• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

(SONY/MICROSOFT Hardware & Services) We WILL Get Mid-Gen Refreshes, But Not What You Might Think...

What type of mid-gen refreshes do you think we'll get this time (select all that apply)?

  • Pro-level models

    Votes: 137 46.8%
  • Slim models

    Votes: 166 56.7%
  • Service-orientated expansions and launches

    Votes: 30 10.2%
  • Nothing; same PS5/XSS/XSX hardware all gen

    Votes: 37 12.6%

  • Total voters
    293
  • Poll closed .

UltimaKilo

Gold Member
I can see slim models coming next year, it just saves them so much money on shipping, components, that it wouldn't make sense to wait much longer than that. As for more powerful revisions of each respective console, I doubt we see that next year, but those things have higher margins and with PSVR2, it's quite possible.
 
They'd be smarter to just bring PS now to cell phones.

I call BS in this. Granted I didn't watch the video.

They can do both. Issue with just relying on PS Now on mobile is, Sony doesn't have any control over the hardware specifications of the mobile market. It would be extremely difficult to count on consistent performance across a range of mobile phones with different RAM, CPU, GPU, and wifi capabilities, as well as varying size and speeds of storage.

Also the fact it messes up vertical integration, and it alone not the best option for possibly getting those PS Now users further into the ecosystem. Again I don't think everything in the video lines up (and as far as the PS5 Pro or even the Xbox Series X2 rumors, I don't think any of that will pan out to be true tbh), but there's a lot that makes some sense when you think about it.

I can see slim models coming next year, it just saves them so much money on shipping, components, that it wouldn't make sense to wait much longer than that. As for more powerful revisions of each respective console, I doubt we see that next year, but those things have higher margins and with PSVR2, it's quite possible.

They do have somewhat higher margins, but their total sales volumes are very low. With the current systems probably not going down in price nearly as quickly as PS4 and especially XBO did (and those were already slowing the trend of price drops on consoles compared to prior generations), I think that the base units will be able to have high margins for longer with slowed-down price cuts thanks to extremely high demand, and sell at volumes magnitudes more than Pro units ever could.

I'm curious if some other type of hardware expansion that doesn't involve a PS5 Pro would be possible for even better PSVR2 performance. Maybe an Elite PSVR2 headset with some extra hardware for processing built into it?
 
Last edited:
They can do both. Issue with just relying on PS Now on mobile is, Sony doesn't have any control over the hardware specifications of the mobile market. It would be extremely difficult to count on consistent performance across a range of mobile phones with different RAM, CPU, GPU, and wifi capabilities, as well as varying size and speeds of storage.

Also the fact it messes up vertical integration, and it alone not the best option for possibly getting those PS Now users further into the ecosystem. Again I don't think everything in the video lines up (and as far as the PS5 Pro or even the Xbox Series X2 rumors, I don't think any of that will pan out to be true tbh), but there's a lot that makes some sense when you think about it.
It would be cloud powered and not locally dependent. The phones would just act as a streaming device.

They have the technology. Microsoft and Google are already doing this. Same with some other companies. It doesn't make sense for them to try to split their development and re-enter the mobile space. It's not really competitive right now.

If they just got on to all mobile devices through PS now it would be a huge win for them without having to develop their own hardware and own ecosystem.
 
It would be cloud powered and not locally dependent. The phones would just act as a streaming device.

They have the technology. Microsoft and Google are already doing this. Same with some other companies. It doesn't make sense for them to try to split their development and re-enter the mobile space. It's not really competitive right now.

If they just got on to all mobile devices through PS now it would be a huge win for them without having to develop their own hardware and own ecosystem.
That's an interesting and logical way of looking at it. All said, they could still conceivably do both and reap the benefits of both. IMO a new Sony portable would just be a way of providing native portable gaming of PS4 games and new cross-gen games that could benefit from a portable option, and offer Remote Play streaming of PS5 games.

I'm not under the belief a new Sony portable would ever be treated as its own wholly unique platform the way Vita or PSP were, i.e there'd be no Sony 1P exclusives for it. But you might still get some 3P exclusives, or at least more PS ports of otherwise Japanese 3P Switch exclusives, which could do well for it in Asian markets among gamers who want a more performant portable device for those games (even if they'd be a minority compared to those still perfectly fine with Switch-tier specifications).
 

UltimaKilo

Gold Member
They do have somewhat higher margins, but their total sales volumes are very low. With the current systems probably not going down in price nearly as quickly as PS4 and especially XBO did (and those were already slowing the trend of price drops on consoles compared to prior generations), I think that the base units will be able to have high margins for longer with slowed-down price cuts thanks to extremely high demand, and sell at volumes magnitudes more than Pro units ever could.

I'm curious if some other type of hardware expansion that doesn't involve a PS5 Pro would be possible for even better PSVR2 performance. Maybe an Elite PSVR2 headset with some extra hardware for processing built into it?
That's why I fully expect "slim" versions to release, well before any "pro" model. The margins on a slim, at this output, are so much higher. $$$
 
That's why I fully expect "slim" versions to release, well before any "pro" model. The margins on a slim, at this output, are so much higher. $$$

Yeah. Not only that; slims would allow them to keep BOM per unit lower, get more profit off hardware sales (due to higher margins, as you, I, and others have said), and more easily satisfy demand.

PS5 and Series X might not even be readily available until the end of the year (which kind of sucks for me), and demand has remained steady, in some cases even increasing somewhat. No need to cut into that so soon with Pro models in 2023, even 2024 might be pushing it. Plus if things like FSR 2.0 and equivalent upscaling/reconstruction techniques end up working incredibly well this generation, that will strongly lessen the need for mid-gen Pro refreshes.
 

PeteBull

Member
To all the naysayers i got one question- do u want matrix demo quality in 20fps(or drops to 20fps) like current ps5/xsx got or u want it in 60fps and some decent res, preferably at least 1440p- coz current machine's max is such res+ graphic bells and whistles at 20fps.

If u think logically u gotta assume u will want at least 2x gpu/cpu power to get better results from actual full fledged game unless resolution gets downgraded to 1080p or maybe even under with 60fps stable mode.

Even fricken gta5 that debuted in 2013 on ps360 consoles has only 1440p60fps mode(coz with rt it drops badly, and that mode is far from pc max settings) or 4k 30fps mode- that one is very stable, but again to get 4k60fps u need at least 2x gpu power, even if u assume game has big cpu headroom which is possible.
Current consoles are about 2x weaker from top end gpu's so something gotta give or we getting midgen upgrades with sizeably more oomph.
 
Last edited:
PeteBull PeteBull i think it's safe to say that consoles and gpus this gen can't seem to get past this native 4k 30fps threshold. Even if they reach it, it's not stable because they are implementing raytracing which again tanks performance by:

lowering resolution
decreasing frame rate.

For this, they have implemented many solutions: Variable Rate Shading, FSR (DLSS for PC), and other efficiency techniques specific to each console-but has barely been touched due to the shackles of cross gen potato jaguar, and time for implementation, due to those techniques being in their infancy stages. For PS5Pro, and XsXPro to reach stable 4K 60fps (not thanks 30fps) with all the bells and whistles like raytracing the following (I think must happen):

1)They have to 'brute force' the specs to reach NATIVE 8k 30fps minimum target. That way gamers can bitch about how PS5Pro and XsXPro are having difficulties reaching that target, while I could give 2 shits because I am lowering the settings (or downsampling) to stable 4K 60fps on my new QD-OLED from Samsung and I am in gaming bliss. Their is always 8k checkerboarding 🤷‍♂️

2)Xbone X, Xbone, PS4, PS4Pro MUST DIE. Or in other words, discontinue support for PS4, PS4Pro, Xbone, XboneX. Dont worry, all your games will be in gamepass, and xcloud (for XsXPro atleast)

3)Incorporate ML cores, or Neural Cores in RDNA3. This is essential. Because if you want DLSS like feature in XsXPro and PS5Pro, the ML cores spares resources/processes of the system so you can spare it towards important priorities like raytracing (which itself should already have a dedicated core to further spare performance)

4) Have better storage medium. Blu-Ray must die. All the big tech giants must come together and decide on a storage medium which is cost effective, gives shit ton of space (think 1 terabyte games), and gives instantaneous game loading (like cartridge).
 
Last edited:
1)They have to 'brute force' the specs to reach NATIVE 8k 30fps minimum target. That way gamers can bitch about how PS5Pro and XsXPro are having difficulties reaching that target, while I could give 2 shits because I am lowering the settings (or downsampling) to stable 4K 60fps on my new QD-OLED from Samsung and I am in gaming bliss. Their is always 8k checkerboarding

My only issue here is that if the midgen refreshes are capable of native 8K 30, that weakens some of the argument for what 10th-gen systems will be able to provide, at least in terms of performance. And that's also assuming they would be able to provide 8K 30 without needing beefier I/O specs, more RAM, larger storage etc. The costs would add up quick especially if they're doing all of that on say 5nm.

2)Xbone X, Xbone, PS4, PS4Pro MUST DIE. Or in other words, discontinue support for PS4, PS4Pr, Xbone, XboneX. Dont worry, all your games will be in gamepass, and xcloud (for XsXPro atleast)

I mean that's a foregone conclusion; 8th-gen hardware will be retired sooner rather than later.

3)Incorporate ML cores, or Neural Cores in RDNA3. This is essential. Because if you want DLSS like feature in XsXPro and PS5Pro, the ML cores spares resources/processes of the system so you can spare it towards important priorities like raytracing (which itself should already have a dedicated core to further spare performance)

This would be a necessity for any mid-gen refresh. However, I'm not 100% convinced they need to increase the raw computational/TF (and therefore, doubling of CUs) for the consoles to do this. If base PS5 and Series systems had ML cores/Neural Cores in them, you'd be getting RT in most games with 60 FPS at 1440p - 4K resolutions.

Which is why I'm still not sold on mid-gen "Pro" style refreshes; smaller Slim-like refreshes adding in some hardware accelerated units like ML/Neural Cores for better RT and image upscaling could do wonders while keeping BOM prices lower and getting much more volume of units out there. Plus they could also use such a refresh to basically phase out the base units altogether, instead of needing to split their production stocks (which could be a logistical nightmare).

4) Have better storage medium. Blu-Ray must die. All the big tech giants must come together and decide on a storage medium which is cost effective, gives shit ton of space (think 1 terabyte games), and gives instantaneous game loading (like cartridge).

I think we've actually talked about this before. TBH they don't need to invent a new storage medium; either for midgen refreshes or 10th-gen systems I can see both Sony & Microsoft switching to cheap 64 GB/128 GB microSD cards with 60 MB/s - 90 MB/s read speeds, to ship physical games out on. So, somewhat similar to what Nintendo does with Switch games, but more storage obviously.

Then they could have those interface with the decompression engines in their consoles to effectively double or triple the bandwidth for compressed data on the cart; suddenly a 64 GB card can store 128 GB worth of content (with lossless compression), a 128 GB card 256 GB worth of data, etc., possibly more. And you're talking bandwidths way more than anything Blu-Ray can offer now or into the future.

MicroSD cards are getting cheaper and cheaper; maybe not cheap enough to do this for mid-gen refreshes, but by the time of 10th-gen I can definitely see all the consoles going this route. Just add a SDXC USH-II microSD port, make it interface with the file I/O subsystem same way M.2 SSDs do, through the FMC (Flash Memory Controller), big problem solved right there.

Sony must he dying to get a slim model out thats cheaper...what are the chances in the next 18 months? I have no idea.

Microsoft will shrink it when they can. I'm interested to see it.

They have that plant I think which was recently set up, which they'll probably start using soon. So the chances of it happening within the next 18 months are actually pretty high.

But, will they also use that opportunity to get PS5 Digital at an affordable BOM so it can be manufactured in real numbers, is the question. I think there is a way they can do it, by keeping the disc drive out, reducing internal SSD storage to 64 GB or maybe 72 GB max (enough for a single massive game in rotation if you compress the data 2:1), decent microSD slot with internal 512 GB or so microSD storage of say 20 MB/s - 50 MB/s. That could serve as cold storage and offer bandwidth twice as fast as UHD Blu-Ray with data 2:1 lossless compressed, and work out to be cheaper than having 825 GB of internal SSD storage the way both PS5 SKUs do it currently.

I think Microsoft are more interested in doing something that's an even cheaper option into Series product family than the S. Like an Apple TV-style box they can sell for $149. Powerful enough to run select GamePass games natively, but stream everything else. They could add in some eMMC Flash storage to interface with the decompression blocks; since it would be focused mainly on streaming they can keep storage amount small, keep memory amount small (4 GB max), the CPU and GPU notably cut-down in size and clocks, require less cooling etc.

Maybe they could upgrade the wifi from Wifi 5 to Wifi 6, though. I guess they chose 5 for cost-saving reasons but with how cheap they could make a Series system acting as an Apple TV-like, they'd have enough saved over to afford Wifi 6 and still keep costs low enough to hit $149 MSRP.
 
Conservatively the PS5Pro and XsX would be Zen3+/RDNA 3 based.

Ideally, I would like PS5Pro and XsXPro to be Zen4/RDNA4 based with 3-D V-cache and other bells\whistles\refinements\enhancements specific to that microarchitecture provided by AMD.

Lastly, the CPU frequency has to reach 4.0Ghz threshold (not talking overclocking cause you cant overlock consoles) and the GPU frequency to reach 2.5Ghz threshold (not talking overclocking).

MS also needs to spend a fuck ton on SDK, API, development tools, development software\compilers, UI to make the game development process as fluid and universal as possible. Perhaps similar to the UWP platform (but for games) with one base source code without too much overhead and being 'close to the metal'. Imagine making Forza Horizon 6, but works on XsX, XsS, XsXPro, PC, and Windows 11 ARM (running on native ARM hardware) utilizing:

Variable Rate Shading, DLSS like upscaling utilizing ML cores, Raytracing, Variable Refresh Rate\Latest Freesync, HDR, Sampler Feedback Streaming, etc.
 
Last edited:

Kokoloko85

Member
PSP3 is dead out the water if it cant play PS5 games downgraded. SteamDeck can play Futre PC games, no reason this cant
 
Conservatively the PS5Pro and XsX would be Zen3+/RDNA 3 based.

Ideally, I would like PS5Pro and XsXPro to be Zen4/RDNA4 based with 3-D V-cache and other bells\whistles\refinements\enhancements specific to that microarchitecture provided by AMD.

Lastly, the CPU frequency has to reach 4.0Ghz threshold (not talking overclocking cause you cant overlock consoles) and the GPU frequency to reach 2.5Ghz threshold (not talking overclocking).

MS also needs to spend a fuck ton on SDK, API, development tools, development software\compilers, UI to make the game development process as fluid and universal as possible. Perhaps similar to the UWP platform (but for games) with one base source code without too much overhead and being 'close to the metal'. Imagine making Forza Horizon 6, but works on XsX, XsS, XsXPro, PC, and Windows 11 ARM (running on native ARM hardware) utilizing:

Variable Rate Shading, DLSS like upscaling utilizing ML cores, Raytracing, Variable Refresh Rate\Latest Freesync, HDR, Sampler Feedback Streaming, etc.

The potential issue here is that you might be asking for systems that end up costing $600/$700, even by 2024 timeframe, because the consoles would need to use at least 5nm (if not 3nm) to deliver that performance in a TDP suitable for a console (not to mention APU size suitable for a console). And in addition to that, you might need more SSD bandwidth in order to feed the (inevitably increased) RAM capacity to keep that type of GPU fed at a rate similar to what the current systems can do with their storage.

It's just probably too much for a midgen refresh, especially considering most of that performance would not be utilized for games specifically targeting it, so again similar issues as seen with PS4 Pro and Xbox One X where they were basically just resolution & framerate boost boxes. I think you can still get a substantial performance increase with new baseline models keeping the same general perf profiles of current systems (same TF, same pixel/texture fillrates, same # of CUs, same clocks, same RAM bandwidth and capacity etc.), while adding in new hardware acceleration features of RDNA 3 like ML cores.

That helps keep BOM prices lower, doesn't risk splitting up production between models in a way that doesn't work out, and lets them wholly replace the base models (eventually) while still giving room for performance benefits. It's like a Slim and Pro rolled into one.

About the Microsoft stuff...well they have already done this with the GDK, no? The GDK was built so that you could develop games across Windows devices in a unified way. They've just been pretty slow with rolling out the updated API features, or maybe it's more that developers haven't had enough time to optimize for them. Though it could also easily just be that such an approach is always going to have a lower ceiling than a leaner, but tighter SDK API package built and optimized for simply one hardware configuration, which would also help with reducing required hardware overhead, etc.
 

PeteBull

Member
The potential issue here is that you might be asking for systems that end up costing $600/$700, even by 2024 timeframe, because the consoles would need to use at least 5nm (if not 3nm) to deliver that performance in a TDP suitable for a console (not to mention APU size suitable for a console).
Entusiasts(who will be target of midgen refreshes, like with ps4pro and one x), will have 0 problem spending 700$/€ in 2024 on the hardware as long as it will be at least 2x as strong as current consoles( basically instead of 4k/30 mode or 1440p/60mode u will get pr0 mode with best of both worlds and let pleb on base consoles have to make tuff choices how they games should run/look :)
There is a reason enthusiast market is also called bleeding edge market, coz u literally gotta bleed money so very often to keep up :)
 
Last edited:

Belthazar

Member
A PS4 Portable would be a ver neat idea if they can keep production costs low for it to be competitive, not to mention managing a good battery and heat levels. Would give them a fighting chance in Japan and the PS5 would still receive the games made for it (as they would be PS4 games) anyway.
 
Last edited:
Entusiasts(who will be target of midgen refreshes, like with ps4pro and one x), will have 0 problem spending 700$/€ in 2024 on the hardware as long as it will be at least 2x as strong as current consoles( basically instead of 4k/30 mode or 1440p/60mode u will get pr0 mode with best of both worlds and let pleb on base consoles have to make tuff choices how they games should run/look :)
There is a reason enthusiast market is also called bleeding edge market, coz u literally gotta bleed money so very often to keep up :)

Yeah but realistically how many of them would be out there right now or in a few years time? Maybe a bit more than usual since a lot of them would prob want the high-end GPUs but can't even find them, but will that situation still be around by 2024? Would R&D costs for such a unit be made up with enough volume of sales at good enough MSRP? What's the real cutoff limit for such a system price-wise for even enthusiasts in comparison to similar or better-performing GPUs?

I just don't really see a path for more "traditional" Pro models this gen, sadly. They might get more mileage by merging some expected Pro features (dedicated RT/ML cores, upsampling tech etc.) into otherwise Slim models that keep the same specifications, but I guess we'll see sooner or later.

A PS4 Portable would be a ver neat idea if they can keep production costs low for it to be competitive, not to mention managing a good battery and heat levels. Would give them a fighting chance in Japan and the PS5 would still receive the games made for it (as they would be PS4 games) anyway.

Absolutely, and that's been the pitch for a new portable I've been rolling with for a long time now. It'd be a portable extension of the PS4, with base PS4 specs, can do Remote Play for PS5 games, native on-the-go gaming for your PS4 library and being a good platform for new smaller games that can also play on PS4 & PS5 consoles natively.

It should be more than possible to do the PS4 APU design on 7nm, and with the relatively low CPU & GPU clocks (even compared to Steam Deck), would have great low power consumption and thermals. Only potential issue could be RAM; I don't know if the system would NEED to use the same 256-bit bus interface as PS4. If not they can cut it down to 128-bit with 4x 2 GB 14 Gbps chips downclocked to 11 Gbps, for 176 GB/s bandwidth (same as base PS4).

If GDDR6 at that pin clockrate is still to power-hungry, they could do HBM2 which would use a lot less power. But the price would probably be a little bit higher than GDDR6. The cost savings could maybe be seen in needing less cooling, however.
 
My predictions are coming through...


AMD is not gonna let MS (highly unlikely) and Sony (more likely) create a Zen 2.5/RDNA2.5 type of console with 'slightly more' this and 'slightly faster' that. PC gaming for RDNA 3 would poo poo on those consoles and make them dead on arrival if it had garbage specs like that.

MS TFLOP jump is pretty significant: 1.2 to 6 to 12
Sony TFLOP jump is meh: 1.8 to 4.2 to 10.28

The bare minimum TFLOP for both PS5Pro and XsXPro would be 20 without Sony and MS doing any customizations/optimizations/tweaking. The RDNA 3 design will just not allow 'slight' or 'minor' upgrades. I dont know if its possible to have RDNA 2.5 even if Sony wanted it, like how they did with PS4Pro where it had some features of Vega.
 
Last edited:
My predictions are coming through...


AMD is not gonna let MS (highly unlikely) and Sony (more likely) create a Zen 2.5/RDNA2.5 type of console with 'slightly more' this and 'slightly faster' that. PC gaming for RDNA 3 would poo poo on those consoles and make them dead on arrival if it had garbage specs like that.

MS TFLOP jump is pretty significant: 1.2 to 6 to 12
Sony TFLOP jump is meh: 1.8 to 4.2 to 10.28

The bare minimum TFLOP for both PS5Pro and XsXPro would be 20 without Sony and MS doing any customizations/optimizations/tweaking. The RDNA 3 design will just now allow 'slight' or 'minor' upgrades. I dont know if its possibly to have RDNA 2.5 even if Sony wanted it, like how did with PS4Pro where it had some features of Vega.

Oh yeah, dat RDNA3 powah lezzgooo!

But...there is one...sort of problem. Power consumption. I've been hearing that power consumption for both the next AMD and Nvidia GPU lines will be high, quite high, and they are already moving onto 5nm (which brings IIRC 40% power consumption savings over 7nm). How can Sony & MS fit 2x TF performance into a mid-gen refresh on 7nm or even 5nm while staying in the same TDP range of the current console GPUs (which I'd figure is about 180 watts)?

Architecture changes could account for some additional power consumption reduction, but it won't be nearly as much as moving from 7nm to 5nm. I don't even think raw TF are going to be what constrains the base consoles as time goes on, it's the lack of dedicated silicon for lower-precision work (currently it all has to be done on the CUs but they aren't optimized for lower-precision math or math matrices) or doing specific tasks with small, low-power silicon budgets.

How much can doubling the TF by doubling the CUs really accomplish there and how much can that be done by staying in a TDP that's console-friendly? At a guess, it'd be with GPUs using ~ 215 watts under full load, then you gotta add in the memory (and the increased wattage due to faster GDDR6 modules to get more bandwidth), potentially more TDP for a faster-clocked CPU...that's easily pushing consoles to 250+ watts under full load, so the PSUs would need to get bigger, the PCB would probably need to be a bit bigger, cooling more elaborate, and again 5nm is more expensive than 7nm and we're talking doubling the compute units.

It's practical for PC because those dudes will pay super-high for the top-end performance, but consoles have to balance out more power with other concerns. Midgen refreshes with 2x perf of base systems in 2024 would probably go for $700, how many people are really gonna buy that? And it sucks that TSMC delayed their 3nm fabbing until 2025 because while that'd be even more expensive wafer-wise, it'd at least bring costs down further.

I'd like to be proven wrong on this but we gotta wait a long time for any midgen upgrade news officially (or even any solid leaks :/ ).
 
Top Bottom