Call it confirmation bias if you want, but I'm starting to believe that this new information confirms my previously held beliefs
Seriously, though, if both this rumour and Emily Rogers' rumour are true (and I'm taking both with a healthy dose of salt), then they both tie into a theory that I had that Nintendo's most sensible route to take with NX hardware is to design a home console that's on the low end of the XBO/PS4 ballpark, and a handheld with a low-res screen but an SoC that's as powerful as possible. Here are a couple of snippets from my posts on the matter over the past year or so:
Thraktor said:
if Nintendo want to make development as simple as possible across multiple hardware configurations, they're going to want to get as close as possible to something that, for the sake of argument, we'll call per-pixel-performance-parity, or PPPP. This means that if, for example, the home console is displaying a resolution four times that of the handheld, it should have a GPU which is roughly four times as powerful. Achieving PPPP would significantly reduce the burden on developers to support both platforms by maximising the amount of graphics code they can re-use across both (eg shaders, lighting code, effects, and anything else which scales roughly linearly with resolution). Assets would still have to be scaled down for the handheld, but this is far less labour intensive with modern tools.
(Link to full post)
Thraktor said:
I think the default option for Nintendo is to use a single ~4.5" 540p capacitive touch screen on their handheld, like the one used on the 2nd gen Moto E, which sells for as little as $75 off-contract. Taking the SoC and RAM out of the picture for a little bit, the BoM for the Moto E and a simple Nintendo handheld wouldn't be all that different. The handheld wouldn't need the modem, and would probably end up with a slightly smaller battery, but would add physical buttons and sticks and would probably have a little bit more internal flash memory. With the Moto E selling for that little, I can't see the total BoM for a straightforward 540p handheld (excluding the SoC and RAM) being much higher than $40-$50.
Even on the worse end of the spectrum for 14LPP yield estimates, you could squeeze a pretty nice 14nm SoC and some LPDDR4 in there in late 2016 without having to lose money at $199. The alternative, of course, is that the "spare cash" goes to a more modest 28nm SoC and some kind of special feature, be it a freeform display, 3D, or whatever else Nintendo may have up their sleeve. Hence, my argument is effectively that Nintendo should forgo any kind of expensive special functionality this time around and focus on designing a straightforward handheld with the highest performing SoC possible.
[...]
A 540p handheld with the same GPU architecture, but running at a quarter the performance of a 1080p home console is pretty much the platonic ideal of straight-forward multi-device development. It doesn't make it trivial to develop a game for both, but with good tools it would bring the additional development work required to support both of the devices down pretty much as low as it can go.
(Link to full post)
Thraktor said:
My logic is basically that Nintendo should want to satisfy these goals:
1. Release a home console that is within touching distance of Xbox One and PS4, at a reasonable price
2. Release a handheld which is relatively cheap (i.e. $200 or under)
and
3. Make developing one game for both the home console and the handheld as simple as humanly possible
The first goal is relatively easy, but satisfying both it and the third goal is easiest if you low-ball the XBO and PS4 a little, i.e. something a little less powerful than the XBO (which should also keep the price low).
The second goal on its own is trivial. They could probably make a handheld that outperforms the 3DS (w/o 3D or anything like that) for $80 if they really wanted to. When combined with the third goal, though, they basically want to have as low a screen res as they can get away with (i.e. 480p/540p) combined with the most powerful SoC they can for $199, and without anything else that would push up the BoM.
Aside from the above, the third goal is most effectively met if they use the same CPU arch (i.e. ARM) and the same GPU arch across both SoCs. AMD and Nvidia are pretty much the only companies that could do both, but it seems like they're going with AMD.
(Link to full post)
While writing those I was under the assumption that AMD would be providing the chips, but the logic still stands if Nvidia is providing the SoCs for each machine (although you can replace 14nm with 16nm). It may even explain
why Nintendo would choose Nvidia over AMD. Even if Tegra didn't work out with 3DS, Nvidia still has a much better history than AMD in this market segment, and if Nintendo's primary goals were the three above, then Nvidia may have been the better option purely on the basis of being able to offer a higher-performance handheld part, or Nintendo having more faith in them being able to deliver a higher-performance handheld part.
As an example, consider the TX1. Test silicon would have been available around the time Nintendo were making these decisions, so it's reasonable to believe Nintendo could have been given it as an example part of what Nvidia could do in this kind of thermal envelope. Of course, Nintendo would have to clock down the TX1 considerably from the 1GHz GPU clock you see in the Shield TV, but when clocked down to around 500MHz
the GPU reportedly draws just 1.5W (just the GPU, not the full chip), which puts it within handheld territory, and even within the ballpark of one quarter of Xbox One's performance. Compared to that, AMD would have handed them Mullins, a binned part which would have provided a fraction of the performance when clocked down to fit a handheld power draw. This isn't to say AMD couldn't put together a competitive chip with Polaris on 14nm if they really tried, but Nintendo may well have seen Nvidia as the lower-risk option at the time, and if their entire strategy with NX hinges on them getting as much performance as possible out of their handheld SoC then I can see them switching vendors to Nvidia to ensure that they get that performance.
Why would their entire strategy hinge on getting as much performance as possible out of their handheld SoC? Well, as much as many people on GAF might like to think, Nintendo's best strategy for bringing more third parties to their consoles and convincing players to move over to their hardware isn't just to bring out the most crazy-powerful home console hardware possible. It didn't work for them with the N64 and Gamecube, and there's no reason to believe it would work for them now. Nintendo should leverage the one advantage they have over their competition; a line of successful handhelds with healthy install bases. Nintendo shouldn't be trying to convince third parties to develop games which run on the successor to a console which sold under 15 million, they should be trying to convince third parties to develop games which run across two devices, the predecessors of which sold over 70 million. And unlike Nintendo's home consoles, where EA, Activision, etc, can reasonably claim that any sales of their games are simply cannibalising sales on other platforms, handhelds would be a genuinely expanded market for them, as many people pretty much only play games on handhelds.
To actually convince third parties, though, they'd need to make porting a game to both devices as simple and cheap as possible. This means making a console that's powerful enough to easily port PS4/XBO games to and making a handheld that's able to play those games with as little work as possible at 480p/540p. Making a console competitive with PS4 and XBO isn't an issue, you just go to AMD and say "build me one of those". Making a handheld that can play games with XBO quality graphics at around a quarter the resolution for $200 isn't. Assuming your home console has precisely XBO performance and games run at 1080p on it (though most multiplats don't) you're looking at a mobile GPU with 325 AMD Gflops of performance to "easily" scale down the games to run at 540p on the handheld. If you're looking to compete with PS4 in home console performance, then 450 AMD Gflops is your target with the handheld. God help you if you try to compete with PS4K.
The point of this is that the higher the performance you can squeeze out of your handheld SoC, the better this strategy works. Pretty much the only limit on home console performance is cost, but handheld performance is intrinsically limited by architecture and manufacturing processes, so the performance of the handheld becomes the variable; the more you push it up the closer you can get to XBO and PS4 with your home console while keeping a small relative performance gap between the two devices. If your handheld is barely more powerful than the Vita, then the strategy simply doesn't work, whereas if you can somehow push it up to one quarter of PS4's performance you'll probably be swimming in third party support. Some games are going to be too CPU limited to run on the handheld no matter what (eg Assassin's Creed), but getting versions of CoD, FIFA, Battlefront, etc. which run on both home console and handheld (in non-gimped form) would be the best chance Nintendo's got of winning over gamers who would choose MS or Sony's consoles by default.
Now, within one day, we have one rumour which indicates the home NX is closer to XBO than PS4 in performance and that there's "a good reason" why Polaris rumours are unfounded, and we have a second rumour saying that Nintendo is using a Tegra chip on the handheld, which in its currently available off-the shelf forms (even clocked down to fit Nintendo's TDP) would provide substantially higher levels of performance than most people would have expected from the handheld. It ties in pretty much exactly with what I had been arguing Nintendo's approach should be, but for the switch from AMD to Nvidia. I'm still hesitant to put too much faith in these rumours, but if they were true it would definitely seem like Nintendo is going all-in on the shared ecosystem between home console and handheld.