Middle Generation Console Upgrade Discussion [Scorpio, 4Pro]

[Deleted member 11852]

There is no actual evidence to support this. We actually heard the first rumor about Scorpio *first* and Phil Spencer was talking about the possibility of mid-cycle upgrades before even that rumor came out.

Yeah, the idea that Microsoft developed Scorpio as reaction to rumours of an upgraded PS4 is hard to swallow. Mark Cerny, in his post-PS4 Pro announcement interview, said the Pro began as an idea shortly after they launched PS4 so November 2013 to November 2016 launch - three years. The first real rumours of a mid-gen PS4 upgrade was earlier this year and while 18 months isn't impossible to get from idea to design to product, it's highly unlikely.

I wish this rumour/speculation would die.

 

[mrcorbo]

Sure, but maintaining goodwill by continuing to support a system you've effectively replaced, costs actual money.

Support in this sense means continuing to come out with games that have a XBOne mark 1 mode. It's going to be a while before that becomes a worse investment than coming out with a game that only runs on Scorpio. And any hardware accessories will work on both.

 

[Rikimaru]

There is no actual evidence to support this. We actually heard the first rumor about Scorpio *first* and Phil Spencer was talking about the possibility of mid-cycle upgrades before even that rumor came out.

We heard Scorpio rumour after Neo already had a devkit and dev conference.

There is a japanese interview in 2015 hinting about Neo.
http://www.dualshockers.com/2015/10...e-release-of-a-ps4-with-improved-performance/

 

[Shifty Geezer]

It's still not a thing. Why was XB1 designed with a VM base? Partly to enable an updated console with full BC. Scorpio (a more powerful console that runs the same software) was always going to happen.

 

[Rikimaru]

It's still not a thing. Why was XB1 designed with a VM base? Partly to enable an updated console with full BC. Scorpio (a more powerful console that runs the same software) was always going to happen.

It was designed this way for security. Windows base has many holes. Even if Windows base is compromised it won't lead to piracy (if hypervisor won't be hacked too of course).

 

[function]

MS wanted to be able to update game and dash independently without compatibility concerns, and also to tightly and efficiently control reliable resource split between the two. Security was probably a factor, but these were also explicitly stated aims.

It's a good thing to have if you plan to be able to move your games across platforms over time.

 

[Vic]

They are blowing out old stock to make way for the new SKU. It's really not negative when your new product is good enough that no one is going to want to buy your old one unless you discount it is it?

I don't agree that they will take on small loss on Scorpio. Those days are over for console makers.

Sorry I meant the price-cuts even since Xbox One was released. Sony meanwhile has done price-cuts as and when it pleased.

 

[mrcorbo]

We heard Scorpio rumour after Neo already had a devkit and dev conference.

There is a japanese interview in 2015 hinting about Neo.
http://www.dualshockers.com/2015/10...e-release-of-a-ps4-with-improved-performance/

That interview came first, then. Spencer's presentation was on March 1st. Neo first leak March 18th.

 

[Rikimaru]

I do not think Kotaku made an article as soon as developer knew about Neo and told them though.
I wonder for how long devs talked about it.

 

[iroboto]

I imagine a world where 2 marketing departments with access to similar data charting it over their BI tools and seeing similar trends.

They both started thinking it was time to up the cadence.

How they go about it after this mid generation will be interesting


Sent from my iPhone using Tapatalk

 

[Silent_Buddha]

I do not think Kotaku made an article as soon as developer knew about Neo and told them though.
I wonder for how long devs talked about it.

And you somehow think that Microsoft didn't start development for Project Scropio before Spencer mentioned that something like that might be a good idea? As a related side note, Dx12 started development well before Dx12 was announced or hinted at. Mantle was announced and a BETA implementation released between when Dx12 started development and Dx12 was actually announced. People ended up erroneously thinking that Dx12 was a panic response from Microsoft to the release of Mantle. That obviously wasn't the case.

The same goes for Project Scorpio. Unless you are hopelessly incompetent or you are crowd funding something you don't announce or hint at something until you've already done significant work to see if a product is viable and has a solid foundation for further development. That applies to both Sony and Microsoft.

It's quite obvious that Microsoft wanted to abstract the software from the hardware as much as possible. Not only to facilitate forwards/backwards compatibility for games, but to accommodate potential changes in hardware. Microsoft isn't a stranger to having to deal with software operating on multiple different hardware architectures as well as multiple different hardware generations. It's been their business for over 4 decades now.

And while x86 has been the foundation for most of their success, you can see them even now gaining experience on running their software on ARM based devices. They certainly aren't a stranger to RISC based devices (DEC Alpha ran a version of Windows NT, Windows CE ran on a variety of RISC based designs, etc.).

While I'm sure they are prepared to stay with x86, even there the groundwork has been laid for a transition to ARM if required as UWP is designed to allow application code to be compiled to run on both x86 and ARM. That said Project Scorpio isn't going to deviate from x86 as games developed for XBO aren't UWP applications. However, at some point after Project Scorpio when XBO no longer needs to be supported and UWP is more universal for game code on Xbox, it's entirely possible they could switch to an ARM based console while still maintaining backwards compatibility with the x86 based consoles.

TL: DR. When Microsoft made the move to the OS model used in the XBO, they were already committed to the idea of a moving hardware target in shorter timeframes than traditional console generations. UWP is work they have put in that synergizes and expands upon that. Not just allowing for updates in the GPU but also allowing for changes in the underlying CPU.

Regards,
SB

 

[Rikimaru]

http://www.i-programmer.info/news/144-graphics-and-games/5744-amd-no-directx-12.html

I'd say AMD definitely knew something. Also do not forget how arrogant was at the time. Windows 8 and Xbox One were the most mismanaged products I've ever seen (and W10 is not good too).

They started making DX12 after DX11 failure on xbo. They even copy pasted some text from Mantle documentation.

UWP is just an OS API, it does not abstract hardware, just OS calls.

 

[Silent_Buddha]

http://www.i-programmer.info/news/144-graphics-and-games/5744-amd-no-directx-12.html

I'd say AMD definitely knew something. Also do not forget how arrogant was at the time. Windows 8 and Xbox One were the most mismanaged products I've ever seen (and W10 is not good too).

They started making DX12 after DX11 failure on xbo. They even copy pasted some text from Mantle documentation.

UWP is just an OS API, it does not abstract hardware, just OS calls.

You do understand that both IHVs and Intel and some software developers assisted Microsoft with the development of Dx12. NVidia has already confirmed that they were working with Microsoft on Dx12 before Mantle was announced or released. Considering that, it's not unexpected that the wording will mimic some of what AMD used for Mantle as AMD assisted Microsoft with the development of Dx12, just like Intel and just like NVidia.

Regards,
SB

 

[goonergaz]

So again I understood you fine it seems. Your saying that IN 2013 gamers bought a ps4. In 2016 gamers will buy a ps4 pro and in 2018 gamers will but a ps5 or ps4 super pro ? In 5 years you think gamers will line up and buy 3 different systems from Sony ? You also think sony will be able to support 3 completely different specced systems ?

Stop reading me wrong, even in your quote from me I clearly say (and please - READ THIS PROPERLY) "coming out in 2019". That's 6 years in case you're unsure, so I'm under the impression we get a refresh every ~3 years with every other being a 'generational leap' and the in-between being a nice upgrade. Clear now!?

And I haven't even said they will be supporting 3 machines so stop putting words into my mouth. Before reveal I thought they may be going with a Apple style system but I think it was fairly clear the generations will remain as such and PS5 will be a new generation.

 

[temesgen]

Yeah, the idea that Microsoft developed Scorpio as reaction to rumours of an upgraded PS4 is hard to swallow. Mark Cerny, in his post-PS4 Pro announcement interview, said the Pro began as an idea shortly after they launched PS4 so November 2013 to November 2016 launch - three years. The first real rumours of a mid-gen PS4 upgrade was earlier this year and while 18 months isn't impossible to get from idea to design to product, it's highly unlikely.

I wish this rumour/speculation would die.

Both Sony and MS working closely with AMD would be familiar with the development of products moving forward from AMD and I would imagine a contingency path for upgraded hardware would have been mapped out at least at a high level if for no other reason than AMD would be presenting the options as a way to gain adoption of newer technologies.

The fact that both platform holders are using the same vendor is going to cause overlap if for no other reason than both had similar challenges due to the limitations of the hardware at the design stage and being caught in an early generation change (1080p to 4k TV) presents similar opportunity for both to provide a stop gap solution as well as test the waters with an iterative hardware approach.

Futhermore XB1 was somewhat a break from MS previous doctrine (powerful hardware) made MS need to provide an even more aggressive hardware spec and simply matching Sony would not have enough and the added complexity of their memory would all favor waiting another 12 months and essentially releasing new platform that can offer software compatibility with XB1 but aggressively positions MS as the premium brand.

 

[Shifty Geezer]

And what will the prices be ? $200 , $300 , $500 ? They will kill their market because they will be seen as rushing consoles. Who is going to buy a ps4 super pro or ps5 in 2018 when they know in just two years sony may release a ps5 pro or ps6 or whatever.

Who's going to but a TV now when a new, improved TV is a year away? Or a new phone when a better phone can be had in two years? Or a new GPU now when if they hold out for a couple of years they'll get a much better GPU?

It will still affect sony buyers as it will just be 3 years after the ps4 pro. Sony will train users to just wait for the ps5 pro in 2021

If users aren't interested in the 4Pro then the time to PS5 will be measured against the PS4 they bought. If 4Pro is bought in large numbers, it'll point to the market being accepting of half-gen upgrades. After all, if someone bought a 4Pro 3 years after their PS4 for its marginally improved visuals, why wouldn't they also buy a PS5 with much improved visuals 3 years later than their last purchase? And then, as they are that way inclined and like the best hardware, get a PS5Pro 3 years later? Makes little sense that someone will want a visual upgrade after 3 years enough to get a 4Pro but then be happy to wait 6 years until their next upgrade (to the 5Pro).

I think for Sony if the pro fails they would best double down on the ps4. Drive costs down as low as they could on the slim and ride out the generation they have a large installed base. Wait till 2020 or 2021 and then launch a true ps5

That's what they're doing anyway. 4Pro isn't a new generation 3 years early. A new console will be introduced as seems fit based on market movements, 2019 (6 year long generation) at the earliest.

 

[sebbbi]

While I'm sure they are prepared to stay with x86, even there the groundwork has been laid for a transition to ARM if required as UWP is designed to allow application code to be compiled to run on both x86 and ARM. That said Project Scorpio isn't going to deviate from x86 as games developed for XBO aren't UWP applications. However, at some point after Project Scorpio when XBO no longer needs to be supported and UWP is more universal for game code on Xbox, it's entirely possible they could switch to an ARM based console while still maintaining backwards compatibility with the x86 based consoles.

Consoles traditionally allow very low level GPU hardware access. Consoles allow you to directly submit GPU shader microcode, GPU command processor code and access memory of GPU acceleration data structures (such as HiZ, color compression, etc). It would be pretty hard to automatically port code like this to a completely different brand GPU, as programs can also generate new GPU code at runtime (by both CPU and GPU). Porting existing shader data files is not enough.

If you take away low level access, you will trade off some performance for better forwards compatibility. You'd also need to be careful not to expose highly platform specific features, such as cross lane operations (64 wide waves). Future hardware from a different manufacturer is likely different in low level details. Consoles have always been playgrounds for new rendering technology research. It would be sad if the GPU feature set was artificially limited to guarantee full forwards compatibility.

X86 vs ARM is not a problem, since most software doesn't runtime generate CPU code and both architectures are little endian CPUs (data is the same). X86->ARM binary (exe) translation could be done fully offline. There are some memory model differences (sequential consistency guarantees, atomics, etc) that would need special care. Porting binary ARM->x86 would be easier, because x86 has super strict memory ordering guarantees.

This presentation contains good info about memory models on different architectures: https://channel9.msdn.com/Shows/Going+Deep/Cpp-and-Beyond-2012-Herb-Sutter-atomic-Weapons-2-of-2

 

[eastmen]

Stop reading me wrong, even in your quote from me I clearly say (and please - READ THIS PROPERLY) "coming out in 2019". That's 6 years in case you're unsure, so I'm under the impression we get a refresh every ~3 years with every other being a 'generational leap' and the in-between being a nice upgrade. Clear now!?

And I haven't even said they will be supporting 3 machines so stop putting words into my mouth. Before reveal I thought they may be going with a Apple style system but I think it was fairly clear the generations will remain as such and PS5 will be a new generation.

but 6 months from Scorpio is 2018 not 2019th and you clearly said 6 months in your post. Even 2019 that's 3 systems in 6 years.

3 separate systems each with different capabilities within 6 years of each other. Ps4 in 2013 , Ps4 Plus in 2016 and Ps whatever in 2019 . That's extremely quick and my point stands

 

[eastmen]

Who's going to but a TV now when a new, improved TV is a year away? Or a new phone when a better phone can be had in two years? Or a new GPU now when if they hold out for a couple of years they'll get a much better GPU?

If users aren't interested in the 4Pro then the time to PS5 will be measured against the PS4 they bought. If 4Pro is bought in large numbers, it'll point to the market being accepting of half-gen upgrades. After all, if someone bought a 4Pro 3 years after their PS4 for its marginally improved visuals, why wouldn't they also buy a PS5 with much improved visuals 3 years later than their last purchase? And then, as they are that way inclined and like the best hardware, get a PS5Pro 3 years later? Makes little sense that someone will want a visual upgrade after 3 years enough to get a 4Pro but then be happy to wait 6 years until their next upgrade (to the 5Pro).

That's what they're doing anyway. 4Pro isn't a new generation 3 years early. A new console will be introduced as seems fit based on market movements, 2019 (6 year long generation) at the earliest.

1) your using markets with clearly established upgrade cycles that have existed for decades. Buying a new phone ever 2 years or so is engrained in the market because of how contracts work. New leases let you upgrade each year now. TV's people buy with new tech in the market. Those who bought hd tvs in the late 2000s will be ready to transition to 4k a decade later.

Sony on the other hand has for decades now established a 6-8 year console cycle which they are now moving to a 3 year console cycle. Its a large move by them

As for your last point. Of course Sony could be the next sega and flood the market with a new system every other year and have a sega cd to 32x to Saturn transition and be done as a company

 

[bunge]

Sony on the other hand has for decades now established a 6-8 year console cycle which they are now moving to a 3 year console cycle. Its a large move by them.

This is a farce. Sony has always introduced we models of the same system through the console's lifecycle.