Do you think there will be a mid gen refresh console from Sony and Microsoft?

And more recent rumors stated 2026. Ok, rumors.

And more recent RECENT rumors stated 2025, that Microsoft will be eliminating generations and offering more incremental upgrades similar to the smartphone model (which could partially explain the ARM SOC route). But yes, rumors.
 
No chance are PS5 and Series S/X going to be on the market for 9 years before they get a replaced.

They're already starting to feel the performance pinch.
With the price of semiconductors, I think that they need all the time possible to get close to 2x the performance of a pro console at an acceptable price.
They are probably targeting 2028 but maybe it will slip in to 2029, like the ps5 that was targeting 2019 but it came out in 2020 (or was that just a rumor?)

Ps: here I'm mostly talking about Sony, who knows what Microsoft wants to do.
I don't know what Microsoft gains from releasing a new console in 2026 when they still have to support series x and s for like 8 years...
 
Microsoft will be eliminating generations and offering more incremental upgrades similar to the smartphone model
That would be really unmicrosoft. After more than 3 years they aren't able to produce a revision model just to reduce the production cost, and should they do a yearly upgrade?
And I'm not talking about a pro or a lite, just components revision to streamline production like sony's doing.
 
No chance are PS5 and Series S/X going to be on the market for 9 years before they get a replaced.

They're already starting to feel the performance pinch.
Quite the opposite - these new consoles are only just recently starting to actually get utilized and they've got plenty more in store.

'Performance pinch' - they're consoles, they're always necessarily limited, even from Day 1. Stop looking at this from a PC gamer perspective and they're plenty capable. I also think some PC gamers still dont grasp that consoles dont 'follow' PC. It's again, the other way around. Just because PC's are leaping forward in power doesn't mean there's any big worry for consoles falling behind, cuz devs are still building games for those consoles first and foremost.

I dont see anything wrong with this generation going nine years. In fact, I think that would be the best and healthiest thing to do for both consumers and developers/publishers. Heck, I think the PS4/XB1 generation could have comfortably had another year left in them before any new consoles came out, and I'd say the lack of actual next gen titles for years supports this. I'm pretty sure devs would love to have some extra time this gen to not only make the most out of these systems, but also, ya know, actually release more than one or two games on them before they have to completely reset again.

We need more time between generations nowadays so we can get proper generational leaps. We wont have the low hanging fruit of going Jaguar->Zen this next time around. There wont be a low hanging fruit in terms of switching from HDD->NVMe SSD's either. So we're looking at a more raw capability improvement, and nothing is going to achieve that except time.
 
That would be really unmicrosoft. After more than 3 years they aren't able to produce a revision model just to reduce the production cost, and should they do a yearly upgrade?
And I'm not talking about a pro or a lite, just components revision to streamline production like sony's doing.

Let's say for the sake of argument that the ARM rumor or the FTC disclosure of it, is true. I could see Microsoft moving to a more streamline system that is more cloud specific... a more hybrid console with cloud compute in mind. Something Microsoft has (is) been aiming for quite sometime. Heck, Series X (and S to a certain extent) has more of a server arch-approach. Microsoft has the resources and infrastructure to accomplish these goals soon or later, more so than Sony. And truth be told, I don't see a future where standalone consoles aren't more of a hybrid design, or leaning closer to cloud compute. As such, starting with a ARM design and having more (quicker) iterations of said design, could potentially get them to their goal of a fully reliant cloud console for better streaming game services. Gamers may laugh, but Nvidia Shield is pretty good at streaming and handling gaming in such a small package (imagine what Microsoft can do if they hit the right stride in cloud console hardware and service delivery).
 
I'll just keep repeating the same point, which you're doing a tremendous job of overlooking. I would add that I also did not make any performance comparison to PC GPUs either, I've only seen posts on that that from you and David Graham. So consider me baffled.


I cannot think of any other way to express the aim was to demonstrate that high-end products. across all types of product categories, exist almost oblivious of economic realities in terms of market share, especially given the overheads of semiconductor manufacture. The only figure I can find that Nvidia have ever hinted at is 100,000 4090s GPU having been sold a month after release, which is. fraction of the number of PS4 Pros sold at launch.

The issue here is the false equivalence you're drawing between the PS4 Pro and xx80 and above class GPU's in terms of market segmentation, seemingly to back up your original point of "but look, it still sold better than high end PC GPU's".

Sure, it sold better than premium priced xx80 class and above GPU's, but that's a meaningless comparison given the price differential, as well as the availability in the PC space of much cheaper, but still performance orientated parts (much more in line with the Pro in both respects), that if included on the Pro side or your equation rather than the base model side, completely changes the conclusion.

If you're argument really is simply that higher priced performance orientated parts always sell worse than lower priced value orientated parts (in which I believe there is some truth) then I'd suggest a re-phrasing of your original point into something a little more reflective of comparable market segment such as "even in the PC space, performance orientated parts in the xx70 and upwards range sell less than the lower end mainstream parts". In this case to the tune of 1 in 3 PC GPU's being of the performance orientated variety vs roughly 1 in 5 Playstations.
 
And more recent RECENT rumors stated 2025, that Microsoft will be eliminating generations and offering more incremental upgrades similar to the smartphone model (which could partially explain the ARM SOC route). But yes, rumors.
I cannot see more concurrent console performance variations being good in terms of the time developers spend to optimise games for the hardware configurations. Perhaps Xbox will transition to having more PC-like settings and users just change these themselves?
 
And more recent RECENT rumors stated 2025, that Microsoft will be eliminating generations and offering more incremental upgrades similar to the smartphone model (which could partially explain the ARM SOC route). But yes, rumors.
The most logical explanation is that whoever started these rumors mixed XSX shrink as next gen, and actual next gen is 2028 as MS documents said
 
I cannot see more concurrent console performance variations being good in terms of the time developers spend to optimise games for the hardware configurations. Perhaps Xbox will transition to having more PC-like settings and users just change these themselves?
This is the whole reason why they started the Xbox

The most logical explanation is that whoever started these rumors mixed XSX shrink as next gen, and actual next gen is 2028 as MS documents said
This makes the most sense to me. The FTC leaks were authentic and revealed alot about the expected next gen launch year which is 2028. They dont just arrive at such a number and likely Sony has the same year considering they're releasing a pro console.
 
I think the following is the essential question. Is it possible that FSR3 only works on Xbox consoles and is it possible to solve 30-60FPS with FSR3? I'm asking this because if it's feasible, then MS doesn't need a pro console for that reason either.
 
I think the following is the essential question. Is it possible that FSR3 only works on Xbox consoles and is it possible to solve 30-60FPS with FSR3? I'm asking this because if it's feasible, then MS doesn't need a pro console for that reason either.
It can support fsr3 technically. Any 5000 series card and above can support it, so Gen 5 consoles should be OK. Whether it comes and at what performance levels, I don’t know.
 
Quite the opposite - these new consoles are only just recently starting to actually get utilized and they've got plenty more in store.

There are already multiple games with native 720p rendering, piss poor image quality and poor performance.

They offer GPU performance that's equivalent to a mid-range PC GPU from 2018.

'Performance pinch' - they're consoles, they're always necessarily limited, even from Day 1. Stop looking at this from a PC gamer perspective and they're plenty capable. I also think some PC gamers still dont grasp that consoles dont 'follow' PC. It's again, the other way around. Just because PC's are leaping forward in power doesn't mean there's any big worry for consoles falling behind, cuz devs are still building games for those consoles first and foremost.

Please see the above, they're already falling behind.

I dont see anything wrong with this generation going nine years.

I see plenty.

In fact, I think that would be the best and healthiest thing to do for both consumers and developers/publishers. Heck, I think the PS4/XB1 generation could have comfortably had another year left in them before any new consoles came out, and I'd say the lack of actual next gen titles for years supports this.

The console not really having the performance for 'next gen' features is also a big factor.

I'm pretty sure devs would love to have some extra time this gen to not only make the most out of these systems, but also, ya know, actually release more than one or two games on them before they have to completely reset again.

This generation needs to used to get RT and next generation geometry ion to games engines and that's it.

We need more time between generations nowadays so we can get proper generational leaps.

Define 'proper generational leaps'
 
I think the following is the essential question. Is it possible that FSR3 only works on Xbox consoles and is it possible to solve 30-60FPS with FSR3? I'm asking this because if it's feasible, then MS doesn't need a pro console for that reason either.

AMD themselves recommend a minimum of a 60fps input for FSR3.

So that pretty much rules it out being used to get 30fps to 60fps in consoles.
 
Last edited:
That cache reduction's interesting. Do you happen to know if it's incurred any performance penalty? It seems odd to reduce IC when the chiplet nature of RDNA3 seems entirely geared around modularity of IC.

Can measure hit rates I guess which would vary application to application but I don't know how you'd quantify performance impact as there is no like for like part otherwise different apart from cache, the architecture itself is even different.

There's likely design considerations in terms of the cache spec. RDNA3's L3 cache is on the MCD along with the memory controller/phys. 96MB works out to a rather simple 16MB per MCD across 6 MCDs for the 7900xtx. It's modular in terms of MCD scaling.

You've made me wonder: maybe adding cache in a mid-gen would be more hassle than it's worth, and that's more within the purview of a new generation and paradigm?

It'll likely be weighed against the costs/logistics of just conventionally memory bandwidth and what the need is to feed the GPU.

A large influence in the choice on the PC was likely that there was basically no real practical way to significant increase conventional memory bandwidth for much of the product stack for the IHVs without significant power draw and cost issues (and also market segmentation, but that's another debate).

But taking the PS5 for example it's only using 14Gbps GDDR6 but 18 Gbps is now commonly available and even 20 Gbps in mass commercial products. The latter would be 42% more memory bandwidth (likely more effective as the CPU utilization percentage would go down) without having to resort to anything else.
 
I cannot see more concurrent console performance variations being good in terms of the time developers spend to optimise games for the hardware configurations. Perhaps Xbox will transition to having more PC-like settings and users just change these themselves?
I've watched my nephew dial in ue5 games on his 2070s and he's not a real tech guy, he just selects medium preset then selects the dlss preset until he gets what ever performance he wants which has been balanced for the 2 things I saw him setting up. If MS drops generations I think the best thing devs could do is just have like 2 options, one would be quality setting with 2 or 3 options low/med/high and a scaling option quality/balanced/performance. That should let the whole product stack use their hardware to achieve the quality they are happy with and also reduce the need for updates/patches as new hardware arrives every 2 years or whatever.

Oh and maybe something for vsync/vrr.
 
An order of magnitude or greater in term of increase in processing power.

That doesn't work in the price range consoles target as we don't see increases in processing power jump that much.

PS5 and Series-X represent the lowest performance jump between generations that we've ever seen from consoles.

And unless Sony or Microsoft are willing to sell consoles in the $700 bracket and use the price increase to get a larger GPU, longer generations won't really work.
 
Last edited:
Back
Top