and applications? That’s a good question and one I’m answering in this guide.
Let me start you off with the Club386 Table of Doom™.
| RTX 5080 | RTX 3080 | Ratio | |
|---|---|---|---|
| Released | Jan 2025 | Sep 2020 | – |
| Codename | Blackwell | Ampere | – |
| GPU | GB203 | GA102 | – |
| Process | TSMC 4N | Samsung 8N | |
| Transistors bn | 45.6 | 28.3 | 1.61 |
| Die size mm² | 378 | 628 | 0.60 |
| CUDA cores | 10,752 | 8,704 | 1.24 |
| Base clock MHz | 2,295 | 1,440 | 1.59 |
| Boost clock MHz | 2,617 | 1,710 | 1.53 |
| FP32 Boost TFLOPS | 56.3 | 29.8 | 1.89 |
| SM count | 84 | 68 | 1.24 |
| RT cores | 84 (4th Gen) | 68 (2nd Gen) | 1.24 |
| RT TFLOPS | 170.6 | 58 | 2.94 |
| Tensor cores | 336 (5th Gen) | 272 (3rd Gen) | 1.24 |
| FP16 Acc TFLOPS | 450.2 | 238.2 | 1.89 |
| FP4 TOPS | 1,801 | – | – |
| ROPS | 112 | 96 | 1.17 |
| Memory GB | 16 | 10 | 1.60 |
| Memory type | GDDR7 | GDDR6X | – |
| Mem. clock Gb/s | 30 | 19 | 1.58 |
| Mem. interface bits | 256 | 320 | 0.80 |
| Mem. bandwidth GB/s | 960 | 760 | 1.26 |
| Board power watts | 360 | 320 | 1.13 |
| Launch MSRP $ | 999 | 699 | 1.43 |
You’re looking at architectures almost 4.5 years apart. Advances in manufacturing process – 8nm to 4nm – enable RTX 5080 to shoehorn 61% more transistors into a die that’s substantially smaller. Aside from behind-the-scenes improvements emanating from architecture, today’s champion x80 card carries almost 90% more FP32 throughput.
It’s abundantly clear to me Nvidia focusses on increasing ray tracing performance by a larger degree. Numbers suggest an almost 3x jump, while FP4 support is a future weapon for speedy AI calculations that are becoming more pervasive across games and applications. RTX 3080, sadly, is well behind in this regard.
There’s also 60% more memory and it’s clocked in almost 60% higher, but I do wonder if Nvidia ought to shift to 20GB or even 24GB in this segment. Memory bus width is one area where RTX 5080 lags – 256 bits vs. 320 bits – and I imagine that’s a move designed to save die space and power.
Shifting down a couple of process nodes usually avails efficiency benefits. Nvidia uses them to ramp up clocks on most meaningful fronts, with the end result being RTX 5080 actually chews through more total board power than RTX 3080. Putting my conjecturing hat on, an RTX 5080-like GPU on the Samsung 8N process would run at over 1,000mm² – making it technically unfeasible with modern reticule limits – and consume 750W. There’s good reason why it didn’t exist back in 2020.
Then there’s price. Nvidia feels absolutely confident in charging 43% more for a card carrying the same class numbering. Part of the reason is to do with less competition from AMD, another is costlier silicon on cutting-edge nodes. Nevertheless, I imagine Nvidia is making a healthy dollop of margin.
All said and done, RTX 5080 is a much better fit for the today’s games that rely on ray tracing and DLSS to look and run at their best. Let me show you by how much.
Performance
RTX 5080 and RTX 3080 duly take centre stage in the latest Club386 platforms. Run with exactly the same supporting hardware, it’s an apples-to-apples comparison.
Our 7950X3D Test PCs
Club386 carefully chooses each component in a test bench to best suit the review at hand. When you view our benchmarks, you’re not just getting an opinion, but the results of rigorous testing carried out using hardware we trust.
Shop Club386 test platform components:
CPU: AMD Ryzen 9 7950X3D
Motherboard: MSI MEG X670E ACE
Cooler: Arctic Liquid Freezer III 420 A-RGB
Memory: 64GB Kingston Fury Beast DDR5
Storage: 2TB WD_Black SN850X NVMe SSD
PSU: be quiet! Dark Power Pro 13 1,300W
Chassis: Fractal Design Torrent Grey
Application and AI
Featuring advanced ray tracing and pushing new cards to their limits, 3DMark Speed Way is a good test to get the ball rolling. RTX 5080’s improved hardware nets an 87% improvement.
Steel Nomad, on the other hand, is better at evaluating pure rasterised performance. I expect older GPUs to do better here as their architectures are primed for FP32 core throughput.
There’s little in it between the two tests, actually, as RTX 5080’s lead reduces from 87% in Speed Way to 85% here.
Blender just loves muscular GPU architectures running at high speeds. A project that’ll take two hours rendering time on RTX 3080 takes one hour on RTX 5080.
Don’t chuck RTX 3080 in the bin just yet; it’s good at simple AI workloads.
Furthermore, RTX 3080 is still potent enough to run a complex large language model locally. 13-billion parameter Llama 3.1 produces near-instant results on a ChatGPT-like application.
Gaming
Deliberately run with maximum image-quality, Assassin’s Creed Mirage is a good proxy for rasterised performance in well-tuned games. Truth is GeForce RTX 3080 returns 4K60 comfortably enough, so I wouldn’t rush out to buy a new card if games like it are your cup of tea.
There’s life in the old dog yet. 4K75 and QHD130 are fine returns for a $649 card from 2020.
Remember I said that games which exact a heavy ray tracing toll expose older architectures mercilessly. This is exactly what happens in Forza Motorsport, where RTX 3080’s 10GB framebuffer and comparatively weak RT performance combine to slow things down. In this regard, RTX 5080 looks like a great upgrade.
Lessing the load again, Mount and Blade II: Bannerlord loves older cards. GeForce RTX 3080 says 4K100, thank you very much.
Eerily similar numbers to Mount and Blade II, Tom Clancy’s Rainbow Six Extraction is easy to run on older cards. Another reason not to upgrade for the sake of it.
Looking to the future, I’m adamant you’ll see games adopt technologies such as path tracing and go big on ray tracing in general. Cyberpunk 2077 Ray Tracing Overdrive preset provides a glimpse of that future.
Both cards are hammered by the extreme load, making the game only playable at FHD on RTX 5080. The older GPU’s 10GB framebuffer gives up the ghost at 4K, resulting in a slideshow returning an average 1fps. Yup, you read that right.
DLSS
Matters improve with DLSS Quality thrown in the mix, though do be aware the game’s not smooth at 4K because the base frame rate the technology is working from is so poor.
Using FSR3 for RTX 3080 helps but it can’t rival RTX 5080’s MFG. Put simply, the Blackwell architecture is built for these workloads.
| Game | 3080 to 5080 % uplift at 4K |
|---|---|
| Assassin’s Creed Mirage | 64.2 |
| Final Fantasy XIV Dawntrail | 49.3 |
| Forza Motorsport | 136 |
| Mount and Blade | 69.6 |
| Rainbow Six Extraction | 66.7 |
| Cyberpunk 2077 DLSS + FG | 303 |
Putting it all together, here’s what you get at 4K. The average is easily skewed by including frame generation or huge amounts of ray tracing. I expect RTX 5080 to widen its lead as newer games come to the fore.
If you’ve found this comparison useful, please do read the following head-to-heads.
GeForce RTX 5080 vs. RTX 5090
GeForce RTX 5080 vs. RTX 4080 Super
GeForce RTX 5080 vs. AMD Radeon RX 7900 XTX
Vitals
A stark difference in performance is met by the same observed system-wide power consumption.
And while both Founders Edition cards use dual-slot coolers, RTX 5080 is far more effective at keeping the GPU cool.
Neither kicks up a fuss when under 10-minute gaming load.
Conclusion
Congratulations to those of you who purchased a GeForce RTX 3080 in 2020. The $699+ graphics card has stood the test of time and come out in 2025 as a great performer at anything other than ludicrous settings installed at a 4K resolution.
Yet time is a great exposer. The latest games exact a large toll on RTX 3080’s ray tracing capabilities, while DLSS upscaling is not compatible with the latest generation. This is not a criticism, you understand, but more an observation.
GeForce RTX 5080 embodies the older card’s ambitions by enhancing every facet that matters. Expectedly better in every metric, there’s particular focus on ray tracing and DLSS ability, with multi frame generation, available on 75 games and apps on launch, proving innovation comes in many forms.
Priced from $999, I don’t feel RTX 5080 is quite the bargain RTX 3080 10GB was back in the day – that remains one of my favourite GPUs. Nevertheless, should you feel the need to scratch the frame-rate itch, upgrading from one to another is no bad idea, especially if your games align well with Nvidia’s latest technologies.
GeForce RTX 5080
“Gear up for game-changing experiences with the Nvidia GeForce RTX 5080 and AI-powered DLSS 4.” Read our review.
GeForce RTX 3080
“The GeForce RTX 3080 graphics card delivers the performance that gamers crave, powered by Ampere—Nvidia’s 2nd gen RTX architecture.
