We've completely revised our test bench to better reflect the future of gaming technology - that means we're zeroing in on titles using key engines and low-level gaming APIs, while we've beefed up representation for ray tracing and image reconstruction. Some of our line-up may need tweaks as the results come in, but right now, we're mostly happy with this more forward-looking approach.
Ray tracing is no longer a second class citizen and with the new wave of GPUs, we're seeing some of the most intensive GPU workloads you can get delivered at really decent frame-rates - even before we factor in image reconstruction technologies like DLSS.
Our benchmarking system below offers a number of ways to get to the data you want, the presentation varying according to the device you're looking at right now. You'll get a basic overview of our findings on mobile, with metadata from the video capture of each GPU being translated into simple bar charts with average frame-rate and lowest one per cent measurements for easy comparisons.
On a desktop-class browser, you'll get the full-fat DF experience with embedded YouTube videos of each test scene and live performance metrics. Play the video, and you'll see exactly how each card handled the scene as it progresses. Below the real-time metrics is a bar chart, which you can mouse over to see different measurements and click to switch between actual frame-rates and percentage differences. All the data here is derived from video captured directly from each GPU, ensuring an accurate replay of real performance.
This is Cyberpunk 2077 in its ultra-high RT mode - the full monty, short of the 'psycho' ray tracing setting but still a hefty workout that brings RTX 3080 and all tested AMD cards to their knees in 4K mode. The issue here with the RTX 3080 trending under the 3080 Ti is seemingly down to lack of VRAM. This shouldn't be happening by the way, but testing and re-testing confirmed here that RTX 3080 Ti is marginally ahead of RTX 3090.
The RTX 4080 powers ahead, beating the similarly priced 3080 Ti by 37 per cent at 4K. Ideally you'd be using DLSS here, and we've got you covered there on page four! First takeaway though? Perhaps disappointingly, the RTX 4080 is closer to the 3090 Ti than it is to the 4090. The RTX 4090 is almost 41 percent faster than the 4080, yet costs 33.3 percent more based on US MSRPs.
CYBERPUNK 2077, ULTRA RT, DX12, TAA
Dying Light 2
The trend of RTX 4090 offering performance uplifts higher than the difference in sticker price persists here in our second demanding RT test: Dying Light 2. This time that 33.3 percent increase in cost is delivering almost 45 percent of extra performance. Ideally, the price/performance ratio should be increasing on the cheaper parts, not increasing.
A 35 to 36 percent increase in performance up against the RTX 3080 Ti is certainly pretty impressive, and this rises to 53 percent up against the card's notional predecessor - the RTX 3080 10GB. But here's the thing, you're paying an extra 71.5 percent for that 53 percentage point boost. This is the primary issue facing the RTX 4080: performance doesn't rise in line with its MSRP based on the notional last-gen equivalent, while the more expensive card is once again offering a higher price/performance ratio.
DYING LIGHT 2 ULTRA RT, TAA, DX12
The pricing issues facing the RTX 4080 abate just a touch in Codemasters F1 22, here benched at maximum settings with the full range of ray tracing features enabled. Here, RTX 4090 costs 33.3 percent more and delivers 37 percent more performance, but again, you're paying a lot more money vs the RTX 3080 than you're getting in terms of a frame-rate upgrade.
Cost vs performance only makes sense if you completely disregard the RTX 3080 and look at comparisons with the higher end GPUs based on the GA102 processor - 3080 Ti, 3090, 3090 Ti. The 3090 class cards could be considered halo products that command their own price-points, but that does not apply to the RTX 3080 Ti which should have offered more. Again, it's only compared against RTX 3080 Ti that the RTX 4080 makes sense and that's not a good look.
F1 22, ULTRA RT, DX12, TAA+FSR SHARPENING
The only benchmarking run we've retained from our old benchmarking suite is Remedy's Control, maxed out this time at 4K resolution. Perhaps it's some kind of driver issue, but this is another title where despite testing and re-testing, the RTX 3080 Ti delivered essentially identical results to the RTX 3090. It's also the most challenging RT workload we've seen yet in terms of output performance from the RTX 4080 stacked up against its rivals. Again, we re-tested this one to be sure, but results fall short of expectations.
We've raised a Roger Moore-esque eyebrow at price vs performance up against RTX 4090 in prior tests, but this time the result is outright alarming, with the flagship card offering an increase of 48 to 49 percent in performance terms up against the new RTX 4080. Meanwhile, the new card only delivers a circa 40 percentage point boost up against its immediate predecessor, RTX 3080. The first rule of benchmarking is that if something doesn't look right, typically there's an error in testing - but once again, we double-checked this one.
CONTROL, HIGH, HIGH RT, DX12, TAA
Nvidia GeForce RTX 4080 analysis
- Introduction, hardware and power analysis
- RT benchmarks: Dying Light 2, Cyberpunk 2077, Control, F1 22 [This Page]
- RT benchmarks: Hitman 3, Metro Exodus Enhanced Edition, Marvel's Spider-Man Remastered
- RT/DLSS vs FSR2 benchmarks: Cyberpunk 2077, Dying Light 2, Marvel's Spider-Man Remastered
- Game benchmarks: Control, Cyberpunk 2077, Doom Eternal
- Game benchmarks: F1 22, Gears 5, Hitman 3
- Game benchmarks: Forza Horizon, Red Dead Redemption 2, Shadow of the Tomb Raider
- Nvidia GeForce RTX 4080: the Digital Foundry verdict
Digital Foundry specialises in technical analysis of gaming hardware and software, using state-of-the-art capture systems and bespoke software to show you how well games and hardware run, visualising precisely what they're capable of. In order to show you what 4K gaming actually looks like we needed to build our own platform to supply high quality 4K video for offline viewing. So we did.
Our videos are multi-gigabyte files and we've chosen a high quality provider to ensure fast downloads. However, that bandwidth isn't free and so we charge a small monthly subscription fee of £4.50. We think it's a small price to pay for unlimited access to top-tier quality encodes of our content. Thank you.Support Digital Foundry