In this game at 1440p, the PS5 outperforms the 6800 at 1440p, which is a first. Relative to AMD, this is probably one of its best results. NxGamer didn't say by how much though but the average run seems comparable with the 6800 peaking higher and reaching the 120fps cap at its highest and the PS5 falling slightly short at 116fps.
Interestingly, however, NVIDIA GPUs beat the brakes off their AMD (more specifically RDNA2) counterparts in this game. So much for AMD hardware being easier to work on because the PS5 also has an AMD GPU.
The 3060 Ti is about as fast or faster than a 6800 at 1440p. I don't think I've ever seen that outside of ray tracing. The 6800 is generally 30% faster than the 3060 Ti, so massive underperformance by AMD there. NxGamer also notes that his 2070 struggles enormously even at 1440p, possibly due to a lack of VRAM, but from what I've heard, the game is quite light on VRAM. I suspect his 2700X he paired with the 2070 has much more to do with the abysmal performance as even the 3600 just crumbles running this game. The PS5 here, at least compared to AMD, performs like a 6800 which is usually around 45-50% faster, so that's pretty shocking. Compared to NVIDIA, the gains are more modest and more in the line of 20%, a tad below TLOU if I'm not mistaken.
Seems to be closer to a 3060 Ti, which I speculated a few days ago. It appears to be slightly behind a 2080 Ti because the 2080 Ti is 10-15% faster than the 6800 in this game. As is the 3070.
Edit: Never mind this part. It shows the PS5 is 22% faster over the course of the run. It’s on par with a 3070 Ti then, so very close to the results of TLOU for NVIDIA and much much better against RDNA2.
Optimizing the settings to be closer to PS5 Performance Mode also improves the performance by around 34% from Ultra. Add around 34% to those benchmarks above running Ultra settings at 1440p and you should get an approximation of how they stack up to PS5's Performance Mode. The HU benchmark deliberately went for the heaviest area Steve could find, so do keep that in mind.
It's also worth noting that AMD has made driver improvements that enormously improve frame pacing and 1% lows.
Here, the average fps increased by just 7% but the 1% lows almost doubled. The frame time graph also doesn't look like a saw anymore. Notice the higher GPU usage as well.
Coupled with the ongoing tessellation issues, missing effects at launch, weird Zen 2 performance, and lack of cubemap ray-traced reflections, this game still needs some time in the oven. It will be interesting to see what they can accomplish with a few more patches. I will say that it is disheartening to get yet another PC game that came out undercooked.
I must have missed that I had no clue it used any type of RT but apparently it's super heavy over 4 ms I haven't played anything this generation at 30 fps so I guess I wouldn't know lol
Would be interesting to know how fast NVIDIA cards can accomplish this using RTX/DXR. 4ms is quite a lot, especially for a 60fps target, but they have much faster RT hardware than the PS5's GPU at the expense of a heavier APU. I'm guessing the 4090 could do it in something like 1ms whereas a 3060 Ti might take closer to 3ms.