NVIDIA GeForce RTX 4080 Review: Ada Lovelace For Enthusiasts
Final Fantasy XIV Endwalker Benchmarks

Final Fantasy Endwalker


This benchmark is somewhat CPU bound at 1440p, but the familiar performance cadence we've seen up to this point continues. The GeForce RTX 4080s remain head-and-shoulders above any previous-generation card, and only the GeForce RTX 4090 is faster.
Assassin's Creed Valhalla Benchmarks


MSI GeForce RTX 4090 Suprim Liquid X Details @ 4K


Forza Horizon 5 Benchmarks
Forza Horizon 5 is the latest edition of the popular racing game franchise. The game features much larger maps than its predecessors, with more detailed models, and higher graphics fidelity in general. Forza 5 leverages the proprietary ForzaTech engine and DirectX 12...


Forza Horizon 5 didn't reveal any surprises either. In this game, the new GeForce RTX 4080 series cards once again finished in the second and third positions, trailing only the GeForce RTX 4090.
Marvel's Guardians Of The Galaxy


MSI GeForce RTX 4080 Suprim X Details @ 4K


With ray tracing enabled in this game, we fall back into familiar territory. The MSI GeForce RTX 4080 Suprim X slightly outruns NVIDIA's Founders Edition card at both resolutions, but neither can catch the more powerful RTX 4090.
Far Cry 6 Benchmarks


MSI GeForce RTX 4090 Suprim Liquid X Details @ 4K


With any ray tracing effects enabled, FarCry 6 is somewhat CPU bound when testing at 1440p. Even then, the game reports similar results to all of the rest, with the GeForce RTX 4090 ruling the roost, and the new GeForce RTX 4080 series cards outgunning all the rest.
F1 2021 & F1 2022 Benchmarks
F1 2022 is Codemaster’s latest Formula One racing simulation, and like previous version of the game, it sports impressive visuals. This latest addition to the franchise supports DirectX 12 with ray tracing, and it incorporates support for a number of AMD's FidelityFX Super Resolution (FSR) and NVIDIA technologies (like DLSS). We tested the games with its Ultra High graphics preset, with ray-tracing and TAA enabled at a couple of resolutions to see what the graphics cards could do.
F1 2021 / F1 2022


In the newer, more taxing F1 2022, the framerates are somewhat lower, but we still see more of the same. The GeForce RTX 4090 retains its overall lead, with the RTX 4080 cards outperforming all of the rest.
DLSS 3 And Latency Testing With The MSI Suprim X
We also tested F1 2022 and a couple of additional games—namely A Plague Tale: Requiem and Fortnite—with the various DLSS 3 modes available. Please note, all of these games were also tested with beta builds, updated to support DLSS 3 with resolution scaling and AI Frame Generation options.

A Plague Tale: Requiem is more CPU bound than F1 2022, so we saw somewhat larger performance gains with AI Frame Generation enabled. Still, the performance trend doesn't change. The GeForce RTX 4080 is clearly faster than the RTX 3090 Ti.

One this to consider with AI Frame Generation at the performance levels offered by the GeForce RTX 4080 is that overall latency may be impacted. When resolution scaling alone, the various DLSS mode ultimately improve latency. With AI Frame Generation enabled, however, only Performance mode offered better latency than the native resolution. Something to keep in mind with AI Frame Generation is that the game isn't responsive to input when the generated frames are inserted in the animation. As such, some latency is introduced.



Finally, we did some DLSS and Latency testing with Fortnite. Please note while looking at these three charts above, that the scales are different between the Ampere-based RTX 3090 Ti and the Ada-based RTX 40 series cards. Framerates are off to to the right, and latency at the various modes are to the left. As you can see, the GeForce RTX 4080 and 4090 cards once again clearly outrun the GeForce RTX 3090 Ti with much better 1% low and average framerates. And the various DLSS modes only increase performance over native. Despite the additional processing going on with DLSS, the increased framerates also end up reducing overall latency.