Raytracing and DLSS
Let’s put it this way, above the Full-HD resolution you can forget raytracing without DLSS on all cards, because it’s not playable anymore. Unless you use DLSS. Then the GeForce RTX 3090’s 22 FPS suddenly turns into 40 FPS overnight at the highest level of “Quality” and even 80 FPS++ at “Ultra Performance”. Whereby even the whole thing looks even crisper with about 55 FPS and “performance” and runs much faster than WQHD without DLSS. As long as you don’t rush to turn around your own axis, because then every now and then crumb effects and artifacts will appear. Only this is the exception.
I myself almost always have my 50 FPS with a MSI GeForce RTX 3090 Suprim X, raytracing on “Psycho” in Ultra-HD and DLSS on “Balanced”, whereby with “Performance” you can also reach almost 60 FPS. Visually, this makes almost no difference and in the end is subjective perception. But I prefer to play native on Ultra-HD and take light crumbs when moving, rather than controlling the monitor in WQHD and upscaling it to a much more muddy state than Ultra-HD with DLSS doping. The normal VGA test system from the last launches with the overclocked Ryzen 9 5950X serves as test system, the AMD cards of the newest generation run with SAM on.
Fact is, DLSS looks even better in places than native Ultra-HD and you can/must state without envy that AMD really needs something similar to be able to keep up with this kind of radiation enhancement. Pure rasterization on “Ultra” looks good, no question, but with raytracing on at least “Ultra” and DLSS from “Performance” and better, everything is even more upbeat and real. However, an RTX 3070 is the minimum requirement here, even if the 8 GB VRAM can already become jam-packed. The GeForce RTX 3060 Ti is here with a little over 30 FPS and DLSS performance” a failure, this is only possible with “Ultra-Performance”, but both it and the RTX 3070 are no Ultra-HD cards anyway. Therefore a shorter benchmark list, which also deliberately dispenses with the older cards. It’s not going to be anything half or full.
Rasterization benchmarks from Full-HD to Ultra-HD
Let’s now leave out raytracing along with DLSS and also the AMD cards at the graphics counter. So we return to the old school style. In “Ultra” this is optically still a fat treat (but without the appetizing blob of irradiated cream on top), which is why I chose this setting. With Full-HD this can be implemented and played quite well for most of the current and somewhat older cards, because it is not a first-person shooter. For a full 60 FPS it should be at least a GeForce RTX 2070 Super, with the Radeon RX 5700 XT you come closest to that with AMD cards.
If one switches the resolution higher, then of course some cards fall through the grid. But even below the “ultra-setting” there are still good chances of survival. But it continues what I already noticed with Full-HD: the significantly worse frame times of the new RX 6000 series compared to the comparable GeForce cards and also the older Radeon generation. Driver? Let’s wait and see.
And Ultra HD? Now it’s going to be really close for most cards, really close. Interestingly, the Radeon RX 6900 XT doesn’t even have the 16 GB VRAM to keep the GeForce RTX 3080 in check, on the contrary. One suddenly falls back even more clearly here. The cache prediction does not yet seem to harmonize optimally.
The implementation by studio and engine have been delivered in an extremely exemplary manner, even with the graphics, Chapeau! As CPU you should plan at least a 6-core, for the cards of the newest generation you may use Intel from the 9th generation on. Everything else will be a drag on the fast cards, at least in Full-HD. Besides, the game really has to be installed on a fast SSD, otherwise loading and reloading becomes a torture.
And finally, because it was really beautiful, a few more impressive pictures:
Infected? I could certainly understand that. Just buy it? More like an inner compulsion… Do it, yes! With or without DLSS. It’s crazy.
Kommentieren