Had to recalculate some things: so basically, gauging the percentage delta between PS5 and XSX depends on which way you work it, i.e if taking PS5's figure to scale up to XSX, or taking XSX's figure and scaling down to PS5, because that affects the percentage.
For example, if you take PS5 figure as baseline, the difference in TF from the peak PS5 TF and XSX is about 18.2%. However, if you take XSX baseline and scale down to PS5 peak TF, the difference is closer to 15.5%. Same logic, if you take the 2% frequency reduction Cerny mentions, that puts PS5 to about 10.07 TF. Scaling with PS5 as base to XSX, that gives you about 20.6%, but if you scale from the inverse it comes out to about 17.5%
All would be right but, IMO, you scale up from the lower figure to the higher one, rather than the inverse, because the whole point of that type of comparison is to illustrate how much more the larger figure is over the smaller one so the smaller figure would be the reference point. So in terms of PS5 and XSX TF delta the correct percentages are
18.2% - 20.6%, and that's accounting for if PS5's GPU doesn't have a reduction in frequency beyond 2% for power load-demanding scenarios (which we won't know for sure until game footage running on hardware actually comes out and can be analyzed minutely).
Which is how you get the XBO < PS4 delta percentage for example (so apologies to Vawn because technically they were correct stating 40% and I was wrong stating 35%).
Ehh... I'm not going to jump into speculation, but either way, PS4 Pro @ 4.2 TFLOPS vs Xbox One X at 6 TFLOPS is a 43% power difference as measured in TFLOPS, and no one really gives a shit about the differences we see outside of pixel peepers and console warriors. Now, if we start seeing differences outside of resolution and some post processing, like 30 fps vs. 60 fps, then I think people might start to care.
TBF, PS4 Pro and Xbox One X were held back by the respective base systems. If we saw games actually developed with the mid-gen refreshes in mind exclusively, differences would've been more pronounced, though maybe not to the degree of certain base PS4 and XBO titles were.
But I do agree that TFs aren't the only metric to be looking at these console through; it's about a ton of other things in addition to that. Even when it comes to TFs, just seeing them only for visual/resolution purposes is limiting the scope of what GPUs will provide next gen in areas such as GPGPU asynchronous compute.