That is what I think has happened. RDNA2 was going to be just like the MS hotchip says without some of its customizations.
While AMD was working with MS on their customizations, it was also working with Sony on an RDNA2 GPU but maybe trying more innovative things or maybe they risked more and it turned out well.
Finally Sony came up with a GPU that pushed quite high with a low TDP and high frequencies and AMD adapted it to RDNA2. This could have happened 1 or more years ago.
MS uses your GPU and Sony uses theirs. However, what makes me think that this happened is that MS has been able to present its Hotchip and Sony has not. Sincerely RDNA2 is the result of a collaboration between Sony AMD and MS has everything that AMD developed but nothing that AMD + Sony developed.
In 10 days we will know I hope all this. However, the fact that Sony has not presented a hotchip only makes me think precisely that AMD opted for AMD + Sony solutions.
However, the MS GPU is RDNA2 because it contains the entire RDNA2 base, however, no AMD + Sony customization, which is what AMD and Sony have adopted for RDNA2.
So they will say RDNA2 based but without infinity cache and possibly the RT is different in the diagram, maybe some improvement in the CU and other series of components. The necessary thing to have clocks at 2.4GHz with 80 CU (a lot).
I'm very curious about the AMD hotchip but especially about the PS5 one I want to see that IO controller and Tempest RDNA2CU based. (It is possible that even the RDNA2CU AMD + Sony is different have some easily customizable layer for some tasks)
I'm dying of curiosity. But it's cool to read you around here to discuss ideas of what may be happening. Keep arguing kindly and with knowledge is very enriching.
I think we should stop with the whole RDNA 2 "based" thing, as of now we don't even now what that means and it could be very inaccurate, what do we even mean by the term RDNA 2 it self? AMD themselves have not even properly outlined and defined it. Because of this, I think a lot of the controversy has arisen especially with regards to both the consoles and the feature set they incorporate.
It's difficult to speculate on the relationship and collaboration between Sony and AMD and Microsoft and AMD. I would think that AMD have their own set of ideas of what they want to achieve with the their GPU's, this is not limited to only their upcoming GPU's but even the ones which will come out within 1-2 years hence the term "AMD Roadmap". Now both Sony and Microsoft have their own objectives of what they want to achieve with their own GPU and yes on many occasions these objectives will be shared amongst AMD, Sony and Microsoft, a good example would be the performance per watt and IPC gains as well as hardware based ray-tracing and VRS, which are all significant factors for all three of the companies and their upcoming products.
Thanks to Mark Cerny we also know that both Sony and Microsoft will also have their own "unique" objectives when it comes to their APU and its feature set, like the cache scrubbers, Geometry Engine and I/O complex of the PS5 or the backwards compatibility and Machine Learning implementation on the Series X. If AMD feel these "features" or "unique objectives" that Sony and Microsoft want to achieve can also be beneficial for AMD's products, they'll also incorporate them into their own Roadmap, this would be part and parcel of the collaboration agreement, this roadmap would include RDNA 2 + 3 and likely even 4. However there definitely will be features discarded from both Sony and Microsoft on their GPU because they may feel these features won't be beneficial for what they want to achieve, I think this is where a lot of the nonsense "RDNA 1.5" stuff comes from because certain people will have a black and white picture of what RDNA 2 is meant to be and do so without any knowledge on the inside workings of the cards as well the collaboration between these companies.
I know a lot of this is already known but it's good to remind people, hopefully we'll get a much clearer picture in 10 days time.