It really depends on what the direction the industry is going, whats going to happen to mega franchises like fortnite, minecraft, etc are they going have "next gen versions" built from the ground up?
When ppl talk about holding the gen back with a weaker base of lockhart, they are talking about some PS5/ANACONDA game with some crazy physics simulation that requires 10tflops @ just 1080p.
But it's a really dumb proposition because whoes going to want to do that? It's like Somone this gen doing the same on x1/ps4 but at 480p.
And I don't know what physics simulation a dev could not do at 4k with zen2 that would be inferior to 1080p one?
If you're talking about GPU sim, there's stuff you could do that you wouldn't necessarily want to do on a Zen2 alone, and that would be feasible to do in a gameplay affecting way on console because of the single memory pool. I wouldn't say too many devs would be wanting to do 1080p games on Anaconda/PS5, but how about 1440p or other resolutions reconstructed to 4K? I think that's quite possible, and quite acceptable, if the dev wants to leave headroom for other non-graphics stuff on the GPU, or, to bump the per pixel complexity. I don't think 4K native is necessarily how every game should be going on Anaconda/PS5 - certainly I'd prefer devs to have the option to explore.
For purely graphics stuff - higher per pixel complexity at non-4K resolutions - of course, with Lockhart, there's nothing stopping devs still doing that on Anaconda, and having a separate render path on Lockhart if it doesn't scale down to a decent enough resolution there. But there may be an aversion to doing anything too fundamentally different per-pixel between the SKUs, or a lean toward sticking to simplicity. It's a complication and it's hard to predict in every case that the dev would do with Anaconda what they would have otherwise done if it was the only SKU to worry about.
