It really doesn't matter if MS can't do something to disrupt the market. If Ms just puts out a mid gen console that is a 10% or even 20% faster than ps5 pro but a year later even with the dev support they have it wont change anything. They need to bring out somethign ground breaking along with software that supports it.
Predict: Next gen console tech (10th generation edition) [2028+]
- Thread starter RDGoodla
- Start date
It really doesn't matter if MS can't do something to disrupt the market. If Ms just puts out a mid gen console that is a 10% or even 20% faster than ps5 pro but a year later even with the dev support they have it wont change anything. They need to bring out somethign ground breaking along with software that supports it.
Johnny Awesome
Veteran
Not really. A more powerful machine with 32 dev teams putting out Day One content on GP is disruptive enough. And if they put those same releases on PS three years later for $60 it doesn't even matter how many hardware units they sell, but I'm positive that the 60 million Xbox ecosystem people will stick around and MS will just rake in the money on PC (same day releases) and PS (delayed releases).
The issue is a lot of the ip is committed to being multiplatform. So I don't see what the difference is if they put out a new machine next year vs a different year. Their 32 dev team games would play best on that hardware anyway unless they some how manage to be a year late and less powerful than the ps5 pro
Enough for what? I think you're talking about keeping XB fans happy, whereas others are talking about winning market share away from Sony. A more powerful machine won't make any difference to people leaving PS. The only that might swing it would be software (and marketing!).
MS aims to have 200 million GP subscribers. Maybe they will create a Premium Game Pass service that includes EVERY game released Day1 only on Xbox console.
Charlietus
Regular
Every game? As in every single game that gets released on Xbox from gta 6 to bunny bunny fights the kaiju's 2?
I don't think people have understood that the whole industry is in maximum profit mode now. And that includes Microsoft.
I don't even think the next Xbox is going to be subsided by Microsoft.
Getting to 200 milion subscribers requires more money than Microsoft is willing to spend after the last year.
PS: welp, @iroboto message wasn't there while I was typing
According to MS, they are preparing for the biggest leap in performance with their next console. And according to the latest information, many of us are guessing that the new Xbox will be here in 2025 or 2026. We know exactly what kind of hardware is available for this console, which will be ready within two years. At a traditional level, the possibilities are limited in terms of an acceptable market price. There are two possibilities, one is that there will be a more expensive than usual Xbox in the product range, which is not reasonable from a business point of view. The other option is AI. The AI built into the hardware can have an impact on real-time graphics and thus increase the quality of the visual display even during gameplay. Is this what MS meant when it said that the new Xbox will be the biggest technical step we've seen in hardware generations, the integrated AI processor?
Here is an article that fits here, emphasizing this part:
"AI algorithms can pore over game data like 3D meshes, textures, audio files, environment geometry, and more to condense them without negatively impacting visuals, sound quality, or player experience. By compressing data file sizes, overall game performance can be improved significantly, with faster loading times and smoother gameplay.
AI can also dynamically adjust in-game resource allocation on the fly by performing real-time performance analyses and delivering resources to game elements as needed. This load-balancing act means games utilize available computing power in the most efficient way at all times for optimal operation."
The article: https://elearningindustry.com/the-f...rtificial-intelligence-is-changing-everything
Here is an article that fits here, emphasizing this part:
"AI algorithms can pore over game data like 3D meshes, textures, audio files, environment geometry, and more to condense them without negatively impacting visuals, sound quality, or player experience. By compressing data file sizes, overall game performance can be improved significantly, with faster loading times and smoother gameplay.
AI can also dynamically adjust in-game resource allocation on the fly by performing real-time performance analyses and delivering resources to game elements as needed. This load-balancing act means games utilize available computing power in the most efficient way at all times for optimal operation."
The article: https://elearningindustry.com/the-f...rtificial-intelligence-is-changing-everything
I can't help but think that a new console in 2025 or 26 is a mistake.
I really don't think that there is any "there" there for improving console performance on raster loads. Current gen consoles are good enough that you can make the kind of game you want to make, and have no horrible glaring flaws. I would personally prefer there to be a long generation, which lasts until the next-gen console can do full RT rendering. I doubt we are there yet in 26, given how slow lithography upgrades are nowadays.
I really don't think that there is any "there" there for improving console performance on raster loads. Current gen consoles are good enough that you can make the kind of game you want to make, and have no horrible glaring flaws. I would personally prefer there to be a long generation, which lasts until the next-gen console can do full RT rendering. I doubt we are there yet in 26, given how slow lithography upgrades are nowadays.
davis.anthony
Veteran
My wish is for next gen to focus on physics.
The way we're going with graphics, we're going to have these game worlds that look photorealistic but have next to no interactivity.
The way we're going with graphics, we're going to have these game worlds that look photorealistic but have next to no interactivity.
Frenetic Pony
Veteran
I think MS's logic is that it "lost" this generation, and if it rushes out a new one that has enough PR it can score a "win".
A "mobile console that can play Series S games" is something I've seen people wish they could buy. Circa 2026 that and more should be possible, but I don't think MS wants "just" a mobile Series S. Defining it as a new generation would allow them to sell it free of any baggage the Series generation has.
Otherwise I don't know that even a $599 console in 2026 would be different enough. Star Wars Outlaws looks great, Hellblade II looks great, what more do you need?
There's possible future tech that could warrant a new console. A Vision Pro level headset you could wear for games could wirelessly connect to a console could wow people with more hardware power than the PS5 has, but I'm not sure VR is flexible enough for more than casual games yet. EG too many people get motion sick if you try sticking them in a virtual moving car; problems other than "headtracking and higher res screens" need to be solved for more than casual and fitness gaming to truly catch on there.
Similarly a "lightfield" (just think holographic) display would just work with games and be obviously way cooler, it also needs at least double the PS5's hardware power to run PS5 looking games (at least @4k). But that tech is stuck in demos like Google's Project Starline for now, and I don't think we'll have mass market adoption within 2 years.
Last edited:
Again, it's next to impossible to talk about hardware predictions if we're not talking about them in the context of the strategies behind them. Otherwise, it'd just be a very generic discussion about what tech will exist in 'x' years time.
B3D used to be able to do that just fine from 20 years ago. Check out this thread and see that the talk about market and strategies was kept to a minimum, a background to people's hardware thinkings:
How B3D used to handle a Hardware Prediction Thread
Last edited:
The idea of a MS handheld has me more interested recently. Such a machine could vary greatly depending on price alone, but I've been thinking of some baseline capabilities that the machine would need at a minimum. I'm assuming here it would be a console (universal precompiled smash-drivery low overhead stuff) rather than a PC (dynamic API stuff and on demand shader compilation).
So as a baseline I think you'd be looking at something like:
- Series S Back compat for instant high quality library
- A monolithic design
- 4 performance cores, 8 lower frequency high density cores
- ~20 CUs, 32 ROPs
- A fast NPU (though god knows MS probably won't use it)
- 16GB of LPDDR6
- 128-bit bus
- 32MB of GPU L3
- 512 GB SSD, 2.4 ~ 3.5 GB/s
- An SD card slot
- 1080p, 120hz IPS LED screen
Cost and power and super critical of course. Using a console OS and console games means higher performance for a given CPU, less power used for a given frame, and more consistent performance do the absence of certain API stages and driver translations and of course shader compilation. More stuff can be done offline because you know the precise hardware configuration beforehand.
4 fast cores and 8 high density cores would seem a good balance between performance for the handful of threads that limit performance and lower cost and power and for the jobs that can be highly threaded. A good NPU would allow for higher performance upscaling with lower power usage. A 128-bit bus again limits power, die area and board complexity, and a 32MB cache while taking die area improves performance and lowers off chip traffic, again saving power.
A 2.4 ~ 3.5 GB/s SSD would allow for perfect BC while only needing 2 PCIe lanes, again saving power and reducing board complexity. It could also allow for external Series console memory cards to be (re)used, though hopefully with newer more power efficient units being available.
Finally, a 1080p screen is plenty of Ps for a 7" display, and 120hz would allow for Series S 120hz games to be instantly available.
If there was the demand MS could also release 1TB and 2TB OLED devices for a higher retail price and perhaps the opportunity for some profit. There's also the option of selling a version of desktop Windows for $100+ opening up the machine for use as a surface tablet and ensuring they aren't making a loss on the device if people end up not buying many Xbox games on the device.
Such a machine would, I think allow MS to leverage many of the strengths at once: Xbox BC, the Xbox teams UI and controller experience, Gamepass, the Surface team's design experience, and even Windows itself.
So as a baseline I think you'd be looking at something like:
- Series S Back compat for instant high quality library
- A monolithic design
- 4 performance cores, 8 lower frequency high density cores
- ~20 CUs, 32 ROPs
- A fast NPU (though god knows MS probably won't use it)
- 16GB of LPDDR6
- 128-bit bus
- 32MB of GPU L3
- 512 GB SSD, 2.4 ~ 3.5 GB/s
- An SD card slot
- 1080p, 120hz IPS LED screen
Cost and power and super critical of course. Using a console OS and console games means higher performance for a given CPU, less power used for a given frame, and more consistent performance do the absence of certain API stages and driver translations and of course shader compilation. More stuff can be done offline because you know the precise hardware configuration beforehand.
4 fast cores and 8 high density cores would seem a good balance between performance for the handful of threads that limit performance and lower cost and power and for the jobs that can be highly threaded. A good NPU would allow for higher performance upscaling with lower power usage. A 128-bit bus again limits power, die area and board complexity, and a 32MB cache while taking die area improves performance and lowers off chip traffic, again saving power.
A 2.4 ~ 3.5 GB/s SSD would allow for perfect BC while only needing 2 PCIe lanes, again saving power and reducing board complexity. It could also allow for external Series console memory cards to be (re)used, though hopefully with newer more power efficient units being available.
Finally, a 1080p screen is plenty of Ps for a 7" display, and 120hz would allow for Series S 120hz games to be instantly available.
If there was the demand MS could also release 1TB and 2TB OLED devices for a higher retail price and perhaps the opportunity for some profit. There's also the option of selling a version of desktop Windows for $100+ opening up the machine for use as a surface tablet and ensuring they aren't making a loss on the device if people end up not buying many Xbox games on the device.
Such a machine would, I think allow MS to leverage many of the strengths at once: Xbox BC, the Xbox teams UI and controller experience, Gamepass, the Surface team's design experience, and even Windows itself.
Last edited:
cheapchips
Veteran
Is that CPU enough to run the equivalent of 16 Zen 2 threads at 3.4Ghz?
The NPU, 16Gb and SD card slot aren't useful for Series S compatibility. They're only worth it if it plays PC games too.
Just in general, power consumption is a huge challenge in reaching Series S performance.
Johnny Awesome
Veteran
Winning market share from Sony is a slow long game IMO. The primary thing for MS is to keep their 60 million fans happy while trying to slowly entice PS users with AAA content and more powerful hardware.
The key thing for MS from a hardware standpoint is to hold the power crown 75% of the time. So they'll lose it for one year in 2024 when the PS5 Pro arrives, but regain it from 2025 to 2028 until the PS6 arrives, and then regain it in 2029 etc... The Xbox fanbase won't care about Sony's temporary power lead if they know that a new powerful Xbox is always right around the corner.
I always find it silly when hardware is discounted as being important. It's almost the only thing people talk about on social media: "I thought the X was 20% more powerful than the PS5. Why does Sea of Thieves have worse shadows on the X? Why does such and such game lose 2 frames on the X every 1000 and the PS5 only lose 1 frame? etc..."
I find that MS only half learns the lessons it needs to. For instance, the OG Xbox generation showed them that hardware power wasn't the ONLY thing that mattered so then they didn't care about hardware power as much even though it was still super important with early adopters. The Xbox One should have been $50 more expensive to produce and have all of it's silicon budget in the APU instead of giving some to Kinect 2. Imagine a world where they sold 100 million Xbox Ones and lost an extra $100 per unit so they could be more powerful than PS4 and split the market 50/50. That would have cost them $10 billion, but they likely would have made up half of that in software revenues and then they wouldn't have needed to acquire ABK and Bethesda for $75B. So to save $5B they spent $75B. It's not a fully fleshed out argument, but you get my point. They scrimped on hardware and ended up spending more in the end anyway.
In 2025 they should put out a $599 machine with twice as much silicon as the X where they lose an extra $100 per unit and try to sell 40 million units in 4 years at a loss of $4B. Who cares? Become hardware dominant again. Sell boatloads of GP subscriptions. Blow people away with Gears 6 and Fable to the point that PS users are crying themselves to sleep at night clutching their puny little PS5 Pros.
Create envy MS! Hardware envy! Games envy! GP value envy!
Also, for the record I like the idea of shrinking the S into a handheld M. It doesn't need to be more than that and might not be possible until 2026 given power consumption issues.
Last edited:
Most games run in 900-1080p on Series S.
With a rendering resolution of 500-600p, the power consumption around 15 watts can be solved.
I expect so. For Zen 4 they call them "Zen 4" and "Zen 4C" cores, with the "4C" being the denser and lower clocked cores.
In the 8500G the Zen 4 and Zen 4C cores top out at 5.07 and 3.73 ghz respectively, and that's with substantial IPC gains over Zen 2. MS could disable the big cores and just run BC games on 8 of the compact cores, or they could probably do some fancy stuff with the scheduler to push the demanding threads onto the big cores and push everything else onto the "C" cores.
You could maybe even match the Series S performance with lower clocks due to much higher IPC, and perhaps even use boost to drop frequency and save power when you didn't need to be going full speed. BC games aren't going to use AVX512 either, so that's power consumption you wouldn't need to worry about.
For next gen games - a 2026 console will need to exist long after the Series S - 16GB has got to be the bare minimum, and a clamshell arrangement wouldn't be good for memory on a handheld anyway. 4 x 4GB seems like a sensible arrangement for a gaming handheld IMO.
The NPU would be for future upscaling tech, or anything else you wanted to run AI stuff for on next gen games. If MS was arsed they could bring an AI upscaler to the series consoles, and run an improved version on a newer handheld SoC with improved hardware.
The SD card slot would be for backing up games you can't fit on the internal SSD, or for transferring games between systems, something you can do with a USB HDD even on current Series consoles. It'd be useful even for Series S BC.
Definitely, which is things like using LPDDR on a modestly sized bus, or throttling down CPU cores when you can, or keeping with a wide-ish GPU at relatively low clocks, seem to be sensible choices (from here in my armchair). Also, moving AI inference models and all that to an NPU instead of doing them on shader cores would seem sensible in terms of energy cost. It's what both Intel and AMD have started to move towards in mobile chips.
Those social media contributors are a tiny minority of people who buy consoles. History has shown time and again a console doesn't have to be the most powerful to sell best, and chasing the power crown doesn't lead inevitably to a significant sales gain. Even this gen, PS5 is notably underpowered, but the market doesn't seem to care much! To get a significant power advantage now costs a fortune in more hardware. I think any and every console company should pursue the most efficient hardware design, a perfect balancing act to land right on the optimal cost/benefit sweet-spot.
Frenetic Pony
Veteran
Power consumption is the one thing that's actually advancing in silicon anymore. A Series S has all of 20CUs at 1.5ghz, in like two months you'll be able to buy an RDNA 3.5 laptop with 16CUs and > 1.8ghz clockspeeds, same thing at 15w this year just in terms of "teraflops".
CPU as well, relatively speaking. Zen 5c at 2.2ghz should have the IPC to match Zen 2 at 3.5ghz for gaming purposes. ROG Ally 2 at Series S level performance here we go!
Last edited:
Similar threads
