The nice thing about this sort of architecture is that the design is modular, 4CUs sound great but considering Nintendo's actual goals and audience (the masses) 2cus would do fine and offer more battery life.
Nintendo GOing Forward.
- Thread starter Wynix
- Start date
The nice thing about this sort of architecture is that the design is modular, 4CUs sound great but considering Nintendo's actual goals and audience (the masses) 2cus would do fine and offer more battery life.
You are completely ignoring the fact that those specs will be piss-weak compared to pretty much every single cheap smartphone released during 2016.
Also AMD will be using 14nm for mobile parts by that point.
Basically you are being overly conservative, even for Nintendo standards, to the point of sounding exaggeratingly pessimistic.
GCN 2.0 would be the current GPU architecture, also while it might seem "weak" cheetah cores are based on ARM8 K12 afaik, and 2016 being 16nm or even 14nm (they are in the same pipeline and it's really up to AMD) would mean great things for the handheld. 4CUs is 256 shaders, 500mhz would be a good way to conserve power but I am not making these chips, so I am basing everything on the mullins chips that exist now with 128 shaders @ or below 500mhz using 2.8 watts. (which still out performs Qualcomm's 805) we are talking about a clear doubling of mullin's performance btw. (without taking into consideration GCN 2.0)
Anything released in 2016? this is what AMD is releasing in 2016 for this TDP. Smart phones btw cost 500-700 dollars, if you go in thinking you can make something better than a smart phone for even 1 refresh, you will make a handheld that no one can afford. If it happens that AMD has some 4CU GCN2.0 APU that runs at less than 2watts in 2016, Nintendo should use that, however this is not an ancient architecture (it's literally same year) and it isn't some low end part (it's a low wattage part which they have to use in this form factor.
Maybe you were thinking of some other architecture, but I don't see where this handheld would be a bad move, you are looking at a very nice upgrade for 3DS that is actually ~twice as powerful as Wii U.
I wasn't clear indeed, as in using the same SoC/tech as mobile/tablet hardware, likely competitive, but it could be a little conservative.
They might be at a cost disadvantage unless they commit to huge volumes.
Unsubsidized prices for the top smart phones are the same, using comparable components. Yet only Apple and Samsung make profits while others like HTC struggle, possibly because they pay higher costs for a fraction of the volume of the other two.
Unsubsidized prices for the top smart phones are the same, using comparable components. Yet only Apple and Samsung make profits while others like HTC struggle, possibly because they pay higher costs for a fraction of the volume of the other two.
D
Deleted member 13524
Guest
4 CUs at 500MHz is the equivalent of a full-power TK1 (950MHz) in 2014 -> probably what will appear in the Shield 2, which is a handheld console.
2 CUs at 500MHz would make a handheld half as fast as the Shield 2. It would be a console launching 2 years later with 1/2 the processing power. In 2016 with matured 20/16nm we're looking at a Shield 4 with 2GHz 64bit Denver CPUs, high-bandwidth memory, >4GB RAM and an iGPU that is probably in the range of a laptop GM107.
Do you understand how this compares to a 500MHz 2 CU GCN part?
The Snapdragon 805 is most probably a bit faster than Mullins (with over twice the memory bandwidth too) and it's still a chip coming out in 2014. I don't know where you got the idea that it would be otherwise.
AMD's plans for 2016 have been laid out. It's the year for AMD's proprietary K12 ARM CPUs, HBM and HSA implementation should be final. There's no mention of iGPU performance but as HSA becomes easier, more powerful and more popular with developers, I guess the amount of die area dedicated to the GPU is bound to increase.
The reason why Mullins kept only 2 CUs is because its architecture is just a small tweak (based more on fab improvements than anything else) over last year's Temash. If Mullins was an entirely new chip, I guarantee you it would have more than 2 CUs.
This is factually wrong and I think you should inform yourself before making such erroneous claims.
Elop invented that idea in order to defend Nokia's suicidal commitment to windows phone (so he could sell the company for peanuts, which he did) and somehow that crap stuck with some users in this forum.
It's a ridiculous statement. Look up for LG, Xiaomi, Lenovo, Huawei and others who have been consistently posting record sales and profits quarter after quarter.
D
Deleted member 13524
Guest
What's your point?
Do Chinese manufacturers that serve the very low end, lower quality, Chinese market really in the same league as Nintendo? Do you really think that applies to Nintendo profitability? Huawei phons are not the epitome of quality, I wouldn't buy one.
And the Korean manufacturers have their own almost incestuous ways. They are intertwined, ginormous, control vast swaths of the supply chain, and almost quasi government backed. It's obviously at times been very effective, but again I dont know if Nintendo can hope for that.
And the Korean manufacturers have their own almost incestuous ways. They are intertwined, ginormous, control vast swaths of the supply chain, and almost quasi government backed. It's obviously at times been very effective, but again I dont know if Nintendo can hope for that.
http://forums.anandtech.com/showpost.php?p=36381763&postcount=8 This is where I found good comparisons between mullins and 805 (the entire thread is pretty good) I don't think what they can do and what Nintendo will do is 100% inline with each other either. Just remember the Wii U speculation threads, we had some assumptions, they could of did more, but they were satisfied with a certain level of performance, I don't think they will stray quite so badly this time on the console side of things, but I do think TDP has to be on their mind and custom hardware is going to force their hand when it comes to handhelds. Sure 6CUs, 8CUs, higher clocks, I'm open to anything once we know the performance per clock of 14nm AMD chips.
Nintendo can't go another generation pursuing the cheap-and-underpowered(-but-still-selling-for-lots) model they've been pursuing for nearly ten years now. That it was (very) successful for several years last decade was more accident than anything else, Nintendo riding a prolonged hype-wave which eventually dissipated.
Another underpowered console like wuu will be as unsuccessful as the wuu, largely because of the same reasons the wuu is unsuccessful - IE, lack of software. 3rd parties are unwilling and unable to cater to a platform which is hugely inferior to the other alternatives out there (and Nintendo is unable to produce enough software by itself), thus ensuring Nintendo's platform's demise.
Another underpowered console like wuu will be as unsuccessful as the wuu, largely because of the same reasons the wuu is unsuccessful - IE, lack of software. 3rd parties are unwilling and unable to cater to a platform which is hugely inferior to the other alternatives out there (and Nintendo is unable to produce enough software by itself), thus ensuring Nintendo's platform's demise.
D
Deleted member 13524
Guest
3dmark with the S805 was the only odd result in the entire benchmark suite:
Anand Lai Shimpy said:
The Snapdragon 805 is quite a bit faster than 801 in 3D games, even using early drivers. Therefore, it'll probably be faster than Mullins.
Yet, too many people seem fixated on the idea that what Nintendo will do during the next 10 years = what Nintendo did in the last 10 years.
The harm is done. The Wii U brought NIntendo its fastest decline ever within a relatively short time span. Even the most hard-headed execs at Nintendo have learned their lesson now. That or the company won't last another 5 years, let alone 10.
They won't admit the reason the Wii U failed was because of piss-poor hardware alienating 3rd parties - because the console is still selling and they won't tell their customers they're buying sub-par hardware.
As long as the console is selling, they will just keep saying the problems came from stuff they're still capable of fixing, like market orientation and 1st party software.
There's also this weird consensus that cost-effectiveness and power efficiency can only be done with underpowered hardware. As if the last 5 years in the ULP SoC and mobile APUs' evolution meant nothing.
Too many people try to treat the console market as if its like the graphics card market. Consoles popularity and success is not based on having the most processing power for the dollar, and you could argue that its irrelevant because of the success of the Wii. Wii wasn't exactly lacking software support for the first 4 years. Wii U isn't alienated because of its being less powerful than the PS4 or X1, but because software sales on the console have been abysmal for third parties. Nintendo has stumbled big time this gen, but seeing as how content is king, and in the console market, software sells hardware, not 3Dmark benchmark scores like in the PC world. Nintendo has failed to bring enough content quickly enough to get consumers interested, and to make that worse, even when they have had excellent content, the marketing behind it was nothing short of atrocious. Wii U as a product is fine. It has some unique features that may or may not be valuable to you, but those features cost money to implement, just like Kinect cost Microsoft money. You may not be a big fan of the Gamepad, but its foolish to argue that it doesn't add a lot of cost to the console over a standard DA controller.
I would say:
Nintendo can't go another generation pursuing the cheap-and-underpowered(-but-still-selling-for-lots) and selling it at a premium model they've been pursuing for nearly ten years now.
Indeed something as underpowered as the wii or wiiu or 3ds won't cut though I don't think they need (or can) match the higher offering. Sony and MSFT whereas they have a more solid ecosystems built around their systems could not do it either this gen.
Anyway I think the fact Apple just jumped into the "closer to the metal API" bandwagon is a pretty bad omen. Google will be the next nullifying some historical console advantages wrt to perfs relatively to the specs. Nintendo will face: volume, way richer environments, and hardware which costs is hidden behind subscription (I speak of the handheld side of things).
The issue is not about having a good hardware but having a selling point has time goes (still 2 years to go till 2016) I think it is going to get tougher and tougher for NIntendo to have any.
Sadly I agree they drove themselves into a corner. I see no light at the end of that tunnel.
Their last chance when they see that the 3ds was not doing that great and the WiiU was dead in the water was to go with a radical move: eol the product ala virtua boy and go for a radical restructuring of the company and try to launch something asap.
By 2016 I think it will be too late.
Except it is because it's not powerful enough to run easy ports. Ergo devs will ignore it. Although a substantial part of the console market doesn't buy based on the best hardware, hardware has to be sufficient to get dev support within a generation of cross-platform titles, or be such a runaway success that devs can afford to go platform exclusive. Wii U is neither of these things and so will suffer due to the inadequate hardware. Even if Wii U had launched with stellar content, it'd have had to sell gangbusters to justify the expense of ports from next-gen hardware which otherwise can fairly simply target the range of 12 CU XB1 up to PC with "same+1" from the starting point. When the performance differential is too great, a platform is sidelined, which is why there's a narrow crossover between generations despite a MASSIVE install-base advantage for the old, too-slow machines.
Well I disagree the WiiU is alienated because it does not sell whatever the reason.
WiiUmote was a bad bet ok, the perfs of the system are not great ok.
To me the most offending point is the price, like kinect the wiiumote could be remove, though I think it would not be enough to make the product attractive: it would still be too pricey for what it does aka delivering, partially as the ecosystem is not on par with what msft or sony provide, what other systems have been delivering for 8 or 9 years.
