Someone asked John if its true the PS5 version runs better (on avg better frame rates, less frame tearing)

John Linneman: "Yes, this is true. I'm not working on that vid, though"

^ that was his reply on twitter.

Edit: Found it:

Yes, this is true. I’m not working on that vid, though.

— John Linneman @dark1x.bsky.social (@dark1x) November 13, 2020

here is the proof:

PS5 spends 98,1% of the time at 16.67 ms or under.
XSX spends 88,92% of the time at 16,67 ms or under.

PS5 keeps 60 fps alot more often than the XSX in this game.

Found this interesting from june. Looks like What devs where claiming about ps5 being easy for development  true after all.

I mentioned one aspect of the system design choices being different. Imagine when people have more time to really optimize their code/data and hit these caches more often. https://t.co/AIpTLZHu0Q https://t.co/6PJJfhpWMc

— Matt Hargett (@syke) November 11, 2020

That is certainly one dimension of reducing inter-fabric latencies. cache latencies/invalidation and inter-core bandwidth are primary problems to solve for 120fps gaming, more so than raw RAM bandwidth IMO. latency on going out to RAM is a killer when your frame time is 8ms.

— Matt Hargett (@syke) June 30, 2020

That's it, someone make Mark Cerny ceo of Sony.

Because hes good at makeing hardware design choices?
Not sure if that actually makes for a good CEO.

But yes, hes able to make a chip thats 20-25% smaller than the Xbox Series X one, perform almost as well as it.

The cost saveings of that, probably allowed for other things, like more expensive Dual Sense controller.

That said, Mark Cerny is someone you can have faith in when it comes to hardware solutions, and software solutions.
The backwards compatability solution was his baby too (hes why we can play more or less all old PS4 games on the PS5).
As a game developer? hes not fantastic...

However give him a game engine, and ask him to help optimise it, and the game, and ask how it would be possible to impliment x,y,z funktions, and the guy will slave away at the problem and probably come up with a creative solution. He also understands game design, and hardware, and has a scope that allows him to "see" possible future issues, and is good about avoiding bottlenecks.

It was joke. Though I'd still give him a chance to run SIE.  

Would be intresting to see what he'd do at the very least.
Guy is like the most super of geeks, even in a field filled with those types.

He clearly loves technology, reading research papers on graphics techniques and technologies.
Hardcore programmer skills, and with his background into the software side of things, it gives him a perphaps slightly differnt perspect, than others in terms of hardware design.

Hes the reason why alot of studios at Sony follow something they call the "the methode proccess".

"Cerny observed that there were completely different approaches needed in the preproduction stage and the production stage of video game development, and that it was impossible to put a timeline on the creative process. He suggested that the pre-production stage should be freeform, allowing the creative persons to explore a game's viability prior to full development. The end product of the preproduction stage under the Method process should be a "publishable first playable" version of the game that can be used to determine the viability of the title. This version does not need to be content-complete but provide enough to be used in playtesting from potential consumers. If the game at this state does not excite players, then the game idea should be set aside before too much effort is put into it. Once the decision is made to move forward on the game, then Cerny recommends by the Method that the typical use of scheduled milestones and deliverables to keep the project on track."

I would recommend a watch of From Bedrooms to Billions documentary. Which goes through PS history. Sony back in the pre PS1 era (Ken Kuturagi) built the original PlayStation on the back of the feedback, set from big third party developers. Much like Cerny did the same with the PS5, as explained in the Road to PS5.

I still find it fascinating that a GPU with 256-bit bus and lower compute can keep up with a 320-bit one with higher CU's and slightly higher CPU speed/s.

If I understand correctly the PS5 and Xbox series X GPU are mostly comparable to a RTX2070S/2080 with the ray tracing capabilities of a RTX 2060S. Which for the price point is great. Ray tracing has its own dedicated hardware within the GPU. So it should not effect the rasterisation process to much. But because ray tracing is so incredibly heavy we can only get good results by decreasing the framerate. That's why both consoles should be able to handle 4k/30 with ray tracing whilst not hitting 4k/60 without like Spider-Man. 

I dont know about that. According to AMD the reconstruction technique they are implementing is not done yet. According to the video they claim all 3 consoles have checkerboard native to the hardware. Amd's reconstruction could be a build upon checkerboard that will hit the consoles sometime next year. if checkerboard looks like a close 1800p while rendering at 1440p, who knows if this checkerboard+ can be something like close to 2000p. Im pretty sure most gamers wont complain with that and having RT + 60fps. They might even have enough juice left over to crank up some settings. 

It is a one time thing. Once you play the game once online, you can play it offline indefinitely, even Game Pass titles. I tested multiple games from different generations, and it worked for all of them. You have to do a day one update for both the Series X and Ps5, so if you don't have internet at all, you're screwed for both.