In a move that really should have happened on Monday, Nvidia have finally released some figures regarding how much better their new Turing RTX 2080 card is compared to their current big frame hitter, the Nvidia GeForce GTX 1080. Some, meaning a handful of very specific numbers for only a select number of games, at 4K, 60Hz, with no mention of graphics quality.
Still! Numbers! And some pretty decent-ish ones too, considering the rather nebulous circumstances. Let’s take a look.
RTX 2080 vs GTX 1080: performance
Of course, without knowing what kind of graphics settings Nvidia used to obtain these figures, the following numbers and stats obviously only paint a very small picture of what the RTX 2080 is truly capable of. Full judgement will have to be reserved until review samples are available, but in the ten games Nvidia did talk about, you can probably expect to see roughly 1.5x the performance of a GTX 1080.
That’s across Epic’s uber demanding Unreal Engine 4 Infiltrator demo, Final Fantasy XV with HDR switched on, Plunkbat, ARK: Survival Evolved, JX3, Shadow of the Tomb Raider, Hitman 2 with HDR, Wolfenstein II: The New Colossus, HDR-ified Middle-earth: Shadow of War, and Mass Effect Andromeda, also with HDR – as illustrated by the graph below (click to enlarge).
You’ll probably notice that some games in that there slide have an extra light green bar on top of them. To make things extra confusing, that’s the kind of performance increase you can expect – around 2x a GTX 1080 – in games that support Turing’s rather cool Deep Learning Super Sampling tech (or DLSS).
Without getting too bogged down in the technical nitty-gritty, DLSS essentially applies the same kind of AI and deep learning know-how to rendering techniques (resulting in those crisp, smooth edges everyone’s so fond of) as Turing’s ray tracing stuff does to, well, ray tracing.
If all that sounds a bit dubious, trust me, it’s not. At an Nvidia press event today I saw DLSS in action, where Nvidia used that aforementioned Infiltrator demo to show how it compares to standard temporal anti-aliasing (TAA). With a GTX 1080Ti running on one side of the screen using TAA, and the RTX 2080Ti running with DLSS enabled on the other, the RTX 2080Ti was indeed running at twice the speed of its GTX predecessor, with only a very marginal, almost imperceptible dip in overall quality to boot.
This would appear to confirm Nvidia’s claims of double performance with DLSS (despite using the Ti version of the RTX 2080 instead of the regular one mentioned in the slide), and would probably go a long way to help justify the card’s enormous price – provided you like/end up playing the games that actually support it.
RTX 2080 vs GTX 1080: frame rate benchmarks
Of course, 1.5x or 2x the speed of a GTX 1080 means nothing without some concrete numbers. Or at least as concrete as you can get when Nvidia only release numbers for a very limited number of games at the moment (which, rather unhelpfully, aren’t quite the same games listed above) with no mention whatsoever of what graphics settings they used to record them, or indeed whether ray tracing was in effect, or DLSS was engaged.
But! If you were hoping the RTX 2080 might finally offer a single GPU solution for playing games at 4K at 60fps in HDR, you may well find these frame rates rather more hopeful than the 60fps 1080p targets Metro Exodus is aiming to hit with ray tracing switched on. Here’s what you’re looking at:
- Final Fantasy XV: 60fps
- Hitman: 73fps
- Call of Duty WWII: 93fps
- Mass Effect Andromeda: 67fps
- Star Wars Battlefront II: 65fps
- Resident Evil VII: 66fps
- F1 2017: 72fps
- Destiny 2: 66fps
- Battlefield 1: 84fps
- Far Cry 5: 71fps
To put at least one of those figures into perspective, you may remember from my Final Fantasy XV PC performance analysis that even the GTX 1080Ti struggled to run it at anything much beyond 45fps at 4K on its Highest preset, which makes the RTX 2080’s figure of 60fps seem rather impressive. That said, the GTX 1080Ti could also reach speeds as high as 55fps when I dropped the things down to Average, which would be pretty damning if Nvidia did indeed get their 60fps RTX 2080 figure using the same graphics setting. We just don’t know at this point.
Still, as an initial indicator, the RTX 2080’s performance seem promising. Perhaps not a grand’s worth of promising, all told, but I won’t be able to say one way or the other until I’ve been able put it through its proper paces. One thing you can guarantee, though, is that I’ll scrutinizing every last figure like a hawk as soon as I’m physically able to do. Stay tuned.