Middle Generation Console Upgrade Discussion [Scorpio, 4Pro]

[iroboto]

He is most likely to speak at a dev conference, GDC at San Francisco is later this month



Why do you think this?

Thanks!

As for why.... hmm...
Mind mapping some reasons to stick with generations here:
a) it's what their customer base is used to, no reason to shake the boat when the generation less method is still unproven; as a leader this is actually your advantage, you can wait and see what sticks and what doesn't.
b) as you said earlier, the hard cut generation can provide bigger return on graphical investment
c) it's easier, and possibly cheaper to do it this way
d) more profitable, new generations signals a time when you can ask the customer to pony up for all new accessories, new plans, new features etc
e) just like not doing the online DRM/sell/lending games commercial, it's appears that it would be a win for Sony (at least for their customer base) to never do what MS does. LOL

 

[KirkSi]

No it won't. A hardware generation sees massive gains across the board in every aspect, pretty much. Scorpio won't have 4x the bandwidth, 4x the memory increase (incredibly small by new gen standards), 4x the CPU power. It has (as per current understanding) a 4x increase in GPU TFs over XB1S. Even factoring in bonus feature like FP16 and maybe a notably better CPU, it's not going to be enough of a hardware advance as we expect from previous generations.

The fact it's only been 4 years as technological progress is slowing means it has to be less. A true 8x increase over XB1 probably isn't possible for a few years. We'd need stacked RAM well beyond 1TB/s and 32+ GB (only a 4x increase but much more if same OS 4 GB reservation).

Are we likely to see hardware develop at the rates that it has historically though? I thought hardware innovation was slowing down and Moore's law being unsustainable. Can't remember what the power increase was between OG XBox and 360, but that was a 4 year increment. Was that more than we'll be getting from XB1 to Scorpio? I think we're going to have to get used to smaller increases in power and personally I'd rather have them more regularly and without the disadvantages of a hard reset than waiting longer for the hardware to be there for a true generational increase.

 

[Deleted member 11852]

Traditional generations also set the install base back to zero again at the start. Can't see Devs being happy about that.

You're confusing devs with publishers.

Beauty of MS's approach is the install base will be the same, if not a bit higher due to the new console sales, and the game will scale across hardware.

The obvious drawback is that leaps forwards in technology are artificially stunted if your game has to run on the last gen. Try to imagine how GTA III would have ended up if Sony mandated it had to run on PSone. Or how LittleBigPlanet would have been if it had to run on PS2.

 

[BRiT]

No it won't. A hardware generation sees massive gains across the board in every aspect, pretty much. Scorpio won't have 4x the bandwidth, 4x the memory increase (incredibly small by new gen standards), 4x the CPU power. It has (as per current understanding) a 4x increase in GPU TFs over XB1S. Even factoring in bonus feature like FP16 and maybe a notably better CPU, it's not going to be enough of a hardware advance as we expect from previous generations.

The fact it's only been 4 years as technological progress is slowing means it has to be less. A true 8x increase over XB1 probably isn't possible for a few years. We'd need stacked RAM well beyond 1TB/s and 32+ GB (only a 4x increase but much more if same OS 4 GB reservation).

In most situations it should have more than 4x memory bandwidth increase (4.7x actually) going by DDR3 speed of 68 compared to new speed of 320.

 

[iroboto]

You're confusing devs with publishers.



The obvious drawback is that leaps forwards in technology are artificially stunted if your game has to run on the last gen. Try to imagine how GTA III would have ended up if Sony mandated it had to run on PSone. Or how LittleBigPlanet would have been if it had to run on PS2.

I would hope that MS has a sort of contingency against that, as in always having 1 base platform coupled with one rolling platform.
If they can win over the public with a good policy on the hardware front, I think generation less is a better model as the pros greatly outweigh the cons.
Unfortunately I'll need to make some major assumptions here.

For instance:
a) the argument that cutting everyone over to the new base system means better looking games
* While we know this is true, this is true only after the launch of the generation.
* At the beginning of every generation we still get that cross platform stuff anyway, and we get many folks who write (this looks like my PS3 etc etc)... well yea, that's because it is a PS3 game running on PS4.
* So this transition period is still here anyway

b) Developers can always target lower spec platform, traditionally with generations, once everyone has moved on you effectively stop coding for the older platform unless you want to do twice the work. This is no longer the case for developers who are interested in this. (mainly for indies and smaller games), this is a big win for consumers because your older hardware is not necessarily ever obsolete.

c) Developers have a much longer time to work with newer production harder. So while devs are still trying to get the most out of XBO and PS4, later this year they can already begin their roadmap of developing their engines for scorpio as a baseline. They would know for instance that in 3-4 years Scorpio would become the baseline, so they can continually add and work on features and look at their customer base as when to switch over. We still don't know when PS5 arrives, but if it's closer to 2020/2021 then Scorpio will already be a mature platform with engines fully tuned for it, and the next MS device will roll in.

d) no one loses their library, and your accessories carry forward.

edit: had to change near end of generation, and switched it to launch of generation

 

[Deleted member 11852]

I would hope that MS has a sort of contingency against that, as in always having 1 base platform coupled with one rolling platform.

What contingency? If you have a hard policy of 'nobody left behind' then newer more capable hardware will realise only some of it's potential until that policy no longer applies. Sony have the exact same problem.

If they can win over the public with a good policy on the hardware front, I think generation less is a better model as the pros greatly outweigh the cons.

I actually agree, but the obvious downside of a generationless platform with the above policy is your hardware will only ever likely realise anything like it's maximum potential years after it's new and when the generation prior to it is cut of. And that means your hardware is now the lower performance tier. People craving the cutting edge will just have jumped on the higher tier. Those types of people will be in a perpetual loop of never having their hardware potential maximised.

And that's fine, if you understand that going in.

I am not a fan of such policies, I'd prefer both Microsoft and Sony allow devs to build games for the leading edge tier where the game would just have to sacrifice too much to ran on the previous tier. It's not something publishers will go for lightly because you are consciously restricting your market but it works on PC.

 

[iroboto]

What contingency? If you have a hard policy of 'nobody left behind' then newer more capable hardware will realise only some of it's potential until that policy no longer applies. Sony have the exact same problem.

Yea the issue of having to support an older platform forever for new titles I mean.
It will be a real problem for XBO and Scorpio since I assume Scorpio is running FL12_0+ with a good/full hardware support of SM6.

Eventually Scorpio must become the base model, or its potential will never be fully utilized.

 

[Deleted member 11852]

Yea the issue of having to support an older platform forever for new titles I mean. It will be a real problem for XBO and Scorpio since I assume Scorpio is running FL12_0+ with a good/full hardware support of SM6.

Eventually Scorpio must become the base model, or its potential will never be fully utilized.

Right. So the best time to buy a Scorpio will be... in 2-3 years time when it's cheaper, but also when the next Xbox is launched? Like high-end PC/Pro owners, Scorpio will be for that select few who are willing to spend to have today's games run a bit better.

That's going to change nothing in the market. It's a great option. As a consumer I like options, but the mass market are budget conscious.

 

[mosen]

Now, if this works as we believe, that there will be full BC/FC on MS platforms going forward, kudos to MS. And they have done something good for the consumer, question is how many actually cares. I understand people not having played GTA V want to play it, sure, but I am guessing there is just a handful of games like that. Does anybody really care about Call Of Duty/Battlefield N releases back?

Now Sony and MS looks to have different views on the matter here, Sony wants to do a generation, they think that to get the generational leap and the best out of the hardware you need to do that. MS looks to be taking the approach that full BC/FC is more important and maybe they even disagree with Sony about the generational leap. IE incremental releases will give big enough leaps on eye candy/features in games.

If Sony is right and you need a massive HW change, then they will for certain cater to the hardcore market (excluding the PC master race here). But will the more casual/mainstream market care? Will they do PS5 or do a Scorpio Refresh model with their library intact? Lot of people jumped from X360 to PS4 this gen (I think that is the assumption at least) and lost their library.

If MS is right that the frequent refresh of hardware is equal to doing generational leaps, then Sony is probably in a world of hurt, due to the perceived extra value the BC/FC will have.

It will be fun to see whom gets it most right.

I think MS strategy is superior since Sony can't define software generations by its own. When the time comes for PS5 launch there are already 100-80m PS4/Pros, 40-50m XB1/S and 5-10m Scorpios (if not more) and millions of PCs with variable specs. So they can't define a new software generation on their own.

Microsoft vision looks to be more future-proof. They want UWP to be the next master race on PC/Consoles and probably mobiles and other Windows devices with different form-factors (set-top box devices based on XB1 for example).

Microsoft needs Xbox consoles to make UWP a successful game platform at first, since they are really weak on PC and mobile. I think after that they will try to bring more cross-platform-play enabled games to Xbox/W10 (by allowing kb/m for playing games on consoles and predicting appropriate solutions for probable problems) and then we could expect to see more UWP/XPA titles on Windows Store. However, they can also encourage developers to make their games cross-network compatible (between Steam and Xbox Live).

If they mange to do this, then it will be very hard for Sony to define new software generation with PS5.

 

[iroboto]

Right. So the best time to buy a Scorpio will be... in 2-3 years time when it's cheaper, but also when the next Xbox is launched? Like high-end PC/Pro owners, Scorpio will be for that select few who are willing to spend to have today's games run a bit better.

That's going to change nothing in the market. It's a great option. As a consumer I like options, but the mass market are budget conscious.

So many dependencies here unfortunately, much as to what Shifty indicated earlier, way too many assumptions we need to make.
The most important being, we have yet to see what a Scorpio edition title looks like yet. I'm not sure how far developers can push the envelope with XBO being the baseline.
I think from what I understand they can leverage quite a bit still, for instance Just Cause 3 uses Conservative Rasterization for GPUs that support it. It certainly won't be the difference of upgrading X360 to XBO is. They are still fundamentally running the same technologies. Though full hardware SM6 support is a pretty big deal imo. SM5.2 has been around for nearly 7 years now.

Ultimately purchasing times depend on the consumer entirely. The hardcore will upgrade every cycle on launch day. The budget minded will wait. People who want to buy into Xbox will probably get a Scorpio if they can't wait. And some existing XBO customers can wait until the platform after Scorpio is released.

 

[Shifty Geezer]

In most situations it should have more than 4x memory bandwidth increase (4.7x actually) going by DDR3 speed of 68 compared to new speed of 320.

Is there a reason you're discounting the ESRAM bandwidth? Total bandwidth for the CPU and GPU to do their jobs is ~250 GB/s.

Are we likely to see hardware develop at the rates that it has historically though? I thought hardware innovation was slowing down and Moore's law being unsustainable.

Yes. Thus you need longer between generations to get a generational improvement.

Was that more than we'll be getting from XB1 to Scorpio?

Yes. Graphics greatly improved in versatility plus unified shaders efficiency. 360 had more than 4x bandwidth available in real terms. Scorpio is architecturally very similar to XB1 simply because there haven't been any major GPU changes, 50% more RAM (going by rumours), 50% more BW in real terms.

I think we're going to have to get used to smaller increases in power and personally I'd rather have them more regularly and without the disadvantages of a hard reset than waiting longer for the hardware to be there for a true generational increase.

I think many might agree. That wasn't what I was arguing though. You suggested Scorpio will be enough of an advance to be viewed and presented as a new gen console. I don't think that's true - the numbers don't add up.

 

[Deleted member 11852]

I think MS strategy is superior since Sony can't define software generations by its own. When the time comes for PS5 launch there are already 100-80m PS4/Pros, 40-50m XB1/S and 5-10m Scorpios (if not more) and millions of PCs with variable specs. So they can't define a new software generation on their own.

Absolutely right, and neither company can directly define a generation. This is partly a progressive technical thing driven by AMD and Nvidia R&D and the ability of fabs to produce chips at a given complexity/pricepoint but it's also partly a consumer thing and their desire to embrace more powerful hardware.

Microsoft vision looks to be more future-proof. They want UWP to be the next master race on PC/Consoles and probably mobiles and other Windows devices with different form-factors (set-top box devices based on XB1 for example).

I'd argue sustainable more than future-proof, but sustainability is the result of restraint. Again, you're back to the same dilemma of new hardware taking years to realise it's potential, not just because of lack of familiarity with the hardware as we have now, but also by an arbitrary support policy. Is that really better for consumers?

The most important being, we have yet to see what a Scorpio edition title looks like yet. I'm not sure how far developers can push the envelope with XBO being the baseline.

For years Scorpio is going to be in the same ballpark as Pro. Anybody who is expecting much more is likely setting themselves up for disappointment. The promise started out that everything would be native 4K but the leaked whitepaper walks back on that. Just took at Rise of the Tomb Raider on PS4 Pro for an idea of what to expect: combinations of higher resolutions (YMMV), 'richer visuals' (thanks Nixxes), higher framerate, longer draw distances. Probably better textures on some titles as well. Scorpio is not magic. If that's all you want then Scorpio will please you greatly.

 

[mpg1]

I don't really ever see the experience of "traditional console hardware leaps" happening again. If you want that experience you will have to spend $5000+ on a PC/monitor every 5-6 years in which you won't have the same economic/heat/energy/form factor constraints.

Unlike years ago AAA third party games are released on PC and console...so anything a put on a new $500 console will be beaten by a new $3000 PC a year or two prior.

 

[BRiT]

Is there a reason you're discounting the ESRAM bandwidth? Total bandwidth for the CPU and GPU to do their jobs is ~250 GB/s.

Worst case scenario is 4.7x faster or more. Unless all the work can be done in simple 32Meg segments. Now you're open to using 12gig of storage at 320 instead of 32Meg at 209. That makes things easier with memory footprint increase of 384x.

 

[Shifty Geezer]

Worst case scenario is 4.7x faster or more. Unless all the work can be done in simple 32Meg segments. Now you're open to using 12gig of storage at 320 instead of 32Meg at 209. That makes things easier with memory footprint increase of 384x.

But it's not that clear cut, and no-one is going to be saying XB1 is achieving its on-screen graphics using 64 GB/s, or even only 32 MBs RAM! Bandwidth and memory comparisons have always been contentious. In terms of how many bytes per second XB1 is processing though, rather than whatever pipes are available, it's going to be 200+ GB/s (ignoring caches). Similarly in terms of bytes stored and accessed from hardware memory, it'll be about 5 GB of data. In terms of bytes per second available for processing on Scorpio it'll be 320 GB/s, and hardware RAM storage will be 9 GBs (assuming same 3 GB OS reservation). So 1.5x more BW available to feed the processors, and 2x the storage for data. I don't see that any other comparison makes sense or meaningful comparisons the show relative power. Worst case scenarios clearly aren't happening in well written games and don't represent the level of attainment in XB1's software library such that Scorpio should be compared to the worst XB1 can manage.

 

[mosen]

I'd argue sustainable more than future-proof, but sustainability is the result of restraint. Again, you're back to the same dilemma of new hardware taking years to realise it's potential, not just because of lack of familiarity with the hardware as we have now, but also by an arbitrary support policy. Is that really better for consumers?

From software/service/game point of view this model is better for consumers and even from hardware point of view this model offers more options for different tastes, so I think it'll be definitely better. However, I'm not so sure about developers.

 

[BRiT]

But it's not that clear cut.

Exactly. That's why I was bringing up the significant increases you seemed to be ignoring. Why were you ignoring the increases that makes development substantially easier?

 

[Shifty Geezer]

Absolutely right, and neither company can directly define a generation.

I'll debate this.

As tech advances, software tends not to use it until a new console gen. That gives devs the incentive to leave the old, outdated ways behind and invest in the latest techniques. On PC that's completely progressive, the market size economically limits devs to using a lower baseline. On consoles if they become progressive and the older tech hangs around as still active, there should be the same economic effect. going with mosen's numbers:

When the time comes for PS5 launch there are already 100-80m PS4/Pros, 40-50m XB1/S and 5-10m Scorpios (if not more) and millions of PCs with variable specs.

With well over 200 million existing customers, why would anyone write a game for PS5? Well the same reason they ignored 200 million PS2 generation consoles to target PS360 and comparable PCs - because it was time to move on with games and make better games, which meant leaving the old machines behind. The economic argument then for ignoring the existing hardware base is that the gaming community also wants to move on and will buy new hardware.

Well, the gaming community isn't going to be so keen about buying new hardware if it plays games capped to the previous-generation as a lowest common denominator. If all a new box does is play the old games at 4K60, is it really worth the $400 entry price? So if Sony present a new generation with next-gen software that utilises it, that gets gamers all excited, it'll provide an economic backbone to a software shift. Devs can look at a new generation that'll have a healthy start and drop lower-spec, older PCs as a baseline. If no-one makes a clean generational break, software progress will likely be retarded as it is on PC, with no-one wanting to commit to bleeding edge tech because it limits their audience.

So I think Sony could define a software generation simply by being the only people to come out with a new machine forcing a clean break. If they announce PS5 and it is significantly advanced, devs will target it as a new audience is guaranteed and PC targets will shift to a new higher minimum.

 

[Deleted member 11852]

Unlike years ago AAA third party games are released on PC and console...so anything a put on a new $500 console will be beaten by a new $3000 PC a year or two prior.

I don't see the relevance of this? People generally buy something new either becuae their old thing broke or becuase they want a better experience. Very few people exhibit that wholly fanboy craving to have a better experience than everybody else, or just mostly everybody else. Most people just want to have fun playing games.

From software/service/game point of view this model is better for consumers and even from hardware point of view this model offers more options for different tastes, so I think it'll be definitely better. However, I'm not so sure about developers.

If consumers care about that. If you're the type of consumer who buys a console cheap, buys games cheap, plays then once and never again, then no. It's too early to say how things will pan out in the long term. You can't even look at the AA/AAA market on the PC as a guide because the whole PC market has long been tethered to the technological trends of the console market. Changing the latter will have an impact but it would be difficult to predict how.

So I think Sony could define a software generation simply by being the only people to come out with a new machine forcing a clean break. If they announce PS5 and it is significantly advanced, devs will target it as a new audience is guaranteed and PC targets will shift to a new higher minimum.

Sony would need that tech to be reasonably widely adopted on PC otherwise it's just Sony. They can't drive a new generation on their own. But new audiences are not guaranteed for new tech. Plenty of tech fails - Kinect being the most recent gaming tech to be dropped.

 

[mosen]

With well over 200 million existing customers, why would anyone write a game for PS5? Well the same reason they ignored 200 million PS2 generation consoles to target PS360 and comparable PCs - because it was time to move on with games and make better games, which meant leaving the old machines behind. The economic argument then for ignoring the existing hardware base is that the gaming community also wants to move on and will buy new hardware.

Well, the gaming community isn't going to be so keen about buying new hardware if it plays games capped to the previous-generation as a lowest common denominator. If all a new box does is play the old games at 4K60, is it really worth the $400 entry price? So if Sony present a new generation with next-gen software that utilises it, that gets gamers all excited, it'll provide an economic backbone to a software shift. Devs can look at a new generation that'll have a healthy start and drop lower-spec, older PCs as a baseline. If no-one makes a clean generational break, software progress will likely be retarded as it is on PC, with no-one wanting to commit to bleeding edge tech because it limits their audience.

So I think Sony could define a software generation simply by being the only people to come out with a new machine forcing a clean break. If they announce PS5 and it is significantly advanced, devs will target it as a new audience is guaranteed and PC targets will shift to a new higher minimum.

PS2 to PS3 is a good example to reject my opinion but I think PS3/360 to XB1/PS4 is a far better and more realistic example which isn't against my suggestion. We have already experienced more cross-generation games and remasters in first 2-3 years of this generation than some unique and generational defining titles. So, it's not Sony's decision to shift overall game quality across industry to the next level by launching a new hardware with zero install base. Most developers tend to develop their games based on the most common dominator hardwares with the most active players. Sony may not force developer to make their games for PS4 but I think they will do that if they think it's better for them just like current generation. However, I don't think Microsoft is going to force them to make games for XB1 either but some of them definitely will support XB1 in the coming years.

Also last generations Microsoft and Nintendo policy were also in favor of Sony (PS3/360/Wii), but this time they are alone.