Digital Foundry: Control PS5 vs Xbox Series

[Pemalite]

SvennoJ said:

DF added there photo mode observations.

https://www.eurogamer.net/articles/digitalfoundry-2021-control-ultimate-edition-photo-mode-rt-benchmark

Running unlocked does show varying degrees of overhead and based on over 20 matched scenarios, Series X does have a rendering advantage, which on average delivers a 16 per cent lead over PlayStation 5. We go through those scenarios in the embedded video on this page and the variation from test to test is significant - so we should emphasise that the 16 per cent figure is indeed a mean average. Some tests show an even wider margin, others see the situation close up significantly.

The rest of the analysis is only in video form, I haven't watched it.

That the Series X has a faster GPU is not surprising. (16% faster in raw flops) However with everything frozen, shouldn't the difference always be the same percentage wise. Perhaps the video reveals why some scenes work better on either console.

Clearly the Xbox Series X is the faster machine, it's got the better specifications.
Trying to correlate that with FLOPS however isn't going to get us very far.

But depending on the scene, the bottleneck may exist on another part of the chip like the CPU, hence why in that 30~ fps scene the difference between the Xbox and Playstation was essentially inconsequential. - A few hundred megaherts on a multi-gigahert chip is more or less a... Who gives a shit.
Doesn't matter on the PC, ain't going to matter on console.

shikamaru317 said:

Flops only give you a general idea of performance, the GPU's in PS5 and Series X are quite different. Though Series X has a flops advantage, PS5's GPU is actually faster in terms of pixel fillrate and triangle fillrate, while the Series X GPU wins out in terms of texel fillrate, ray-tracing fillrate, video RAM speed, and Bus width. PS5's GPU has less cores at a higher clock speed while Series X has more cores at a lower speed (some engines favor clock speed while others favor more cores). Different scenes and sequences even with the same game would favor each GPU's strengths differently, leading to the variation. There is also the variable nature of PS5's GPU to take into account, it doesn't run at it's boost clock speed all of the time, whereas the Series X GPU has a locked clock speed.

Flops tells us absolutely nothing of performance.
It's theoretical, not real world.

Pixel Fillrate can be just as dubious as using FLOPS to gauge capabilities... It's a theoretical unit of measurement (Rops*Core Clock), not a real-world one. (As it doesn't take into account the effect other parts of the GPU have on it such as culling.)

Pixel Fillrate can sometimes be bottlenecked by memory bandwidth, in those instances, the Xbox Series X can potentially take the lead there.
Plus there is a shift happening towards hardware Ray Traced shadows, lighting and reflections right now... And that changes things up even more.

Lets take Shadow Volume Rendering, which is very Pixel-Fillrate demanding, with next-gen engines and games, that is likely to be put on the backburner as we start using Ray Traced Shadows, meaning that pixel fillrate just became less important in shadow-heavy games.