Nintendo's hardware choice philosophy *spawn

[ERP]

Anyway, Super mario Galaxy is the perfect example of a bueatiful wii game that was clearly coded in assembly from scratch using every ounce of the wii pipeline. No thirdparty company is going to waste their time doing that. Frankly nintendo doesn't care if thirdparty ports suck.

I have seen no evidence to support the fact that Nintendo has in anyway better developers than anyone else.
And no one writes games entirely in assembly anymore, it's both not worth it from a performance standpoint and not practical from a software engineering standpoint.
First parties do however have the advantage of being able to target a single platform, though it matters a lot less than it used to.

I think Nintendo saw an opening, consider the tablet to be a clear differentiator and figured they could ship something that could get 3rd party support through PS360 ports for a couple of years, hoped that would be enough to reach some sort of tipping point in installed base so 3rd parties couldn't just ignore them when PS4/720 shipped.

 

[owen]

@Exophase It should be possible to build a 10x faster ps2 console which is backwards compatibile. PC hardware has been upgrade the same stuff for YEARS! A upgraded console could have solved the first party exclusives problem.

@ERP Software engineering a broad word when refering to console games. And when I say assembly I really mean "to the metal". And the whole first parties have more advantages is over shadowed by the release of multi-platform games like Call of Duty. Nintendo is doing the same thing it did with the Wii; cheap hardware, optimised games. Somehow people always seem to focus on the obvious gimick when its clear that they don't want to sell hot consoles with high failure rates.

 

[tongue_of_colicab]

I'm no programmer but I doubt SMG is anything special from a technical POV. The art and design is very good, pleasing the eye. But generally there isn't a whole lot of geometry or enemies/moving stuff on screen at once. And as soon as you come close to objects most of the textures look really bad as well.

 

[milk]

I'm no programmer but I doubt SMG is anything special from a technical POV. The art and design is very good, pleasing the eye. But generally there isn't a whole lot of geometry or enemies/moving stuff on screen at once. And as soon as you come close to objects most of the textures look really bad as well.

As the game is not the technical marvel some seem to advertise, rather a extremelly competent pairing of great tech, art, and a style that plays to the system's strenghts.
Yet, some nice tech is in there, specially in the sheer amount of geometry detail present in the game.
On this video for exemple, most of the detail of the candy wraping and the chocolate inside, that seem to employ some kind of bump mapping, is achieved through actual geometry. And I suspect even the flat wrapers used geometry to make the clean vector like graphics printed on them, but I never looked at them from up close enough to see if I could detect aliasing typical polygon intersections.

http://www.youtube.com/watch?v=g49ByfSp7lE

 

[BRiT]

Alright, enough about game discussions. Lets get back to the topic at hand, Nintendo's hardware choice philosophy.

 

[tongue_of_colicab]

Okay. So Nintendo's philosphy changed to building cheaper (and weaker) hardware as they think competing on specs alone isn't getting them anywhere. They have a fair point and so far you can't blame them.

Then there is actual value of hardware and perceived value of the console/handheld.

As for perceived value, Wii and DS offered a lot for your money I think. Sure, they hardware could have been better but at 250 and 150 euro the systems weren't expensive even if the hardware wasn't top notch. Basically for this price you can pick up the system and it will offer good value for money even if you just play Nintendo's own collection of tripple A games you know the system will get.

So no issues there imo. I picked up a wii at launch, at 250 it's just good value. Same for the DS and DS lite I bought. Always thought they offered great value for money.

3DS, probably somewhere in the middle. The 3D screen is probably costing a fair bit and ''wasting'' gpu power on something most people probably don't care about too much if any at all. I always see people playing with 3D turned off. BTW, this is after the initial price drop, at 150 the system is OK I think, both for hardware costs and perceived value. Before it was too expensive for what you got I think.

Wuu OTOH is terrible imo. It's slow just like Wii but unlike Wii it isn't cheap. I don't care what Nintendo says but the wuublet can't be more than 50 euro's to build. You can get tablets that got much more hardware for 80 at retail.

Even if the wuublet was 100 to build, you can buy a A6-5400k (with a igp that is probably better than what the wuu has), a 7750HD 1gb, 4gb ram, a mainboard and a case+psu for 250 euro's at retail. Actual manufacturing costs, especially after consolizing this as most parts on the mainboard and gpu are redundant or not needed in a console, and I can't believe this will cost more than 150 euro's. And this will be way faster than what the wuu has for hardware. Actually, they could make money on this at retail while nintendo now claims they need to sell 1 game with a wuu to make a profit. How the hell they are doing that is beyond me.

So it's bad value for money in the hardware department.

Perceveived value, atleast for me, isn't great either. Again, pricing is too high and so far ports seem to be as good as current gen consoles in the best case, usually worse. I know these are launch games but I doubt devs will be spending a whole lot of time optimzing their current engines to run as best as they can on wuu. Why bother with ps4/x720 coming out? Most probably won't go beyond ''good enough''.

So you are paying more but probably 3rd party games won't really be any better than on a much cheaper system.

The wuublet doesn't really give me a warm and fuzzy feeling either. The wiimote was something really new, offered many new possibilities, it was exciting. Now 3rd party devs didn't bother with it and even Nintendo didn't make that great of use in it, atleast not in their core games.

But the wuublet really feels like a gimmick to me. I'm sure you can do some great things with it, but the question is: Will devs actually do it? Most failed doing something decent with the wiimote, the wuublet might not be any better.

More important, the wuublet isn't really anything new. People are used to touchscreens now. It doesn't really have a wow factor imo.

At 200 ~ 250 the wuu would be like wii, great value for money. Even if the wuublet fails you will be getting those Nintendo games and ~250 isn't a bad deal for just those games.

But current pricing just makes it look like an overpriced, ''slower'' ps360 with a tablet tacked on.

 

[owen]

My only point is that nintendo's hardware philosophy might not be solely focused on affordable/cheap hardware but on unique advantages which the hardware provides. Software optimization versus hardware optimization. It could be that nintendo purposely chooses hardware that its developer team like to use.

 

[Betanumerical]

My only point is that nintendo's hardware philosophy might not be solely focused on affordable/cheap hardware but on unique advantages which the hardware provides. Software optimization versus hardware optimization. It could be that nintendo purposely chooses hardware that its developer team like to use.

Can you give us some examples of the unique advantages that this hardware provides?.

 

[Sonic]

My only point is that nintendo's hardware philosophy might not be solely focused on affordable/cheap hardware but on unique advantages which the hardware provides. Software optimization versus hardware optimization. It could be that nintendo purposely chooses hardware that its developer team like to use.

Unfortunately there aren't many "unique advantages" WiiU will have compared to MS and Sony's next offerings. The only advantage it might actually has is having Nintendo's games on it. Other than that there's little reason to purchase one. It's gimped hardware in the name of being cheap.

 

[owen]

Can you give us some examples of the unique advantages that this hardware provides?.

Without insider knowledge I can only speculate as to why there was an apparent gap in the quality output between first party and third party content in the previous generation. They were obviously doing something special. So I think their hardware philosophy is to only produce hardware that is optimized "enough" to satisfy internal developers with no desire for bleeding edge general purpose use-cases.

 

[Sonic]

Nintendo's overall design philosophy regarding Wuu has left me rather perplexed. I see little to no reason in gimping the hardware on purpose in order to run at half the power envelope of the current machines. That was stupid of Nintendo to take that on as a main design challenge. If they aimed for the same power envelope we may have ended up with an more powerful console and not the Nintendo Wii60. If they went with the traditional type of power draw from a new console then we'd be getting an actual next gen console. They decided not to do that, but to also throw in a tablet controller that siphons performance from the console as well.

The lack of CPU grunt is also something that already seems to be biting them in the rear. Sure it has a primitive/limited form of OOE and gets more work done in less cycles than Cell's PPE or Xenon, and plus the fact that the GPU can do GPGPU and some DSP to offload work done totally makes up for weaksauce that is WiiU's CPU. The CPU is an important part of the machine, and to design other components around its shortcomings is no excuse for having a crappy CPU. Why is hardware backwards compatibility so important? Emulation would have served fine.

And then the tablet controller, which is too limited of a tablet to of general use, and likely to to be a distraction while gaming. They could have launched a tablet on its own and had a new handheld that fits in with the current mobile space and also could have designed a more powerful WiiU that could turn a nice profit at its current price. Instead the tablet is nice for what it is, but is primarily tied to the console and may have extremely limited range given one's living situation. It isn't revolutionary in the same sense that motion controls were and is not as innovative as others make it out to be.

Nintendo's hardware philosophy is one of cheapness and great engineering of said cheapness to go with it. The company believe it can go its own way and will always continue to be able to do so all the while ignoring market dynamics. I hope the machine sells well so the company continue to flourish, but I wouldn't be surprised if in a few years Nintendo is in serious financial trouble. As for giving the company the benefit of the doubt, that went out the window after confirmation that it is lacking severely.

 

[Betanumerical]

Without insider knowledge I can only speculate as to why there was an apparent gap in the quality output between first party and third party content in the previous generation. They were obviously doing something special. So I think their hardware philosophy is to only produce hardware that is optimized "enough" to satisfy internal developers with no desire for bleeding edge general purpose use-cases.

That is not unique hardware, that is being as cheap as they can afford to be.

 

[Squilliam]

I have a question: How much of say the Xbox 360's current CPU workload is unnecessary when taking into consideration having a relatively capable GPU in the system? I'm not talking about GPGPU but things that are done on the CPU to assist the GPU in rendering such as vertex culling and animation which a more modern and capable GPU architecture wouldn't need?

Overall it seems that console design has typically had a relatively weak CPU with a much stronger GPU. Isn't the current generation the odd one out in terms of the CPU/GPU balance and the strong rumours of next generation consoles also point to relatively strong GPUs and relatively weak CPUs so in this context Nintendo's decisions regarding the CPU fit in with the other designs but perhaps the other choices such as the low memory bandwidth do not.

 

[Exophase]

Overall it seems that console design has typically had a relatively weak CPU with a much stronger GPU. Isn't the current generation the odd one out in terms of the CPU/GPU balance and the strong rumours of next generation consoles also point to relatively strong GPUs and relatively weak CPUs so in this context Nintendo's decisions regarding the CPU fit in with the other designs but perhaps the other choices such as the low memory bandwidth do not.

PS3 and XBox 360 had CPUs that were optimized for high peak vector throughput and sucked at everything else. Not as much then as Wii U's CPU sucks today, but still pretty sucky.

This was not at all a new precedent for consoles. Several older consoles used a relatively weak CPU (compared to PCs of the time) augmented with either a strong vector coprocessor(s) or separate autonomous vector units - the only one I can think of that comes close to being an exception is the original XBox. PSP had its VFPU, PS2 had its vector units, Dreamcast its fast dot-product instruction, N64's RSP, Saturn's DSP, and 3DO's math coprocessor. The other variation is where the system didn't so much have traditional programmable vector processing but heavy fixed function geometry operations, or something in between - like the GTE on PS1 and geometry engines in NDS, Gamecube, and Wii (and PSP).

Even before 3D this wasn't unprecedented. SNES helped make up for a weak CPU by adding dedicated external multiply and divide circuits and coprocessors on carts (although as you can see by my earlier posts I still think the system was too compromised)

 

[Squilliam]

PS3 and XBox 360 had CPUs that were optimized for high peak vector throughput and sucked at everything else. Not as much then as Wii U's CPU sucks today, but still pretty sucky.

This was not at all a new precedent for consoles. Several older consoles used a relatively weak CPU (compared to PCs of the time) augmented with either a strong vector coprocessor(s) or separate autonomous vector units - the only one I can think of that comes close to being an exception is the original XBox. PSP had its VFPU, PS2 had its vector units, Dreamcast its fast dot-product instruction, N64's RSP, Saturn's DSP, and 3DO's math coprocessor. The other variation is where the system didn't so much have traditional programmable vector processing but heavy fixed function geometry operations, or something in between - like the GTE on PS1 and geometry engines in NDS, Gamecube, and Wii (and PSP).

Even before 3D this wasn't unprecedented. SNES helped make up for a weak CPU by adding dedicated external multiply and divide circuits and coprocessors on carts (although as you can see by my earlier posts I still think the system was too compromised)

I guess I was mistaken on that part but I still wonder about how the design choices from Sony/Microsoft seem to follow the same dimensions at least in terms of relative CPU performance. Even if they go with 8 jaguar cores we're probably looking at something like a 70/30 Power/Performance split between the GPU and CPU and perhaps even more which does seem to follow Nintendo's own design philosophy. I feel the significant differences in Nintendo's design are actually the overall performance of the system and the level of memory bandwidth and flash storage and not their relatively weak CPU.

 

[Exophase]

I guess I was mistaken on that part but I still wonder about how the design choices from Sony/Microsoft seem to follow the same dimensions at least in terms of relative CPU performance. Even if they go with 8 jaguar cores we're probably looking at something like a 70/30 Power/Performance split between the GPU and CPU and perhaps even more which does seem to follow Nintendo's own design philosophy. I feel the significant differences in Nintendo's design are actually the overall performance of the system and the level of memory bandwidth and flash storage and not their relatively weak CPU.

What do we really know about what MS's and Sony's next consoles will be using? Jaguar cores is certainly not the upper end here.. if the x86 rumors are true then it could at least be Piledriver or Steamroller.

Let's say it's 70/30 afterall. Wii U is considerably more extreme than that, at least in area. Area's probably the best metric for evaluating how much the console maker prioritized one part over the other, especially on something like Wii U which is far from pushing thermal limits in its form factor (despite how much Nintendo talked up power consumption). Here the ratio is over 80/20 at first glance, but actually it's much more than that given that TSMC 40G is quite a bit denser than IBM's 45nm. More than the node names alone would suggest.

Nonetheless, I will grant you that if newer consoles are just following the old precedents then it'd make sense for them to have wimpy CPUs while cutting at least part of the non-GPU vector processing, because a huge amount of that processing was there in place of vertex shading that is done on the GPU now. Since XBox 360 and PS3 already had vertex shading it may be a fair argument that they dedicated more than their equivalent share to non-GPU compute. But I don't think MS and Sony were being totally superfluous with it (well, in Cell's case it is helping a weaker GPU), and I think the problems we're seeing with Wii U attest to that. It looks like game developers did find things that benefit from CPU-style FP SIMD yet aren't great fits for GPU compute, and ports are struggling with that. Maybe Nintendo didn't find those uses because they haven't put in strong CPU-style compute coprocessors ever since they moved to dedicated geometry acceleration (or if you'd prefer, never outside of N64).

 

[Squilliam]

What do we really know about what MS's and Sony's next consoles will be using? Jaguar cores is certainly not the upper end here.. if the x86 rumors are true then it could at least be Piledriver or Steamroller.

Jaguar seems like a good fit for now. It fits with TSMC and their GCN architecture as they both belong with the same foundry and it combines excellent performance per watt/area. Given the fact that the software is tailored to the application then having weak single threaded performance is less of a burden than with the PC.

Let's say it's 70/30 afterall. Wii U is considerably more extreme than that, at least in area. Area's probably the best metric for evaluating how much the console maker prioritized one part over the other, especially on something like Wii U which is far from pushing thermal limits in its form factor (despite how much Nintendo talked up power consumption). Here the ratio is over 80/20 at first glance, but actually it's much more than that given that TSMC 40G is quite a bit denser than IBM's 45nm. More than the node names alone would suggest.

80/20 is probably the better fit than 70/30. Though with respect to the GPU I would be tempted to include some of the GPU area as part of the memory sub system given the inclusion of eDRAM. I would still say that the ratio of 80/20 is about right even discounting the eDRAM.

I guess we cannot forget that the console is designed by Nintendo for Nintendo. They simply do not make any games which require a fast processor so the console doesn't have one.

 

[Entropy]

Even if the wuublet was 100 to build, you can buy a A6-5400k (with a igp that is probably better than what the wuu has), a 7750HD 1gb, 4gb ram, a mainboard and a case+psu for 250 euro's at retail. Actual manufacturing costs, especially after consolizing this as most parts on the mainboard and gpu are redundant or not needed in a console, and I can't believe this will cost more than 150 euro's. And this will be way faster than what the wuu has for hardware. Actually, they could make money on this at retail while nintendo now claims they need to sell 1 game with a wuu to make a profit. How the hell they are doing that is beyond me.

So it's bad value for money in the hardware department.

Isn't that the way of just about all new electronics, entertainment stuff in particular?
I bought my new digital camera half a year back, when it was fresh on the market, and today I would pay $200 less. Should I be upset? If you buy last years TV model, do you pay the same price as it was at introduction?
Of course the WiiU is bad value for money in terms of hardware - released just before Christmas how on earth could anyone expect otherwise? They are obviously skimming the cream, and by appearances, they have managed to balance availability and price quite well.

At 200 ~ 250 the wuu would be like wii, great value for money. Even if the wuublet fails you will be getting those Nintendo games and ~250 isn't a bad deal for just those games.

But current pricing just makes it look like an overpriced, ''slower'' ps360 with a tablet tacked on.

The advantage of Nintendos approach is that they can drop prices as the market requires in order to keep sales at a decent level. This is good for them, obviously, but also for their business partners who know that they are capable of doing a 3DS drop if need be, and it is even good for the early adopters, because they can have some confidence in a reasonably healthy eco system. Nobody wants to be in the Playstation Vita situation. Manufacturers, publishers, early adopters - all share the pain.
The WiiU will get to the levels you describe, the question is only when.

 

[AzaK]

Unfortunately there aren't many "unique advantages" WiiU will have compared to MS and Sony's next offerings. The only advantage it might actually has is having Nintendo's games on it. Other than that there's little reason to purchase one. It's gimped hardware in the name of being cheap.

Well I think Nintendo are relying on exclusives, the Tablet and the non-gaming features of the system to differentiate it from the competition.

1) Exclusives are always an advantage no matter what the other consoles contain.

2) The Tablet is an advantage and I don't think it should be underestimated. It's a viable browser, TV remote and game controller. It's not looking like Sony or MS will copy this. Smartglass is about as close as MS will get and that's not practical for gaming scenarios that require both button/stick control AND screen touching at the same time. It's also incapable at the moment of doing full 3d rendering for off TV play.

3) The non gaming features like Nintendo TVii are pretty unique....at the moment. I imagine both Sony and MS will do something like this and they will do it on a much larger scale than Nintendo. Wii U will control the cable box, but it's looking like Durango will actually connect to it. MS are also getting into TV stuff.

Overall I think Wii U's success rests on the Tablet controller selling people and exclusives.

Nintendo's overall design philosophy regarding Wuu has left me rather perplexed. I see little to no reason in gimping the hardware on purpose in order to run at half the power envelope of the current machines. That was stupid of Nintendo to take that on as a main design challenge. If they aimed for the same power envelope we may have ended up with an more powerful console and not the Nintendo Wii60. If they went with the traditional type of power draw from a new console then we'd be getting an actual next gen console. They decided not to do that, but to also throw in a tablet controller that siphons performance from the console as well.

I agree to some extent. There were two things that I thought Nintendo needed to ensure when deciding on the tech in the Wii U.

The first was that it could receive downports from Durango and Orbis. It's hard to predict what the other makers will do but Nintendo could have been pretty certain it's be a seriously good bump over 360. Nintendo seem to have picked the smallest possible jump over current gen that *MAY* make it possible to handle downports but it's looking shaky when you compare known/leaked/speculated specs. RAM, GPU + APU, Speeds etc just seem like a large jump over Wii U.

Second, from a marketting perspective I think having ports from 360/PS3 that ran at least a bit better even if it was just 720p for all games would have helped. As it stands, in the mind of the enthusiast gamer that Nintendo said they want (I'm starting to doubt their actual sincerity here) they needed to ensure those launch ports were really solid and preferably better.

To both these ends I think they should have invested a little more money in the core tech, passing that onto us if they wanted, or at least taking that loss for the first year until they can get manufacturing costs down.

Overall regarding tech, I'm not impressed with what I'm seeing but I do think we'll see games noticeably better than average current gen, especially as engines are retooled for Wii U/Orbis/Durango more modern feature sets. However, the big concern for me is whether they will get those Orbis/Durango downports. If they can't/don't get that, then we will see Wii again I feel. Possibly successful for Nintendo if they can sell the console to the mainstream but potentially barren for us as Nintendo console gamers. What that means is that we then go and buy another console and buy games for that, which means less money for Nintendo.

Nintendo have a very very important year next year and they need to nail it and secure those ports and install base. Next year could start to go totally pear shaped.

 

[owen]

How important are ports to Nintendo as a company? Ports are good for third parties and people who don't want to buy the lead console. But to nintendo itself it seems less important. Suppose there are no ports at all? it would be like the gamecube days. lol.