Using a frame cap in every game, as I do. I can't tell the difference between my 5800X3D+7900XTX and my 7800X3D+9070XT 95% of the time.
I have not tried enabling FSR 4 on RDNA 3 yet, so I can obviously tell if I use upscaling. The only other time is when I'm using RT in certain games, e.g. Hogwarts and the Spider-man games. AM4 might drop below the frame cap here and there, when AM5 does not. Memory bandwidth probably, occasionally, comes into play, as well.
Nothing that ruins the experience, or makes me think the 5800X3D is aging out. In some titles, like racing Sims, it can still outperform all vanilla AM5, last time I checked. Take a look at GN's results. Prefacing the charts: They hold Intel back with ram speed. However, given pricing and availability today? Perhaps their choice is more appropriate now than then? Not a big proponent of how they test games, but for the high-end stuff it does not matter much. Not going to have game performance and rendering, multitasking, or audio issues the way the low-end stuff can. 1s and .1s would probably better for all the CPUs though, because of testing long enough for any background game demands to even out over time.
Regardless, for gamers like myself, the multicore performance is a non-starter. So, if the cost of the Zen 3 3D is the cheapest and easiest upgrade i.e. drop-in, it is still compelling.
FFXIV
F1 24
2077