I actually realized after the post that a PS5 Slim with RDNA 4 doesn't make too much sense when it'd require a redesign of a lot of parts of the architecture, so I've stepped back from that specific part. The other aspects though? I don't see how these are controversial. We know that faster and better modules of GDDR6 will come about over the next 3-4 years so the stack as a whole shifts in terms of costs to secure them. In other words Sony and Microsoft will be able to afford a 14 Gbps or 16 Gbps GDDR6 module for cheaper by 2023/2024 than they were able to in 2020, especially at mass volumes.
Addressing the GPU generation skip...again while I've since cut back on that idea at least regarding a PS5 revision (if it is indeed a Slim-style model), I based that on the release timings established so far for RDNA 1 to RDNA 2, which was about 16 months. Supposing that occurs again we would be able to expect RDNA 3 by March 2022, and RDNA 4 by July 2023. That assuming there's no major shakeups in AMD's roadmap or some change in architecture altogether that could throw off that release schedule. It's the best we can work with at this time in guessing when next generational releases for RDNA GPUs will occur, and it's based off a current pattern that's already been established.
So going with that in mind, neither mid-gen refresh/revision would actually release before the PC counterpart for this hypothetical release schedule.
I agree that crazy speculation happened in the last next-gen spec thread but I have deliberately tried avoiding that here and will continue to do so. Like let's take this SSD you bring up; my proposal here in Sony's case is that the drive is custom-built, it's basically them repurposing what's already soldered on the PS5 motherboard as something to an M.2 standard-issue form factor, that should also allow compatibility with 3P M.2 NVMe drives.
Sony is not Samsung; Samsung is selling that drive for profit, at least 2x the profits over assembly BOM and packaging/distribution costs (if not more). Sony would be designing such a drive for simply one product: a PS5 Slim revision. And that is a product with a business model that doesn't require high profit margins directly from the hardware sales. So IMO these aren't comparable scenarios.
Once again I scaled back on the suggestion of PS5 Slim being RDNA 4 as I had some more time to think about it and realized it made more sense to stick with the current architecture for ease of porting and production costs reasons. However, technically speaking, I'm not 100% sure each iterative RDNA gen can be seen as a "generation" in such a way RDNA is viewed compared to GCN, or even something like Vega compared to earlier GCN. PS4 Pro and One X were a full 2 (2.5 in PS4 Pro's case IIRC) generations (I get in the PC space gen version steppings are considered in generational terms but that lingo isn't always used on the console gaming side of these discussions) removed from their respective base systems, so there's precedent.
In fact, and this is more directed to those questioning if a PS5 Slim would have any spec upgrades over PS5, there is actual precedent of Sony having done this going as far back as the PlayStation 1. Here's a quote from Wikipedia regarding one of the early PS1 model revisions and its hardware improvements over the original launch model (December 1994):
Bolded emphasis mine. But that's something of an aside more to address any doubts on any chances of a hypothetical PS5 Slim doing anything as I suggested in the OP (outside of the RDNA 4 stuff, which I no longer think fits that design, and possibly scaling back on the storage capacity from 1.536 TB to 1 TB). I've never said these speculations were perfect or are even a guiding stone to what Microsoft and Sony do for mid-gen refreshes and revisions (if they even do anything, tho I think they will do at least something). However I feel rather confident at least in the level of research and thought put into this and feel similar with the 10th-gen system speculations which I'm hoping to put up soon.
Of course, folks are always welcome to disagree; that's part of what makes these discussions interesting because it leads to exchange of ideas
It's really hard to see a business justification for Sony doing a PS5 Pro this time around. Like was mentioned earlier, the base PS5 is pretty well-built for more streamlined VR, and there's no major revolution in TV display technology on the horizon. Those were the biggest factors for the original PS4 Pro, but they don't really seem to exist this time.
That said, there's still room (and prior precedent) for Sony to basically merge some performance gains into a PS5 Slim; the base PS5 (going by some of the Oberon revision listings) has support for up to 512 GB/s of memory bandwidth. So at some point, Sony were considering this but eventually went with slower 14 Gbps chips. They could afford 16 Gbps chips if the costs are right and not need to do any revisions to the PS5's memory controllers.
From what we can see on AMD RDNA 2 GPUs, there is room for them to also increase the GPU clocks further while sticking with the same GPU design, provided they provide the additional cooling to handle the upclock. That upclock increases TFs only slightly, but other things like cache speeds, pixel fillrate, texture fillrate etc. get some noticeable bumps. Even geometry/culling rates increase a small bit, and they can maybe do this within a smaller power budget than base PS5. The potential is easily there.
3nm is an option, though a small one. Apparently some people have found Sony booking 5nm wafers for 2023. I only found this out after posting this stuff because I think word started getting around over the weekend. 3nm would be more expensive than 5nm (which is more expensive than 7nm) and I dunno how that becomes an affordable option, though if Sony's mainly porting the PS5 design to the new process that creates saved R&D costs that can be used to offset the extra costs for going 3nm over 5nm....we would just need to see some evidence/proof...even rumors...of Sony booking 3nm wafer production. I don't see it happening though.
I don't know if a full-on switch away from silicon will ever happen because any new material has to prove it can both bring massive performance gains and do so at lower costs than silicon to justify retooling every part of the pipeline (including the fabs) to accommodate it. Maybe, some parts of some chips will use new materials while other components stick with silicon and some new packaging techniques come about. Seems like that will be the better option for the next decade in terms of keeping things simple from production POV, pricing POV etc.
But to a hybrid format, or a handheld that can be used in a docked-ish way to an output display, perhaps it might be possible. Obviously not 2021, but we're talking 2023 here, tech should progress further by that point to make it more possible.
If not, then they may be able to come somewhat near it. A "good enough" equivalent by such a time but, then again, I tend to be an optimist.
Depends on what you mean by "power", because TFLOPs aren't the only way to do it. The things you explicitly mention can be done with hardware acceleration, which I'm sure AMD are going to work on for RDNA 3 and future designs.
For what you mention, I think Microsoft have more of a business-oriented reason to pursue a "traditional" style mid-gen refresh. They'll want to increase the streaming fidelity for GamePass Xcloud while keeping costs manageable, which can best be done through a more capable APU design that, hey, you can also spin off as a mid-gen refresh that can possibly target some performance areas Microsoft arguably are a bit behind Sony on (pixel fillrate, GPU cache speeds, potentially need for cache scrubbers assuming AMD makes that standard going forward or Microsoft just wants to commission that for a new APU, etc.).
Sony has less a need for that, but they'd want to edge out a bit more performance while really targeting the main criticisms of PS5: system size, storage capacity, and cumbersome compatibility with 3P storage drives. A PS5 Pro doesn't really hit those notes, or at least better to say, it could do so in a way that's considered overboard versus what a Slim could provide. Meanwhile, there's always an argument that "hey, gamers will want more power and could go off to Microsoft or PC!", but that was the same argument for the last mid-gen refreshes and yet when you look at PS4 Pro and One X sales ratios, they are
very low compared to the base systems.
So
maybe that fear was unwarranted.
