Next Generation Hardware Speculation with a Technical Spin [2018]

[MrFox]

I disagree with most claims related to small features (from both MS and Sony) having a major impact. Not without some benchmark.

They didn't bring an AMD design much further than what the silicon was capable of to begin with, and they didn't break any price/performance/launch figure. Proof is in the pudding, not in PR.

MS could have chosen a 399 4.2TF in 2016 with RPM, as much as Sony could have chosen a 6TF for 499 without RPM in 2017. Everything else haven't yet been proven to be more than small tweaks to what AMD had to offer them.

Sony's "cleverness" seems to cost some die area, and I am beginning to suspect RPM is not power efficient. It didn't come for free. ID buffer is unclear. MS "cleverness" cost them a year an 100 higher BOM.

What I am curious about is whether devs eventually manage to use more RPM in the future. Is it something next gen can use more and more to improve efficiency, or is there is a logical limitation which will prevent getting more than a few percent overall? We have seen small pieces of code showing a great improvement, but not enough to have an overall big impact on the full pipeline.

And I'm also curious if AMD will solve their power efficiency issues, at least in comparison to nvidia.

 

[Tkumpathenurple]

Anyway, I keep coming back to a 2 tier launch, and wondering how it could best be handled.

Provides 7nm comes along well, I'm thinking 2019 goes something like the following:

--PS4 Super Slim--

- $150

- No Blu-ray drive.

- No HDD. 256GB SD card in its place.

- Becomes the 720p-900p console over the coming years.

- If it's more economical (and technically possible,) a 4c8t Zen in place of Jaguar. I say this because there's every chance Sony, Microsoft, and AMD wouldn't want to port Jaguar to 7nm when it's all but dead, and Zen is here for years to come.

- Same sort of principal with GDDR6: they'll use it if it's cheaper, or will effectively become cheaper over the next few years. Also, if it's even technically possible to use 4*2Gb GDDR6 chips instead of 8*1Gb GDDR5 chips and maintain requisite bandwidth. I'll have to Google-fu around and get myself some numbers for that though.

--PS4Pro Slim--

- $250

- Blu-ray drive.

- 1TB HDD.

- Becomes the 1080p console over the coming years.

- Same as above: Zen and GDDR6 if more economically pragmatic.


--PS5 Mini--

- $350

- UHD Blu-ray drive.

- 1TB HDD.

- 4K focused, but with plenty of games sacrificing resolution for other effects.

- 8 core Zen CPU. Maybe 16 threads, maybe not. It depends how valuable that would be to developers.

- 16GB GDDR6 for 576GB/s bandwidth.

- 24GB NVME.

- >7.36TF GPU. Enough to run every 1080p PS4 game in native 4K, with some extra juice.

--PS5--

- $500

- UHD Blu-ray drive.

- 1.5TB HDD

- Focused on native 4K.

- 8 core Zen CPU. Clocked higher than the base unit's. Maybe with SMT if disabled in the base unit.

- 24GB GDDR6 for 864GB/s bandwidth. Maybe even HBM Low Cost for some megaballz bandwidth.

- 48GB NVME.

- >16.8TF GPU. Just like the PS4Pro's butterfly wing design. Enough to render every PS4Pro game in native 4K, with some extra juice.


How would anyone feel about something like that? A console refresh and a new console line, each consisting of higher and lower tiers.

 

[mrcorbo]

Anyway, I keep coming back to a 2 tier launch, and wondering how it could best be handled.

Provides 7nm comes along well, I'm thinking 2019 goes something like the following:

--PS4 Super Slim--

- $150

- No Blu-ray drive.

- No HDD. 256GB SD card in its place.

- Becomes the 720p-900p console over the coming years.

- If it's more economical (and technically possible,) a 4c8t Zen in place of Jaguar. I say this because there's every chance Sony, Microsoft, and AMD wouldn't want to port Jaguar to 7nm when it's all but dead, and Zen is here for years to come.

- Same sort of principal with GDDR6: they'll use it if it's cheaper, or will effectively become cheaper over the next few years. Also, if it's even technically possible to use 4*2Gb GDDR6 chips instead of 8*1Gb GDDR5 chips and maintain requisite bandwidth. I'll have to Google-fu around and get myself some numbers for that though.

--PS4Pro Slim--

- $250

- Blu-ray drive.

- 1TB HDD.

- Becomes the 1080p console over the coming years.

- Same as above: Zen and GDDR6 if more economically pragmatic.


--PS5 Mini--

- $350

- UHD Blu-ray drive.

- 1TB HDD.

- 4K focused, but with plenty of games sacrificing resolution for other effects.

- 8 core Zen CPU. Maybe 16 threads, maybe not. It depends how valuable that would be to developers.

- 16GB GDDR6 for 576GB/s bandwidth.

- 24GB NVME.

- >7.36TF GPU. Enough to run every 1080p PS4 game in native 4K, with some extra juice.

--PS5--

- $500

- UHD Blu-ray drive.

- 1.5TB HDD

- Focused on native 4K.

- 8 core Zen CPU. Clocked higher than the base unit's. Maybe with SMT if disabled in the base unit.

- 24GB GDDR6 for 864GB/s bandwidth. Maybe even HBM Low Cost for some megaballz bandwidth.

- 48GB NVME.

- >16.8TF GPU. Just like the PS4Pro's butterfly wing design. Enough to render every PS4Pro game in native 4K, with some extra juice.


How would anyone feel about something like that? A console refresh and a new console line, each consisting of higher and lower tiers.

While B&M is becoming less and less of a factor, I think available shelf space is still somewhat of a limiting factor that makes that many SKUs a problem. Add to that the potential for consumer confusion. Add to that the increased difficulty maintaining a supply chain with that many variations.

It doesn't make a lot of sense to me. Consoles are all about simplicity and presenting consumers with that many distinct SKUs works against that in a way that would be a turn-off for many.

 

[iroboto]

MS could have chosen a 399 4.2TF in 2016 with RPM, as much as Sony could have chosen a 6TF for 499 without RPM in 2017. Everything else haven't yet been proven to be more than small tweaks to what AMD had to offer them.

Sony's "cleverness" seems to cost some die area, and I am beginning to suspect RPM is not power efficient. It didn't come for free. ID buffer is unclear. MS "cleverness" cost them a year an 100 higher BOM.

That's fairly reductive. Both companies have been developing their hardware for years.
You should judge hardware by what they aimed to do, not by what they aimed not to do.

MS wanted to make a 4K console, most of their games reach 4K whether there are caveats around it, they've hit their objective with numerous AAA titles under a single year, and hit their objective with quite a few previously released titles. Even now we see new patches moving from earlier X1X enhanced, 1720p, now realizing full 4K (https://www.epicgames.com/fortnite/en-US/news/v2-5-0-patch-notes) As the platform matures i only expect this number to get more consistent, not less.

In retrospect, Sony never sold PS4 Pro as a 4K machine. That wasn't their goal, their goal was a better PS4 experience which it certainly is, and a price point that many want to buy it at and in a way that should have fit their long term strategies. Both companies knew what they were trading off to make their targets, the minor changes to both platforms were designed with these goals in mind. To call it anything less is reductive imo. They aren't the same machine, and their performance and price profiles clearly shows.

I do not think any company is just throwing hardware together and guessing performance will be in x range.

The process in which MS used to make Scorpio will undoubtedly be used again; they will use the game code from 1X Enhanced games and use that to profile their next console. It's only logical. The question is what their next goal will be. 60fps? 4K native? More 4K dynamic ? etc.

I don't know, but I don't think it's as easy as just choosing parts off a shelf and waiting for the right price points and calling it a day and moving on to developing the subsystems and OS afterwards.

 

[MrFox]

That's fairly reductive. Both companies have been developing their hardware for years.
You should judge hardware by what they aimed to do, not by what they aimed not to do.

MS wanted to make a 4K console, most of their games reach 4K whether there are caveats around it, they've hit their objective with numerous AAA titles under a single year, and hit their objective with quite a few previously released titles. Even now we see new patches moving from earlier X1X enhanced, 1720p, now realizing full 4K (https://www.epicgames.com/fortnite/en-US/news/v2-5-0-patch-notes) As the platform matures i only expect this number to get more consistent, not less.

In retrospect, Sony never sold PS4 Pro as a 4K machine. That wasn't their goal, their goal was a better PS4 experience which it certainly is, and a price point that many want to buy it at and in a way that should have fit their long term strategies. Both companies knew what they were trading off to make their targets, the minor changes to both platforms were designed with these goals in mind. To call it anything less is reductive imo. They aren't the same machine, and their performance and price profiles clearly shows.

I do not think any company is just throwing hardware together and guessing performance will be in x range.

The process in which MS used to make Scorpio will undoubtedly be used again; they will use the game code from 1X Enhanced games and use that to profile their next console. It's only logical. The question is what their next goal will be. 60fps? 4K native? More 4K dynamic ? etc.

I don't know, but I don't think it's as easy as just choosing parts off a shelf and waiting for the right price points and calling it a day and moving on to developing the subsystems and OS afterwards.

I'm not saying the work was easy, just that the possible choices were obviously similar and most of the credit should go to AMD. Not all of it, but most of it. They both like to amplify what their contribution was, but I still see nothing magical about it.

The different direction was a business decision. The tweaks were certainly logical based on their respective targets. Long process, years of work, but logical and safe choices.

 

[iroboto]

I'm not saying the work was easy, just that the possible choices were obviously similar and most of the credit should go to AMD. Not all of it, but most of it. They both like to amplify what their contribution was, but I still see nothing magical about it.

The different direction was a business decision. The tweaks were certainly logical based on their respective targets. Long process, years of work, but logical and safe choices.

I agree, semi custom business heh.

But to be fair to both Sony and MS, some of their customizations do make it back to the core lines at a later time. They're also not semi conductor companies, so I guess their involvement is going to be limited, they're not Apple.

 

[Tkumpathenurple]

While B&M is becoming less and less of a factor, I think available shelf space is still somewhat of a limiting factor that makes that many SKUs a problem. Add to that the potential for consumer confusion. Add to that the increased difficulty maintaining a supply chain with that many variations.

True - physical, visible presence is important in retail. But how important is that in an age where online is flourishing, whilst brick and mortar stores are going out of business left and right?

Consumer confusion is probably the most important part. But there, I think the demarcation of PS4 and PS5 would do a good job of clarifying platforms. And I think a lot of the market is used to higher and lower tier versions of the same product, something which I think has largely been popularised by the iPhone.

As for the matter of maintaining a supply chain: also true, I'm sure that would be tricky. But AMD have been talking about chiplets for a while now, so entire GPU's and CPU's aren't rendered redundant because of one faulty CU/core.

I should imagine that makes it more feasible than ever to maintain a higher+lower tier supply chain: the better chiplets, which clock higher, go into the higher tier console.

It doesn't make a lot of sense to me. Consoles are all about simplicity and presenting consumers with that many distinct SKUs works against that in a way that would be a turn-off for many.

I agree that it's a risk. But I think it's a risk that's largely mitigated by the presence of generations. As long as the same games play on the same generation of consoles, people need only all themselves if they're willing to spend an extra $150 for prettier graphics.

With PS4 and PS4Pro, we've heard that approximately 20% of PS4 sales are PS4Pro. I don't know what the XBoxOne and X1X ratio is like, but, going by the sales thread, it seems a much greater ratio.

So there's a willing market out there. But it's a market that's been established mid-generation, so yeah, it's difficult to know how the market would react to such a launch.

 

[eastmen]

So anyone here expecting an Xboy or psp 3 ? Amd is set to announce Great Horned Owl next week at Embedded world it seems. 2 zen cores + 11GCN cores and it sounds like a TDP starting at 12 watts for embedded systems s;; all the way up to 4/8 zen cores and 11GCN. What are the chances of sony or ms taking a 7nm edition of this and strapping on 8 gigs of HBM2 and releasing a portable system capable of playing X0/ PS4 games ? A 4core zen with 8 threads should be able to run code designed for 8 jaguar chips and the 11 vega gcn cores should be enough to hit 1080p.

 

[iroboto]

So anyone here expecting an Xboy or psp 3 ? Amd is set to announce Great Horned Owl next week at Embedded world it seems. 2 zen cores + 11GCN cores and it sounds like a TDP starting at 12 watts for embedded systems s;; all the way up to 4/8 zen cores and 11GCN. What are the chances of sony or ms taking a 7nm edition of this and strapping on 8 gigs of HBM2 and releasing a portable system capable of playing X0/ PS4 games ? A 4core zen with 8 threads should be able to run code designed for 8 jaguar chips and the 11 vega gcn cores should be enough to hit 1080p.

What’s the business case for it? The technical stuff can be worked out but what’s the expected ROI? What’s the strategy here? How much investment of resources will you be taking away from their main business?

What are their focuses for this 2018+ and beyond?

When we frame it like that, mobile seems like a no go for MS. Maybe a go for Sony but they just closed up Vita.

 

[eastmen]

What’s the business case for it? The technical stuff can be worked out but what’s the expected ROI? What’s the strategy here? How much investment of resources will you be taking away from their main business?

What are their focuses for this 2018+ and beyond?

When we frame it like that, mobile seems like a no go for MS. Maybe a go for Sony but they just closed up Vita.

Why ? the strategy is the same as with the console a 30% cut of game sales and 100% of first party titles. It would also enable more subs to xbox live / psn and in MS's case the game pass.

Resources shouldn't be a huge hit since its x86/gcn it shouldn't be much work to get the current OS running and the majority of the games should work with little to no patching.

This isn't like the psp or vita. A system like this would be able to play games from both 2013 and 2016/17 consoles , it would just simply scale based on the hardware as they are doing now. They wouldn't really have to maintain any diffrences. For MS in this case the XO original was 1.23 tflops the GHO APU should be able to hit that . Don't forget for MS with that they can sell the xbox , xbox 360 and xbox one game line up. With the XBOX X out they will continue making games for the system for a few more years which means more releases and revenue from the xboy. The ps4 and xbox one both sold well and will continue to get 3rd party titles well into the future.

The portable would be a great way to keep people in your ecosystem too. Yes the switch is a great on the go console but its power is lacking for many modern games. A GHO protable would solve that problem and with something like play anywhere on MS's platforms you could buy on your xbox one x , play on your xboy and even your windows tablet or laptop. The ryzen 2700u is already in that 1tflop + range. In 2018 MS could adopt GHO in their surface line and unite xbox on home consoles , PC computers , mobile computers and a dedicated portable

 

[HBRU]

I still think a downclocked ps4pro GPU to fit original PS4 Tflops (at 7nm) and modified to use less power hungry memory chips (and CPU clock still at 1.6 ghz) can base a portable PS4.... For MS may be even easyer to just use an Xbox One GPU at 7nm for creating a portable Xbox ... Interesting is to integrate Phone capabilities in this. Project Andromeda is on the way...

 

[fehu]

So anyone here expecting an Xboy? Amd is set to announce Great Horned Owl next week at Embedded world

Conspirationally, today's Bing image is a Great Horned Owl

 

[Shifty Geezer]

Why ? the strategy is the same as with the console a 30% cut of game sales and 100% of first party titles. It would also enable more subs to xbox live / psn and in MS's case the game pass.

But do people want a powerful handheld that plays 'AAA' games? Switch is selling well mostly because its games are unique. If all it did was play Skyrim and COD, would people have bought it? That's unproven, although having it TV-out capable probably makes it a lot more attractive.

When we frame it like that, mobile seems like a no go for MS.

I think this makes more sense for MS. Create a Switch-a-like with TV out and Play Anywhere, you naturally appeal to existing customers who can get a second console for the house that'll also go on holiday, which'll maybe be shown round people's houses. It'd drive adoption of their ecosystem and growth of the Windows store.

 

[iroboto]

But do people want a powerful handheld that plays 'AAA' games? Switch is selling well mostly because its games are unique. If all it did was play Skyrim and COD, would people have bought it? That's unproven, although having it TV-out capable probably makes it a lot more attractive.

I think this makes more sense for MS. Create a Switch-a-like with TV out and Play Anywhere, you naturally appeal to existing customers who can get a second console for the house that'll also go on holiday, which'll maybe be shown round people's houses. It'd drive adoption of their ecosystem and growth of the Windows store.

Agreed.

My general feeling is that MS is dedicated towards "your software on your devices anywhere" vision statement from Satya . Cloud is the largest and will continue to be the largest investment they will continue funding in the near future, and I would assume a mobile strategy would be generic (much like Play Anywhere), as opposed to releasing an entirely new segment of device which could (would likely) flop in the face of strong competition like the switch. A whole new slew of games would have to be made, so much more investment to do to make a dedicated platform to work. The mobile gaming space is already massive, leaving little room except to compete directly with Nintendo which individually comes no where near the profits of mobile gaming.

Assuming MS release their own mobile device in the near future say later this year, and that something would be large enough to support mobile gaming, you are likely to see Xbox happen on that device before any dedicated Xboy handheld.

 

[Shifty Geezer]

Why would new games be needed? It'd be sold as a portable XBox playing the same games, playing them portably on your TV, and open to many indie titles better suited to mobile play than the AAA X1 blockbusters. Plus PUBG on the go.

Hell, they could stick Windows on it and go that route rather than XBox, which is better for their platform. Just include dedicated gaming controls on a mobile device. Or even just support wireless controllers and TV out, with a dock for mobile controls - a cute Windows portable that plays games a la Switch, runs Office through Continuum, plays Halo on TV in the hotel. The main focus would be to add another access vector to get people onto Windows Store and grow that ecosystem.

Unlike the Switch or other consoles, it wouldn't need critical mass to secure long-term stability. If it plays the games on Xbox already, it'll naturally continue to get games. So there's no minimum need for success other than to recover costs and pay for updates.

 

[Shifty Geezer]

ARM Ares next-gen Cortex A75 *successor* will be unveiled in May. Clocked at 3.0 GHz it will be 2-to-3x of Jaguar while offering far superior perf/watt and perf/mm2 metric over Ryzen, amongst a host of other advantages including the ability to produce a *true* 8 core console, thanks to Dynamic IQ, and be coupled with low power OS cores...ARM not Ryzen will be the true next-gen console core, mark my freaking words! I can post the tweets where I was 'wrong' before way before what the majority where saying.

The issue of BC/emulation came up, and this recent news reveals why people shouldn't assume things from unreleased tech...

• x64 apps are not supported. This is something we’ve known, but Windows 10 on ARM does not support emulation of x64 apps. Microsoft is planning to support these in the future at some point, though.
• Certain games and apps don’t work. Microsoft says that games and apps that use a version of OpenGL later than 1.1 or that require hardware-accelerated OpenGL won’t work on Windows 10 on ARM. Games that use anticheat technologies also won’t run on Windows 10 on ARM.

Not having access to 64bit apps makes Windows 10 ARM horrifically gimped and pointless. Using ARM could easily break BC of X1 games. Why bother if there are x64 options available?

 

[Entropy]

Why would new games be needed? It'd be sold as a portable XBox playing the same games, playing them portably on your TV, and open to many indie titles better suited to mobile play than the AAA X1 blockbusters. Plus PUBG on the go.

Hell, they could stick Windows on it and go that route rather than XBox, which is better for their platform. Just include dedicated gaming controls on a mobile device. Or even just support wireless controllers and TV out, with a dock for mobile controls - a cute Windows portable that plays games a la Switch, runs Office through Continuum, plays Halo on TV in the hotel. The main focus would be to add another access vector to get people onto Windows Store and grow that ecosystem.

Unlike the Switch or other consoles, it wouldn't need critical mass to secure long-term stability. If it plays the games on Xbox already, it'll naturally continue to get games. So there's no minimum need for success other than to recover costs and pay for updates.

The problem I see with this analysis is that it presupposes that at the time when a portable XBox1 would be possible, say from early 2020 onward, the original XBox1 will still be a viable platform, attractive enough that the XBoy wouldn’t have to be a viable platform in its own right.
I don’t think that’s a valid assumption.

 

[Shifty Geezer]

It'll be as much a viable platform as that level of PC is. Unless games stop getting made for PC that'll run on Intel integrated graphics, it'll be able to run software. The important thing for MS is to get games onto Windows Store instead of Steam. They need to give devs reasons to release on Store, and gamers reason to buy on Store. What better incentive than buying an XBox game and getting to play it on your PC and on your portable?

 

[eastmen]

But do people want a powerful handheld that plays 'AAA' games? Switch is selling well mostly because its games are unique. If all it did was play Skyrim and COD, would people have bought it? That's unproven, although having it TV-out capable probably makes it a lot more attractive.

I think this makes more sense for MS. Create a Switch-a-like with TV out and Play Anywhere, you naturally appeal to existing customers who can get a second console for the house that'll also go on holiday, which'll maybe be shown round people's houses. It'd drive adoption of their ecosystem and growth of the Windows store.

the question isn't if people want it , because I want it and i'm people and so do a lot of people I know … and they are people. The question is how many people want it. So I mean obviously it would depend on pricing and features.

However I think a portable Xbox one that can also dock to your tv and play the current xbox one games at the same or better performance levels can be appealing to a lot of people . Skyrim and Doom are popular , I am sure there are a lot of people who would want to play PUBG while on holiday and so on. Also like I said for ms this would be a continuation of play anywhere and expanding that xbox brand. They can also make a minimum xbox one certification for laptops and tablets. Anything that meets the xboy standards for playing xbox one titles can get a new logo for a box and those games can appear in the windows store for people.

 

[iroboto]

Why would new games be needed? It'd be sold as a portable XBox playing the same games, playing them portably on your TV, and open to many indie titles better suited to mobile play than the AAA X1 blockbusters. Plus PUBG on the go.

I think given the criticism of MS library of games today on console they would fare worse with mobile. Mobile gaming is different and I don't think many of the games we see on xbox today lend itself well to the type of gaming mobile would offer. Mainly, a majority of xbox and xbox exclusives are highly MP based and Gaas. Both require that online always type of behaviour. I'm just not entirely sure that the existing library of xbox games would be good for portability. Thus my belief a dedicated handheld would require additional investment into first party studios.

That being said, I do believe in a docking device, but it's unlikely a docking device can output the power of Xbox One S for a long time. It would be easier to stream to a docking device than for that device to run natively.

I'm all for seeing the future of streaming to a docked device in the near future, or supporting more xbox play anywhere type games (Cuphead + Indie games etc) I'm not banking on high probabilities of seeing AAA xbox one titles being on a portable device for many years.