I don't understand though, apart from a power difference, aren't these two consoles very similar and based on the same architecture? Why should one be more easy than the other to work on?
The hardware may be quite related but there could be other things at play:
1. The SDK or 'Software Dev Kit' which developers use in order to build their games on this hardware. It could be that one company's software is easier to use and also 'lower level' which means that they can achieve a certain level of performance against that hardware, and it might be easier vs the competition. To put it in different terms, let's say that Box A can push
20TF but Box B can only push
17TF. Box A is harder to work with,
and the APIs/SDK aren't quite as performant as Box B. So it takes the developer more time to work on and ultimately they get, let's say, on average
80% hardware utilization on their game.
That's
16TF of performance.
Box B, while being weaker, is way easier to use and its APIs are much more streamlined. That box is able to hit
95% hardware utilization on average for the game. That's
16.15TF of performance.
So in this case the weaker box, because it is easier to develop on and has thinner APIs, has more real world performance than the stronger box. The other box, however, has cost the developers a lot more head ache and ultimately might be struggling to reach performance targets.
There's also the reality that they may be forced to make sure their game is running acceptably on a weaker iteration of Box A designed to target lower resolutions. This problem may be because the hardware of this weaker version of A isn't quite up to snuff to do an easy downport or it might be because the developers weren't given an SDK specifically to target that box. Instead, they just have the regular Box A SDK with a mode or setting to see how their game works on this lower spec machine. If the implementation of this mode/setting is subpar then that could be giving them a headache where optimization is concerned.
Look, we both know what I'm saying with 'Box A' and 'Box B'. But I'm not trying to push a narrative that this is fact, its just possible based on
some (not all! This is a two way street) of the rumors and grumblings we've been hearing.
At the end of the day, Box A could have APIs just as robust as Box B and still retain the benefits of being the stronger console.
Full disclosure: I fully expect
Box A to have stronger real world performance than
Box B. I'm less certain this real world performance will be in every use case though, and that the gap is as clear cut as what we see on paper.
2. Box B has more dedicated silicon to offloading tasks (I.O., Audio, cache scrubbing, whatever) so the developer's pool of resources to create the game on is basically uncontested by these tasks. So they may have proportionately more to work with and less to worry about. It's 'invisible' to them, no toil necessary. The other box might not offload as much so they have to spend more time ironing out idiosyncrasies. Although I wouldn't expect this to be a huge shift for them, as the last generation surely offloaded a far lesser amount of work then this current gen on either system.
So there are various reasons why there could be problems on one machine vs the other despite sharing the same DNA of hardware.