Viability of a new portable PlayStation

[Tkumpathenurple]

Back on topic. Now, I may look like this:

... but I'm going to go for it anyway.

The following site conducted a test on the power draw of a laptop with an AMD Ryzen 2500U - 4 core, 8 thread Zen core, clocked between 2.0-3.6GHz ; 8 Vega CU's clocked at 1.1Ghz.

https://www.pcper.com/reviews/Mobil...-Raven-Ridge/Power-Consumption-and-Software-C

The screen was off, so we need to take that into consideration, but the highest power draw came in at 43.1 watts.

We've already discussed a portable PS4, and reached the conclusion that it's possible, although memory would be an issue. So, I thought, what does that figure look like with PS4Pro performance?

2500U power draw / 2500U clockspeed * PS4Pro clockspeed
=
43.1 / 1100 * 911
=
35.69

The PS4Pro has 4.5x more CU's, so:

35.69 * 4.5
=
160.61

Which is too high for even a laptop. But that isn't factoring in GlobalFoundries' claims for 7nm: a greater than 60% reduction in power.

160.61 * 0.4
=
64.24

Which seems quite doable.

So, I say Sony release a PS4Micro next year on an immature 7nm node. The year after they release the PS5 in the brain and shell format discussed previously.

 

[Picao84]

Which is too high for even a laptop. But that isn't factoring in GlobalFoundries' claims for 7nm: a greater than 60% reduction in power.

160.61 * 0.4
=
64.24

Which seems quite doable.

So, I say Sony release a PS4Micro next year on an immature 7nm node. The year after they release the PS5 in the brain and shell format discussed previously.

Is that PS4Micro supposed to be mobile like Nintendo Switch? If so, not a chance a 64W chip would be used for such a thing....

 

[Deleted member 13524]

I think a PS4 Portable is not only possible in 7nm but also very probable, as long as Sony is willing to spend money on a HBM stack (or someone has secretly been evolving Wide IO).

But a PS4 Pro portable seems very unlikely to be honest.

 

[Tkumpathenurple]

May I just contest the movement of these posts please? Most of my post was referencing the power consumption of a PS5 brain, running PS4Pro code. I know that wasn't especially clear (I was typing it in the train to work) but I do think it's relevant, and I would prefer to carry on the discussion in the next gen thread.

Is that PS4Micro supposed to be mobile like Nintendo Switch? If so, not a chance a 64W chip would be used for such a thing....

I think a PS4Micro should be sold as a brain + a simple dock. The simple dock consisting of 3 or 4 USB 3.1 type-C ports, the camera port, HDMI and AV out, and power.

Separately, they should sell a portable shell. IMO, a laptop form factor would be best, as it could include a ~10" screen, a 2.5" HDD, dual cameras, and hardware for wireless VR transmission. Most importantly, it could contain a big ass battery.

AMD's 11CU APU already reaches 1.4TF at laptop power draw, so near-PS4 performance is already possible.

Then comes the PS5, which I think should follow a similar route to the Switch and the hypothetical PS4Micro. Vega 64 seems like a reasonable target, as that GPU at 14nm is something like 450mm2.

The PS5 brain runs all CU's at full speed when docked. When placed in the laptop shell, it can operate in one of a few configurations:
PS4 compatibility mode - caps the clockspeeds and number of CU's at base PS4 levels.
PS4Pro compatibility mode - same principal as the above.
PS5 portable mode - all/most CU's running at a much lower clockspeed.

Don't mandate any of the above to PS5 developers, just let them use whichever configuration they want. Give them the flexibility to, in essence, take their PS4+Pro code and use that as the PS5's portable mode code. Or, if they wish to do so, create a more bespoke version for the latter configuration.

 

[DieH@rd]

But a PS4 Pro portable seems very unlikely to be honest.

The only thing that matters is running PS4 software on the go, which makes Pro Portable not only unlikely, but also completely unnecessary.

But some remnants of Pro OS stack could survive even on portable. If they can achieve higher clocks for CPU/GPU, they could give users option to activate "boost mode", and see that at least some performance stats go up.

 

[Tkumpathenurple]

I agree that the highest priority is a portable PS4, and that Pro code doesn't matter for that device. But, I think it would be an excellent move for the PS5 to run Pro code when portable, and would help a narrative surrounding its power.

 

[Deleted member 13524]

But some remnants of Pro OS stack could survive even on portable. If they can achieve higher clocks for CPU/GPU, they could give users option to activate "boost mode", and see that at least some performance stats go up.

If they can achieve that then why not.

But TBH I'd rather have full, constant PS4 performance and more battery life than trying to boost the base PS4.

 

[DieH@rd]

PS5 to run Pro code when portable

PS5 will be a traditional beastly home console. As for BC, I expect native code execution so it will easily run old games that are booted in their Pro mode [and for the games without the Pro mode, I expect even better boost mode [since clocks will be higher on everything]].

 

[phoenix_chipset]

Even if Sony were to make a ps4 go it'll be years before mobile hardware reaches that point. Not that I think they'll even attempt it after vita. So i'm just wondering when Nintendo can match base ps4 performance... probably 4 or 5 years but with more memory at least

 

[Deleted member 13524]

Even if Sony were to make a ps4 go it'll be years before mobile hardware reaches that point.

It will be possible later this year with an optimized 7nm shrink of Liverpool (+ integrated southbridge with the ARM CPU) and a single HBM2 stack.

No need for interposer, just a single HBM2 stack clocked at 1.4Gbps (total 179GB/s) on top of the SoC using TSV to save power, like the Vita.
Jaguars in Liverpool are already very modestly clocked and GCN CUs should be sipping power at 800MHz, if Raven Ridge is anything to go by.
Cooling for 15W on a thick 8" tablet and then it'd need something like 40Wh worth of batteries.
Since LPDDR5 is aiming for 6400MT/s, they could go with 192bit LPDDR5 (total 179 GB/s, 6 x32 chips). Though that certainly wouldn't be available in time for 2018, even 2019 could be hard, the PCB space required for 6 RAM chips wouldn't be great for a mobile solution and since manufacturers probably won't be making LPDDR5 chips with less than 16Gbit capacity (or even more), so the console would end up with 12GB RAM, 4 of which completely unused I guess..


I'd pay ~500€ for that. Day one adopter.
That would be for the 256GB eMMC 5.1 storage version, with a SD card slot (you heard that Sony? Secure Digital cards, not your overpriced proprietary crap that is worthless everywhere else!)

No need for dock other than providing its own power source for external USB power, the PS4 camera connector and ethernet controller+port, though this could be sold as a ~50€ extra, to be honest.

 

[Deleted member 13524]

Bringing the discussion to its proper place.

Even if Sony were to make a ps4 go it'll be years before mobile hardware reaches that point. Not that I think they'll even attempt it after vita.

It will be possible later this year with an optimized 7nm shrink of Liverpool (+ integrated southbridge with the ARM CPU) and a single HBM2 stack.

No need for interposer, just a single HBM2 stack clocked at 1.4Gbps (total 179GB/s) on top of the SoC using TSV to save power, like the Vita.
Jaguars in Liverpool are already very modestly clocked and GCN CUs should be sipping power at 800MHz, if Raven Ridge is anything to go by.
Cooling for 15W on a thick 8" tablet and then it'd need something like 40Wh worth of batteries.
Since LPDDR5 is aiming for 6400MT/s, they could go with 192bit LPDDR5 (total 179 GB/s, 6 x32 chips). Though that certainly wouldn't be available in time for 2018, even 2019 could be hard, the PCB space required for 6 RAM chips wouldn't be great for a mobile solution and since manufacturers probably won't be making LPDDR5 chips with less than 16Gbit capacity (or even more), so the console would end up with 12GB RAM, 4 of which completely unused I guess..


I'd pay ~500€ for that. Day one adopter.
That would be for the 256GB eMMC 5.1 storage version, with a SD card slot (you heard that Sony? Secure Digital cards, not your overpriced proprietary crap that is worthless everywhere else!)

No need for dock other than providing its own power source for external USB power, the PS4 camera connector and ethernet controller+port, though this could be sold as a ~50€ extra, to be honest.

 

[HBRU]

wow... another way to extend PS4 commercial span and so also life... yes sure Sony is interested in whatever helps... Also a lot of new (and old) titles can be produced and then sold into another market... It makes a lot of commercial sense... Also to keep feeded the old PS4 installed base with new titles...

 

[Tkumpathenurple]

Bringing the discussion to its proper place.



It will be possible later this year with an optimized 7nm shrink of Liverpool (+ integrated southbridge with the ARM CPU) and a single HBM2 stack.

No need for interposer, just a single HBM2 stack clocked at 1.4Gbps (total 179GB/s) on top of the SoC using TSV to save power, like the Vita.
Jaguars in Liverpool are already very modestly clocked and GCN CUs should be sipping power at 800MHz, if Raven Ridge is anything to go by.
Cooling for 15W on a thick 8" tablet and then it'd need something like 40Wh worth of batteries.
Since LPDDR5 is aiming for 6400MT/s, they could go with 192bit LPDDR5 (total 179 GB/s, 6 x32 chips). Though that certainly wouldn't be available in time for 2018, even 2019 could be hard, the PCB space required for 6 RAM chips wouldn't be great for a mobile solution and since manufacturers probably won't be making LPDDR5 chips with less than 16Gbit capacity (or even more), so the console would end up with 12GB RAM, 4 of which completely unused I guess..


I'd pay ~500€ for that. Day one adopter.
That would be for the 256GB eMMC 5.1 storage version, with a SD card slot (you heard that Sony? Secure Digital cards, not your overpriced proprietary crap that is worthless everywhere else!)

No need for dock other than providing its own power source for external USB power, the PS4 camera connector and ethernet controller+port, though this could be sold as a ~50€ extra, to be honest.

Largely agreed. The dock would be fine as an extra, but I think an HDMI out port would be wise too, so the portable PS4 could fully replace the current base model.

I think it'd be great to see that this year, along with a revised Pro. Zen cores instead of Jaguar would probably make a big enough difference to cooling and noise.

 

[DieH@rd]

I would very much like PS4 Portable, but I won't be willing to pay 500€.

 

[Panino Manino]

Apparently Sony is thinking how they can re-enter the handheld market with a Vita successor.
If this does happen, what are the chances of Sony exploiting that AMD+Samsung partnership regarding the GPU? For example, a Vita with Exynos Mx CPU and Navi GPU?
An ARM CPU would still make sense instead of low clocked Zen 2 cores?

 

[MrFox]

Apparently Sony is thinking how they can re-enter the handheld market with a Vita successor.

Citation needed.

 

[Shifty Geezer]

Apparently Sony is thinking how they can re-enter the handheld market with a Vita successor.
If this does happen, what are the chances of Sony exploiting that AMD+Samsung partnership regarding the GPU? For example, a Vita with Exynos Mx CPU and Navi GPU?

I think you really need to rewind to 'what are the chances of Sony re-entering the handheld market with a Vita successor?" You'd need to know Sony's intentions to consider what the hardware might be.

 

[DieH@rd]

Apparently Sony is thinking how they can re-enter the handheld market with a Vita successor.

The easiest way they can do that is with a portable PS4.

 

[manux]

copy quest and make a standalone vr headset with no wires. That would be awesome in so many ways and also circumvent the "but I already have a phone and games on it" argument.

 

[Rootax]

VR brings nothing for a lot of games.

The main question for me is, will their new portable be compatible with the existing ps4 librairy, or will they start from scratch.