What's the current thinking on what PS4K's silicon will be? This thread is moving so fast it's difficult to keep up.
I see three possibilities:
1) Polaris. This seems the least likely, as, at least to some extent, development would need to be adapted to a moderately different architecture. I doubt Sony nor developers would find that palatable. It would also increase the difficulty of ensuring all PS4 titles are fully compatible.
2) Shrink the current PS4 APU down to a 14/16 nm process, stuff in more CPU cores and compute units, and potentially upclock. This seems pretty likely - it's probably the simplest way to obtain a significant boost in power while requiring the least amount of coding changes. Going forward, developers can code for both OG and 4K consoles in accordance with the way they target low, medium, and high quality presets for PC games.
3) '2x' or SLI. This is where I get confused; I would have figured (2) would have been far likelier but we've now got insiders implying some sort of a dual chip configuration. Zoetis specifically mentioned SLI. Is the logic that Sony would shrink the current PS4 APU down to 14/16 nm, and then stuff a second shrunken APU in the box? This just doesn't make a great deal of sense to me. Option (2) would afford a similar increase in power without the fuss of connecting and coordinating operation of two APUs on separate packages. Then again, this wouldn't require AMD to go to the trouble of redesigning the APU with ~2x power on a 14/16 nm process as would be the case with option (2).
Regardless, we're pretty confident the PS4K silicon will be on a 14/16 nm process, right?