Predict: Next gen console tech (9th iteration and 10th iteration edition) [2014 - 2017]

Status
Not open for further replies.
The problem WRT consoles is that it's still quite expensive.

The major benefit that it has over NAND is that speed isn't reliant on number of channels, hence a 32 GB caching drive (like the retail Optane unit) doesn't suffer drastically reduced speed as NAND based drives do when compared to much larger capacity drives.

There are certainly benefits for console use, but until the price comes down significantly, it's not really an option.

Regards,
SB
Right now it's about $78 for 32GB option...in a few years it could be half that.

Anyway I hope it's something the console makers look into.
 
Hm....
so here's Raven Ridge @ 210mm^2

40x4iIi.jpg


0SeS0uB.jpg


This is pretty big imo :/
Plus they had to halve the L3 cache compared to desktop Ryzen...
 
GPU L2 cache size is also likely much smaller as well.

That said, they'd really really really really want to launch on a later node.
 
How does that bode for a portable PS4 once on 7nm?

That power consumption makes it look quite promising IMO, for a large enough device anyway - a chubby 8" tablet.
 
What more can Sony do to show that they really don’t care about hardware-based backward compat?
announce a partnership with nvidia and intel? That would both drive the internet crazy.
A loss in BC, a gain in performance/watt. The ultimate feedback hamburger, good/bad/good

"We're Going Nvidia!" -- Yay!!
"But you Lose BC" -- BOOO!!
"But we're going Intel!" -- Yay!!
 
Are you one of the Faceless Men? It was me that asked, you plant pot :p

If the PS5 contains an AMD APU, and it's backwards compatible, Sony will have to emulate Jaguar with Zen. In this case, a portable PS4 would be possible with the Raven Ridge APU if it contained more CU's.

Although it does get me thinking: if they were to go such a route, maybe it would be worthwhile to just release a new portable that's a stripped back version of the PS5. So, it'd be capable of playing any PS4 game, and have the same architecture as the fully fledged PS5, so any game developed for the portable would automatically run on the home console.

They'd tackle the Japanese market and the West simultaneously, and the "is this title cross-play, cross-buy, and does it have cross-save?" needn't apply. It would eliminate the problem that Sony's portables have always faced: too few units.
 
Are you one of the Faceless Men? It was me that asked, you plant pot :p

If the PS5 contains an AMD APU, and it's backwards compatible, Sony will have to emulate Jaguar with Zen. In this case, a portable PS4 would be possible with the Raven Ridge APU if it contained more CU's.
You must not emulate Jaguar on Zen. Zen can do everything jaguar can do but much faster.
Problem could be the GPU as developers use a lower level optimization for PS4 than on xbox. Small changes could get big impacts that way. If the software-level is big enough it should be possible to just rewrite the "drivers" and everything works again.
 
You must not emulate Jaguar on Zen.

Yeah, well what are you gonna do about it if I do huh?! :p

Zen can do everything jaguar can do but much faster.
Problem could be the GPU as developers use a lower level optimization for PS4 than on xbox. Small changes could get big impacts that way. If the software-level is big enough it should be possible to just rewrite the "drivers" and everything works again.

Is there any way that such low level optimisations could be emulated, or is it a hardware limitation? If the latter, could hardware be included to facilitate running in a PS4 mode?

Failing that, maybe Sony could follow the route of X360 backwards compatibility on the XBoxOne, or PS2 backwards compatibility on the PS4: convert the software. Only I'd imagine converting from one console to another would be a doddle when they're almost exactly the same architecture - if they could automate it and consequently make it dirt cheap, that would be an effective way of padding out the launch library pretty quickly.

If it was also up to publishers/developers to conduct the conversion, there would also be an automatic quality filter, as only the publishers/developers that give a shit and are still operational would be able to conduct the conversion.
 
Are you one of the Faceless Men? It was me that asked, you plant pot :p

If the PS5 contains an AMD APU, and it's backwards compatible, Sony will have to emulate Jaguar with Zen. In this case, a portable PS4 would be possible with the Raven Ridge APU if it contained more CU's.

Although it does get me thinking: if they were to go such a route, maybe it would be worthwhile to just release a new portable that's a stripped back version of the PS5. So, it'd be capable of playing any PS4 game, and have the same architecture as the fully fledged PS5, so any game developed for the portable would automatically run on the home console.

They'd tackle the Japanese market and the West simultaneously, and the "is this title cross-play, cross-buy, and does it have cross-save?" needn't apply. It would eliminate the problem that Sony's portables have always faced: too few units.

Would feel like a huge waste of silicon to me, specifically WRT the CPU. Wouldn't you end up with a "portable" PS4 in this configuration that is more expensive than the existing PS4?

Wouldn't it be cheaper (not necessarily in terms of the engineering costs to design it, rather the price at retail) to just port the PS4 APU onto the 7nm process node? Surely this would produce a chip that consumes way less power than your solution (dual-module Jaguar @ 7nm uses significantly less transistors than a Zen CCX), would be physically smaller and thus cheaper to manufacture.

Surely, that would be more appropriate if your goal is to produce a PS4 console in a portable-level power envelope.

Edit:

Out of interest, does anyone know the silicon footprint of PS4 Pro's Jaguar to compare it to the 4 Zen cores in RR @ 14nm.

If RR can consume as low as 9W, with a 4 core Zen (assuming an 8 core Jaguar is significantly smaller) then perhaps a 18CU GPU paired with 8 core Jaguar @ 14nm is approaching the level of possibility for a portable PS4... leading me to believe that a 7nm PS4 APU consuming less than around 7W would not only be possible, but very very doable.... Do it Sony!
 
Last edited:
Would feel like a huge waste of silicon to me, specifically WRT the CPU. Wouldn't you end up with a "portable" PS4 in this configuration that is more expensive than the existing PS4?

Wouldn't it be cheaper (not necessarily in terms of the engineering costs to design it, rather the price at retail) to just port the PS4 APU onto the 7nm process node? Surely this would produce a chip that consumes way less power than your solution (dual-module Jaguar @ 7nm uses significantly less transistors than a Zen CCX), would be physically smaller and thus cheaper to manufacture.

Surely, that would be more appropriate if your goal is to produce a PS4 console in a portable-level power envelope.

Edit:

Out of interest, does anyone know the silicon footprint of PS4 Pro's Jaguar to compare it to the 4 Zen cores in RR @ 14nm.

If RR can consume as low as 9W, with a 4 core Zen (assuming an 8 core Jaguar is significantly smaller) then perhaps a 18CU GPU paired with 8 core Jaguar @ 14nm is approaching the level of possibility for a portable PS4... leading me to believe that a 7nm PS4 APU consuming less than around 7W would not only be possible, but very very doable.... Do it Sony!
Portable PS4 ... dream on ... ;)
The PS4 SOC at 14nm consumes about 60-80W. I don't think we could get it below 30W with all this memory you would need even on 7nm.

It is like the XB360 APU. You can shrink it, but at some point, you won't get it that much better because of the shrinks. You must alter the silicon/architecture to actually save some energy. Else MS could easily release the xb360 mobile on 14nm, but it just isn't that easy.
 
Was the PS4 GPU/CPU architecture THAT inefficient?

I'm gobsmacked that a PS4 APU (18CU GPU and 8c Jaguar) at 14nm consumes 60-80W and a RR APU (10CU GPU and 4core Zen) consumes 9-15W....

That's a HUGE difference in power consumption for a CPU that's probably twice as fast and a GPU half the size (which is based on Vega—which according to the internet is the anti-christ of power inefficiency).
 
Was the PS4 GPU/CPU architecture THAT inefficient?

I'm gobsmacked that a PS4 APU (18CU GPU and 8c Jaguar) at 14nm consumes 60-80W and a RR APU (10CU GPU and 4core Zen) consumes 9-15W....

That's a HUGE difference in power consumption for a CPU that's probably twice as fast and a GPU half the size (which is based on Vega—which according to the internet is the anti-christ of power inefficiency).

Yeah... the 60-80W range for PS4 soc seems fishy to me.
 
Status
Not open for further replies.
Back
Top