Sony PS6, Microsoft neXt Series - 10th gen console speculation [2020]

Is anyone else thinking that this upcoming generation will be far more impressive than the current one? 3nm will help though we don’t know how much yet. But the biggest difference will be that new consoles are likely targeting 1080p at best and leaving the rest to ML upscaling. So target resolution isn’t going to move much if at all. That should hopefully leave more headroom to produce better visuals.
I hope so! If Sony releases the PS6 in late 2028 then its more possible for this to happen. The 3 year gap between now and then will provide enough technical advancements.
 
Games that are 10+ years old are actually only of interest to 1-2% of potential users. MS is targeting the 100+ million gamers with their next gaming model who are mostly interested in new games.

I've been an Xbox fan since the beginning, collecting games with the idea that I'd play them all someday. Well, since then I've realized that I'm only interested in current games and I only play new ones.

Furthermore, if I still want to play an old game, there are emulators that are constantly improving.
Backwards compatibility is a huge advantage though, Sony proved as much with the PS2 and PS3. Being able to play your library of previous gen games, at least, should be a goal for both MS and Sony. The alternative is expecting your customers to throw out their old games and buying “remasters”. And now that I typed that out, of course that’s the route they’ll take if they can in this day and ages corporate environment…

Hardware wise it shouldn’t be a huge issue for Sony if they go with an iterative AMD solution to keep compatibility. If MS goes ARM they could still theoretically make a “Rosetta for games”.

On a different note, when do we expect the next consoles to arrive? 2-3 years? That’d narrow down what tech we can feasibly imagine them to be based on.
 
Backwards compatibility is a huge advantage though, Sony proved as much with the PS2 and PS3. Being able to play your library of previous gen games, at least, should be a goal for both MS and Sony. The alternative is expecting your customers to throw out their old games and buying “remasters”. And now that I typed that out, of course that’s the route they’ll take if they can in this day and ages corporate environment…

Hardware wise it shouldn’t be a huge issue for Sony if they go with an iterative AMD solution to keep compatibility. If MS goes ARM they could still theoretically make a “Rosetta for games”.

On a different note, when do we expect the next consoles to arrive? 2-3 years? That’d narrow down what tech we can feasibly imagine them to be based on.
late 2028 for Sony
 
When PS6 comes out, it's not just the hardware that has to be ready (complete ray tracing architecture that speeds up most of the pipeline + a better architecture for ML tasks), but the software too.

FSR 4 is already really good, but they will have to develop a denoiser + ray reconstruction solution and an updated frame generator. It's a lot of software to make, so maybe 2028 is more realistic than 2027 for a launch. Then again, they made FSR 4 really fast, so who knows.
 
When PS6 comes out, it's not just the hardware that has to be ready (complete ray tracing architecture that speeds up most of the pipeline + a better architecture for ML tasks), but the software too.

FSR 4 is already really good, but they will have to develop a denoiser + ray reconstruction solution and an updated frame generator. It's a lot of software to make, so maybe 2028 is more realistic than 2027 for a launch. Then again, they made FSR 4 really fast, so who knows.
What updated frame generator? Multiframe? Based on code FSRFG has been built for it since day 1 and is it really something anyone wants or needs? The latency doesn't go down and TVs don't go over 120 Hz to begin with.
 
I was thinking that they might want to upgrade the frame generator with ai since now they are all in on that.
They probably will, but only once they have enough TOPS to run upscaling, denoising, and frame generation simultaneously. And instead of pursuing MFG at fixed generation factors, I believe AMD should go for Adaptive Frame Generation like Lossless Scaling has, where it always ouputs at screen refresh rate regardless of input frame rate. That's probably Nvidia's next step anyways.
 
When PS6 comes out, it's not just the hardware that has to be ready (complete ray tracing architecture that speeds up most of the pipeline + a better architecture for ML tasks), but the software too.

FSR 4 is already really good, but they will have to develop a denoiser + ray reconstruction solution and an updated frame generator. It's a lot of software to make, so maybe 2028 is more realistic than 2027 for a launch. Then again, they made FSR 4 really fast, so who knows.
Yup thats why I think end of 2028 is the most likely outcome for a release date.
 
Yup thats why I think end of 2028 is the most likely outcome for a release date.
So that's a few years out. Development would already have been ongoing in the preliminary stage. I assume Sony and AMD are deeply engaged with each other already, and if it follows a similar path as the PS5 I suppose the PS6 would sport a Zen ~6-7 core depending on development and release date, and a custom RDNA 5 based core? Whatever that will look like. Storage might get a tad quicker too I suppose. Feels like it could well be a PS5 2.0. Not terribly exotic until you get into finer details.

In either case I'm going to assume that studios will target a range of resolutions from 1080p to1440p or thereabouts, and rely on up-scaling to get the rest of the way to 4K in most - if not all - cases.

Microsoft seems like the wild card, considering talk of both handhelds and speculation of standard-based boxes in lieu of their own. Certainly going to be interesting to see what happens there. Could Sony follow suit and make their own portable? If they want to be wild I wouldn't mind seeing them make their own version of the Meta Quest. Go all in on portable VR with a cable to use the PS6 for a massive performance boost?
 
If next-gen has 3D V-Cache that be fairly exotic compared to anything else AMD has put in a console. There were rumors that the Xbox would have an NPU too. If either console does, it would be interesting to how developers juggle having both an NPU and a GPU with matrix units (and a CPU with AVX-512 as well).
 
I’ll preface this with what might become obvious when I ask my questions: I have no idea how NPU’s work.

A) what would they contribute in a game?

B) would the hardware be useable for other tasks or would it be so much dead silicon unless “AI” tasks were asked for?
 
I don't even see what an NPU adds compared to AI acceleration inside the compute units like in Nvidia and AMD modern GPU's. If anything, it sounds like it would increase latency.
 
I don't even see what an NPU adds compared to AI acceleration inside the compute units like in Nvidia and AMD modern GPU's. If anything, it sounds like it would increase latency.
I'm pretty sure if given the choice, developer would rather run for example local language models on separate NPU leaving GPU for graphics related tasks.
 
Unless a NPU is significantly more transistor/power efficient than running it on the GPU it seems like you'd just be trading off GPU performance for it.

On the other hand having it separate would give developers a more consistent AI performance target as well as a use it or lose it incentive to leverage it versus ignoring it for graphics if shared with the GPU.
 
It's also a marketing bullet point. Blast processing 10.0!

Ironically we seem to be headed to more bespoke hardware next gen for upscaling, ML etc.... I do think there will be a cloud-compute component to the next Xbox as well. Hard to say what high-latency things can be offloaded that way, but I think it will be the test case for the fully in the cloud Gen11 or Gen12 Xbox.
 
Making a console dependent on cloud computing would really alienate rural communities and parts of the world with less than stellar internet connectivity. Or what I’d call a dick move personally.
 
So that's a few years out. Development would already have been ongoing in the preliminary stage. I assume Sony and AMD are deeply engaged with each other already, and if it follows a similar path as the PS5 I suppose the PS6 would sport a Zen ~6-7 core depending on development and release date, and a custom RDNA 5 based core? Whatever that will look like. Storage might get a tad quicker too I suppose. Feels like it could well be a PS5 2.0. Not terribly exotic until you get into finer details.

In either case I'm going to assume that studios will target a range of resolutions from 1080p to1440p or thereabouts, and rely on up-scaling to get the rest of the way to 4K in most - if not all - cases.

Microsoft seems like the wild card, considering talk of both handhelds and speculation of standard-based boxes in lieu of their own. Certainly going to be interesting to see what happens there. Could Sony follow suit and make their own portable? If they want to be wild I wouldn't mind seeing them make their own version of the Meta Quest. Go all in on portable VR with a cable to use the PS6 for a massive performance boost?
They already made a portable with the portal. I think they'll build on that as a complement to the PS6. Your PS6 can be your dedicated game server and you can play from anywhere in the world. For the hw architecture its still too early to tell but it will have the latest CPU and GPU architectures of the year it releases in.

Xbox seems to be flirting with becoming a third party publisher. This news just came out today:


So you can imagine the next gen Xbox may not be a closed system
 
Back
Top