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

Status
Not open for further replies.
The worst decision Sony can make is take a different way from MS backward compatibility.... that would even be worst for Sony having a huge installed base. I know Cerny sayd he belives in the console generations... but probably PS4 owners not so much.

PS4 owners already bought the console that did not offer backwards compatability
 
New I wonder, has the cooling issue for 3D stacked ASICs been solved?
I don't think it has been resolved yet, if so I think the industry would be leaning hard in this direction.

I guess to quote Elon Musk, there are few things that are inevitable the rest is just possibility. By doing certain actions you can increase the probability of the outcome.

I think looking at this, 3D stacking is one possibility, on a similar vein multichip solutions is another solution (2x APU). Separation of CPU and GPU is another. Perhaps the inclusion of different accelerators.

I think at best 7nm will get APUs to about 12TF GPU performance. And that will take 2-3 years from now to happen at a price point that will be reasonable.

If that isn't enough then 2x that APU would be ~24TF. But the price point may not make any sense and heating and power becomes an issue.

Separation of CPU and GPU you run into a similar problems as multi-APU solutions.

3D stacking has its issues as well, and what would you 3D stack? GPU on top of GPU? CPU on top of GPU?

None of these issues have been solved yet, and from what I can see there is no clear front runner either.

When I look at what MS is doing with DX12 I lean towards multi GPU. Explicit and Implicit GPU support is there, Asynchronous and execute indirect supports multiGPU. So you can have execute indirect draw a scene on 1 GPU and after it's complete have it trigger he other GPU to do post processing for instance without CPU intervention.

So for me I see a software stack that supports a Big/Small model, or a big/big model etc.

If I were to not get too ahead of myself some form of multichip is likely to be the next generation if 12TF is not enough, otherwise 12TF will be next generation.
 
Last edited:
Stacking doesn't solve the problem of cost. Moore's Law meant transistor count increased without increasing cost (amount of silicon), so you could either have bigger chips for the same price or cheaper chips of the same performance. An end to transistor shrinking means you are capped to how many transistors per mm^2 of silicon, requiring more silicon for more power. eg. Let's say Moore's Law ran out with the intro of PS4 and there were no more shrinks. 1.8 TF cost say $50 in silicon. 2 years later there's no cost reduction. 4Pro is introduced at $100 silicon for 3.6 TF (no more room to clock higher unless you add more cooling). PS5 aims for 12 TF - that'd require 6.7x as much silicon as PS4, so $333 for the chip. Whether that's stacked or multi-chip or whatever, it's going to cost lots and lots of money in theory.
 
I find the notion of a ps4.75 to be the less likely of pretty much any options available.

The cost to basically just beef it up to the required amount compared to using newer IP for the time would be pretty high by then I suspect.

PS5 being BC to ps4 isn't as guaranteed as I once thought, even if ps5 is x86 & gcn based.
The level of work to develop and verify it is a lot higher than I thought it would be. So Sony may deem it as a low priority in the grand scheme of their requirements.

The landscape has changed with digital though, maybe they'll let you stream your digital content (without having to buy it again etc), if bc isn't implemented.
 
The worst decision Sony can make is take a different way from MS backward compatibility.... that would even be worst for Sony having a huge installed base. I know Cerny sayd he belives in the console generations... but probably PS4 owners not so much.

So we should be talking the latest PS3 with Cell 4K edition instead of the PS4 Pro then ;-)

I think Sony would rather be as far away technologically speaking as they can from being part of the PC continuum rather than trying to match MS in that en-devour. They obviously want a toolchain that makes porting multi-platforms not cost prohibitive but they also want to be different enough to create a gaming experience that the PC may not be able to touch at 3 times the price if at all with current gaming production constraints at least as a goal. Someone mentioned 32GB of DDR6 for a PS5 which is the kind of thing PC games should never have to rely on if they wanted to make a profit as an example.

Now BC has value and MS is probably doing itself a favor with trying to leverage the PC ecosystem into their product but it also constrains the Xbox as well. How much value BC has is the question and what is Sony or the consumer willing to pay for it. I would almost rather see a PS5 and PS5 BC on sale with the BC version being say 499 to soak up folks who care enough about it and are willing to pay leaving a more pristine PS5 ready to break a few boundaries game experience wise at 399. Again this is aspirational and I don't know what Sony will actually end up doing.

Maybe add a BC "slice" as an addon to a PS5 to throw on the slanted stack or whatever geometric shape they come up with :D
 
I don't think it has been resolved yet, if so I think the industry would be leaning hard in this direction.

I guess to quote Elon Musk, there are few things that are inevitable the rest is just possibility. By doing certain actions you can increase the probability of the outcome.

I think looking at this, 3D stacking is one possibility, on a similar vein multichip solutions is another solution (2x APU). Separation of CPU and GPU is another. Perhaps the inclusion of different accelerators.

I think at best 7nm will get APUs to about 12TF GPU performance. And that will take 2-3 years from now to happen at a price point that will be reasonable.

If that isn't enough then 2x that APU would be ~24TF. But the price point may not make any sense and heating and power becomes an issue.

Separation of CPU and GPU you run into a similar problems as multi-APU solutions.

3D stacking has its issues as well, and what would you 3D stack? GPU on top of GPU? CPU on top of GPU?

None of these issues have been solved yet, and from what I can see there is no clear front runner either.

When I look at what MS is doing with DX12 I lean towards multi GPU. Explicit and Implicit GPU support is there, Asynchronous and execute indirect supports multiGPU. So you can have execute indirect draw a scene on 1 GPU and after it's complete have it trigger he other GPU to do post processing for instance without CPU intervention.

So for me I see a software stack that supports a Big/Small model, or a big/big model etc.

If I were to not get too ahead of myself some form of multichip is likely to be the next generation if 12TF is not enough, otherwise 12TF will be next generation.

12TFLOPs will be plenty for a PS5, especially with efficiency improvements in the GPU architecture and a nice new CPU based on Zen. Video game software production costs are always gonna be the limiting factor going forwards, as fewer and fewer publishers are able to continue reasonably budgeting their games with the ballooning development costs. I imagine the entire industry is going to be making a concerted effort not to push pure graphical fidelity as the primary justification for the next hw generation. Especially because the jump to native 4k rendering is going to demolish a huge chunk of the performance improvement going from PS4 to PS5.

Beyond PS5 is much more of a question mark. And I suspect both Sony's and MS' approaches to PS5 and XboxTwo will hint towards their plans concerning future hw generation upgrades (if they even continue doing them).

Stacking doesn't solve the problem of cost. Moore's Law meant transistor count increased without increasing cost (amount of silicon), so you could either have bigger chips for the same price or cheaper chips of the same performance. An end to transistor shrinking means you are capped to how many transistors per mm^2 of silicon, requiring more silicon for more power. eg. Let's say Moore's Law ran out with the intro of PS4 and there were no more shrinks. 1.8 TF cost say $50 in silicon. 2 years later there's no cost reduction. 4Pro is introduced at $100 silicon for 3.6 TF (no more room to clock higher unless you add more cooling). PS5 aims for 12 TF - that'd require 6.7x as much silicon as PS4, so $333 for the chip. Whether that's stacked or multi-chip or whatever, it's going to cost lots and lots of money in theory.

This was my thinking too. Maybe graphene will save us from a post-5nm-silicon apocalypse.

PS5 being BC to ps4 isn't as guaranteed as I once thought, even if ps5 is x86 & gcn based.
The level of work to develop and verify it is a lot higher than I thought it would be. So Sony may deem it as a low priority in the grand scheme of their requirements.

The landscape has changed with digital though, maybe they'll let you stream your digital content (without having to buy it again etc), if bc isn't implemented.

BC through streaming is a nice idea. Not sure if it would end up being cost effective for Sony to enable free streaming of games in your digital PS4 library. Perhaps through a low monthly subscription.

In reality, I suspect BC will be abandoned as it was coming into this gen.. We said the same thing last time about how important BC was with our digital libraries, and Sony still killed it with a non-BC console out the gate that allowed them to resell you up-ports of the very best older games.

Given that most games released each gen are sequels in established series, the very best titles would most likely be better served being ported and re-released on new consoles (see TLOU and the monster that is GTAV), with the rest being offered through other avenues like streaming solutions.

Additionally, with PS4 doing so well you might argue that Sony has an additional incentive to continue selling PS4s after PS5's release, therefore abandoning BC in PS5 will retain PS4's exclusivity in access to its vast content library.
 
I don't see a PS4 PRO 2 but I could seeS4 PRO VR introducing a 2nd gen PSVR system that would iterate on the PRO to get enough performance to really make PSVR pretty smooth and maybe even wireless

With my flight of fancy now heading towards a messy ending on the side of a mountain I am also entertaining an idea that given AMD's new I/O tech and Sony's penchant for proprietary memory tech we could see a "cartridge" based PS5 or maybe later with a huge amount of say crossbar memory ( 2021 or the like timeframe ) that can easily slurp up whatever data is on the cartridge with some more SSD floating around for digital downloads and just going to town on rendering huge amounts of whatever with the textures just laying out in vast rows to be streamed quickly or just a huge database of procedurally generated game world that just goes on and on and on ....
 
Last edited by a moderator:
PSVR should get an update when PS5 arrives. Much denser display, foveated rendering and support for 2 cameras [for better tracking].
 
PSVR should get an update when PS5 arrives. Much denser display, foveated rendering and support for 2 cameras [for better tracking].
Oh yeah all that stuff fir the ps5 but if there was another rev of the PS4 that would be a thing to do to not piss off as many pro owners. Most likely the ps5 gets psvr2 or whatever.

Sony the display company could really clean house with that display but it would cost a lot.

Sent from my Nexus 5X using Tapatalk
 
Forward compatibility... and improving games on newer consoles... ?!? uhm uhm uhm :D ...

I guess ---the only way-- is by making NEW games that dedicate part of the original PS4 GPU to CPU-LIKE tasks (for instance half of the PS4 GPU dedicatet to physic, enemy IA, environment animations) and so obtain maybe just an 720p with the rest of GPU left free.... so 720p(HDR) on PS4, a nice 1440p->4K(HDR) on PS4pro, and a full 4K@60fps(HDR) in the new PS4-12TFplops (or PS5) that keeps the same old Jaguars.

Same approach not possible for MS...
 
Forward compatibility... and improving games on newer consoles... ?!? uhm uhm uhm :D ...

I guess ---the only way-- is by making NEW games that dedicate part of the original PS4 GPU to CPU-LIKE tasks (for instance half of the PS4 GPU dedicatet to physic, enemy IA, environment animations) and so obtain maybe just an 720p with the rest of GPU left free.... so 720p(HDR) on PS4, a nice 1440p->4K(HDR) on PS4pro, and a full 4K@60fps(HDR) in the new PS4-12TFplops (or PS5) that keeps the same old Jaguars.

Same approach not possible for MS...
You're going to have to revise this post, it's really hard to read what your write with respect to context and arguments. It almost looks like you responded to the wrong thread or meant to reply someone but we don't know who.
 
I thought the main idea behind not having generations was to stop playing russian roulette at every console release and see whether you ended top seller, but instead keeping & growing your market over time...
I don't think backward compatibility alone would do that, well at least it didn't work for Nintendo which was doing it since NGC...
 
Nintendo sold about 5 or 6 times as many Wiis as Gamecubes, and the WiiU was poorly designed and widely, wildly undesirable.

So the first instance of home console backwards compatibility correlated with higher sales and the second instance didn't; the Wii sold largely to casual gamers, whereas the WiiU sold largely to Nintendo fans; the first instance didn't feature a digital library, the second did, but necessitated a small payment (I think, apologies if I'm wrong on that point.)

The PS2 was fully backwards compatible and outsold its predecessor, whereas the PS3 was only backwards compatible in the beginning, when it was markedly more expensive than the other consoles, and was generally bettered by the X360. The PS4 and XBoxOne both launched without backwards compatibility and, before backwards compatibility was announced for the XBoxOne, the PS4 had built up enough of a lead to become entrenched.

I think it's difficult to know how much backwards compatibility effects sales. The number of backwards incompatible consoles that have been sold suggests that it doesn't negatively effect sales, but it's always talked about positively and enthusiastically by those who's console has it, which at least suggests that it adds value. Also, with digital stores, publishers may forever have their products on sale without having to invest in porting them to a new platform. So there's potential financial incentive for the console manufacturers and definite incentive for publishers.
 
BC isn't just about playing your old titles anymore. It's mostly about an ever expanding library that isn't restarted every gen. In the past you could probably go and buy a new BC enabled console and take advantage of an ever shrinking supply of old titles. But DD in conjunction with robust BC can create a reality where a 10 year old in 5-10 years can still readily experience games they have never played like Horizon, UC4 or Bloodborne without having to dust off some old PS4 or go bin hunting.

There are a ton of movies and TV shows that I readily and thoroughly enjoy today that were released through out my lifetime and back before I was born. You know how many I would have interest in watching if I had to breakout a VCR or a film projector? Practically none. And that's what consoles have mostly given us. A bunch of fond memory of classic titles with no easy and practical way to re-experience those titles unless you hold onto some old hardware and software or like indulging in emulators on the PC or mobile hardware.

DD allows BC to be far more exploitable than in the past. When MS talks about Netflix for games, I am sure BC titles are a large part of that vision. Netflix enjoyed very cheap licensing fees in the beginning because no one saw much value in offering people a bunch of old content through a streaming VOD service. MS simply hopes that extends to games. Your average attached rate for consoles is less than 1% of your typical console's library which means there is a ton of content that gamers might find valuable if packaged correctly. With BC, a console manufacturer offering such service doesn't have to worry about that service being suddenly impractical on a new platform at the start of a generation because there is no content to offer.
 
The problem with BC is that we are looking at it through a PC lens. You bought an old game on Steam 6 years ago and played it at medium to high? And now you got a brand new PC? Well you get a new experience by maxing it out completely

But on consoles, backwards compatability is much more like the smartphone model. Do any of you go back to playing your old games when you get the latest iPhone? Im sure some do for nostalgia but i suspect the vast majority just forget about them and pick up new games with newer graphics

I do think BC is important but as others have mentioned how important?
 
The problem with BC is that we are looking at it through a PC lens. You bought an old game on Steam 6 years ago and played it at medium to high? And now you got a brand new PC? Well you get a new experience by maxing it out completely

But on consoles, backwards compatability is much more like the smartphone model. Do any of you go back to playing your old games when you get the latest iPhone? Im sure some do for nostalgia but i suspect the vast majority just forget about them and pick up new games with newer graphics

I do think BC is important but as others have mentioned how important?

Well, this is "sort of" what is happening with XBO BC: better performance, locked V-Sync (and 16XAF with Scorpio).
 
The problem with BC is that we are looking at it through a PC lens. You bought an old game on Steam 6 years ago and played it at medium to high? And now you got a brand new PC? Well you get a new experience by maxing it out completely

But on consoles, backwards compatability is much more like the smartphone model. Do any of you go back to playing your old games when you get the latest iPhone? Im sure some do for nostalgia but i suspect the vast majority just forget about them and pick up new games with newer graphics

I do think BC is important but as others have mentioned how important?

Do you think most people only buy games that were released after their current smartphone?
 
Status
Not open for further replies.
Back
Top