Equally amusing that they gloss over the fact that internal resolution for NMS Series X version bottoms out at 1080p vs 1296 for base PS5. I guess they are no longer bewildered by the console with the bigger GPU being outperformed by the "inferior" gpu in the PS5.
As the PR goes (and DF has mentioned that before too) that advantage 'only' applies to 'last-gen' centric renderers and rasterization focused games.
That's why especially RT path is faster with UE5 titles like The First Descendant on the PS5... oh wait...
Anyway - the hypothetical RAM discussions are always fun.
4GB PS5 would have gotten a lot more flack than the damage control XBox got for the first 2 years. It's a much easier mouthpiece to criticise textures without even looking at media, we've had entire generations of that already.
For me the more interesting question would be what would a 256MB 360 turn out like. Optionally against 512MB PS3 we got.
But much better question - how would that same 360 stack against the
192MB PS3 that was also supposed to launch in 2005 (yes that's not a typo - but
all 192 would be eDram - so the bandwidth disparity would have easily been 10:1, or worse). And had both HD consoles launched in 2005 - what would that mean for the Wii...
The 8MB GS PS2 would also be an interesting thing to see (worse hw-shortages and more money bleeding for Sony in year 1 - but then they did far worse for the PS3, so it's not even that unfathomable).
I love how Alex is finally acknowledging that it is primarily the new BVH8 engines and possibly other unnanounced hw allowing PS5 Pro to reach 2-4x RT performance. I'm sure he'd like everyone to forget that he proposed the additional CUs alone were responsible for the RT boost.
It's a bit of a mish-mash - he's careful enough to acknowledge that feature list is not contextualized enough to infer much from it (unless you know the terms explicitly - and he admits not to).
But on the flipside he's dancing around with the assumption that 2-4x faster RT refers to
a) relative to PS5
b) a subset of RT pipeline
Now - both 'might' be true - but the leaks don't confirm or deny any of it. Which raises the question if he's referencing things he's seen on DevNet (Richard is inferring they may have seen the contents themselves - which in of itself is a bit touchy thing to go on record with) - or he's just taking most convenient guess.
Wording as has been leaked to me at least negates b) - I'd read it as 'RT end-2-end' is X times faster - but I'm not pretending I've had access to the above.
For a) - harder to say, I've seen examples of both in PS4 Pro documentation - where we explicitly had statement to 'pipeline X is not Y times faster than before - and that was NOT relative to running on base PS4' and then of course there were statements on 'X times faster than PS4' as well for some elements. So