Support NeoGAF

[AmyS]

PS5
November 2020 -

TSMC second gen 7nm node.

20 TFLOP Navi GPU architecture - using several GPU "chiplets" with Sony's specified customisations.
12-core / 24 thread Zen-2 CPU (again, several CPU "chiplets") @ 3.5 GHz
32GB HBMx providing 1.0 ~ 1.2 Terabytes/sec memory bandwidth.
Would use one of the newer mass storage technologies.
$499

Main point for using a number of CPU and GPU "chiplets", as AMD terms them, is so that everything does not have to be crammed onto a single piece of silicon, reducing the manufacturing cost and greatly improving yields.
Everything would come together on a rather large "Mega APU" as a significant portion of future AMD PC products will go this route.

See this article: https://www.overclock3d.net/news/cp...a_exascale_mega_apu_in_a_new_academic_paper/1

One of the largest issues comes when manufacturing large CPU/GPU dies, with yields decreasing and costs rising as you create larger products. Imagine a silicon wafer and imagine that a single wafer has a certain number of defects, each wafer creates a certain number of chips, which means that only a small number of chips will be affected in the whole batch. When creating products with large die sized the number of chips per silicon wafer decreases, which means that defects can destroy a larger proportion of the products in a single silicon wafer.

According to this paper, AMD wants to get around this "large die issue" by making their Exascale APUs using a large number of smaller dies, which are connected via a silicon interposer. This is similar to how AMD GPUs connect to HBM memory and can, in theory, be used to connect two or more GPU, or in this case CPU and GPU dies, to create what is effectively a larger final chip using several smaller parts.

In the image below you can see that this APU uses eight different CPU dies/chiplets and eight different GPU dies/chiplets to create an exascale APU that can effectively act like a single unit. If these CPU chiplets use AMD's Ryzen CPU architecture they will have a minimum of 4 CPU cores, giving this hypothetical APU a total of 32 CPU cores and 64 threads.

Right now this new "Mega APU" is currently in early design stages, with no planned release date. It is clear that this design uses a new GPU design that is beyond Vega, using a next-generation memory standard

This design method could be the future of how AMD creates all of their products, with both high-end and low-end GPUs being made from different numbers of the same chiplets and FUTURE CONSOLES, desktop APUs and server products using many of the same CPU or GPU chiplets/components.

And there will be no reason or need for a PS5 Pro three years later.

I believe the PS4 Pro mid-gen upgrade is probably a one off for Sony. It won't happen every generation. Sony will only do this where it makes sense, and PS5 will be powerful enough to run next-gen game engines w/ much higher rendering quality, physics, etc at native 4K resolution, and also be much better suited for VR.

PS5 will be at the core of Sony's nextgen ecosystem for the 2020s, with a new controller interface (not simply a Dualshock 5) and a next-gen PSVR headset. Cloud will also play a much bigger part of Sony's ecosystem, but won't replace physical games -- That won't happen until 2030+ - well beyond the PS5 cycle

 

[LAM09]

I wouldn't expect any sort of announcement till next E3 at the earliest with possible March 2019 release.

 

[Last_colossi]

I would fully welcome a return to the 5-6 year console cycle. The 7-8 years of last gen was just too long.

.

I wouldn't be surprised to see Sony announce the PS5 early next year for a late 2018/early 2019 release. I really don't think Sony will want the Scorpio to lead (performance-wise) for too long.

 

[RootCause]

November 2020 - TSMC second gen 7nm node.

20 TFLOP Navi GPU (using several GPU "chiplets" with Sony's specified customisations.
12-core / 24 thread Zen-2 CPU (again, several CPU "chiplets") @ 3.5 GHz
32GB HBMx providing 1.0 ~ 1.2 Terabytes/sec memory bandwidth.
Would use one of the newer mass storage technologies.
$499

Main point for using a number of CPU and GPU "chiplets", is so that everything does not have to be made on one die, reducing the manufacturing cost and greatly improving yields.
Everything would come together on a rather LARGE APU / MCM (Multi Chip Module) as a significant portion of future AMD PC products will go this route.

See this article: https://www.overclock3d.net/news/cp...a_exascale_mega_apu_in_a_new_academic_paper/1









And there will be no reason or need for a PS5 Pro three years later.

I believe the PS4 Pro mid-gen upgrade is probably a one off for Sony. It won't happen every generation. Sony will only do this where it makes sense, and PS5 will be powerful enough to run next-gen game engines w/ much higher rendering quality, physics, etc at native 4K resolution, and also be much better suited for VR.

PS5 will be at the core of Sony's nextgen ecosystem for the 2020s, with a new controller interface (not simply a Dualshock 5) and a next-gen PSVR headset. Cloud will also play a much bigger part of Sony's ecosystem, but won't replace physical games -- That won't happen until 2030+ - well beyond the PS5 cycle

That's looks like a machine worth upgrading to.

How would a 2019 ps5 look like?

 

[FIREKNIGHT2029]

Not happening, you can quote me on that.

Earliest we'll see a PS5 announcement is E3 2019.

PS4 is selling way too good for Sony to sabotage their sales like that.

 

[Mrbob]

When the PS4 pro released it basically put the PS5 at 2020. So 2018 is not happening.

Plus with the PS4 selling well there is less incentive for Sony to push a new system.

 

[metalslimer]

Why in the hell would Sony cut off a console at the knees to start over with even more expensive development costs until the console looks to slow down signficantly

 

[psynergylover]

anythings possible but doubt we'll get any real news of a PS5 until 2019 at the earliest

 

[rokkerkory]

This would be stupid move. They dont need to do this at all.

 

[Duxxy3]

I wouldn't mind it. Let Microsoft run their generationless attempt. I don't think it worked with the PS4 Pro because the PS4 is already a capable machine. Maybe the Scorpio will succeed because of the XB1's lack of performance.

I'd also like to see Sony move on from the Dual Shock 4. Move on from that design entirely. It's long overdue.

I think as long as it sticks with the same x86 architecture, developers will be fine.

 

[Sid]

PS5
November 2020 -

TSMC second gen 7nm node.

20 TFLOP Navi GPU architecture - using several GPU "chiplets" with Sony's specified customisations.
12-core / 24 thread Zen-2 CPU (again, several CPU "chiplets") @ 3.5 GHz
32GB HBMx providing 1.0 ~ 1.2 Terabytes/sec memory bandwidth.
Would use one of the newer mass storage technologies.
$499

Main point for using a number of CPU and GPU "chiplets", as AMD terms them, is so that everything does not have to be crammed onto a single piece of silicon, reducing the manufacturing cost and greatly improving yields.
Everything would come together on a rather large "Mega APU" as a significant portion of future AMD PC products will go this route.

See this article: https://www.overclock3d.net/news/cp...a_exascale_mega_apu_in_a_new_academic_paper/1









And there will be no reason or need for a PS5 Pro three years later.

I believe the PS4 Pro mid-gen upgrade is probably a one off for Sony. It won't happen every generation. Sony will only do this where it makes sense, and PS5 will be powerful enough to run next-gen game engines w/ much higher rendering quality, physics, etc at native 4K resolution, and also be much better suited for VR.

PS5 will be at the core of Sony's nextgen ecosystem for the 2020s, with a new controller interface (not simply a Dualshock 5) and a next-gen PSVR headset. Cloud will also play a much bigger part of Sony's ecosystem, but won't replace physical games -- That won't happen until 2030+ - well beyond the PS5 cycle

All this in 3 years for just $500?

 

[photogaz]

I wouldn't mind it. Let Microsoft run their generationless attempt. I don't think it worked with the PS4 Pro because the PS4 is already a capable machine. Maybe the Scorpio will succeed because of the XB1's lack of performance.

I'd also like to see Sony move on from the Dual Shock 4. Move on from that design entirely. It's long overdue.

I think as long as it sticks with the same x86 architecture, developers will be fine.

Errr, no thanks

 

[randomengine]

No.

Playstation -> PS2 = 5 years
PS2 -> PS3 = 6 years
PS3 -> PS4 = 7 years

Since PS4 is successful, there is no reason for a PS5 any time soon. My guess is 2020.

 

[Sjefen]

Bring it, a game developed only on Ps5 with 8+ terabytes would look amazing. And a Ps5 Pro in 2021.

 

[Spyderist]

The only way I can see it being acceptable is that if it is a huge graphics leap and the system has backwards comp. I dislike how I have to still keep my ps3 around to play a far larger selection of games next to my ps4.

 

[PeterLegend]

Nope, the leap isn't big enough.

2020 sounds about right, Sony knows the sweet spot is $399 for new hardware, will they have the testicular fortitude to do $499? I don't think so after experiencing the type of consumer response with PS4.

 

[SlickShoesRUCrazy]

2020 at the earliest imo. With mid gen upgrades, probably extend out the gen a bit longer.

I Assume that the pro version replaces the PS4 slim at some point closer to the ps5 release, meaning no more game support outside of maybe smaller indie games for the base PS4.

Same with xb1 and Scorpio.

 

[Eylos]

Nope, the leap isn't big enough.

2020 sounds about right, Sony knows the sweet spot is $399 for new hardware, will they have the testicular fortitude to do $499? I don't think so after experiencing the type of consumer response with PS4.

I'm not a Math Guy, what would be the PS4 launch price today? considering inflation

 

[Suburban Thug]

I'm still very happy with my base PS4

 

[Crayon]

Like I said before. It's not impossible. But they would have to have a specific strategic reason to do it when they can just ride out a winning generation a little longer.

 

[hanspampel]

PS5
November 2020 -

TSMC second gen 7nm node.

20 TFLOP Navi GPU architecture - using several GPU "chiplets" with Sony's specified customisations.
12-core / 24 thread Zen-2 CPU (again, several CPU "chiplets") @ 3.5 GHz
32GB HBMx providing 1.0 ~ 1.2 Terabytes/sec memory bandwidth.
Would use one of the newer mass storage technologies.
$499

Main point for using a number of CPU and GPU "chiplets", as AMD terms them, is so that everything does not have to be crammed onto a single piece of silicon, reducing the manufacturing cost and greatly improving yields.
Everything would come together on a rather large "Mega APU" as a significant portion of future AMD PC products will go this route.

ambitious
before the Xbox Scorpio tech reveal i would have said "possible"
but now i say "no way"

you mention state of the art high end PC tech for that year,
that will sadly not make it into a console and obviously not for $499

so either later than 2020 or less and lower performance

 

[e-gamer]

If scorpio would cost 599 then a Ps5 next year would cost 899.

 

[sense]

2019 would be a lock. Why wait till 2020 and go head to head against ms next machine as that would be 3 years from Scorpio. Ms would not repeat the blunder of this gen and would probably go the more powerful route and if they both launch at same time, comparisons will be made and ms has advantage to gain more in that scenario, price and early mover advantage will benefit ps5 as I don't see ms releasing a new console in 2019, just 2 years after Scorpio.

 

[Marvel]

Absolutely not.

 

[yanipheonu]

A new CPU would get me interested.

I'm down with people making the '60' version, a la Nintendo 64. Make a console that is designed for solid 60 and I'm immediately down.

 

[LeleSocho]

2020 might be too early? The thing came out in 2013 ffs, it's been showing it's age for a while now.

5 years is long enough for a gen when the companies cheap out on hardware so much

Yes, i think 2020 might be too early.
Unless you want games to look essentially undistinguishable to PS4 Pro and Scorpio you want a machine with 10-12 TFlop/s but also you don't want this thing to be a 600mm2 monster die only for the gpu so considering how slow the process nodes are evolving to have a ~400mm2 (PS4P and Scorpio size) with good cpu and said gpu you might need to wait 2020-2021.

 

[Raven77]

I would fully welcome a return to the 5-6 year console cycle. The 7-8 years of last gen was just too long.

Agree 100%. That gen nearly made me give up gaming in the later years of the cycle.

 

[Planet]

Playstation -> PS2 = 5 years
PS2 -> PS3 = 6 years
PS3 -> PS4 = 7 years

Not fully accurate! PS1 was released first in December 2014, the PS2 followed in March 2000, the PS3 in November 2006 and the PS4 in November 2013. Thus it is a bit more precise:

5 years, 3 months
6 years, 8 months
7 years

The PS2 era was almost as long as the PS3's, so I never got why people insisted on saying it was overly prolonged. Surely this makes 2020 historically more probable, though I don't think this is the only aspect in planning the next generation's timing.

Still it matches other circumstances like I mentioned above, predominately the availability of 7nm process components to make a huge performance jump feasible for a reasonable price. I can't see Sony trying to aim for more expensive machines in the future. Especially since the reasonable midrange worked so well with PS4.

 

[RPGamer92]

Please no, 2020 seems like a good time for PS5 to come out.

 

[Gradly]

Yes, I totally expect that. Things have changed, Sony wants a powerful device and this is the reason they brought PS4 Pro as well. Some people forgetting that we had no Pro version of PS3 or PS2 yet they still compare PS5 to the older devices in terms of year of service. Therefore a new device expected every so often.

I really expect the new generation of consoles to be like smartphones, a new more powerful device that runs all previous games and coexists with the previous generation until those generations got retired. That's it, a PS5 in 2018 that runs all released games on PS4, and PS4 will still run new games but with low settings until 2-3 more years of support

And there is Scorpio coming, I don't think Sony ignores the competition

 

[farisr]

Holiday 2019 at the earliest.

 

[Kirye]

I sincerely doubt it. The PS4 is still going strong and the Pro hasn't even been out for a year. I think earliest I could see PS5 is around Holiday 2020.

 

[metalslimer]

Also a PS4 in 2018 would not be a huge step over Scorpio, in which case there is no point to it unless they are going for something different.

 

[Huge Succeeded]

If this generation begins to wind down to a close in 2018, then this will have been the lamest, most middling generation in gaming.

 

[Sjefen]

Also a PS4 in 2018 would not be a huge step over Scorpio, in which case there is no point to it unless they are going for something different.

Doesnt need to be a huge step as long as devs are developing games exclusivly for PS5. Much like Pro, Scorpio will be a "upscaler" console with no exclusives.

 

[nOoblet16]

Why in the hell would Sony cut off a console at the knees to start over with even more expensive development costs until the console looks to slow down signficantly

Oh they are spending money at developing PS5 right now, you can be sure of that at the very least.

 

[Zaru]

We need at least twice as strong CPUs for that TDP, and a graphics card that can handle that year's average games at 4k without being at the limit. Preferably on HBM2 (although I'm not sure how/if that would work with the unified RAM idea)

That's not going to happen at any console-reasonable price point until 2019 at the earliest, and nobody needs compromises when Pro and Scorpio are still fresh.

 

[nOoblet16]

Not fully accurate! PS1 was released first in December 2014, the PS2 followed in March 2000, the PS3 in November 2006 and the PS4 in November 2013. Thus it is a bit more precise:

5 years, 3 months
6 years, 8 months
7 years

The PS2 era was almost as long as the PS3's, so I never got why people insisted on saying it was overly prolonged. Surely this makes 2020 historically more probable, though I don't think this is the only aspect in planning the next generation's timing.

Still it matches other circumstances like I mentioned above, predominately the availability of 7nm process components to make a huge performance jump feasible for a reasonable price. I can't see Sony trying to aim for more expensive machines in the future. Especially since the reasonable midrange worked so well with PS4.

Because the Xbox 360 came out a year before. When people say it was too long they are talking about the generation being too long not the time it took between PS3-PS4.

 

[LelouchZero]

2019 is the right year in terms of the generation being "the right length", not as long as the horrendously drawn out previous generation, but not so short that games didn't have time to set themselves up.

2019 won't be where they want it, technically. The mid-gen refreshes only make sense if they're planning to draw it out longer than 2018, and probably more like 2020-2021. Even if we assume sub-native 4k is the target, as it is on PS4 Pro, that still means that you need essentially 16TF to get a 4x increase over PS4 Pro. If you say "the next gen will do 4k native" the comparisons get even worse since a portion of that 4x figure will be eaten by the increase in rendering resolution, and the actual power left over for other graphical things will be highly underwhelming for a generational transition.

In 2017, we have Nvidia chips doing 14tf. In 2018, we'll probably end out the year with the "Titan" level Nvidia chip pulling 20tf if it over-clocks well. But in terms of when that level of power will be available on a $399 console, the answer is 2020 and beyond.

Indeed, with maybe 2019 at the earliest for something around 16TF which won't be where they want it as you said.

The next generation consoles are almost destined to have capable CPUs by today's standards as they'll likely be utilizing 6+ AMD Zen cores, with 8 cores at 2.5GHz+ they'll offer a sizable leap over the previous generation machines, even more so if they were to have more than 8 cores.

So if they fall short on GPU power hopefully they'll have capable CPUs going for them.

 

[AmyS]

Yes, i think 2020 might be too early.
Unless you want games to look essentially undistinguishable to PS4 Pro and Scorpio you want a machine with 10-12 TFlop/s but also you don't want this thing to be a 600mm2 monster die only for the gpu so considering how slow the process nodes are evolving to have a ~400mm2 (PS4P and Scorpio size) with good cpu and said gpu you might need to wait 2020-2021.

But as I posted, there won't be a monster die, that is the point of AMD's future APUs being constructed from a number of smaller chips, CPU chiplets and GPU chiplets, each manufactured separately, and only then being put together as an APU with stacked DRAM memory (whether that is HBM3 or something else, AMD isn't saying yet).

I don't think PS5 in 2020 is too early, but certainly 2018 is, and probably 2019 also. That said, I think 2021 is too far away, that's 8 years after PS4, the same amount of time there was from Xbox 360 to Xbox One, and a year more than PS3 to PS4 (7 years).

 

[Shin]

Unless Sony changed it's mind from their previous statement then we should expect shorter console cycles: https://www.google.com/search?q=sho...rome..69i57.5940j0j7&sourceid=chrome&ie=UTF-8

 

[nOoblet16]

Yes, i think 2020 might be too early.
Unless you want games to look essentially undistinguishable to PS4 Pro and Scorpio you want a machine with 10-12 TFlop/s but also you don't want this thing to be a 600mm2 monster die only for the gpu so considering how slow the process nodes are evolving to have a ~400mm2 (PS4P and Scorpio size) with good cpu and said gpu you might need to wait 2020-2021.

10-12TF in 2020 at console level die size is totally in the realm of possibility. We have Scorpio at 6TF in 2017, 3 years is enough time to double it if not triple it even. PS4 was 1.8TF in 2014 in 3 years we are going for more than 3x of that.

 

[Humdinger]

Well, they're DF, so they're just looking at it from a tech perspective. From a tech perspective, is it possible to release a thing in 2018 called the PS5? Sure, it's possible.

But from a business perspective, it would be stupid. PS4 is hitting its stride, selling extremely well, doing better than ever. You don't cut the legs out from under something like that when it's riding high.

I expect 2020. Maybe 2019, but I'd bet on 2020.

 

[AmyS]

Fall 2020 - a new U.S. President, and a new PlayStation.

Sounds good to me

p.s. PS5 won't be anything less than a 12 TFLOP console.

 

[DirgeExtinction]

I'd be interested in upgrading from a Pro maybe by 2019.

 

[LeleSocho]

But as I posted, there won't be a monster die, that is the point of AMD's future APUs being constructed from a number of smaller chips, CPU chiplets and GPU chiplets, each manufactured separately, and only then being put together as an APU with stacked DRAM memory (whether that is HBM3 or something else, AMD isn't saying yet).

Ps4 and Xbone are already APUs, saying that the next consoles will use APU means nothing if you want certain performance you always have to occupy a certain amount of area on the die regardless on how you want to call it CPU, GPU or APU.

10-12TF in 2020 at console level die size is totally in the realm of possibility. We have Scorpio at 6TF in 2017, 3 years is enough time to double it if not triple it even. PS4 was 1.8TF in 2014 in 3 years we are going for more than 3x of that.

Process node roadmap are slooooooow and each gain are worse than the precedent, if the trends continues you won't get that far.
Also in my opinion the next round of things the cpu in the apu might need more die area to have a balanced system so that means less space for the gpu and therefore needs to be more efficient than usual to reach the same levels.

I stand by the very late 2020 to first half 2021 timeframe for a potential next gen console.

 

[ZSaberLink]

Way too early imo. I'd wait for a much larger generational leap at the least. Current-gen (at least PS4 & XB1) have only finally hit their stride and I don't feel like the current hardware has really brought out enough experiences pushing the platforms to their limits or anything. Instead they're so focused on stuff like 4K instead of spending those resources on a better framerate at 1080p or just more intense gameplay and the like. I feel like something releasing in 2018 will just seem like a slight improvement over Scorpio.

 

[Eylos]

I calculated 400 dollars in 2013 is the same as 420 today


My guess is a 425-450 dollars ps5

 

[hanspampel]

10-12TF in 2020 at console level die size is totally in the realm of possibility. We have Scorpio at 6TF in 2017, 3 years is enough time to double it if not triple it even. PS4 was 1.8TF in 2014 in 3 years we are going for more than 3x of that.

not really
https://www.semiwiki.com/forum/content/6713-14nm-16nm-10nm-7nm-what-we-know-now.html
https://community.cadence.com/cadence_blogs_8/b/breakfast-bytes/archive/2017/03/22/tsmc2

TSMC
7nm vs 16nm FF
density factor 1.63
performance 15 - 20% gain
power consumption 35 - 40% reduced

performance and power consumption is EITHER not both


i see double or triple is only possible for 7nm+ including EUV (risk productions starts H2 2018)
that node will likely not be ready for a mass console launch in 2020. but 2021 should be feasible

 

[Kagoshima_Luke]

If Sony releases a PS4 Pro+ under the name of PS5, I would be majorly disappointed. If they do so and fail to make it backward compatible, I would be done with PlayStation.