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

[Remij]

Could ML incorporate something like a lower resolution rendered frame to help in the frame gen process? Perhaps at a 360p intermediate frame and use the data from it to help generate the rest of the frame.

What you're describing is what AI reconstruction does before frame generation. Frame gen already analyses multiple full frames to predict what it should look like when it generates the in-between frames.

Reconstruct from lower res frames > analyze full frames > generate in-between frames

 

[Ikshvaku]

With 3D stacked EDRAM, two SOCs makes even more sense IMHO. So certainly hope that's on the cards. A big video RAM boost, latency and bandwidth, would benefit all GPU and AI work a great deal methinks.

Is there any chance of HBM being cost effective and be in the running for inclusion in PS6.

 

[RDGoodla]

As much as I'd love that, a PS5-level portable isn't on the cards any time soon. Even on TSMC's A16 it would be in the realm of 46W.

The trouble though, is bandwidth. You'd need 12Gbps LPDDR6 on an enormous 336 bit bus. Maybe LPDDR7 will be viable in a few years, but I can't even find any rumours about it.

The 4nm handheld APU has 8 cores ZEN 4 and 12CUs.

CPU operates at 3.8 GHz and GPU can run at 2.4 GHz with 30W power. Assuming GPU using 15 W and scaling to 36 CUs, 4nm APU consumes 60 W at PS5 level.

It is very likely TSMC A14 can lower the APU power to less than 30W.

Memory bandwidth is indeed a concern. It needs 18.7 Gbps LPDDR 6 at 192 bits. This year we already have 10.7 Gbps LPDDR 5X and we have 4 to 5 years to release PS5 level handheld.

 

[mpg1]

The 4nm handheld APU has 8 cores ZEN 4 and 12CUs.

CPU operates at 3.8 GHz and GPU can run at 2.4 GHz with 30W power. Assuming GPU using 15 W and scaling to 36 CUs, 4nm APU consumes 60 W at PS5 level.

It is very likely TSMC A14 can lower the APU power to less than 30W.

Memory bandwidth is indeed a concern. It needs 18.7 Gbps LPDDR 6 at 192 bits. This year we already have 10.7 Gbps LPDDR 5X and we have 4 to 5 years to release PS5 level handheld.

Launching a device on a bleeding edge process node that still uses ~30W is probably still a non-starter in terms of cost and battery life.

 

[QPlayer]

I would like to remind people that the same graphics with a handheld console do not require the same bandwidth, because the rendered resolution will be much lower due to the few inches of screen. While the XSX or PS5 runs at 1080-1440p, a mobile version will settle for 540-720p native rendering and increasingly better image scaling techniques.

So if you look at it this way, within 1 or 2 years there will be a technology that can deliver current generation graphics quality to a handheld console at an affordable price. LPDDR6, which will debut next year, will be perfectly fine for this.

 

[Dictator]

I would like to remind people that the same graphics with a handheld console do not require the same bandwidth, because the rendered resolution will be much lower due to the few inches of screen. While the XSX or PS5 runs at 1080-1440p, a mobile version will settle for 540-720p native rendering and increasingly better image scaling techniques.

So if you look at it this way, within 1 or 2 years there will be a technology that can deliver current generation graphics quality to a handheld console at an affordable price. LPDDR6, which will debut next year, will be perfectly fine for this.

I agree with that, only issue is Backwards compatible titles on PS5 that render at high res. Sony has yet to post-release edit Games to change render res, I am Not even sure If they can.

 

[RDGoodla]

In that case would Sony mandate PS6 games to be compatible with their handheld?

PS6 games all have mobile versions but not all features are ported, such as split screen.

Launching a device on a bleeding edge process node that still uses ~30W is probably still a non-starter in terms of cost and battery life.

ROG ALLY X uses 80Wh battery. And it may
only increase BOM less than $50. It should be cheaper 4 years later.

I agree with that, only issue is Backwards compatible titles on PS5 that render at high res. Sony has yet to post-release edit Games to change render res, I am Not even sure If they can.

It is very hard to patch all PS5 games . So a handheld running PS5 code is still very useful.

And again it raises a question. Will Sony just wait so long to release a handheld in 2028? Or will Sony release a handheld in 2025 with new AMD mobile APU, which runs some PS5 games with patch and every PS4 game?

 

[Tkumpathenurple]

And again it raises a question. Will Sony just wait so long to release a handheld in 2028? Or will Sony release a handheld in 2025 with new AMD mobile APU, which runs some PS5 games with patch and every PS4 game?

The Switch launched in 2017, and its successor is launching next year, so it'll be around for a while. That will be Sony's competition; a PS4 kind of portable would have enough power to go toe to toe with it, current manufacturing processes can make it, current memory can feed it, and it needn't cost the earth.

18CU's and 8c/16t could clock at 1.115GHz and 1.75GHz respectively when portable, and 2.233GHz and 3.5GHz when docked. This would mean any developer that can get their game running in portable mode would be able to run it at 60 when docked, double the resolution on the PS5, and PSSR to 4K60 on the Pro.

My only concern is whether a Zen 2 at 1.75GHz be able to match the Switch 2's CPU in portable mode?

 

[raytracingfan]

And again it raises a question. Will Sony just wait so long to release a handheld in 2028? Or will Sony release a handheld in 2025 with new AMD mobile APU, which runs some PS5 games with patch and every PS4 game?

If Sony was going to release a new console SKU next year, we would already know about it. It was already known that the PS5 Pro was in development last year. We have no reason to believe that Sony is going to muddy the waters by releasing a third hardware target that is weaker but more expensive than the base PS5 just when developers have finally abandoned the PS4. The PlayStation Portal is selling great, Sony doesn't need another portable device this generation.

 

[davis.anthony]

With how slow technology is these days lets hope Sony release PS6 with Emotion Engine 2.0 and Graphics synthesizer 2.0

Emotion Engine 2.0

5Ghz clock speed
16 VU's
Ray intersection units a part of the VU's
Dedicated BVH generation hardware
256MB onboard unified cache

Graphics synthesizer 2.0

8GB EDRAM
10TB/s bandwidth
Texture compression

It's Monday and I'm bored.....

 

[Charlietus]

With how slow technology is these days lets hope Sony release PS6 with Emotion Engine 2.0 and Graphics synthesizer 2.0

Emotion Engine 2.0

5Ghz clock speed
16 VU's
Ray intersection units a part of the VU's
Dedicated BVH generation hardware
256MB onboard unified cache

Graphics synthesizer 2.0

8GB EDRAM
10TB/s bandwidth
Texture compression

It's Monday and I'm bored.....

Reject AMD

Bandwidth

 

[fehu]

This is something more doable by microsoft than sony, but what about a traditional amd console, and an arm portable?

 

[McHuj]

Is there any chance of HBM being cost effective and be in the running for inclusion in PS6.

There's always a chance, but probably unlikely unless there's a major cost reduction that happens. Many of the memory companies are exploring HBM without a silicon interposer (using glass for example) to reduce the cost, but unless something like that hits, i doubt it happens.

What's an interesting though exercise is at what point does using HBM break even with the cost of GDDR. With GDDR7, you're looking at 8-10 chips on a 256-320-bit bus giving you somewhere in the 1-1.5 TB/sec bandwidth. That all could be replaced by a single H

 

[Cappuccino]

I’m not sure why everyone is convinced we need console handhelds again. Steam Deck sold a few million units, which is good but it’s clearly a niche product. For reference even the ‘loser’ console (Series X/S) sold over 20 million. Where is the demand for this class of product that would justify the amount of R&D required to make it?

 

[fehu]

Switch

 

[MrSpiggott]

Switch is aimed at a different market though isn't it? Often bought by parents who don't want the family TV tied up or have their children disappearing into bedrooms on their own for hours on end. I wonder how many actually leave the house? I haven't seen many.
Is there a call for a very powerful, and therefore very expensive, handheld console? Steam Deck would suggest the market is quite small.

 

[davis.anthony]

Switch

Switch didn't really need any R&D from Nintendo as they used an SOC that cost Nvidia to R&D.

 

[QPlayer]

I’m not sure why everyone is convinced we need console handhelds again. Steam Deck sold a few million units, which is good but it’s clearly a niche product. For reference even the ‘loser’ console (Series X/S) sold over 20 million. Where is the demand for this class of product that would justify the amount of R&D required to make it?

1. Steamdeck =/= all handheld PCs
2. Steamdeck has only been on the market for 2.5 years
3. Switch ~ 150 million
4. With a modern handheld console, a new generation can be launched with a much more popular price, considering the current production costs.

Presumably, the console manufacturers have much more insight into these numbers.

 

[Cappuccino]

1. Steamdeck =/= all handheld PCs
2. Steamdeck has only been on the market for 2.5 years
3. Switch ~ 150 million
4. With a modern handheld console, a new generation can be launched with a much more popular price, considering the current production costs.

Presumably, the console manufacturers have much more insight into these numbers.

1. Sure, but it’s easily the most successful and the best of the bunch. You can add up all the different models and it’s all an order of magnitude less than say PS5 I would imagine.

2. Even if you chart the sales linearly to match the 4 years Xbox and PS5 have been out it’s still a small fraction.

3. The Switch didn’t sell 150 million units because it was a handheld and I suspect you know this. This is like Sony making Move in response to the Wii without understanding that it wasn’t motion controls that made the Wii what it was, it was Nintendos first party output and popularity among families. An Xbox or PS handheld isn’t going to have the same appeal.

4. I don’t think cost is the issue here.

I think the main thing that makes an Xbox or PS handheld silly is adults buying these units have no reason to buy it. When I was a kid I had a DS so I could play with friends on the bus or at recess, what is the adult analog to this? That’s why these devices will always remain niche, and of the adults that I know have a SD most sit unused because most of the time if they have some time to game they are also about 10 feet from their normal gaming setup.

I think Xbox and PS focusing on handhelds over improving the value proposition of home consoles will just accelerate the death of traditional home consoles.

 

[Cappuccino]

Switch

My above post kind of already responded to this but to be very clear on this: trying to replicate the success of Switch by just making an Xbox handheld will fail, because the reason Switch succeeded is because kids love Mario and handhelds. Nobody is going to buy a device to play Senua’s Sacrifice and Halo Infinite on the go.