Predict: Next gen console tech (10th generation edition) [2028+]

But! Next Xbox, if it launches in 2026, could be a 20CU 6tf machine for $399, then one with a $599 60Cu/24tf chip and a vastly more capable cooling system for when you plug it in. This gives customers of the bigger one a fast stationary machine and a mobile one at the same time.
I also find the same solution suitable. A dual system that practically copies the model of the N Switch, only in a much stronger design. However, it will be at least a 50-60 TFlops console in docked mode, as MS talked about a significant performance jump.
 
The docking part contains a high-end GPU and plenty of fast RAM, to which the basic mobile unit is connected via an ultra-fast PCI express connector.

The mobile Xbox can come with a normal price, for example $399, which together with the docking version can be up to $799.
 
You're never going to grt away from a dock needing it's own RAM are you? That would probably make it a challenge to have it cost much less than a dedicate home console.
 
You're never going to grt away from a dock needing it's own RAM are you? That would probably make it a challenge to have it cost much less than a dedicate home console.
The mobile unit would have the CPU with a variable clock signal and 8 GB DDR6, which would be enough for a 600p render (with FSRX) resolution graphics intended for a 7.5-inch 900p screen, with a quartered texture resolution. And the docking unit has a powerful GPU and 24GB GDDR7. Smaller textures will probably be enough as AI can quadruple the resolution of textures when the console is docked. Thus, the games can fit in a much smaller space, which will be needed in the future.
 
Last edited:
Putting anything in a dock, even the dock itself, is unnecessary. Chips aren't big, cooling used to be big but now there's solid state coolers that can cool 30w+ and are tiny. People would lose the dock, you'd have to do a ton of complicated engineering to make anything useful be in the dock that doesn't just copy what the mobile console does.

You can put a PS5 Pro into a steamdeck or smaller, chipwise, today. By 2026 you should be able to put a cooling system into one as well.
 
Putting anything in a dock, even the dock itself, is unnecessary. Chips aren't big, cooling used to be big but now there's solid state coolers that can cool 30w+ and are tiny. People would lose the dock, you'd have to do a ton of complicated engineering to make anything useful be in the dock that doesn't just copy what the mobile console does.

You can put a PS5 Pro into a steamdeck or smaller, chipwise, today. By 2026 you should be able to put a cooling system into one as well.
The dock is also necessary because a modern console GPU uses at least 200 watts, and even then we are talking about modest numbers compared to expensive VGAs. A 200-watt GPU cannot be included in a mobile unit, it is impractical just because of the power supply. Or do you think that by 2026 it will be possible to scale an extremely powerful GPU using 200 watts at 4K resolution, which consumes only 15 watts at 600p render resolution? And can all this be solved with GDDR7 memory on board in a mobile console?
 
The dock is also necessary because a modern console GPU uses at least 200 watts, and even then we are talking about modest numbers compared to expensive VGAs. A 200-watt GPU cannot be included in a mobile unit, it is impractical just because of the power supply. Or do you think that by 2026 it will be possible to scale an extremely powerful GPU using 200 watts at 4K resolution, which consumes only 15 watts at 600p render resolution? And can all this be solved with GDDR7 memory on board in a mobile console?

The problem is though, if you have such a massive power disparity between docked/undocked 'mode', then what have you gained exactly by having it be sold as 'one' console?

You've not making developers lives significantly easier if the docked mode ramps up the power 10X relative to the handheld, you've just in effect created 2 seperate consoles that will basically require separate ports to actually take advantage of them.
 
The problem is though, if you have such a massive power disparity between docked/undocked 'mode', then what have you gained exactly by having it be sold as 'one' console?

You've not making developers lives significantly easier if the docked mode ramps up the power 10X relative to the handheld, you've just in effect created 2 seperate consoles that will basically require separate ports to actually take advantage of them.
The games that run on 7900XTX are not completely separate ports on Steam Deck either, but the same game runs on both systems.
 
PS5 Pro has made me skeptical about the next gen consoles.

If Sony can only offer a 45% increase in the 4 years between base PS5 and PS5 Pro.

What can they realistically offer in the ~4 years between PS5 Pro and PS6?

Another 45% again?

The 7900XTX is 83% faster than the 7700XT and if the PS5 Pro uplift is anything to go by, PS6 will be lucky to get 7900XTX performance.

I think consoles are in trouble after this generation when it comes to worthwhile performance increases and maintaining a reasonable price.
 
PS5 Pro has made me skeptical about the next gen consoles.

If Sony can only offer a 45% increase in the 4 years between base PS5 and PS5 Pro.

What can they realistically offer in the ~4 years between PS5 Pro and PS6?

Another 45% again?

The 7900XTX is 83% faster than the 7700XT and if the PS5 Pro uplift is anything to go by, PS6 will be lucky to get 7900XTX performance.

I think consoles are in trouble after this generation when it comes to worthwhile performance increases and maintaining a reasonable price.
My hope is PS6 can shoot for 2028 and get on N2P with backside power delivery for higher speed and efficiency as well as chiplets joined with infinity fanout links to drastically cut cost. Then pair with Zen 6 & "RDNA6". They could also bolt 3D cache on which'll probably be a lot cheaper than making a bigger die.

In addition, another 4yrs maturation/development in techniques, dev experience, hardware effectiveness and model training for ML/AI, allowing for 4x scaling with better-than native results and 2x frame-gen with minimal latency; all with lower frame-time costs than we're seeing for this implementation.

I think it's going to be a case of multiple factors compounding each other.

It's also likely to be the first gen with no jump in target display resolution (4K will still be the main res for almost everyone). Subsequently -- along with the maturer ML scaling -- power will be going towards FX, visual quality and general compute rather than just more and more pixels. And hopefully the system as a whole can start doing more interesting things such as simulation and pushing fundamental game design. RT, mesh shaders, micro geometry will all be much maturer concepts too.

I think this gen has been very transitional on multiple fronts as well as having multiple impediments (pandemic, prolonged cross-gen and the holdover of having developed for an underclocked circa 2011 netbook CPU for the last 10yrs).

But I think a lot of it is going to converge next gen, so I'm actually very optimistic.
 
Last edited:
PS5 Pro has made me skeptical about the next gen consoles.

If Sony can only offer a 45% increase in the 4 years between base PS5 and PS5 Pro.

What can they realistically offer in the ~4 years between PS5 Pro and PS6?

Another 45% again?

The 7900XTX is 83% faster than the 7700XT and if the PS5 Pro uplift is anything to go by, PS6 will be lucky to get 7900XTX performance.

I think consoles are in trouble after this generation when it comes to worthwhile performance increases and maintaining a reasonable price.
It's because PS5 Pro does not use the best process available. Therefore PS6 needs to use the most advanced process and it can easily have 2x improvement.
 
It's because PS5 Pro does not use the best process available, while PS4 Pro used the best 16 nm process in 2016.
Therefore PS6 needs to use the most advanced process and it can easily have 2x improvement.

Is 2x enough though?

That'll barely put it above the 7900XTX, which you can buy right now.

Then when RDNA4 releases at the end of this year, you'll potentially be able to build a PC faster than PS6, 4 years before even PS6 releases.

And by the time PS6 releases in 2028, would could be looking at RDNA6 releasing and suddenly that 2x increase is looking terrible.

This is what I'm talking about, next gen is going to be so rough for consoles (And Sony/Microsoft) as it takes such as long time these days to get past generation enthusiast performance, down to entry level pricing that the consoles need.

How long will take that $250-300 GPU market that the consoles target to reach 7900XTX performance?
 
you'll potentially be able to build a PC faster than PS6, 4 years before even PS6 releases.

Same performance predictions had been thrown around with kepler, pascal and vega in relation to consoles coming in next 3 years. As usual, whatever comes out this year will be featureless useless space heater by then, especially in current machine learning/raytracing age.
 
In addition, another 4yrs maturation/development in techniques, dev experience, hardware effectiveness and model training for ML/AI, allowing for 4x scaling with better-than native results and 2x frame-gen with minimal latency; all with lower frame-time costs than we're seeing for this implementation.
All super lame though. I remember 10x everything Moare!! Those were console generations. The spirit of console generations is definitely dead and the magic and excitement of new hardware and a new level of gaming experience is something future generations will never know.
 
Do console gamers really care?

If next gen machines deliver full RT w/4x ML upscalling and 3x pixel power for $599, it'll be plenty.

That's the issue though.

3x pixel power in the GPU tier consoles typically shop in is wildly unrealistic, as that performance tier doesn't increase enough gen on gen to get to 3x.

In the $300 to $500 GPU range, a 40% increase is higher than it has been for years now, but let's role with it.

40% increase in the mid-range with RDNA5 and another 40% increase when RDNA6 lands.

That's only 80% across 2 generations and only lands the pixel power at 7900XTX level.

3x base PS5 should be possible, but is that enough as that's even less than a 7900XTX.

Could there be some breakthrough that has huge gen on gen increases? Of course there could.

But looking at how poorly that midrange tier has scaled on PC in terms of gen on gen performance gains over the past generations it seems unlikely.
 
Back
Top