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

[Rangers]

In away this saddens me. No more risk takers.



Yes, but it made things more interesting. The pixel counting debate(s) got real old, real thin, really quickly. What's next for console gaming? Counting vegetation plants on proving the more robust system?

FYI: I'm not advocating a truly exotic hardware design for PS5/XBRedux... but something that wasn't so off-the-shelf and could match (visually and performance wise) a mid-end PC for about 2-3yrs.

The problem is AMD, Intel, Nvidia, Qualcomm, have collectively hundreds of thousands of engineering man-years, not too mention hundreds of patents, on this stuff. There's simply no way to catch up. when AMD build a new GPU they're simply iterating on tens of thousands of man years (AMD easily employs 1000 engineers for how many years? 40?) and the patents are sorted.

For many years now it's been the case nothing exotic can come close to competing for raw performance. The last gasp of exotic technology was Cell in PS3, and it was still a half-attempt at best (cell was only the CPU, and it was still partly designed by IBM). I am quite sure Sony does not like this state of affairs, they would rather design their own. But it's simply not possible, unless they want to be drastically underpowered. The idea (allegedly considered) of a multi-cell, no GPU, PS3 would have been an example of such a thing. Would have been an utter disaster next to an Xbox 360 in real software.

You could match a mid pc for 2-3 years fairly easily...just spec high enough in off the shelf (mostly) stuff. For example lets say your console was releasing in January 2016. Dual Fury X's (8 teraflops apiece) ought to do it...it's just a matter (mostly) of cost, cooling, etc. It's a matter of want to, not possibility.

 

[Phil]

Even if Sony's next platform was fully capable of being backwards compatible... they wouldn't do it (not immediately anyhow). Too much money being made on remakes and PS+. Anyhow, Sony massive lead isn't just based on hardware alone... Sony aces will always be their first party IPs.

I'm not so sure about that. You are basically facing a technological challenge (is a backwards-compatible console technically feasible from a tech and cost pov) against what might give them the edge on the market. You see, the problem is; When you start each console cycle at zero, you need to find ways to lure the people who supported you last generation to stay on your console the next time around. You either do that by binding them to your eco-market and giving value to past purchases (this is where backwards-compatibility comes in) or you simply hope that your console is better at the right price. The latter can't be ensured on any level - not when you're building consoles years in advance to hit some day x in the future and hope your competitor has put lower targets. Logically, you want to have a failsafe option in case that isn't the case.

This will get more important as we make progress. Milking old games might seem attractive now, but it's not something that will make your userbase grow faster than your competitors. And the incentive to play the games of last generation are smaller than perhaps the wish to play 2 or 3 generation old games that look really bad by todays standard on revamped graphics. The games of this generation are quite solid from an IQ point of view - and resolution will likely not increase much into next, even if 4k TVs to become the norm. The real benefit of 4k is low, not last because at the distances we view the screen, the eye can't distinguish the pixels and all that detail. So 1080p is probably where we are probably going to be heading into next generation. And the graphics of games will probably be more limited due to asset creation and budgets than outright technology, so there is a limit on how much better you can make current generation games look on next generation hardware with limited amount of resources. A bit better textures here and there, more resolution and maybe better framerate. That's it. And the difference will probably be rather small to the average player.

Look at GTA5 and you might see one of the view games that really benefited from a remaster (thanks to the PC version). I'm doubtful that a game targeting this generation consoles could make the same jump through a remaster on a next gen console. So remakes will probably decrease as a result because the chance of them milking the current base is smaller.

If I were a console maker, I'd not only be tackling questions of how much bang for buck I can hit on day x for my next generation consoles, I'd also be thinking hard about how I get those million of platform owners to buy my next generation console again. And that means binding them to your console through every means possible.

 

[Deleted member 11852]

Yup. For some, being able to sell your hardware but keep playing your games will incentivise early adoption. Or not just incentivise, where money is the barrier to entry, to make possible.

 

[DieH@rd]

I wonder will Sony deliberately look for specific PS5 [still X86] hardware so that it can help them with the emulation of PS3.

 

[Deleted member 11852]

You mean PS4?

 

[Lalaland]

I wonder will Sony deliberately look for specific PS5 [still X86] hardware so that it can help them with the emulation of PS3.

Do you mean another x86 module or a custom ASIC or FPGA daughter chip or module? I can't see them doing that personally as it would seem to add an awful lot of cost for relatively modest gains. With PS5 they'll have a much more straight forward path for back compatibility to PS4 and if their cloud platform hasn't imploded by then they have that for PS3. If it has collapsed by then we'll have more mature, if legally shaky, emulation on the PC side to take learnings from to try s/w emulation (although I'm not sure the next gen would have the CPU grunt to make that happen).

Could 'full HSA' (ie full arbitrary access to shared RAM by both GPU and CPU) help with this by allowing devs to 'fake' SPUs by using GPU threads? Or are the ALUs in the GPU too restricted to emulate all the functions of an SPU? Or are the exclusive registers the SPUs had a hill that can never be climbed w/o brute forcing it on a CPU?

 

[colinisation]

I wonder will Sony deliberately look for specific PS5 [still X86] hardware so that it can help them with the emulation of PS3.

I wonder if both console manufacturers will stay with x86, only reason I am wondering is AMD has not updated it's cat cores in a while and I would imagine Zen would not be suitable for a console (too big & too power hungry)

 

[Kaotik]

I wonder if both console manufacturers will stay with x86, only reason I am wondering is AMD has not updated it's cat cores in a while and I would imagine Zen would not be suitable for a console (too big & too power hungry)

To my understanding Zen has been designed to cover everything from low power to high end, kinda like Skylake (which is atm available from 3.5/4.5W to 91W)

 

[Rangers]

I wonder if both console manufacturers will stay with x86, only reason I am wondering is AMD has not updated it's cat cores in a while and I would imagine Zen would not be suitable for a console (too big & too power hungry)

All parties will likely be sorely tempted by ARM this go round. It was close this past time.

 

[3dilettante]

I wonder if both console manufacturers will stay with x86, only reason I am wondering is AMD has not updated it's cat cores in a while and I would imagine Zen would not be suitable for a console (too big & too power hungry)

Vgleaks had a leaked version of an older SoC for the PS4 based on 4 AMD desktop cores running twice as fast as the Jaguar cores, which would have yielded generally equivalent throughput absent unknowns about the power and manufacturing of the chip.

It's not a space that cares too much about the lower end of Jaguar's power band, and modern large chips idle decently.
Later Puma variants improved on Jaguar in terms of power management and clocks, so the consoles didn't get the best low-power solution anyway.
Zen has been disclosed as having SMT, so it could split its throughput between multiple threads as well.

 

[Shifty Geezer]

All parties will likely be sorely tempted by ARM this go round. It was close this past time.

What's the benefit of ARM in a static console environment? x86 works and provides the easiest development + multiplatforming. I don't see an obvious reason to drop it.

 

[colinisation]

Vgleaks had a leaked version of an older SoC for the PS4 based on 4 AMD desktop cores running twice as fast as the Jaguar cores, which would have yielded generally equivalent throughput absent unknowns about the power and manufacturing of the chip.

It's not a space that cares too much about the lower end of Jaguar's power band, and modern large chips idle decently.
Later Puma variants improved on Jaguar in terms of power management and clocks, so the consoles didn't get the best low-power solution anyway.
Zen has been disclosed as having SMT, so it could split its throughput between multiple threads as well.

I remember seeing that article from vgleaks, but I believe larger CPUs are inappropriate for consoles, cos in TDP limited scenarios to maintain low average power draw they will throttleor boost depending on how you look at it. Anandtech ran some benches on Core M in an ultrabook form factor and what I took away was that if you hammer the CPU the clockspeed would drop to about the 1Ghz mark. This makes sense in home computing as the majority of time you are idling so boosting intermittently nis not an issue as you will drop speed in a couple of seconds anyway so average power will remain low.
Consoles on the other hand must maintain predictability and thus one constant clockspeed. Secondly inbthe grand scheme of things the CPU is the secondary partner when it comes to dividing resources in consoles - be those resources power draw or silicon area the GPU is the price.ry target. My guess for 9th gen consoles would be low power 8 core with some kind of SMT think jaguar+ or A72+

 

[Shifty Geezer]

TDP isn't an issue. They can (and I'm pretty sure do) fix the clockspeed during a game and get a known heat output, which the custom cooling solution can be engineered to deal with. The only way you'd have/want throttling is if your cooling solution can't deal with max heat. This happens in ultrabooks and mobiles because there isn't room for a decent cooling solution, but not consoles.

 

[function]

Yeah, for the 360S MS went nuts with the thermal testing. They showed in one of their (hotchips I think it was) presentations about the SoC that they'd tested the chip and its cooling using a power virus running on both the CPU and GPU components simultaneously.

 

[steveOrino]

All parties will likely be sorely tempted by ARM this go round. It was close this past time.

Doubt it. Where would they get the GPU from? Maybe Nvidia might have something ARM based for consoles in the future but right now AMD is the only solution I can see, especially if Zen is competent.

 

[Shifty Geezer]

How would choice of CPU affect GPU vendor?

 

[AlNom]

Where would they get the GPU from?

IMG

 

[upnorthsox]

How would choice of CPU affect GPU vendor?

Well it would be quite odd to go with a Zen cpu and a NVidia gpu.......but yea I think we're talking about the other way around where the gpu selection is driving the cpu selection.
In that way a NVidia gpu would certainly put an ARM cpu in play.

Personally, I think its a console lock for AMD unless their existence is in doubt.

 

[upnorthsox]

What's the benefit of ARM in a static console environment? x86 works and provides the easiest development + multiplatforming. I don't see an obvious reason to drop it.

Price probably the biggest, if you want to go with a different gpu than amd then that may drive adoption of an ARM cpu also.

That said, I think AMD has a pretty compelling sales pitch. It would take a full and significant solution to beat it.

 

[steveOrino]

IMG

Do they have anything competitive now or in the pipeline that would be fit for a console?