Broken Hope
Member
The flash RAM is likely cheap and slow, that's how Nintendo tends to do things. I doubt it would speed things up much.
WiiU "Latte" GPU Die Photo - GPU Feature Set And Power Analysis
- Thread starter Thraktor
- Start date
- Status
The flash RAM is likely cheap and slow, that's how Nintendo tends to do things. I doubt it would speed things up much.
Broken Hope
Member
By the time the next consoles are out we should have stacked RAM being used.
JordanN
Banned
Eh, I would be more concerned if the thing is even as powerful as Xbox One.
Nintendo's home consoles have gotten increasingly conservative not less. For all we know, they could pull a Wii again and overclock Latte and Expresso by 50%.
JordanN
Banned
I'm not sure where you're getting struggling from. Are you sure you're not overestimating the hardware like it has some secret power a la PS2?
Developers may not bother with it if the resources needed are not worth the returns.
Their next console will be a hybrid/portable so we'll be lucky if it's Wii U level. Maybe a little more like 4GB RAM, 4 core, but knowing Nintendo and the fact that they truly seems to believe that diminishing returns has kicked in, it won't be a big jump and it wouldn't surprise me if it's below what Sony and MS do this generation.
Only for Nintendo. This whole concept of >4GB of addressable RAM might catch them off guard like HD development did this time around. I mean, it's a pretty complicated subject.
We are. I see sharper textures in general all around, even in games like Lego City
Full ray tracing is probably impossible until a major breakthrough in electronics. Hybrid systems are already in use and will probably just get better and better.
TheGreyHulk
Member
They can't do this because they're learning themselves. LOL. Seriously, they're having issues with the transition to HD development. Hence all the delays to their games.
No such limit exists, so there should be no presumption of architecture change based on this argument alone.
lightchris
Member
In terms of "graphical effects" you can already do pretty much anything in DX9, though as you stated some things may not be efficient or only with less precision). Essentially the shaders are already freely programable.
Obviously without geometry shaders you can't generate new vertices and without a tesselation unit you can't do tesselation on the GPU. These are examples where DX10/11 adds more flexibility to the graphics pipeline; Diagrams like this might make things a little more clear.
Even then you don't have to do this on the GPU. GT6 does tesselation on Cell.
As for what specific new features DX10.1 introduced I found this: http://www.codermind.com/news/Microsoft-announces-March-2008-DirectX-SDK-DX-10-1-RTM-4100035.html
I think the most obvious instant benefit as a developer would be that you don't have to fight for every MB as much any more, thus making development easier. When you already have those games optimized for the smaller memory pools of PS360 though it might not be that a big thing though. In this case the easiest way to make use of the additional RAM would be to use higher resolution textures. Even if only a little this requires some extra work and apparently it's enough for some devs to just do the simple 1:1 port.. call it lazyness if you like, or just a very restricted budget for the Wii U version.
What do you mean no such limit exists? 32-bit CPUs can only use 32 bits to address memory which amounts to 4 GB.
Using PAE (Physical Address Extension) you can address a 64bit memory space from a 32bit system. Way back when, we used to use to use the PAE version of Linux on 32bit Xeons to give servers an extra boost in memory without needing to migrate services to new 64bit boxes.
I'm not saying it's ideal, but the on-paper limitation of 32bit addressing was overcome a long time ago to the point that it's an irrelevant limitation and is simply now a choice. Therefore this reason alone would not mean having to completely change CPU architecture.
ambientmystic
Member
Perhaps he was referring to Physical Address Extension, which does not apply to the IBM CPU used in the Wii U.
Or does it? PAE is only applicable to Intel and AMD x86 CPU architectures AFAIK.
I think this rumor has been already officially dismissed?
What I expect is that they will use similiar architectures though.
That may be correct, but there comes the point where you have to do the switch (and memory addressing is just one limitation) when PS4/XB1 are 64-bit systems from this year on.
To be honest I never thought of this limitation of Espresso before, just recently because of some posts of a pCARS dev.
I think ARMv8 could be a good choice because AMD will already have experience with it when 3DS/WiiU successors land, but I don't know how important backwards compability will be the next time (if they drop the Gamepad not at all).
Some ARM chips support variants of it as well. Bearing in mind how much Nintendo have customised their CPU line I would still consider this a choice on their part to add support rather than throw out the entire architecture due to wanting more memory.
Certainly don't disagree with that - but after seeing what they were willing to do to the GPU I wouldn't rule out Nintendo's stubbornness on any matter
lightchris
Member
I was under the impression that, besides not being supported on the CPU, PAE comes with a perceptible performance hit. However having searched a bit this doesn't need to be the case.
Still, I don't think it'd be a good idea to much longer stick which such an ancient CPU design. Instead of having to try to add a few new features here and there it'd be more wise to benefit from all the research that has since gone in more recent architectures.
Still, I don't think it'd be a good idea to much longer stick which such an ancient CPU design. Instead of having to try to add a few new features here and there it'd be more wise to benefit from all the research that has since gone in more recent architectures.
freezamite
Banned
WiiU has bigger chips than the GC had at its day, so no, Nintendo's home consoles were conservative during the DS/Wii era, not now.
The same for 3DS. The difference between PSP and NDS was MUCH bigger than the one between 3DS and PSVita, and being PSVita 1 year newer.
This generation of Nintendo consoles is not as underpowered as the past one. FACT.
For curiosity, which are those benefits from all the research you mention? The only real problem this CPU has regarding its architecture is its short pipeline that prevents Nintendo to clock it higher, but this also makes this CPU ultra efficient and that's what Nintendo wanted from their console to begin with.lightchris said:
This really isn't the problem, Jaguar has a much longer pipeline and doesn't clock that much higher either. I also heavily doubt that WiiU successor's CPU will clock higher than ~2GHz.
BTW: We are becoming OT in here, there is a CPU thread too.
lightchris
Member
What do you think why new architectures are even being developed if it had no advantages over the old ones? Why doesn't anyone use old IBM designs in smartphones or the like if it really were a competetive solution in terms of low power efficieny?
This goes deeper than just counting pipeline stages. Look at the evolution Intel's Core CPUs have gone through for example. The performance gains are significant and unrelated to clock speeds.
That seems to be about file system access. Nintendo apparently uses an unusually large block size, and it seems some developers ignored or simply didn't know that, which increased load times. It was also mentioned in the fmod changelog:The other part of the Project Cars stuff I wanted to understand was:
Does that actually divulge anything or is this again already a given based on what we know?
Any clever people have time to explain?
"KB" is obviously a typo, should be just "B". It's almost certainly about the same thing.
Mr. Garton was asking for not posting any of these logs outside of WMD forums, may as well ban users not keeping it that way. JFYI.
BTT: TellTales should have listened to Nintendo, Lego Cities' load times come straight out of 8 byte block size hell
I think you meant to quote The_Lump?
freezamite
Banned
Precisely speaking about that, I see on the WiiU's CPU really recent technologies like huge L2 eDram caches, multi-core instead of single core, increased registers... those are for sure changes that go beyond a clock increase, and there we have them.
I'm not by any means an expert regarding CPUs architectures, so still for curiosity, which are those innovations impossible to implement on the current architecture and that, for example, are found on the Jaguar cores the PS4 has?
Thank you!
Think of it this way: The (computing/graphic) processing unit works at full speed calculating the workload it has readily at hand. Once it's done it needs the next job to do. Ideally the memory offering that data is at the same speed as the processing unit so that the latter does not need to wait. Memory with that kind of latency is very expensive though so it's very small. So in reality there are many cache levels between the processing unit and the originating memory holding all the data (i.e. the DVD/BR disc or the hard drive/SSD/flash). With each cache level there is a need to balance the size and the latency. The bigger the size the less the developer needs to worry about managing the data to make the cache work well at all time (whenever data can't be found on a particular cache level it need to step one cache level lower which comes with the respective time penalty), but the lower the latency the less the processing unit is idling doing nothing after completing stuff.
More like we shouldn't expect ray tracing in games ever as long as power is limited and rasterization offers the same visual results at a much lower power cost. Keep in mind that games are real time while CG movies could theoretically opt for as much time and power as necessary, but they still go for the rasterization shortcut.
Same issue like with GC and Wii really. And Wii showed that market share didn't push developers to put in the extra time either. Nintendo needs to offer ready made optimized tools for developers to reuse in Wii U games, which is kind of hard when Nintendo itself is still in the progress of catching up.
KB make more sense for file io/device block sizes. You normally see block sizes like 4K on Windows for instance. 8 or even 64 bytes would be utter madness. A quick look at a 2010 Blu-Ray paper here shows that the block size is 64K.
For a developer to use something else, and load times to be extremely slow I would suggest one of three things happened.
1) Nintendo told them some completely silly number in the docs by mistake. ie. 'Block sizes are 32 bytes'. This would cause the dev to use that for their buffer sizes when doing file I/O, and then having to call the system API's thousands of times to get the data for one block from the lower level read cache. Unlikely, as any dev with 1/2 a brain would go WTF??
2) Nintendo has a file-system or drive with no/little actual read caching. Therefore when you read from it you better damned well read the block size, or else you'll be waiting for the disc to spin around again to read from the block is just read from again. Couple this with an expensive user->kernel mode transition for instance and you could have even worse problems.
3) TT put their data for their files randomly scattered all over the disc which requires seeking around like a mother-fucker in order to read it.
#2 is probably more likely IMO if we attribute the TT developers with some level of competence.
Agreed. A block size measured in bytes on a filesystem that supports terabyte+ partitions is unthinkable (unless they've gone for a really crazy FS design). A 64K block size makes so much sense it shouldn't make anyone blink.
Has anyone confirmed any details about the WiiU filesystem?
regarding the next generation for Nintendo, I'm curious what they'll do with their current policy of supporting the previous generation for backwards compatibility.
With the WiiU, next generation would have to support the gamepad, and thus still have all that communication built in. I guess it wouldn't come with a gamepad, just for backwards compatibility, but since they don't sell them separately (yet), does that mean there will be a lot of used WiiU's next gen without gamepads?
Regarding specs, I think they'll have to go 64bit, probably stick to the same processor family (power), hopefully they're already working on a variant of Power 7 or whatnot, that they can use for the next 3 generations. Probably a reduced cache, and maybe modified cores to make the die size as small as they can go to conserve cost.
I wouldn't expect 10nm or anything close to that, Nintendo has always been very behind the times with die shrinks, GC, Wii, WiiU, they tend to choose very mature processes. Probalby be something like 22nm or thereabouts
I'd hope for 16 GB's of ram, but they'd probably do 8GB's. Probably be somewhere in line with the PS4/XBone for graphics. I'd hope for stacked, Nintendo seems to be very memory centric in their thinking, GC's memory type and caches were ahead of it's time, the Wii Continued that, the Wii U, I guess continued it, but seems pretty conventional. Maybe Nintendo will work with some obscure tech company for some neato technology.
BluRay again, no video playback, hopefully double layer.
Next Gen Portable probably highly integrated, maybe that's how they'll support WiiU Backwards compatibilty with regards to the gamepad.
With the WiiU, next generation would have to support the gamepad, and thus still have all that communication built in. I guess it wouldn't come with a gamepad, just for backwards compatibility, but since they don't sell them separately (yet), does that mean there will be a lot of used WiiU's next gen without gamepads?
Regarding specs, I think they'll have to go 64bit, probably stick to the same processor family (power), hopefully they're already working on a variant of Power 7 or whatnot, that they can use for the next 3 generations. Probably a reduced cache, and maybe modified cores to make the die size as small as they can go to conserve cost.
I wouldn't expect 10nm or anything close to that, Nintendo has always been very behind the times with die shrinks, GC, Wii, WiiU, they tend to choose very mature processes. Probalby be something like 22nm or thereabouts
I'd hope for 16 GB's of ram, but they'd probably do 8GB's. Probably be somewhere in line with the PS4/XBone for graphics. I'd hope for stacked, Nintendo seems to be very memory centric in their thinking, GC's memory type and caches were ahead of it's time, the Wii Continued that, the Wii U, I guess continued it, but seems pretty conventional. Maybe Nintendo will work with some obscure tech company for some neato technology.
BluRay again, no video playback, hopefully double layer.
Next Gen Portable probably highly integrated, maybe that's how they'll support WiiU Backwards compatibilty with regards to the gamepad.
Fourth Storm
Member
I know my opinion is probably controversial, but I think Wii U will be the end of the line for Nintendo's traditional consoles. With the home console market the way it is, I just cannot foresee a future in which Nintendo makes a huge box with the capabilities of a PS4 or Xbone. For one, they are still a relatively small company, and their ability to constantly pump out AAA games for such a device is questionable. They also don't seem to have a real interest in chasing such horsepower (as seen with 3DS and Wii U). Few of their games would actually end up utilizing it anyway (Zelda, Metroid, maybe 3D Mario), so what is the point? Addionally, the home console market is moving towards convergence devices - basically PCs for the living room and even some cable box functionality thrown in there in Xbone's case. Nintendo have a long journey into uncharted territory if they are to follow suit.
What I do see happening is a complete paradigm shift in their device strategy. Basically, I can see them imitating Apple in sticking to one "platform"/OS and having a variety of devices which function on that platform. So, they could still release a dedicated portable and home console, and add even more devices in there like a tablet, but they would mostly play the same games. The architectures of such devices would be largely the same - most likely ARM-based CPU and some type of mobile GPU like Tegra or SGX. What would vary between devices is the number of cores, RAM, clocks, screen size, internal storage, etc. I picture their next home console as a small Apple TV (or Wii)-like device, eschewing horsepower for ease of use, unintimidating form factor, and low price. A variety of controllers would connect to it as well as their tablet and portable devices for dual screen gameplay.
The benefits of this approach are:
a) A streamlined output of their main franchises, preventing series fatigue.
b) Freed up resources to create more experimental titles
c) A wider combined userbase for most titles
d) Less risk. If one device bombs, little is lost as the games and architecture would be common to all devices.
e) By refreshing form factor periodically (as they already did with the DS and GBA line), they can keep up with consumer electronic design trends without splitting the userbase
Wii U BC could be supported in one SKU of the home console device by including the chipset, if it's deemed worth the effort. It would need to be distributed in appropriate quantities and sold either at profit or at cost, so that little risk is associated with it, and then likely phased out over time completely.
Smurfman256
Member
In regards to sales, Pikmin 3 seems to be selling fairly well in Japan now and is helping the Wii U gain some more sales.
Fourth Storm
Member
Actually, looking at AMD's ultramobile line, extending the Wii U architecture into a true tablet may not be as far off as I thought. They were putting 80 shaderGPUs and dual core CPUs into 4.5 watt tablets already last year and on a 40 nm process! O_O
Smurfman256
Member
holy shit...
Correct me if i'm wrong but doesn't Nintendo have more cash in the bank then Sony's entire stock price and company value?
I honestly don't buy this whole 'Nintendo are a small company' argument.
They have something like 10 billion USD in the bank.
They have thousands of staff
Hundreds of millions to billions in hard and soft assets
Own significant infrastructure globally
Some of the most valuable IPs in the industry
The DS and Wii were highly successful
The real issue is Nintendo have failed to invest in modernizing their business for over a decade. Iwata has basically destroyed the company through his fiscal tight ass and reserved management. The company has stagnated while the rest of the industry has moved on. Nintendo are now so far behind the competition they're going to be forced to go down the direction you've suggested if they want to remain in the hardware business.
And if Nintendo go down the path you've suggested, i will not be buying another one of their platforms ever again.
ArchangelWest
Member
Hopefully this insight helps guys.
http://hdwarriors.com/general-impression-of-wii-u-edram-explained-by-shinen/
If someone wants to make a thread about it go ahead. I don't know that it's my place.
Snippet:
'We use the eDRAM in the Wii U for the actual framebuffers, intermediate framebuffer captures, as a fast scratch memory for some CPU intense work and for other GPU memory writes.
Using eDRAM properly is a simple way to get extra performance without any other optimizations.
In general, development for Wii U CPU+GPU is simple. You don’t need complicated setups or workarounds for things like HDR(High Dynamic Range)or linear RGB (Color Modeling). What we also like is that there are plenty of possibilities for speeding up your rendering and code, but you don’t have to dig deep for them to get proper performance.
For instance, all of our shaders used in ‘Nano Assault Neo’ are not really optimized. We just used the first iteration of them because they were already fast enough. We looked later through the shaders dis-assembly and noticed we can make them 30-40% faster by better pipeline usage or better hints for the shader compiler. '
http://hdwarriors.com/general-impression-of-wii-u-edram-explained-by-shinen/
If someone wants to make a thread about it go ahead. I don't know that it's my place.
Snippet:
'We use the eDRAM in the Wii U for the actual framebuffers, intermediate framebuffer captures, as a fast scratch memory for some CPU intense work and for other GPU memory writes.
Using eDRAM properly is a simple way to get extra performance without any other optimizations.
In general, development for Wii U CPU+GPU is simple. You don’t need complicated setups or workarounds for things like HDR(High Dynamic Range)or linear RGB (Color Modeling). What we also like is that there are plenty of possibilities for speeding up your rendering and code, but you don’t have to dig deep for them to get proper performance.
For instance, all of our shaders used in ‘Nano Assault Neo’ are not really optimized. We just used the first iteration of them because they were already fast enough. We looked later through the shaders dis-assembly and noticed we can make them 30-40% faster by better pipeline usage or better hints for the shader compiler. '
ColdBlooder
Banned
Bein filthy rich and a big/small company are 2 different things. Nintendo has roughly 5000 employees. MS as ~90.000. Sony has ~160.000
Compared to MS and Sony, their tiny.
ArchangelWest
Member
It also gives a tiny bit of insight into XBone's use of it.
'Even if you dont use MSAA (MultiSample Anti-Aliasing) you already need around 16Mb just for a 1080p framebuffer (with double buffering). You simply dont have that with XBOX360 eDRAM. As far as I know Microsoft corrected that issue and put also 32MB of Fast Ram into their new console. '
'Even if you dont use MSAA (MultiSample Anti-Aliasing) you already need around 16Mb just for a 1080p framebuffer (with double buffering). You simply dont have that with XBOX360 eDRAM. As far as I know Microsoft corrected that issue and put also 32MB of Fast Ram into their new console. '
Some interesting things about the interview:
- We've discussed the possibilities in this(and the WUSTs) threads, but this seems to be the first real conformation that Espresso has direct access to Latte's eDRAM. (question: Was this possible - or even practical - to do with X360's design?)
- Apparently Shin'en didn't even use the system's DSP. Obviously, they didn't put all the cores to work fully, so there wasn't a need to further complicate things by introducing the DSP, but perhaps this could have been one of the system's resources that - according to them - weren't initially accessible? In any case, this bodes well for the performance of future software.
- Intermediate buffer captures? Is that related to a frame buffer for the gamepad?
- I like the fact that they see "plenty of possibilities" for optimization with Wii U's architecture. People - based on popular reactions - act as though every console has huge potential to show a performance boost as time goes on, except for Wii U, which apparently - in their sights - is already close to it's max. Thankfully, that perception did lose a bit of steam after E3, but it's still prevails.
- We've discussed the possibilities in this(and the WUSTs) threads, but this seems to be the first real conformation that Espresso has direct access to Latte's eDRAM. (question: Was this possible - or even practical - to do with X360's design?)
- Apparently Shin'en didn't even use the system's DSP. Obviously, they didn't put all the cores to work fully, so there wasn't a need to further complicate things by introducing the DSP, but perhaps this could have been one of the system's resources that - according to them - weren't initially accessible? In any case, this bodes well for the performance of future software.
- Intermediate buffer captures? Is that related to a frame buffer for the gamepad?
- I like the fact that they see "plenty of possibilities" for optimization with Wii U's architecture. People - based on popular reactions - act as though every console has huge potential to show a performance boost as time goes on, except for Wii U, which apparently - in their sights - is already close to it's max. Thankfully, that perception did lose a bit of steam after E3, but it's still prevails.
Pointless comparison.
NOt all 90 or 160 thousand employers of Sony and Microsoft work in fields even related to gaming.
How many staff do MS and Sony have for their Xbox and PS divisions.
Which again highlights how Nintendo are not 'small'
Nintendo is smaller than EA
ColdBlooder
Banned
Yo DO know that they have like 3 headquaters filled with alot of people aren't working in the software development department?
And EA twice the size, has no hardware department...
Why would they be aiming for a "true tablet" when they were never aiming for a tablet to begin with? I thought it was stated and reinforced many times that the bases for the Wii U Gamepad was the DS which released before any of the popular tablets, and that they decided on going with the touchscreen controller for their next console in 2007 which was also before the tablet boom occurred. Miyamoto stated clearly that the Wii U having a touchscreen controller and the tablet boom occurring were pure coincidence. The focus of the Wii U as console is dual screen gameplay. If they made the console into a tablet then that would defeat the entire purpose of its existence. It would just be single screen DS.
I don't see Nintendo ever making a tablet. They are already have a touchscreen hardware design that has dominated the gaming market for 2 generations now. I also dont' buy into this handheld only business that people are insisting on with Nintendo. Did any of the Ipads or Androids ever break 150 million in sells?
They just came out of releasing the second best selling console in history as well. The console market has always been their primary focus even if their handhelds have overtaken it in sales number. I fail to see how more success in one market than another that you are still profitable in would lead to you dropping the latter. Dropping from the pure home console market would just be giving vantage to their competition.
This new info from Shin'en is interesting as well. They're shading system in the launch Nano Assault Neo is already exceeding what I've seen in XBLA games and they state they can get 30%-40% more performance just by optimizing the pipelines. We aren't even getting into other optimizations. This means they were just wasting shading before and still beating the 360.
This throws more weight behind the probability that the console has more than 160 shaders for me. Correct me if I'm wrong, but wasn't one of the critical pointed for explaining the improved shading in some Wii U games over their 360 counterparts that the 360 wasn't using all 240 in said games, but that the Wii U was some how getting near full efficiency out of its 160? This Shin'en comment just blows that out of the water. Fromo what I'm garnering from not only them, but Miyamoto's recent comment as well, getting full efficiency out of the Wii U's shaders wasn't even possible at launch and still doesn't seem to be likely.
In fact, I think most of the Wii U's functionality isn't even usable. All of these performance increasing updates are making that seem more and more likely. They even announced that the next big update is going to add functionality.
On another note, aren't Xenos's 240 shaders stored in 5 ALU's?
ArchangelWest
Member
I think the 'handheld only' idea at this point is attached to the idea that the handheld would be something in form similar to the Wii U gamepad, and that in some way it could also be used as a console to play your games on your tv. Streaming tech should be sufficient and inexpensive enough to do that by the time Nintendo releases their next console I think.
Smurfman256
Member
If the adoption rate is high enough by at LEAST 2016, how likely do you guys/gals/amorphous beings think that the next Nintendo console will support 4K resolution?
shakin_haitian
Banned
Considering that consoles are meant to be played on the couch and not two feet away from the screen, I really doubt they'd optimize their system for 4k. They'd probably consider that overkill.
The PlayStation?
w1gglyjones
Member
Nintendo is "small" in the sense that they are very efficient and expanding their human resources is not easy. I'm not saying that they don't need to expand, they do, but it's inevitable that as they expand, their efficiency and quality standards will suffer.
Smurfman256
Member
Well, Miyamoto said that he would have loved PIkmin 3 to have been running at 4K, so they may consider it (although he said that it really wouldn't be that helpful for other games. On a side note, is it just me, or is 4K really just an RTS player's wet dream?).
You can actually directly compare AMDs newest tablet parts with the XB1/PS4 since they're based upon the same architecture.
For example:
Xbox One
CPU: 8 cores @ 1.6 GHz (Cinebench score: 3.3 pts)
GPU: 12 CUs @ 800 MHz (768 cores, 1240 GFLOPS)
TDP: 100W
AMD A4-1200
CPU: 2 cores @ 1 GHz (Cinebench score: 0.52 pts)
GPU: 2 CUs @ 225 Mhz (128 cores, 58 GFLOPS)
TDP: 3.9W
Smurfman256
Member
I just had an idea; would anyone REALLY be THAT disappointed if the next console dropped BC? ...Scratch that, people would be PISSED.
Fourth Storm, Your idea is close to mine except for one big difference.
Just some quick facts:
1. Iwata already announced that they are making 1 architecture going forward and that their handheld and console hardware teams are now together.
2. Iwata stated that they are not making a hybrid device, but in fact possibly expanding their platforms to add a 3rd and possibly more devices once they are all under 1 architecture.
3. This team is already working on their next handheld and console as we speak.
4. Nintendo has their developers creating tools for Wii U's GPU and CPU architecture, building 3rd party engines to run on Wii U specifically (Unity for instance) and building tools to easily port games designed for mobile devices to Wii U.
Having said all of that, the difference between what you think and what I think is that they won't be abandoning Wii U's Architecture going forward. They will almost definitely use a more modern GPU again, with more power for their console. Almost certainly target sub 50watt device again, (which at 20nm or less is a significant upgrade from Wii U)
Now let me talk about what I think they will do with their next hardware cycle, they will likely start with
DS4:
The current handheld doesn't support important 3rd party engines like Unity and will likely be replaced ~2016 with the relatively simple to implement Wii U hardware though it will not be a 1:1 transition most likely. You might see them drop the clock speed back down to 1GHz tricore cpu with 400MHz GPU. System ram might be 3DS's F-RAM(?) for backwards compatibility as they will likely have the 3DS chip on board in an powered down state. (this was still enough to show off the bird demo at e3 2011 on the show room floor @ 720p and 30fps) I think they will go for 480p screen on the bottom and 540p 3D screen on the top, because they are both very standard resolutions and this device would be plenty powerful enough to handle it. (BTW it would be either 256gflops or 128gflops depending on if Wii U's GPU has 320 or 160 ALUs, not that this matters much)
Wii U's successor:
The Wii U would be replaced by a successor a year after DS4 comes onto the market, it is likely what the hardware team is focused on and is already securing the design. I personally think it will be roughly ~4x Wii U, since they are the same basic architecture (Though likely more modern "DX 11.2 or DX 12 or whatever AMD is currently working on for HD 9000 series+)
This might sound a bit too good for Nintendo, but being realistic lets look at the specs: 1280 ALUs (similar to PS4, though more here) or 640 ALUs (if 160ALUs is right for Wii U) this puts the console at 2.56 TFLOPs or 1.28TFLOPs in ~2017 with a more modern GPU, you'd also likely only see the CPU increase to 6 cores but the clock speed could be at 1.6GHz or more, I think this CPU would likely be very similar to Wii U's espresso, meaning 4MBs L2 cache, 1.5 for just the main core and .5 for all remaining cores. edram might not change on the GPU side as 32MB is still enough for 1080p and it is expensive. 8GB of DDR4 memory is also probably where they will end up.
I don't think they will change the controller much, it will still likely have a screen on it but they might add support for 4 of them, as well as possibly making them sleeker. It could have extra functionality like an outward facing camera to add AR elements, maybe built in vitality sensor or it might have a good enough camera to do this with the one pointing at your face. Hopefully a longer wireless range to encompass the house as well.
New hardware?:
This new hardware cycle may see Nintendo expand into new platforms, maybe we will see Virtual Boy 2 or whatever you'll want to call it, which could receive it's signal similarly to the gamepad, and allow you to play from a comfortable distance with the wiimote and nunchuk (which will hopefully also see improvement to refine their orientation sensors)
Another device that would be a new platform for Nintendo could be either some sort of eye wear instead of the VB2 or a phone like device. The way I imagine this device is a rectangle with 2 shoulder buttons on one side and a + where you'd find the little Samsung or Apple button at the bottom of a phone. This would obviously work as a d-pad and allow it to also function as the phone's recall button, which would basically be a 2 stage digital click, with the 2nd click on a rocker to indicate direction. It would also come with a stylus and hopefully NinjaGaiden DS sequel will come. This device would be similar to the DS4 and might actually have the same hardware inside of it, the problem here is obviously if Nintendo can partner with a phone manufacture or at least implement the proper antenna for such a device. I am just taking a wild shot in the dark, and obviously all specs are just what I figure they will need going forward for these devices.
Edit: Obvious benefit of using the Wii U architecture going forward, is that it is ready when they launch DS4, Wii U's successor and any other device they come up with, they could extend the Wii U's life through DS4. They could let 3rd parties target all Nintendo platforms with a single game / code / assets (which would be DS4's as the lowest point in their power envelope) Meanwhile Nintendo would still release handheld titles and console titles separately, but they would come much faster. Something like Mario Kart could be released as 1 title, same with the next smash but they could also release a game for the handheld as well as a game for the console 6 months to a year before or after it. Something as different as OoT to Majora's Mask, or the 2 yoshi's island games in development right now, the difference is Nintendo would only need to use 1 team and instead of creating DLC for the next 6 months to a year, they simply release a new game for the other platform. This staggered delivery would allow owners to avoid droughts far more often, not have a drought upfront and also give 3rd parties a massive user base in this cycle that could easily see anywhere from 100-250 million units of hardware.
Just some quick facts:
1. Iwata already announced that they are making 1 architecture going forward and that their handheld and console hardware teams are now together.
2. Iwata stated that they are not making a hybrid device, but in fact possibly expanding their platforms to add a 3rd and possibly more devices once they are all under 1 architecture.
3. This team is already working on their next handheld and console as we speak.
4. Nintendo has their developers creating tools for Wii U's GPU and CPU architecture, building 3rd party engines to run on Wii U specifically (Unity for instance) and building tools to easily port games designed for mobile devices to Wii U.
Having said all of that, the difference between what you think and what I think is that they won't be abandoning Wii U's Architecture going forward. They will almost definitely use a more modern GPU again, with more power for their console. Almost certainly target sub 50watt device again, (which at 20nm or less is a significant upgrade from Wii U)
Now let me talk about what I think they will do with their next hardware cycle, they will likely start with
DS4:
The current handheld doesn't support important 3rd party engines like Unity and will likely be replaced ~2016 with the relatively simple to implement Wii U hardware though it will not be a 1:1 transition most likely. You might see them drop the clock speed back down to 1GHz tricore cpu with 400MHz GPU. System ram might be 3DS's F-RAM(?) for backwards compatibility as they will likely have the 3DS chip on board in an powered down state. (this was still enough to show off the bird demo at e3 2011 on the show room floor @ 720p and 30fps) I think they will go for 480p screen on the bottom and 540p 3D screen on the top, because they are both very standard resolutions and this device would be plenty powerful enough to handle it. (BTW it would be either 256gflops or 128gflops depending on if Wii U's GPU has 320 or 160 ALUs, not that this matters much)
Wii U's successor:
The Wii U would be replaced by a successor a year after DS4 comes onto the market, it is likely what the hardware team is focused on and is already securing the design. I personally think it will be roughly ~4x Wii U, since they are the same basic architecture (Though likely more modern "DX 11.2 or DX 12 or whatever AMD is currently working on for HD 9000 series+)
This might sound a bit too good for Nintendo, but being realistic lets look at the specs: 1280 ALUs (similar to PS4, though more here) or 640 ALUs (if 160ALUs is right for Wii U) this puts the console at 2.56 TFLOPs or 1.28TFLOPs in ~2017 with a more modern GPU, you'd also likely only see the CPU increase to 6 cores but the clock speed could be at 1.6GHz or more, I think this CPU would likely be very similar to Wii U's espresso, meaning 4MBs L2 cache, 1.5 for just the main core and .5 for all remaining cores. edram might not change on the GPU side as 32MB is still enough for 1080p and it is expensive. 8GB of DDR4 memory is also probably where they will end up.
I don't think they will change the controller much, it will still likely have a screen on it but they might add support for 4 of them, as well as possibly making them sleeker. It could have extra functionality like an outward facing camera to add AR elements, maybe built in vitality sensor or it might have a good enough camera to do this with the one pointing at your face. Hopefully a longer wireless range to encompass the house as well.
New hardware?:
This new hardware cycle may see Nintendo expand into new platforms, maybe we will see Virtual Boy 2 or whatever you'll want to call it, which could receive it's signal similarly to the gamepad, and allow you to play from a comfortable distance with the wiimote and nunchuk (which will hopefully also see improvement to refine their orientation sensors)
Another device that would be a new platform for Nintendo could be either some sort of eye wear instead of the VB2 or a phone like device. The way I imagine this device is a rectangle with 2 shoulder buttons on one side and a + where you'd find the little Samsung or Apple button at the bottom of a phone. This would obviously work as a d-pad and allow it to also function as the phone's recall button, which would basically be a 2 stage digital click, with the 2nd click on a rocker to indicate direction. It would also come with a stylus and hopefully NinjaGaiden DS sequel will come. This device would be similar to the DS4 and might actually have the same hardware inside of it, the problem here is obviously if Nintendo can partner with a phone manufacture or at least implement the proper antenna for such a device. I am just taking a wild shot in the dark, and obviously all specs are just what I figure they will need going forward for these devices.
Edit: Obvious benefit of using the Wii U architecture going forward, is that it is ready when they launch DS4, Wii U's successor and any other device they come up with, they could extend the Wii U's life through DS4. They could let 3rd parties target all Nintendo platforms with a single game / code / assets (which would be DS4's as the lowest point in their power envelope) Meanwhile Nintendo would still release handheld titles and console titles separately, but they would come much faster. Something like Mario Kart could be released as 1 title, same with the next smash but they could also release a game for the handheld as well as a game for the console 6 months to a year before or after it. Something as different as OoT to Majora's Mask, or the 2 yoshi's island games in development right now, the difference is Nintendo would only need to use 1 team and instead of creating DLC for the next 6 months to a year, they simply release a new game for the other platform. This staggered delivery would allow owners to avoid droughts far more often, not have a drought upfront and also give 3rd parties a massive user base in this cycle that could easily see anywhere from 100-250 million units of hardware.
- Status