Nintendo's next gen strategy for home & mobile

[Melqart]

The briefing makes it sound like they're taking an iOS-like route for future platforms:

Last year we also started a project to integrate the architecture for our future platforms. What we mean by integrating platforms is not integrating handhelds devices and home consoles to make only one machine. What we are aiming at is to integrate the architecture to form a common basis for software development so that we can make software assets more transferrable, and operating systems and their build-in applications more portable, regardless of form factor or performance of each platform. They will also work to avoid software lineup shortages or software development delays which tend to happen just after the launch of new hardware.

That last sentence is one of their biggest problems with the launches of 3DS and Wii U (among other problems obviously). People like to say that the casual audience has moved on to tablets and smartphones which is true in some cases but not all of them. The Wii sku of Just Dance 2014 wouldn't be one of the best-selling games right now if that was true.

Nintendo mostly stopped supporting the Wii in 2011 to focus on 3DS and Wii U development. That was a mistake. Even with the aging hardware and high rate of piracy, there was still an audience. The casual demographic targeted by the Wii isn't one who likes to buy new devices every few years (a problem Apple also has).

The Wii U has a lot of problems, don't get me wrong. Price is a major issue. Regardless of inflation rates, $300-$350 is not a mass market price point for the demographic they're chasing. $250 at the most, $200 bundled with a game being the sweet spot.

The HD port of the Zelda title was developed in six months while SquareEnix has been working on a port of FFX for years in comparison due to difference in architecture (Wii U has new versions of Wii components). If the Wii U came out minus the gamepad as-is in 2011 for $200 and all new games were basically upscaled Wii games (with some new effects, textures, etc.) that worked on both platforms, I think it would have been successful. It would have been a nice compromise to ensure the old platform still got games. Exclusives for Wii U could have a different case or something.

If Nintendo's next platforms share a common software architecture regardless of hardware or form factor differences, then they can do something like this (which iDevices already do).

I don't expect Nintendo to ever compete in the enthusiast console space ever again. The Wii U could have been 80% of what the Xbone is and still be in the same sales position. The main reason they shifted focus was because the Gamecube was a failure and the DS was successful. There are already two players in that space competing for razor thin margins and Nintendo likes to see long-term growth. If the next platform doesn't work out, I can see them leaving the market altogether and going back to love hotels and stuff. Maybe try slot machines like so many other former Japanese heavy-hitters.

 

[Commenter]

One factor Nintendo doom-sayers are not taking into account. Console generations last a lot longer now than in the days of PS 1 and 2. PS3/360 had unchallenged lifespans of roughly 8 years, before a new generation was released. This generation may well need to last ten years before consumers will demand a new set of consoles. Do not assume that what applied before necessarily applies to today's market.

 

[BRiT]

Do not assume that what applied before necessarily applies to today's market. The same reasoning could be used to make the point that console generations are moving to be shorter. They could possible be on a 4-5 year cycle. There is a new Phone/Tablet model released every 6-12 months as evidenced by the very popular iPhone, iPad, Android Phones and Tablets, and Kindle Fire Tablets.

 

[RudeCurve]

Does anyone know if it's trivial to recompile ARM code to MIPS?

 

[Grall]

Is 4K going to be required for a next gen console?

You know, that's a really interesting question!

If the "trend" started with xbone/PS4 (two individual consoles, one generation of hardware; not much of a trend I admit) continues, I doubt 4k would be reliably reachable in the traditional console generation timespan. As we all (?) know, "Moore's Law" is busy stalling out right now, we're not getting double the density every 36ish months anymore, and since 4K is 4x the pixels of 1080P we'd need at least two doublings unless a rendering breakthrough occurs that allows us to draw more using less hardware. Unlikely, at this stage, and probably unlikely for a console where momentum would favor current tools and procedures over something new, immature and unknown.

So assuming silicon integration technology that offers 4x the amount of transistors in 8ish years, it would be a new process, comparatively unreliable to what is used today for current new-gen consoles, with probably poor or poor-ish yields, meaning a big christmas launch would be painful or even impossible. In other words, consoles targetting 4K could end up being really expensive.

Also, it would take time for 4K to make an impact. The consumer electronics industry seems to be embracing 4K with gusto, because they fairly desperately need something new to sell, but since current HDTVs have only become ubiqutous fairly recently not everybody is going to be eager to jump onto the next, new thing immediately. Especially if price is going to be high initially. A 40+ inch HDTV can be had for ridiculously little money today (even if it isn't that good), 4K is obviously going to be more costly for quite a while, and the middle class is getting more and more squeezed financially all the time.

The economics just might not be there for 4K next generation.

Does anyone know if it's trivial to recompile ARM code to MIPS?

If written in high-level language, it should be. Code has been re-compiled between radically different architectures for decades now, so no reason there should be any major issues. Porting between 32- and 64-bit architectures is more involved from what I have read, but I don't exactly know why. Apparantly unexpected things can happen in a program if variables are suddenly double the size or whatever. *shrug* I'm not a programmer, I'm sad to admit.

 

[Shifty Geezer]

Does anyone know if it's trivial to recompile ARM code to MIPS?

Swapping CPU makes little odds as long as the compiler is good is the processors are comparable. With so many variables though, you can't be sure swapping a CPU in a system won't come with issues.

I'm not sure of the relevance of your question though. Are you thinking MIPS in the home console, ARM in the handheld, running the same game?

 

[RudeCurve]

No I'm thinking MIPS for both home and portable going forward. Imag Tech purchased MIPS so I see a lot of future potential with that architecture.

http://www.imgtec.com/news/Release/index.asp?NewsID=804

 

[Shifty Geezer]

Then what does it matter if ARM code can be recompiled for MIPS then? Surely the new platform will get new software written in the native ISA.

 

[liolio]

It's a good news they are heading in the right direction.
I wish Nintendo had its Sony moment and that they will get their acts together.
I think nobody wants or expect them to launch powerhouse be it in the home or handheld console realm, though they need to do better, the good news is that there is a lot room to improve.
I think Nintendo should stick to its view on gaming devices, they are indeed toys:
Toys that need to be both cheap and good.

Now will Nintendo manage to break away from their hold habits? Tough to guess.
The best Nintendo could would be to focus on software (/OS) and a forward compatible light weight API

For the hardware say they should pass on anything fancy/costly, which for me means passing on big brands:
No AMD, no Nvidia, no Qualcomm, no powerVr, etc.
I think their best should be to stick to ARM (CPU and gpu) and go with either Korean or Chinese integrator (or integrate ARM IP them-sleves).
I don't think that going for MIPS OS a good idea, they would bond themselves to a single vendor.

AMD could be an option if and only if they make Nintendo a crazy good deal, I would pass on x86cores.

 

[tuna]

Then what does it matter if ARM code can be recompiled for MIPS then? Surely the new platform will get new software written in the native ISA.

I do not believe anyone (except for OS devs) will write stuff in MIPS assembly.

But anyway, the compilers and debuggers are probably worse for MIPS than ARM/x86.

 

[Shifty Geezer]

I do not believe anyone (except for OS devs) will write stuff in MIPS assembly.

I meant complied for whatever CPU is in the new hardware. Devs will target the new hardware with whatever ISA it has, regardless of whether they actually code to that level or not, and not care one jot if it's cross-compatible with other processor architectures.

 

[Sonic]

It's a good news they are heading in the right direction.
I wish Nintendo had its Sony moment and that they will get their act together.
I think nobody wants or expect to launch powerhouse be it in the home or handheld console realm, though they need to do better, the good news is that there is a lot room to improve.
I think Nintendo should stick to its view on gaming devices, they are indeed toys:
Toys that need to be both cheap and good.

Now will Nintendo manage to break away from their hold habits? Tough to guess.
The best Nintendo could would be to focus on software (/OS) and a forward compatible ligjt weight API

For the hardware say they should pass on anything fancy costly, which for me passing on big brands:
No AMD, no Nvidia, no Qualcomm, no powerVr, etc.
I think their best should be to stick to ARM (CPU and gpu) and go with either Korean or Chinese integrator.
I don't think that going for MIPS OS a good idea, they would bond themselves to a single vendor.

AMD could be an option if and only if they make Nintendo a crazy good deal, I would pass on x86cores.

I disagree with your statement that Nintendo should refrain from using a known and proven tech whether it's AMD, Intel, ImgTec, Nvidia, etc. It would be pointless not to explore options of people who have experience designing SoC's in a cost effective manner. No sense in not seeing if one of these companies is able to design an architecture that fits within Nintendo's budget.

 

[liolio]

Wrt to hardware it might be misconception of mine but I noticed that Chinese integrators usually stick to ARM set-ups, it seems to me that mediatek also does so on its lower end soc, so in my mind there might be a reason behind that matter of fact, my assumption is that one way or another it costs less.
So when I say no big brand I mean stick to arm IPs, though ARM is brand too and a big one.

 

[AlNom]

Then what does it matter if ARM code can be recompiled for MIPS then? Surely the new platform will get new software written in the native ISA.

Maybe for 3DS library? (And/or ease of porting from iOS/android )

 

[Shifty Geezer]

Maybe for 3DS library?

That library will need a complete recompile for a new graphics architecture, no? Or are they going to put DS's graphics chip in there as well (wouldn't surprise me!)?

 

[Deleted member 13524]

AMD could be an option if and only if they make Nintendo a crazy good deal, I would pass on x86cores.

Why pass on x86 cores? Intel has been making quite a solid competition to ARM in the Atom department, and AMD is definitely going after handhelds whenever they get the chance.

 

[Grall]

x86 has tons of legacy garbage in its ISA, and it's inherently less efficient because of that and because of fewer GPRs and so on. Why go with x86 when it quite literally has zero advantages to ARM and other ISAs and a number of drawbacks?

x86s one redeeming feature that has kept it alive for over thirty years now is backwards compatibility, which isn't worth shit in the mobile space. You're never going to run decades-old DOS, win3.x, win9x etc apps on a portable system.

So there's no need for portable x86, really. Intel and AMD just needs portable x86 for their own future survival, that is all.

 

[arijoytunir]

i think iwata wanted to share the same hardware architecture on both home and handheld console so that they can released different versions of the same game which will significantly ease up developement time on the developers end and will help to avoid the software drought after the launch.

About the hardware
They can use arm and kepler gpus as both will be matured enough in the time frame of 2015-17.
they may choose 8-4 64bit cores for their handheld and home console. And also use kepler in both hardwares as they both scale perfectly.
Ram wise they will use ddr4 with around 8 gb or more . Possibly an edram.

Also i think they will make their os more multimedia focused like sony's vita os.

 

[RudeCurve]

Why go with Nvidia when they could go with PowerVR's TBDR and not need eDRAM? All of Apple's portable devices use PowerVR so it's well understood and performance is proven let alone multicore scaling for a home console.

 

[arijoytunir]

can powervr scale like kelper ?