Nintendo GOing Forward.

[Syferz]

I don't think it's that far fetched. I believe IBM has stated that they are willing to license designs so I bet Nintendo could acquire the IP and manufacture on their own. From a die size, the WiiU CPU is aorund ~28mm on a 45nm process. That would be probably around 10mm on 28nm. And since it's a low clocked part, it should be portable to other manufacturers.

However, I'm not sure it's needed. The question I have would be possible to customize (or add some sort of hardware assist) to an existing ARM core to help with backwards compatibility. It's a simple RISC core without any fancy vector extensions/coprocessor so it may be feasible and cheaper.

I think the WiiU's GPU with it's embedded RAM is a bigger obstacle to backwards compatibility.

The way I see it, Nintendo's next handheld will launch on a very cheap and mature 28nm. I don't believe 20SOC will be cheap enough nor mature enough in the time frame that Nintendo want to launch (late next year, speculation).

In May, Iwata said that it would take them about 2 years before they are ready, so 2016 for their next handheld is expected. I think by then, 20nm will be plenty mature, a lot of people pretend that Nintendo doesn't understand tech, or know how to invest in it, but the gamecube and wii were both launched on the modern process when they came out. I don't think the benefits of 20nm is going to fall on deaf ears with Nintendo, especially since there are already 20nm chips in production today.

 

[McHuj]

I don't think 20nm will have enough industry support to ever be viable for low cost parts. (For instance, there's already plenty of rumors that NVIDIA is skipping 20nm and going directly to 16FF due limited 20nm benefits and cost).

If Nintendo waits until late 2016, 14/16FF should be a viable option, if they want to release sooner 28 is the only economical choice, imo.

 

[Fourth Storm]

McHuj may have a point. I believe I read that 20nm is the first node where costs may not actually be reduced relative to the larger process. It could be mature enough in a couple years, though. I don't think we'll see the next handheld next year. Although I anticipate 3DS releases to dry up significantly, we're at least getting S.T.E.A.M. and another Zelda (Majora's Mask remake?). I can see them doing a Pokemon X/Y-2 as well. Between that and Nintendo's teams being tied up on Wii U and QoL, I think spring/summer 2016 is more likely.

FinFETs present a whole different set of challenges, though. I don't know if that will be in Nintendo's price range in 2016. It's too bad FD-SOI is going nowhere. Seems like a good low-power tech and from what I've read, has comparable costs to FinFET on bulk.

 

[Syferz]

McHuj may have a point. I believe I read that 20nm is the first node where costs may not actually be reduced relative to the larger process. It could be mature enough in a couple years, though. I don't think we'll see the next handheld next year. Although I anticipate 3DS releases to dry up significantly, we're at least getting S.T.E.A.M. and another Zelda (Majora's Mask remake?). I can see them doing a Pokemon X/Y-2 as well. Between that and Nintendo's teams being tied up on Wii U and QoL, I think spring/summer 2016 is more likely.

FinFETs present a whole different set of challenges, though. I don't know if that will be in Nintendo's price range in 2016. It's too bad FD-SOI is going nowhere. Seems like a good low-power tech and from what I've read, has comparable costs to FinFET on bulk.

Considering the chips will be AMD made, and likely not quite as custom as we are use to seeing from Nintendo, I think they will use whatever process AMD is using for those chips, we should also remember that 20nm has reduced power draws too.

 

[Fourth Storm]

A few interesting articles relative to our discussion:

This article paints a more optimistic portrait of the 20nm node. Considering AMD plan on using it for Project Skybridge (which Nintendo may use a semi-custom variation of), I think it will become popular enough to be a potential candidate.

Also, it appears the IBM/GF deal is off. With IBM still in the manufacturing biz, perhaps Espresso could just be stuck onto an MCM, as it is now, in Nintendo's next home console. They could use it for bg downloads and perhaps some other rudimentary tasks besides BC and then scrap it in later revisions.

 

[Goodtwin]

A few interesting articles relative to our discussion:

This article paints a more optimistic portrait of the 20nm node. Considering AMD plan on using it for Project Skybridge (which Nintendo may use a semi-custom variation of), I think it will become popular enough to be a potential candidate.

Also, it appears the IBM/GF deal is off. With IBM still in the manufacturing biz, perhaps Espresso could just be stuck onto an MCM, as it is now, in Nintendo's next home console. They could use it for bg downloads and perhaps some other rudimentary tasks besides BC and then scrap it in later revisions.

Honestly, even though the tech heads here would hate the idea, Nintendo theoretically stick with Espresso design if they want. A dye shrink would allow the processor to be clocked higher, probably approaching 2 Ghz, and go with 6-8 cores. The Wii U's CPU has been a scapegoat for bad performing games, but the reality is the problem was always within the code, and not a hardware problem. When Warriors Orochi launched on Wii U the framerate issues were pegged as a CPU problem, but if Hyrule Warriors comes out and has just as many enemies on screen with no such performance issues, then its hard to really say the CPU was ever the problem, it was a software problem that could have been fixed, but required more work than the investment was deemed to be worth. If Nintendo were to go with a hybrid system like I think they will, a 6 core 1.8 Ghz Espresso CPU would be just fine for the Nintendo and 2nd party exclusives they would develop, and the Indy community that seem to have embraced this gen.

 

[Deleted member 13524]

There are two major architectures being used for games nowadays. PowerPC isn't one of them.

Could Nintendo keep using a ~20 year-old architecture for their next console? They could, but they'd be really stupid in doing so IMO.

 

[function]

When Warriors Orochi launched on Wii U the framerate issues were pegged as a CPU problem, but if Hyrule Warriors comes out and has just as many enemies on screen with no such performance issues, then its hard to really say the CPU was ever the problem

You can't infer such things based on the performance of two entirely different pieces of software.

 

[Fourth Storm]

Honestly, even though the tech heads here would hate the idea, Nintendo theoretically stick with Espresso design if they want. A dye shrink would allow the processor to be clocked higher, probably approaching 2 Ghz, and go with 6-8 cores. The Wii U's CPU has been a scapegoat for bad performing games, but the reality is the problem was always within the code, and not a hardware problem. When Warriors Orochi launched on Wii U the framerate issues were pegged as a CPU problem, but if Hyrule Warriors comes out and has just as many enemies on screen with no such performance issues, then its hard to really say the CPU was ever the problem, it was a software problem that could have been fixed, but required more work than the investment was deemed to be worth. If Nintendo were to go with a hybrid system like I think they will, a 6 core 1.8 Ghz Espresso CPU would be just fine for the Nintendo and 2nd party exclusives they would develop, and the Indy community that seem to have embraced this gen.

While Espresso is known to "punch above its weight," I agree with those who say it's time for a change. Even with a couple of node shrinks, the core is unlikely to clock much higher, due to the very short pipeline. Also, the out of order execution is pretty rudimentary compared to what's out there now, with only a single reservation station per execution unit.

Performance issues can be attributed to unoptimized code to a certain degree, but what developer is going to have the time to rewrite their engines for Nintendo's next gen machine? Why should Nintendo be so arrogant as to ask them to? I give the developers credit for getting games like Assassin's Creed and Mass Effect to even run on the core. Nintendo's market has shrunk dramatically in the last few years, and thus, they need to do everything possible to make developers' lives easier. Paired singles were good for their time and still yield surprising results, but they pale in comparison to the SIMD extensions on modern AMD and ARM processors.

One then has to wonder how much R&D would be spent integrating Espresso onto an SoC and if it would truly be worth the effort. Combine this with the fact that it's a 32-bit core and can only address up to 4 GB or RAM and the negatives really start to mount up.

 

[Goodtwin]

You can't infer such things based on the performance of two entirely different pieces of software.

Its the same development studio in the exact same genre... Yea, I think you can make some observations. Its the difference between ground up software compared to porting code that was designed for different hardware. Its about as good of a comparison as you can get. Im not saying the CPU wasn't the limiting factor with that game though, but if you can get the same result with better performance writing the code differently, then that shows that it can be done, but needs to be done differently. Two ways to skin a cat type of thing.

I agree that it shouldn't be the CPU Nintendo uses in its next console, but with Nintendo you never know.

 

[wco81]

Mario Kart boosted sales of the Wii U but it still lags behind the other consoles. Nintendo reported a loss.

Now what?

 

[Wynix]

Went off-topic with my speculation in another thread so I will just dump it here instead.

[speculation]
Nintendo will announce the 3DS successor at E3 next year, release in Japan in early October(obviously after TGS) and release in US/EU mid November, $169.99, drops to $149.99 after ~1yr.

Some may think it's too early but it's not; The 3DS launched in Feb/March 2011, if the [4DS] launches in Oct/Nov 2015 that would be 4 years & 8 months between launches.
Now add in the rapid decline of the 3DS and it is almost a sure thing.

Late 2015 + AMD(based on rumours/hints) = 20nm quad core CPU(ARM seems more likely) + 128-256 GPU cores IMO.
[/speculation]

 

[Grall]

Nintendo's already said their next consoles will be unified architecture and that they won't be ready to show anything for a couple years, meaning '16 at the earliest. Four years for a nintendo handheld isn't much, and 3DS isn't selling disastrously poorly either (it's making profits, unlike wuu), so I don't see why they need to hurry out a replacement.

 

[Wynix]

Nintendo's already said their next consoles will be unified architecture and that they won't be ready to show anything for a couple years, meaning '16 at the earliest.

Companies lie, also plans can change.

Four years for a nintendo handheld isn't much

Like I said ~4 years and 8 months, it's barely short of the magical 5 year mark and is just in time for the holiday period; Which would be more beneficial than launching Q1/Q2 in 2016.

3DS isn't selling disastrously poorly either (it's making profits, unlike wuu), so I don't see why they need to hurry out a replacement.

While the current sales aren't disastrous, It's obvious that the 3DS has shifted into a fast decline of sales, give it another year and the console will be competing with the Wii-U during NPD.
Nintendo need to get a successor out to make up the revenue lost from the declining 3DS.

 

[Grall]

You're speaking of future events that haven't happened yet as if they're historical fact and set in stone. Not so.

Nintendo hardware sales are largely driven by nintendo software quantity and quality. The hardware itself has nearly zero customer attraction power other than for a relatively few die-hard fans (such as myself, sigh), so as long as they can continue to pump out great games for 3DS it will sell at least sufficiently for the foreseeable future.

 

[Fourth Storm]

Yeah, even if 3DS software dries up next year, I can't see them releasing a replacement until 2016. Not only will they be selling QoL as new hardware, but I can imagine they will want to convert some of the 3DS users over to Wii U, which is looking to have a fairly strong first party lineup in 2015. They could also drop the price of Wii U by $50 or so and get 2DS under $100 MSRP by Christmas 2015. They could release a Greatest Hits 3DS line. There are ways to stretch it, and they'll have to, because the next handheld simply won't be ready with enough software by next year I reckon.

 

[wco81]

Even with MK, they lost money.

Can they really continue on the present course? Financially the losses aren't huge and they have a lot of cash.

But will they be satisfied with not losing huge amounts of money?

 

[Goodtwin]

Of course Nintendo isn't "happy" with losses, but they have to be happy with the turnaround from last year. Last year Q1 had about 160k in Wii U sales, Q1 this year accounted for over 500k. Wii U will be a holiday seller, selling upwards of 75% of its yearly sales in November and December. Mario Kart 8 was only out for one month of Q1 this year, what did the other two months provide? Nothing, and thanks to Mario Kart sales aren't falling back into the cellar. For the fiscal year, I find it hard to believe they wont make money. Wii U hardware is finally a positive impact instead of a negative, and the lineup of software starting in September is very strong. Hyrule Wariors, Bayonetta 2, and Smash Bros coupled secure a very profitable fiscal year. If Nintendo can sell a couple million units the holiday season at a profit, then games like Smash Bros and Mario Kart will pad Nintendo's pockets nicely. Even if you despise how Nintendo operates, you would have to concede that Nintendo will move some hardware and software this holiday season. Im guessing a fiscal year profit of about 350 million dollars.

 

[Entropy]

For sanity, I feel that this direct quote from Iwata in the beginning of 2014 needs to be posted.

In terms of our platform integration, as I explained to you a short while ago, we are not saying that we are planning to integrate our platforms into one. What we are saying is that we would like to integrate software development methods, operating systems, and built-in software and software assets for each platform so that we can use them across different machines. This means that if we manage to integrate our platforms successfully, we may in fact be able to make more platforms.

At the moment, we only have our current handheld devices and home consoles because if we tried to make more platforms, our development resources would be spread too thinly. The more we can share software across different platforms, the more development resources will be left for something else. Platform integration does not mean creating one type of platform, but the point is that the united method of software development will enable us to share our most precious software assets across different hardware.

In short, unified OS and services. Perhaps, although unclear, shared graphics API or at least middleware. And that's it.

 

[Shifty Geezer]

They should use Unity.