Fourth Storm said:
the pessimistic specs look pretty good to me (minus swapping out the K12s for stock ARM A57s). I don't think they need to worry about matching Wii U on the handheld, though. Yes, it would probably help for VC games, but who knows what type of abstraction layer they have those games running on. I can see them putting the major 3DS SoC blocks on the handheld die and Espresso on the console die.
One thing that must be acknowledged, however, is that Nintendo still haven't fundamentally changed their philosophy on hardware. Certain things are part of their "DNA" as Genyo Takeda would put it. Just looking at die size, the Xbone/PS4 SoCs are massive - Xbone's being 363mm2 and PS4's only slightly smaller. Nintendo believe in efficiency over sheer horsepower. I'd expect the Wii U successor's SoC to be roughly half the size of their competitors' and somewhere between the two in GPU GFLOPs.
There's also the reality that eDRAM is not available at the foundries they would be using at the 20nm node. They'll have to use eSRAM, which will take up a great percentage of die space, as it does on Xbone.
DDR4 can probably happen given the timeframe, but I personally doubt Nintendo will opt for a 256-bit bus like the others. They can get something like 50 GB/s with a 128-bit bus, maybe more if faster clocked DDR4 becomes readily available. The narrower bus would help reduce board complexity and cost. The eSRAM will have to pick up the rest of the slack.
I think the more interesting question is: If they went with ARM in their next handheld, would they necessarily have to for their next home console?
Kinda reminds one of the present situation at another certain three letter company...
Yeah they might go with 128bit memory, that could even go as high as what? 90GB/s.
I do think that it is worth it for them to bring espresso for the seamlessness of the ecosystem they are trying to build, it could handle the os and that would be a big help to them imo, not to mention they have been spending all this extra resources on emulators that work with Espresso, eshop titles they could keep selling as well as their history with the processor for all sorts of future applications, if they put it in one device they will likely put it in another, it's also ridiculously tiny and cheap for them.
AMD is the obvious partner going forward, they have just caught up with HD development, they intimately know that hardware family and AMD is the most price aware of the companies they could use. As for the pessimistic view, I simply took an already existing product on the low end for each device, in 2016 the handheld will be a bit behind their newer apus in this formfactor (Mullins is a 2014 part afterall)
I've been reasonable before, over 3 years ago I thought Nintendo would use a trinity apu quad core 3ghz with 384 shaders @ 800MHz (gpu) for project café, and would just use their gecko cpu for bc, it sounded fairly reasonable as does this.
However I think there is a clear change between what they were doing then and what they are trying to do now. I wouldn't say they were completely oblivious to the market changes with mobile, it just moved faster than they could anticipate, plan for and execute. I imagine that Wii U was actually suppose to come sooner. I'd say what they eventually brought to market isn't even their full vision for the device, Iwata hinted in his wii u ask piece that Wii U was originally going to be more powerful but they thought the power envelope was more important. Japan's console market crashed completely during Nintendo's planning of the Wii U and they didn't under spend on the components, they just focused on tdp far too much.
Their goals with these next devices are different which is why I think they will go with AMD, and the relative power in both platforms is more of a byproduct of AMD than Nintendo's goals. I also think they are planning a more immediate and closer answer to mobile than their handheld, I think when Iwata was talking about combining their architecture going forward and that it would lead to a 3rd pillar, they were thinking of a platform that could more readily compete with mobile, something lacking most buttons and in a single screen formfactor that is small and sleek while still offering inputs that mobile can't.
Also it's not true that people don't have room on them for a dedicated gaming device, people on average carry about 3 tech devices on them according to the research so Nintendo needs a device people are more willing to carry around. I personally believe that their handheld won't fill this void but a smaller device similar to an iPod touch, could.
Edit: the 3rd device I imagine would come after the other 2, have 2 shoulder buttons that look like volume rockers, a dpad that functions like an iPhone home button, and possibly a couple action buttons, a stylus wouldn't hurt. Releasing a device like this under $100 that can also run mobile games but is sort of a undercover gaming device would have a fair chance in the market to end up in every gamer's pocket, casual or core.