Nintendo DX GPU?

[MisterLilBig]

NX is 28nm. The CPU is a quad-core Cortex-A15 at 2.1GHz for the console and 800MHz for the handheld. The console has 384 GCN 1.2 SPs, and the handheld has 64. They're aiming for $130 for the basic handheld, $160 for the XL version, and $200-230 for the console. RAM isn't finalized, but both will have DDR3 (LPDDR3 for the handheld, or maybe for both), probably 2GB for the handheld and 8GB for the console.

Not sure why they would go for an A15 tho, I'd expect atleast an A57.

A glorified "smartphone" SoC is what makes the most sense for Nintendo. And if it is being made by AMD, is has to be the A57. I expect 8 A57 cores.

If they don't go ARM, they SHOULD stay with what they already have on the Wii U, PowerPC and shrink it.

The Wii U CPU and GPU are manufactured in 45nm. The Wii U uses LESS than 35 watts while doing everything. If the NX is really a Two-In-One system like a Tablet/Console, I don't see why the whole thing should be that expensive.


If Nintendo goes x86 it will fail. It does not benefit them. At all.

 

[RussianSensation]

If they don't go ARM, they SHOULD stay with what they already have on the Wii U, PowerPC and shrink it.

Absolutely not. This is the worst idea ever. A clean slate design needs to be done where Nintendo needs to abandon outdated and old architectures/instruction sets. This should be a non-starter because an x86 APU from AMD would beat any PowerPC alternative in price/performance, perf/watt and ease of coding. The fact that Sony/MS ditched PowerPC already means all of this analysis was performed by them, which means Nintendo doesn't even need to entertain this idea.

If Nintendo goes x86 it will fail. It does not benefit them. At all.

I disagree with this completely. I am of the view that Nintendo NOT going x86 will guarantee their next home console will be a failure.

1) Not having x86 architecture automatically means developers can't do easy PC/PS4/XB1 ports. That means out of the gate, they would be setting an extremely unfavourable/harsh 3rd party developer environment for their NX home console. Even though Nintendo has struggled with 3rd parties, why should they give up completely and literally start off with a design that is the most likely to cause them to have the worst 3rd party line-up of any Nintendo console ever made because it'll require specific coding and not easy x86 porting? This is the complete opposite of making console development seamless, which is what Nintendo needs and what Sony/MS did with PS4/XB1.

For some people in this thread stating Nintendo should even consider PowerPC, that's a ridiculous theory. To even propose this is admittance to not even understanding all the reasoning for why MS/Sony abandoned PowerPC in the first place.

2) Even the best ARM graphics are still well behind compared to even XB1's GPU. If Nintendo launches a console that can't even have graphics as good as XB1, not even a $199 price will save it. You cannot take a risk of launching such an under-powered console in 2016 when your primary competitors will be well into the 3rd year of their life-cycle. ARM CPU/GPU SOC would have such an inferior performance basis, that the 2016 NX would be outdated on Day 1 of release. Why would Nintendo make such a risky move? If Nintendo's core CPU/GPU benchmark is an underpowered ARM SoC (all of them are compared to x86 APUs), then why would the consumer want to buy a Nintendo home console when they can just play Nintendo's mobile smartphone/tablet games? The best ARM SoC is so far behind x86 APUs that going with an ARM SoC and abandoning the Wii U completely undermines the reasoning for abandoning the Wii U in the first place:

--> Wii U was too underpowered and cost too much at the same time. This caused poor graphics and lack of 3rd party support. An ARM SoC doesn't fix the issue of NX remaining underpowered. That means logically a pure ARM SoC powered NX cannot be a successful solution since it doesn't solve the underpowered scenario. Secondly, what is the selling point of an underpowered ARM console that has graphics barely better than the Wii U? Why would consumers care when they didn't care about the Wii U? Price? A low price ($199) alone is going to signal that this console is aimed at casuals again and that will only alienate the majority of console gamers. Casuals would rather play free/cheap smartphone/tablet games than purchase a $199 ARM console.

3) You are ignoring the possibility of an option of having an x86 APU + ARM SOC. It shouldn't be that expensive to incorporate both instruction sets into the console considering the level of GPU/CPU horsepower in XB1's APU could probably be purchased for $60 at most by 2016. That leaves plenty of room for an ARM SOC if they wanted that. Combining ARM + X86 APU allows for the greatest compatibility with their handheld but it doesn't throw the home console under the bus for 3rd party cross-platform x86 ports.

4) Given how well XB1/PS4 are selling, and many reasons of why MS/Sony switched to x86 with this generation, it stands to reason that the most logical outcome of this is that XB2/PS5 will also have x86 APUs. MS/Sony switched to x86 to ensure game development can be as seamless/smooth as possible for game developers (a goal Nintendo must align themselves with as well) and since x86 R&D development has way less risk than PowerPC or otherwise, it means MS/Sony won't need to be scrambling for custom designs when it's time to release XB2/PS5. Also, semi-custom x86 APUs are far more cost effective from a price/performance point of view. x86 is the best of all worlds for consoles since it allows Intel/AMD/NV to fund the entire R&D and then Nintendo/MS/Sony just buy these semi-custom designs (i.e., it cost so much money to develop the fully custom Cell PowerPC, that it's a solid enough lesson to see how something like that is a failed strategy). PS4 is already crushing PS3 as far as profitability goes and ease of use/coding. It's a brilliant alignment of synergies and price/performance -- something PowerPC or ARM SoC cannot offer. The former has horrible perf/watt, is expensive, not easy to code for since it's not an industry standard, and has an extremely weak road-map. The latter is way too underpowered to make any sense as the primary backbone for CPU/Graphics, while also breaking seamless x86 ports. Taking all of that into account, transitioning to x86 ensures backwards compatibility for XB2/PS5 with current consoles. That means for Nintendo not go to x86 as well is suicidal akin to Nokia adopting Windows smartphone OS because they would be going completely against the industry trends.

5) We do not need to assume that if the New 3DS successor has an ARM SoC that a home console NX system also must have a powerful ARM SoC. This may or may not be a required condition and certainly it wasn't a requirement for integration of PS4 + PS Vita.

We don't even need to look far. NV Shield TV has a very cutting edge ARM SoC, it's a spec power house for HTPC and yet that thing barely sells for $199 with a controller. It exhibits ALL of the signs of why strictly using an ARM SoC is the most risky strategy for home consoles:

--> Under-powered => crappy graphics. This alienates mainstream gamers who use XB1/PS4 as a benchmark for comparison.
--> Extremely poor 3rd party support because it's not using x86 instruction set. Not many want to spend $200 on a console that has limited game support from 3rd parties when the competitor is selling their console for $299 and it has 3rd party support in spades.

That means Nintendo's key to success has to be differentiation and/or making the console at least reach parity with XB1 and make 3rd party ports as seamless as possible.

Finally, we have to actually look at the rumors and key statements being made:

"Last December, Devinder Kumar (CFO of AMD) revealed that they are currently working on a semi-custom processor he added, "I will say that one is x86 and the other is ARM, and at least one will be beyond gaming, right. ... They [the customers] are going to announce it and then ... you will find out that it is AMD's APU that is being used in those products."
http://www.crossmap.com/news/ninten...ts-spotted-first-is-a-miiverse-type-sys-22090

Let's think about that for a second. Since Sony/MS already have an AMD APU and there is no other large customer in the world that has shown interest in a semi-custom AMD APU, what is this AMD APU that Kumar is talking about? The only logical conclusion is Nintendo's NX home console. Now the question is whether the NX home console will have x86 APU + ARM or just x86 APU while the handheld controller will have ARM SoC (i.e., the 2nd semi-custom design AMD was discussing).

Summary:

ARM SoC/PowerPC do not solve the biggest issues of the Wii U: (1) lack of CPU/GPU horsepower, (2) ability for PC/PS4/XB1 developers to do seamless cross-platform ports. ARM SoC - too underpowered which means NX will get steam-rolled by 2019 by PS5/XB2. It also guarantees that the home console NX would launch as an outdated console on day 1. PowerPC is an inferior solution in price/performance, perf/watt, ease of coding and has questionable future road-map which suggests ensuring backwards comparability with the NX2 in the next decade would be next to impossible.

This is like a repeat of predictions for PS4/XB1 where people were predicting ARM SoC or PowerPC but the most obvious solution was x86 APU. What has changed since 2013? Nothing. X86 APU still crushes ARM SoC and PowerPC in all key metrics that matter and it also provides the most seamless and least risk prone succession for backwards compatibility for next gen XB/PS/NX consoles in the next decade, etc.

 

[jpiniero]

Instruction set compatibility isn't that big of a deal. Most games these days are probably C++ exclusively, so recompiling on ARM would be not a big deal. Dev kit compatibility is a bigger issue... and Nintendo is known for having terrible dev kit tools. That being said, if the NX sells decently at launch, third parties will make games for it. Even if means hurting the potential of the XBO/PS4 version.

MS btw is pretty close to giving up on the XBone, so it seems pretty unlikely you will see another console from them.

 

[NTMBK]

2) Even the best ARM graphics are still well behind compared to even XB1's GPU. If Nintendo launches a console that can't even have graphics as good as XB1, not even a $199 price will save it.

You know that AMD can make ARM chips too, right? They had a 20nm A57 + GCN tablet chip designed and ready as part of Project Skybridge, which got cancelled due to GloFo's failure to deliver (again). That means that the hard work of getting ARM IP and AMD IP to play together nicely is done.

There's no reason why they couldn't get a PS4 level APU, but with the Jaguar cores swapped out for A57/72.

 

[Snafuh]

ARM SoC/PowerPC do not solve the biggest issues of the Wii U: (1) lack of CPU/GPU horsepower, (2) ability for PC/PS4/XB1 developers to do seamless cross-platform ports.

Lack of GPU power has nothing to do with the CPU design. Nintendo could ask AMD or NVidia to build a 300mm ARM APU with more GPU power than the PS4. Depending on the core design a ARM CPU will not be slower than the PS4/Xbox One CPU.
x86 doesn't lead to seamless cross-platform ports. The 3 consoles still use different graphics APIs. You have to build a seperated render path for all of them no matter if x86, ARM or PowerPC. If cross platform ports would be easy on x86 we would see more Linux games.
And I don't see why Nintendo need cross-platform ports. There are already 40 million consoles out there which can play the latest Assassin's Creed, Battlefield, Fifa etc. Nintendo will not get people to buy their console with cross-platform ports.
They need a strong exclusive line up and Nintendo's relations to third parties are good enough for this. Most major publishers are willing to develop games for Nintendo consoles if they see the opportunity for good sales figures

 

[Techhog]

Not sure why they would go for an A15 tho, I'd expect atleast an A57.

A glorified "smartphone" SoC is what makes the most sense for Nintendo. And if it is being made by AMD, is has to be the A57. I expect 8 A57 cores.

If they don't go ARM, they SHOULD stay with what they already have on the Wii U, PowerPC and shrink it.

The Wii U CPU and GPU are manufactured in 45nm. The Wii U uses LESS than 35 watts while doing everything. If the NX is really a Two-In-One system like a Tablet/Console, I don't see why the whole thing should be that expensive.


If Nintendo goes x86 it will fail. It does not benefit them. At all.

They'll want to stick with an older CPU because it's more "proven." Also, it's not a Two-In-One. This has been said many times, yet people still assume that's what it'll be even after the late Iwata denied it. It's two seperate hardware units with similar architectures and running the same OS so that they can easily share games.

 

[Techhog]

Absolutely not. This is the worst idea ever. A clean slate design needs to be done where Nintendo needs to abandon outdated and old architectures/instruction sets. This should be a non-starter because an x86 APU from AMD would beat any PowerPC alternative in price/performance, perf/watt and ease of coding. The fact that Sony/MS ditched PowerPC already means all of this analysis was performed by them, which means Nintendo doesn't even need to entertain this idea.



I disagree with this completely. I am of the view that Nintendo NOT going x86 will guarantee their next home console will be a failure.

1) Not having x86 architecture automatically means developers can't do easy PC/PS4/XB1 ports. That means out of the gate, they would be setting an extremely unfavourable/harsh 3rd party developer environment for their NX home console. Even though Nintendo has struggled with 3rd parties, why should they give up completely and literally start off with a design that is the most likely to cause them to have the worst 3rd party line-up of any Nintendo console ever made because it'll require specific coding and not easy x86 porting? This is the complete opposite of making console development seamless, which is what Nintendo needs and what Sony/MS did with PS4/XB1.

For some people in this thread stating Nintendo should even consider PowerPC, that's a ridiculous theory. To even propose this is admittance to not even understanding all the reasoning for why MS/Sony abandoned PowerPC in the first place.

2) Even the best ARM graphics are still well behind compared to even XB1's GPU. If Nintendo launches a console that can't even have graphics as good as XB1, not even a $199 price will save it. You cannot take a risk of launching such an under-powered console in 2016 when your primary competitors will be well into the 3rd year of their life-cycle. ARM CPU/GPU SOC would have such an inferior performance basis, that the 2016 NX would be outdated on Day 1 of release. Why would Nintendo make such a risky move? If Nintendo's core CPU/GPU benchmark is an underpowered ARM SoC (all of them are compared to x86 APUs), then why would the consumer want to buy a Nintendo home console when they can just play Nintendo's mobile smartphone/tablet games? The best ARM SoC is so far behind x86 APUs that going with an ARM SoC and abandoning the Wii U completely undermines the reasoning for abandoning the Wii U in the first place:

--> Wii U was too underpowered and cost too much at the same time. This caused poor graphics and lack of 3rd party support. An ARM SoC doesn't fix the issue of NX remaining underpowered. That means logically a pure ARM SoC powered NX cannot be a successful solution since it doesn't solve the underpowered scenario. Secondly, what is the selling point of an underpowered ARM console that has graphics barely better than the Wii U? Why would consumers care when they didn't care about the Wii U? Price? A low price ($199) alone is going to signal that this console is aimed at casuals again and that will only alienate the majority of console gamers. Casuals would rather play free/cheap smartphone/tablet games than purchase a $199 ARM console.

3) You are ignoring the possibility of an option of having an x86 APU + ARM SOC. It shouldn't be that expensive to incorporate both instruction sets into the console considering the level of GPU/CPU horsepower in XB1's APU could probably be purchased for $60 at most by 2016. That leaves plenty of room for an ARM SOC if they wanted that. Combining ARM + X86 APU allows for the greatest compatibility with their handheld but it doesn't throw the home console under the bus for 3rd party cross-platform x86 ports.

4) Given how well XB1/PS4 are selling, and many reasons of why MS/Sony switched to x86 with this generation, it stands to reason that the most logical outcome of this is that XB2/PS5 will also have x86 APUs. MS/Sony switched to x86 to ensure game development can be as seamless/smooth as possible for game developers (a goal Nintendo must align themselves with as well) and since x86 R&D development has way less risk than PowerPC or otherwise, it means MS/Sony won't need to be scrambling for custom designs when it's time to release XB2/PS5. Also, semi-custom x86 APUs are far more cost effective from a price/performance point of view. x86 is the best of all worlds for consoles since it allows Intel/AMD/NV to fund the entire R&D and then Nintendo/MS/Sony just sell them the designs. It's a brilliant alignment of synergies and price/performance -- something PowerPC or ARM SoC cannot offer. The former has horrible perf/watt, is expensive, not easy to code for since it's not an industry standard, and has an extremely weak road-map. The latter is way too underpowered to make any sense as the primary backbone for CPU/Graphics, while also breaking seamless x86 ports. Taking all of that into account, transitioning to x86 ensures backwards compatibility for XB2/PS5 with current consoles. That means for Nintendo not go to x86 as well is suicidal akin to Nokia adopting Windows smartphone OS because they would be going completely against the industry trends.

5) We do not need to assume that if the New 3DS successor has an ARM SoC that a home console NX system also must have a powerful ARM SoC. This may or may not be a required condition and certainly it wasn't a requirement for integration of PS4 + PS Vita.

We don't even need to look far. NV Shield TV has a very cutting edge ARM SoC, it's a spec power house for HTPC and yet that thing barely sells for $199 with a controller. It exhibits ALL of the signs of why strictly using an ARM SoC is the most risk strategy for home consoles:

--> Under-powered => crappy graphics. This alienates mainstream gamers who use XB1/PS4 as a benchmark for comparison.
--> Extremely poor 3rd party support because it's not using x86 instruction set. Not many want to spend $200 on a console that has limited game support from 3rd parties when the competitor is selling their console for $299 and it has 3rd party support in spades.

That means Nintendo's key to success has to be differentiation and/or making the console at least reach parity with XB1 and make 3rd party ports as seamless as possible.

Finally, we have to actually look at the rumors and key statements being made:

"Last December, Devinder Kumar (CFO of AMD) revealed that they are currently working on a semi-custom processor he added, "I will say that one is x86 and the other is ARM, and at least one will be beyond gaming, right. ... They [the customers] are going to announce it and then ... you will find out that it is AMD's APU that is being used in those products."
http://www.crossmap.com/news/ninten...ts-spotted-first-is-a-miiverse-type-sys-22090

Let's think about that for a second. Since Sony/MS already have an AMD APU and there is no other large customer in the world that has shown interest in a semi-custom AMD APU, what is this AMD APU that Kumar is talking about? The only logical conclusion is Nintendo's NX home console. Now the question is whether the NX home console will have x86 APU + ARM or just x86 APU while the handheld controller will have ARM SoC (i.e., the 2nd semi-custom design AMD was discussing).

Summary:

ARM SoC/PowerPC do not solve the biggest issues of the Wii U: (1) lack of CPU/GPU horsepower, (2) ability for PC/PS4/XB1 developers to do seamless cross-platform ports. ARM SoC - too underpowered which means NX will get steam-rolled by 2019 by PS5/XB2. It also guarantees that the home console NX would launch as an outdated console on day 1. PowerPC is an inferior solution in price/performance, perf/watt, ease of coding and has questionable future road-map which suggests ensuring backwards comparability with the NX2 in the next decade would be next to impossible.

This is like a repeat of predictions for PS4/XB1 where people were predicting ARM SoC or PowerPC but the most obvious solution was x86 APU. What has changed since 2013? Nothing. X86 APU still crushes ARM SoC and PowerPC in all key metrics that matter and it also provides the most seamless and least risk prone succession for backwards compatibility for next gen XB/PS/NX consoles in the next decade, etc.

You're assuming that Nintendo cares about third-party ports at this point. That ship has sailed. Being able to share games between the console and handheld is more important for them. Thinkb of the NX handheld as the iPhone and the NX console as the iPad. That's the kind of set-up they're going for. It needs to be easy to port games between them, so the need to have the same hardware architecture and OS. I know that you think that the NX is just a console, but it's not. It's a platform encompassing a handheld unit and a home unit (though the units are separate and it's not just a handheld with a dock for home use like some people here think).

 

[monstercameron]

You're assuming that Nintendo cares about third-party ports at this point. That ship has sailed. Being able to share games between the console amd handheld is more important for them. Thinkb of the NX handheld as the iPhone and the NX console as the iPad. That's the kind of set-up they're going for. It needs to be easy to port games between them, so the need to have the same hardware architecture and OS. I know that you think that the NX is just a console, but it's not. It's a platform encompassing a handheld unit and a home unit (though the units are separate and it's not just a handheld with a dock for home use like some people here think).

I would advise against making such an absolute statement about what the NX will/not be. Also the differences in hardware has not stopped ninty from porting games to mobile.

@RS I wouldn't count arm out. There is plenty of knowledge and experience with the isa -ask exophase and he could expound on such a topic. Not to mention that arm currently has cores with higher ipc than anything amd has released so far. I wouldn't be surprised if they didn't have 16nm ff macros for chip designer to just run with. Also you under estimate what a company like imagination could pull together in terms of a gpu. Let's not forget nvidia, I don't know if they have a semicustom (or what ever parlance) division but they have the IP and talent to pull off an arm based console chip.

On another note I think I read on b3d that jaguar is harder to get the most utilization out of because it is an OoO design vs the previous gen's ppc based chips. If I'm not misrepresenting or just misremembering this, wouldn't nvidias Denver cores be a better fit for the consoles due to -among other things- its in-order design?

 

[Silverforce11]

Instruction set compatibility isn't that big of a deal. Most games these days are probably C++ exclusively, so recompiling on ARM would be not a big deal.

Tell that to the guys who ported Batman to PC.

Also it just isn't coding language and shader language differences. It's memory access, shared resources in an APU design like on the Xbone/PS4.

In the gamedev circle, this was touted as a major reason for difficulties in porting Batman since it was designed to not stream textures in from system ram (PC or any separate CPU/GPU system), but to call directly from the shared high speed memory pool.

The last thing Nintendo wants are big AAA studios to shun porting to their crap consoles. They'll be stuck with their own ecosystem games again, minus the charm of gimmick since it's already been done prior, they won't have anywhere as good a success this time around going for weak/incompatible hardware + gimmick.

 

[Techhog]

I would advise against making such an absolute statement about what the NX will/not be. Also the differences in hardware has not stopped ninty from porting games to mobile.

@RS I wouldn't count arm out. There is plenty of knowledge and experience with the isa -ask exophase and he could expound on such a topic. Not to mention that arm currently has cores with higher ipc than anything amd has released so far. I wouldn't be surprised if they didn't have 16nm ff macros for chip designer to just run with. Also you under estimate what a company like imagination could pull together in terms of a gpu. Let's not forget nvidia, I don't know if they have a semicustom (or what ever parlance) division but they have the IP and talent to pull off an arm based console chip.

On another note I think I read on b3d that jaguar is harder to get the most utilization out of because it is an OoO design vs the previous gen's ppc based chips. If I'm not misrepresenting or just misremembering this, wouldn't nvidias Denver cores be a better fit for the consoles due to -among other things- its in-order design?

What I said is true. If you believe otherwise you haven't been paying attention. Again, I'll use the analogy: NX handheld is to NX console what iPhone is to iPad.

 

[RussianSensation]

That being said, if the NX sells decently at launch, third parties will make games for it. Even if means hurting the potential of the XBO/PS4 version.

3rd party developers are not going to be as likely to make games for a console that requires additional work. In this case, doing any other instruction set besides x86 means a lot more work than porting x86 games from PC/PS4/XB1. Why in the world would Nintendo want to do that? What advantage would the ARM SoC provide over x86 CPU? I guess to play NX handheld games on the NX home console?

MS btw is pretty close to giving up on the XBone, so it seems pretty unlikely you will see another console from them.

Source?

"Xbox One is the most important gaming product at MS. Team Xbox is 100% committed to making it our best console generation ever.
12:46 AM - 25 Oct 2015

Asked whether there will be another generation for Microsoft’s consoles or not, Spencer’s answer was definitely positive: "@captaingamepad I fully expect there will be."
http://www.dualshockers.com/2015/10...talks-about-new-xbox-one-experience-and-more/

Xbox One is crushing Xbox 360 sales. Just because PS4 is outselling XB1 doesn't mean XB1 is not the most successful Xbox console ever up to now.

I'll bet anyone on this forum that there will be a successor to the Xbox One. The winner gets a brand new $700 flagship generation GPU in the year the successor of Xbox One comes out.

You know that AMD can make ARM chips too, right? They had a 20nm A57 + GCN tablet chip designed and ready as part of Project Skybridge, which got cancelled due to GloFo's failure to deliver (again). That means that the hard work of getting ARM IP and AMD IP to play together nicely is done.

There's no reason why they couldn't get a PS4 level APU, but with the Jaguar cores swapped out for A57/72.

1) So then you expect this PS4 level APU with ARM CPU cores to be made on 14nm/16nm node then?

Nintendo is already sending out NX development kits. So how do you expect that NX would have 14nm/16nm APUs with ARM cores when no such products even exist as of right now?

2) What advantage would Nintendo gain by going with ARM CPU cores + GCN over X86 CPU cores + GCN?

If ARM CPU cores were superior to x86 CPU cores, why didn't MS/Sony utilize them? You state that AMD couldn't get ARM APU on 28nm nor on 20nm, so how are they going to do it for the NX?

3) Did you guys know that existing consoles already use ARM as secondary/supporting CPUs for background tasks?

Wii U:
"The Latte chip also includes a secondary custom ARM9 processor with 96 KB of SRAM memory that handles system tasks in the background during gameplay or while the system is in sleep mode, and dedicated hardware audio DSP module.[76]"

PS4:
"PS4 includes a secondary ARM processor (with separate 256 MiB of RAM) to assist with background functions and OS features.[40]"

But why is it that the ARM is a secondary not the primary processor? It's because it's not good enough to be the primary, that's why.

Lack of GPU power has nothing to do with the CPU design. Nintendo could ask AMD or NVidia to build a 300mm ARM APU with more GPU power than the PS4. Depending on the core design a ARM CPU will not be slower than the PS4/Xbox One CPU.

That's not the point I am making at all. Read my responses above in this post. You state that an ARM CPU design would not be slower than an x86 CPU design based on what basis? When was the last time AMD has successfully manufactured a good ARM CPU core? When does AMD plan to introduce its ARM cores/APUs? 2017, not 2016:

"dvanced Micro Devices on Wednesday said that it will delay high-volume shipments of microprocessors powered by its custom ARMv8-compatible “K12” cores to 2017. The company did not reveal any reasons behind the postponement.

AMD’s “K12” core is the company’s first 64-bit ARMv8-A-compatible microprocessor engine designed entirely in-house. Development of the core is led by Jim Keller, who also heads development of “Zen” micro-architecture.

http://www.kitguru.net/components/c...introduction-of-k12-based-processors-to-2017/

x86 doesn't lead to seamless cross-platform ports. The 3 consoles still use different graphics APIs. You have to build a seperated render path for all of them no matter if x86, ARM or PowerPC. If cross platform ports would be easy on x86 we would see more Linux games.

Nintendo choosing AMD for the second time goes beyond AMD’s traditional strong points with APUs. It allows all the developers that worked on games for the PC, PS4 and XBOX ONE to carry over their experiences to the Nintendo NX home console (not NX handheld) without having to go through another the hurdle of working on a different architecture like IBM’s PowerPC/ARM.

Therefore, I disagree with your point. X86 leads to the most seamless cross-platform ports compared to any other solution Nintendo can choose. The only way for your statement to be true is if cross-platform ports of x86 PC/PS4/XB1 games would not be any harder/more costly to perform than to perform than same task onto an x86 AMD-powered AMD APU. There is no evidence to support this claim. In fact, the opposite is true as we've seen with Ouya and NV's Shield / TV that x86 game ports to ARM isn't as favourable.

And I don't see why Nintendo need cross-platform ports. There are already 40 million consoles out there which can play the latest Assassin's Creed, Battlefield, Fifa etc. Nintendo will not get people to buy their console with cross-platform ports.

Nintendo will not get the mass market to buy their console without cross-platform ports either. But here is a better question, why design a console from the get go that makes things more complicated? What does the NX stand to gain to have an ARM Soc or ARM + GCN over an x86 APU + ARM Secondary processor - the exact strategy utilized by PS4 with its integration with the PS Vita?

They need a strong exclusive line up and Nintendo's relations to third parties are good enough for this. Most major publishers are willing to develop games for Nintendo consoles if they see the opportunity for good sales figures

Good sales figures do not come from thin air. Do you think when Nintendo was making "lessons learned" documentation on the Wii U and they reached out to 3rd parties, they didn't ask them what Nintendo did wrong with the Wii U and what 3rd parties would like to see from the NX to fix things?

So let's see these developers are making PC/XB1/PS4 games. Why would they recommend Nintendo to make an ARM SoC powered NX home console when all 3 primary platforms they make games for are x86? Look at the abysmal level of PC-cross-platform software support available for Android gaming (ARM) and NV's Shield TV? That should already tell you that porting x86 games to ARM SoCs is not as seamless as a situation if the NX had an x86 APU similar to the ones found in XB1/PS4.

Another way to look at it is name 1 successful home console that's powered by an ARM SoC? There is no such console. NV's Shield TV has failed to capture console gamers despite having cutting edge ARM SoC and an extremely attractive price.

One more point, an x86 APU would have more horsepower to provide emulated backwards compatibility to older Wii U games which can be achieved by a significant performance jump that the new x86 APU is bound to bring. MS has already shown that it's possible to achieve this goal with Xbox 360 -> XB1 emulation. Can you perform this task with an ARM SoC?

You're assuming that Nintendo cares about third-party ports at this point. That ship has sailed. Being able to share games between the console and handheld is more important for them. Thinkb of the NX handheld as the iPhone and the NX console as the iPad. That's the kind of set-up they're going for. It needs to be easy to port games between them, so the need to have the same hardware architecture and OS. I know that you think that the NX is just a console, but it's not. It's a platform encompassing a handheld unit and a home unit (though the units are separate and it's not just a handheld with a dock for home use like some people here think).

I actually don't. I think there will be distinct NX devices, such as the home console and the NX portable/handheld. But let's look at what you are assuming:

- You are assuming that Nintendo has completely given up on 3rd party support. In that case, there is even more reason for them to not have abandoned the Wii U so quickly
- You are assuming a primary x86 APU with a secondary ARM supporting SoC is not a better / viable option
- You are assuming the NX home console may require/is better off with an ARM SoC vs. an x86 APU when PS4 integrated easily with PS Vita. So why would we assume that it's better to have an ARM SoC as the primary backbone of the home console over an x86 APU that's been proven to be extremely successful for XB1/PS4?

I am willing to entertain the idea of an ARM CPU + GCN GPU but I won't accept that the NX will be solely powered by an ARM SoC. There is no ARM SoC powerful enough to have any shot of even matching the XB1 in 2016, and for sure not any such product from AMD. However, it would be more even more effective to go just go X86 APU + secondary ARM (aka PS4's approach).

What I said is true. If you believe otherwise you haven't been paying attention. Again, I'll use the analogy: NX handheld is to NX console what iPhone is to iPad.

This is where we have the disagreement. Your analogy suggests the NX handheld and NX home console are both portable devices then (iphone and iPad are, which is the only way your analogy would work). I think the NX handheld and NX home console are going to be more like PS Vita + PS4, respectively. In other words, the NX handheld + NX home will form the NX eco-system. The question is what's better, ARM CPU cores + GCN or x86 APU + ARM cores as secondary (the PS4 approach)?

@RS I wouldn't count arm out. There is plenty of knowledge and experience with the isa -ask exophase and he could expound on such a topic. Not to mention that arm currently has cores with higher ipc than anything amd has released so far. I wouldn't be surprised if they didn't have 16nm ff macros for chip designer to just run with. Also you under estimate what a company like imagination could pull together in terms of a gpu. Let's not forget nvidia, I don't know if they have a semicustom (or what ever parlance) division but they have the IP and talent to pull off an arm based console chip.

Fair enough. The thing is when was the last time AMD designed a good ARM core/SoC? Never. Based on what I provided earlier, K12 isn't even slated until 2017. So what ARM CPU core is AMD going to provide Nintendo with? Off-the-shelf cookie-cutter version?

Look at what we had in ARM SoC land in 2013:

Samsung S4 Option 1:
Qualcomm MSM8974 Snapdragon 800, Quad-core 2.3 GHz Krait 400

Samsung S4 Option 2:
Exynos 5 Octa 5410, Quad-core 1.6 GHz Cortex-A15 & quad-core 1.2 GHz Cortex-A7

Why didn't MS/Sony just ask AMD to use off-the-shelf ARM designs for the CPU component and add GCN for the graphics back in 2013?

Look at this:

E8870MXM and E8870PCIe
12 Compute Units; 1.5 TFLOPS single precision (Peak)
4GB GDDR5 Memory; 128-bit wide
< 75W Thermal Design Power
Dual HD decode of H.264, VC-1, MPEG-4 and MPEG-2
AMD Eyefinity technology for up to 6 display outputs
Support for DirectX 12, OpenGL 4.5, and OpenCL 2.0
http://www.techpowerup.com/216403/amd-expands-embedded-graphics-lineup.html

Nintendo can up the memory from 4GB to 8GB GDDR5 and play around with clock speeds to drop the TDP closer to 60W and still beat XB1's APU. They can do all of that without requiring a complex combination of a very custom ARM CPU + GCN APU design.

As far as the GPU component goes, there is nothing in ARM land that is even on the map as a substitute for GCN.

MS's surface Pro 3 wipes the floor with iPhone 6S's graphics performance and the Surface Pro 3 would get destroyed by an HD7790.
http://images.anandtech.com/graphs/graph9662/77654.png

Therefore, for the NX home console, PowerVR, Imagination, etc. none of those are even options. NV graphics wouldn't be considered for the NX since they'll never hit the necessary price/performance which means that option is out automatically.

 

[jpiniero]

Also it just isn't coding language and shader language differences. It's memory access, shared resources in an APU design like on the Xbone/PS4.

It's definitely a one-chip design, so those limitations shouldn't be a factor. As mentioned, AMD's already done the hard work on this on failed projects. If Nintendo wanted to (or really wanted to pay TSMC $100+ a chip) they could easily build an ARM+GCN console which was much faster than the PS4.

Source?

IIRC, MS recently said they were going to stop reporting console sales. You don't do that unless things have gotten bad. They'll milk the XBone as much as they can and then get out.

2) What advantage would Nintendo gain by going with ARM CPU cores + GCN over X86 CPU cores + GCN?

I already explained that in my first post in this thread.

If ARM CPU cores were superior to x86 CPU cores, why didn't MS/Sony utilize them?

No 64-bit designs were available at the time. Although if Nintendo is using A15, that's not 64-bit either...

 

[RussianSensation]

It's definitely a one-chip design, so those limitations shouldn't be a factor. As mentioned, AMD's already done the hard work on this on failed projects. If Nintendo wanted to (or really wanted to pay TSMC $100+ a chip) they could easily build an ARM+GCN console which was much faster than the PS4.

Alternatively, it should be possible to build an even faster X86+GCN APU. Considering the ARM gaming eco-system is utter garbage (Ouya, NV Shield, any Android device), and there is no proof whatsoever that ARM CPU cores would be faster/better than an X86 CPU design. Also, since AMD got the NX design win, that means for your statement to be true, when you say "they could easily build", you are referring to AMD. Follow this logic, you just made a claim that AMD could easily design an ARM CPU design that beats their best X86 CPU designs. Do you now how insane that sounds? If AMD could have easily built an ARM CPU design (your claim above) that would beat their X86 designs in performance or perf/watt, why haven't they ever made such a product? Because they can't, that's why. If AMD could have designed an ARM CPU core that was better than their Jaguar, Excavator cores, we would have seen it already.

IIRC, MS recently said they were going to stop reporting console sales. You don't do that unless things have gotten bad. They'll milk the XBone as much as they can and then get out.

You are just making these assumptions. If any company whether it's Intel or AMD or NV start reporting financial performance of various functioning groups differently, it does not mean they will discontinue/abandon such products. Also, it doesn't even make sense what you are saying. If you suggest MS will just "milk" Xbox One, you are implying they are going to make as much $ on Xbox One as possible and then get out. If they are making $ on Xbox One (i.e., milking it), why wouldn't they want to make $ again with XBox Two? Your argument isn't logical.

I don't know if you want the Xbox to disappear or if you are a Playstation user, but MS's long-term Xbox strategy isn't at all aligned with your statements.

I already explained that in my first post in this thread.

All I see if you are stating 4xA72 cores + 640 GCN without any reasons given as to why.

This APU you proposed is worse than XB1's APU and requires custom design, which makes it even more costly than an already designed off-the-shelf XB1 APU.

A72 is not some kind of an X86 CPU killer. Its integer and floating point performance is barely an improvement over A57 cores.

You guys keep ignoring this point and I'll re-iterate it again:

But with the 2012-era Nintendo Wii U, IBM provides a PowerPC processor and AMD provides the graphics chip. With Nintendo, AMD would have to come up with an APU that handled both the CPU and GPU functions and be able to handle the PowerPC processing as well (in order to run older Nintendo games).

To create a backward-compatible console for Nintendo with the Wii / U, AMD would have to enable a way for the new console to use software translation in order to run older Wii and Wii U games. Since a new APU from AMD would be much more powerful, that software translation seems like it would be easier to do with modern chip technology (i.e., X86). Backward compatibility is thus a hurdle for AMD to overcome with Nintendo, but it is would be far harder to do with an ARM A72 cores than a well-known X86 core AMD designed itself. AMD is not an expert in ARM CPU design but they know their X86 CPU designs inside and out.

We have also seen for 100% fact that it's possible to emulate PowerPC games on an x86 AMD APU (Xbox 360->XB1). Have you seen an ARM SoC that's powerful enough to emulate PowerPC games?

Secondly, you still have not explained why X86 APU GCN + ARM (PS4 approach) is not a better solution than ARM + GCN.

Finally, there is another huge elephant in the room --- Memory Bandwidth. The A9X is only estimated to have 50GB/sec of memory bandwidth. That's the fastest ARM SOC in the world as of right now but even the XB1 trumps this easily with 68GB/sec. Therefore, trying to design an ARM CPU + GCN APU and feeding it the necessary bandwidth to not bottleneck the GCN component could become an additional challenge that makes the design even more complex. The integrated memory controller inside the ARM core may not just easily accept whatever memory standard you pair it with just like you can't just add GDDR5 8000mhz chips to Fermi or R9 290 graphics card and assume it'll just work seamlessly.

And the more GCN cores Nintendo wants to add, the worse this memory bandwidth bottleneck becomes. This limitation doesn't exist with AMD's X86 CPU designs (PS4 is 176GB/sec). Let's assume ARM 72 has 128-bit quad-channel memory controller and uses LPDDR4 3200mhz memory. We still only have 51.2GB/sec total memory bandwidth to the GCN GPU -- cripple fest. In other words, yet another underpowered Nintendo console in 2016, almost 3 years behind PS4. Sounds like another Nintendo fail if true. Going with an X86 APU solves this memory bandwidth problem; and once again this design is already proven in PS4 so the risk of a successful implementation is lower.

 

[Snafuh]

ARM APU

The difference between ARM and x86 for porting is minimal. As long as it's not an exotic design like PS3's cell processor the architecture doesn't make much of a difference.

Modern ARM cores can definitely compete with the PS4 and Xbox One CPU.

The Nintendo 3DS, a ARM device, sold 53 million units. Capcom, Ubisoft, Square Enix, Sega, Warner Bros, Bandai Namco, Activision, all major publisher release games on this console. ARM is not a show stopper.
The mass market bought the (3)DS because of the good exclusive software.
The Wii U failed because of the bad start line up (still no Zelda, Mario Kart and Super Smash Bros was late) and the bad marketing. Many people still think the Wii U is an extension for the Wii and not a new console.
That's why Nintendo wants to unify the mobile and stationary environment to overcome software shortage.

I am not saying x86 is not an option but ARM is a very good option, too. It wasn't an option for the PS4/Xbox one because 64bit ARM was not ready yet.

As far as the GPU component goes, there is nothing in ARM land that is even on the map as a substitute for GCN.

GCN is in ARM land.

We have also seen for 100% fact that it's possible to emulate PowerPC games on an x86 AMD APU (Xbox 360->XB1). Have you seen an ARM SoC that's powerful enough to emulate PowerPC games?

Secondly, you still have not explained why X86 APU GCN + ARM (PS4 approach) is not a better solution than ARM + GCN.

The dolphin emulator is available on Android. Powerful phones can play many Gamecube games.
ARM + GCN is a good approach because they might be better suited for a <5W environment (handheld). I think Nintendo wants to go for one scalable design. For the handheld they could just remove GPU shaders and reduce the clockspeed. Many games could be ported to the handheld just by reducing the resolution, framerate or settings like the maximal player count.
BTW the ARM CPU in the PS4 is only used for booting if a new update is available.

 

[jpiniero]

All I see if you are stating 4xA72 cores + 640 GCN without any reasons given as to why.

It's just an guess. We'll have to see what they end up using.

This APU you proposed is worse than XB1's APU and requires custom design, which makes it even more costly than an already designed off-the-shelf XB1 APU.

As mentioned, AMD's already done most of the work on this on other products. The design costs wouldn't be that bad (esp since it's on 28 nm); the manufacturing costs will be substantially lower since it'd be a smaller chip due to less GCN cores.

Finally, there is another huge elephant in the room --- Memory Bandwidth. The A9X is only estimated to have 50GB/sec of memory bandwidth.

The XBone is using simple DDR3, you know. It does have the ESRAM to add some more bandwidth though. Given that the target resolution of this is probably going to be 720p, it's not such a deal breaker.

 

[monstercameron]

@RS read this, http://wccftech.com/amd-8-core-arm-cpu/ will add more color.

 

[Silverforce11]

@RS read this, http://wccftech.com/amd-8-core-arm-cpu/ will add more color.

Designed for servers.

This would be the el-cheapo NX variant if Nintendo doesn't care about grunt.

https://www.youtube.com/watch?v=v6Ry9ct7ymY

Not sure why some of you consider the latest ARM cores to be cheap, when they are reserved & often only used in top of the line mobile devices that costs twice as much as a typical console.

The PS4/Xbone was released not long ago, as some of you may be aware, it was designed in 2009.

If the NX is releasing in 2016, assuming Nintendo is faster at hardware design (for whatever reason), they would have designed it in 2012/2013. Also, dev kits being sent to studios would imply production early this year.

For the time-frame we're looking at in its inception stage, AMD's x86/GCN SOC is proven tech for the console industry. If Nintendo wanted el-cheapo weak hardware, AMD can scale down, or if they want more powerful than Xbone/PS4, simply launching later would assure that given improvements in uarch on x86/GCN since those former SOC went into production.

The DS is still very popular, I don't see the NX replacing that for Nintendo's hand-held approach because its design is well loved by hand-held gamers. The "Wii U's gimmick controller ipad" thing if they go with that again for the NX is just too clunky to replace DS or DS-next-gen.

 

[Zstream]

If the Nintendo system can't get basic ports of madden, Fifa, cod, bf then expect it to flop. These are blockbuster games and not having them on the system has tarnished Nintendo more than anything.

 

[Techhog]

If the Nintendo system can't get basic ports of madden, Fifa, cod, bf then expect it to flop. These are blockbuster games and not having them on the system has tarnished Nintendo more than anything.

It won't get those ports no matter how powerful it is. Wii U can already handle Madden and FIFA, but EA is pissed at them and that won't change anytime soon (which also prevents them from getting any DICE games). All other games are not ported more due to sales than power. Nintendo could put a 5960X and Titan X in there and it still wouldn't get any ports. That ship has sailed, so it makes more sense for them to make a weak, super cheap console that they can easily and cheaply develop for and support with their own games since nothing else sells for them.

 

[Snafuh]

It won't get those ports no matter how powerful it is. Wii U can already handle Madden and FIFA, but EA is pissed at them and that won't change anytime soon (which also prevents them from getting any DICE games). All other games are not ported more due to sales than power. Nintendo could put a 5960X and Titan X in there and it still wouldn't get any ports. That ship has sailed, so it makes more sense for them to make a weak, super cheap console that they can easily and cheaply develop for and support with their own games since nothing else sells for them.

Stop spreading this nonsenseor provide any sources for your statements.
EA also abandoned the PS Vita although there are not "pissed" at Sony. EA stopped releasing games for Nintendo platforms because the games didn't sell well. That's the only reason. Fifa 15 was even release for the Wii but not the Wii U.
The lack of games for the Wii U is not an indicator for bad relations between Nintendo and publishers.

 

[readers]

Lol, I actually have played more and more "kids" games the older I get.

Other way around for me, I couldn't stand tale of abyss when it came out because the char and voice acting are too kiddie to me, never thought it would happen but it did.

 

[Techhog]

Stop spreading this nonsenseor provide any sources for your statements.
EA also abandoned the PS Vita although there are not "pissed" at Sony. EA stopped releasing games for Nintendo platforms because the games didn't sell well. That's the only reason. Fifa 15 was even release for the Wii but not the Wii U.
The lack of games for the Wii U is not an indicator for bad relations between Nintendo and publishers.

You seem to have forgotten the whole "unprecedented partnership" thing from before the Wii U's release. According to rumors, EA was pissed off because they couldn't agree to get Nintendo to use Origin for online services.

 

[Snafuh]

The whole unprecedented partnership thing was post launch. EA was disappointed with sales numbers and the bad integration of the eshop (for DLCs).
Neither Microsoft nor Sony use Origin for online services.

 

[NTMBK]

1) So then you expect this PS4 level APU with ARM CPU cores to be made on 14nm/16nm node then?

Nintendo is already sending out NX development kits. So how do you expect that NX would have 14nm/16nm APUs with ARM cores when no such products even exist as of right now?

Not necessarily, no. A72 cores can also be manufactured on a 28nm process: https://benchlife.info/samsung-exynos-7650-7880-will-take-28nm-to-fight-snapdragon-10192015/

And "dev kits" doesn't necessarily mean "final hardware". The first 360 dev kit was a PowerMac G5, despite the fact that it used a completely different CPU from the final console.

2) What advantage would Nintendo gain by going with ARM CPU cores + GCN over X86 CPU cores + GCN?

If ARM CPU cores were superior to x86 CPU cores, why didn't MS/Sony utilize them? You state that AMD couldn't get ARM APU on 28nm nor on 20nm, so how are they going to do it for the NX?

There's no reason AMD couldn't get one on 28nm, or on a TSMC 20nm or 14nm process, or a Samsung 14nm, or etc. They designed for GloFo 20nm due to the WSA, and then GloFo failed to deliver on their promises and the design was canned. Same thing happened to the 28nm Bobcat shrink that was meant to ship on GloFo 28nm- didn't stop a 28nm Jaguar.

As for why the PS4 didn't use ARM- as Mark Cerny himself has said, it is because at that point ARM did not have 64-bit. That no longer applies, as several 64-bit ARM designs are available.

3) Did you guys know that existing consoles already use ARM as secondary/supporting CPUs for background tasks?

...

But why is it that the ARM is a secondary not the primary processor? It's because it's not good enough to be the primary, that's why.

Those are completely different ARM CPU designs, tiny little things designed for low cost and minimal power consumption. That is like saying "look at how rubbish the performance of this 486 is, x86 is not good enough".

An A72 core has higher performance than an AMD Jaguar core, while also having 100% compatibility with the much lower power A53 core which could be used in a handheld. That is why I think they will go ARM- because AMD don't have an x86 CPU that scales down well to handheld gaming device power consumption (especially at 28nm), while they can use an off-the-shelf ARM design which gives them precisely what they need.

I'm not saying this is a certainty of course! There's all sorts of other possibilities- ARM in handheld and x86 in console, as you said, or they could go with NVidia and get a Denver CPU for the console. Nintendo could even ignore ARM entirely and get a MIPS design with PowerVR graphics. It wouldn't be the first time they changed their mind at the last minute- the 3DS was originally meant to use an Nvidia Tegra 2, but they switched to a different vendor partway through development.

EDIT:

Another way to look at it is name 1 successful home console that's powered by an ARM SoC? There is no such console. NV's Shield TV has failed to capture console gamers despite having cutting edge ARM SoC and an extremely attractive price.

Back in 2010, try naming a successful home console powered by an AMD APU. There weren't any! Every single console was on PowerPC at that point. But technology changes over time, and the designs adapt.

As for the NVidia Shield TV, the explanation is simple; NVidia don't have the games they need. They have a handful of severely belated ports (like Half Life 2, a 10 year old game), and lots of them like Borderlands: Presequel still aren't out. Where are the console-selling exclusives? The Uncharted, the Halo, the Zelda? It's got even fewer cross-platform ports than the Wii U.

And it's not competitively priced- it's far slower than an XBox One, while only costing ~£50 less for the 500GB model which actually has enough storage to install games. A much better option is to buy an XBox 360, which is far cheaper, roughly as fast, has a massive games library, fantastic online multiplayer service, and all the video streaming apps you could need.

 

[MisterLilBig]

This should be a non-starter because an x86 APU from AMD would beat any PowerPC alternative in price/performance, perf/watt and ease of coding. The fact that Sony/MS ditched PowerPC already means all of this analysis was performed by them, which means Nintendo doesn't even need to entertain this idea.

Nintendo is Nintendo.

1) Not having x86 architecture automatically means developers can't do easy PC/PS4/XB1 ports.

By that logic, ARM is the most important architecture for Nintendo and overall.

2) Even the best ARM graphics are still well behind compared to even XB1's GPU.

Graphics in general is about scaling. Mali and PowerVR can scale beyond products that consumers get.

--> Wii U was too underpowered and cost too much at the same time.

And the controller costed them almost as much as the whole console. Being underpowered was not Nintendos failure.

3) You are ignoring the possibility of an option of having an x86 APU + ARM SOC.

I only pointed out the main idea and/or important part of it. What a developer can utilize to do game development in.

4) Given how well XB1/PS4 are selling, and many reasons of why MS/Sony switched to x86 with this generation, it stands to reason that the most logical outcome of this is that XB2/PS5 will also have x86 APUs.

They can and it is fine for them. But, like Mark Cerny has always mentioned...ARM did not have 64bit at the time.

5) We do not need to assume that if the New 3DS successor has an ARM SoC that a home console NX system also must have a powerful ARM SoC. This may or may not be a required condition and certainly it wasn't a requirement for integration of PS4 + PS Vita.

Keep It Simple, Superman. =)

We don't even need to look far. NV Shield TV has a very cutting edge ARM SoC, it's a spec power house for HTPC and yet that thing barely sells for $199 with a controller. It exhibits ALL of the signs of why strictly using an ARM SoC is the most risky strategy for home consoles:

--> Under-powered => crappy graphics. This alienates mainstream gamers who use XB1/PS4 as a benchmark for comparison.
--> Extremely poor 3rd party support because it's not using x86 instruction set. Not many want to spend $200 on a console that has limited game support from 3rd parties when the competitor is selling their console for $299 and it has 3rd party support in spades.

It's Nvidia. If Nintendo had released that, I would have bought it.

That means Nintendo's key to success has to be differentiation and/or making the console at least reach parity with XB1 and make 3rd party ports as seamless as possible.

Nintendo already has that key to success. And already is supporting Unity and Web Tech Game Development. Everything Nintendo can do to improve has nothing to do with CPU Architecture, but if they want to make some developers life a bit easier, sticking to what they already use, ARM or PowerPC, is a slight benefit.

ARM SoC/PowerPC do not solve the biggest issues of the Wii U: (1) lack of CPU/GPU horsepower, (2) ability for PC/PS4/XB1 developers to do seamless cross-platform ports. ARM SoC - too underpowered which means NX will get steam-rolled by 2019 by PS5/XB2. It also guarantees that the home console NX would launch as an outdated console on day 1. PowerPC is an inferior solution in price/performance, perf/watt, ease of coding and has questionable future road-map which suggests ensuring backwards comparability with the NX2 in the next decade would be next to impossible.

This is like a repeat of predictions for PS4/XB1 where people were predicting ARM SoC or PowerPC but the most obvious solution was x86 APU. What has changed since 2013? Nothing. X86 APU still crushes ARM SoC and PowerPC in all key metrics that matter and it also provides the most seamless and least risk prone succession for backwards compatibility for next gen XB/PS/NX consoles in the next decade, etc.

You need to get your facts straight. The only thing x86 had over ARM was 64-Bit support. Now it has nothing, ARM is completely superior. More games are made for, tested and proven to run on ARM than x86. Using ARM is a good business case. Which matters.

wouldn't nvidias Denver cores be a better fit for the consoles due to -among other things- its in-order design?

In that case, "better" means to optimize to, but slow to start off. Watching Mark Cerny's (an actual console engineer, PS4)talk would help in clearing the majority of the nonsense in the thread.

What advantage would the ARM SoC provide over x86 CPU?

Proven consumer hardware, "tested" billions of times per day on everything that a software company would care about.

So how do you expect that NX would have 14nm/16nm APUs with ARM cores when no such products even exist as of right now?

They can test on a bigger model and have the shrink for the actual consumer edition...? Knowledge is Power.

What advantage would Nintendo gain by going with ARM CPU cores + GCN over X86 CPU cores + GCN?
If ARM CPU cores were superior to x86 CPU cores, why didn't MS/Sony utilize them?

They already have a bunch of developers with ARM experience...right? Extremely proven for the consumption of games and entertainment...right?

And as the Architect of the Playstation 4 said, ARM was not ready, it had no 64-Bit support. (It's surprising that you can use Sony as an example and have no clue of why Sony ended up with what they did...very "surprising".)

In fact, the opposite is true as we've seen with Ouya and NV's Shield / TV that x86 game ports to ARM isn't as favourable.

All NV powered. And why the focus on ports? This is Nintendo. They frown upon ports that do not take advantage of the hardware features.

But here is a better question, why design a console from the get go that makes things more complicated?

That is not a better question. Why is it more complicated? What makes it more complicated? But really, that question makes sense towards the Wii U.

Look at the abysmal level of PC-cross-platform software support available for Android gaming (ARM) and NV's Shield TV?

Why do you care about PC to ARM support? And again, those companies are not Nintendo. And Android and iOS are a huge gaming success. Which is the main thing to matter to a software company.

Another way to look at it is name 1 successful home console that's powered by an ARM SoC? There is no such console. NV's Shield TV has failed to capture console gamers despite having cutting edge ARM SoC and an extremely attractive price.

Well, considering the time and place for that was not so long ago...that's like asking 12 years ago to Name one colored US President. And like I said, that is Nvidia. NV is not MS, Sony, Nintendo, Google, Amazon or Apple.

The thing is when was the last time AMD designed a good ARM core/SoC? Never. Based on what I provided earlier, K12 isn't even slated until 2017. So what ARM CPU core is AMD going to provide Nintendo with? Off-the-shelf cookie-cutter version?

First, AMD has ARM SoC. Second, it does not need to be the K12. Third, why are you surprised? It's not like all the ARM cores used in any console or handheld are custom...in fact, is there even one custom cpu core out there on a gaming system?

Why didn't MS/Sony just ask AMD to use off-the-shelf ARM designs for the CPU component and add GCN for the graphics back in 2013?

No 64-bit support...for the 100th time. *mind blows*

Considering the ARM gaming eco-system is utter garbage (Ouya, NV Shield, any Android device), and there is no proof whatsoever that ARM CPU cores would be faster/better than an X86 CPU design.

That's $30 billion worth of "utter garbage" in 2015.

Secondly, you still have not explained why X86 APU GCN + ARM (PS4 approach) is not a better solution than ARM + GCN.

Because it's more expensive. *Ninja Slash*

Finally, there is another huge elephant in the room --- Memory Bandwidth.

Don't remember Mr.Cerny worried about that.

Not sure why some of you consider the latest ARM cores to be cheap, when they are reserved & often only used in top of the line mobile devices that costs twice as much as a typical console.

There are many reasons for that and most if not all will be related to the end product.