The discussion on getting multiple Wii U tablets working on one console had my clutching my iPhone furiously because the technicalities of the discussion were missing some key points.
Right off the bat, the "bandwidth" of HDMI: 1920 x 1080 x 60 (fps) x 3 (24 bits of color / 8 bits per byte) = 355.96 megabytes per second. If I remember correctly, the Wii U controller is exactly a quarter of that (960x540). 5ghz wi-fi tops out at around 112 megabytes/sec. (assuming one hell of a MIMO setup getting 900 megabits, which is beyond fantasy), but neither one of these totals really matter.
So right off the bat, let me make it clear that nothing that even VAGUELY resembles a raw video stream is going to that controller. In fact, any "display" hardware on that graphics chip probably isn't even in charge of that. Fact of the matter is, what Nintendo is shooting for has already been done by Apple (look up Airplay mirroring), Intel (look up Wi-Di), and Sony (PSP Remote Play). Basically the system, either through brute force on the CPU (Sony) or a custom piece of silicon on the CPU (Intel) or GPU (Apple), will convert the framebuffer (basically the piece of memory that your game's image is on) to a video stream.
Keep in mind that this process doesn't need to involve video output of any kind. e.g. they could pick any chunk of memory to send to the controller. This is why you can have different images on the TV and on the controller itself, and in theory, you could send images to as many controllers as you want. Bandwidth on an encoded video stream is ORDERS OF MAGNUTIDE lower than raw bandwidth. Heck, Blu-Ray does 1080p (once again, enough for FOUR controllers) and tops out at 36 megabits.
Well, that's the theory, anyway. In reality, there's a far bigger limitation: processing power. Anyone who's seen the three implementations I mentioned (Sony/Apple/Intel), will note one major issue: They suck. Seriously, they generally fail on all three counts: Bad input lag, crappy visual quality, and a low framerate. Basically, Nintendo's trying to do what all three of these companies have on the market with AN ORDER OF MAGNITUDE better quality. And realistically, this is what will limit them the most.
First off, the conversion, unlike Sony's method, will most certainly not involve the CPU. It's far too much of a performance hit and unlike Sony, Nintendo has the distinct advantage of planning this feature out on their side. So it's probably gonna be part of their custom graphics chip. In fact, it'll probably be the feature that deems it a custom chip, and not a run-of-the-mill Radeon HD 4990 in the first place.
Now, given that the issue of getting the video small, smooth, and quickly to the tablet is non-trivial, you're more than likely going to see the limit of one controller for cost reasons. The hardware needed to get the speed and quality levels Nintendo wants is probably going to be a MUCH LARGER cost (in terms of the space used up on that graphics chip) than any of those previously mentioned methods. In fact, it wouldn't be trivial to implement something like that unless you're SERIOUSLY planning it in advance of a system's release. With that in mind, we have a few possibilities as to what Nintendo will do:
- On the one hand, the video streaming will be set to exactly the tablet screen's specs and that'll be the end of it. The console won't be PHYSICALLY capable of streaming more than one video.
- On the other hand, it might be set to stream a whole 1080p image, which means in theory, by intelligently splitting the image, they could stream it to 4 tablet controllers (remember, each one has a quarter of 1080p's resolution).
- And finally, and this is the caveat I'd be willing to deal with, it could the former while supporting a "party mode", which would essentially send a quarter of that smaller resolution to up to 4 controllers. Basically, you'd get up to 4 players going, but each one would halve a slightly blurry/pixelated image. Think DS games on 3DS. This would be more than acceptable for things like the "dual play screens" for games like Madden. This last one is the most likely to happen.
Anyhoo, that's my 2 cents on the discussion (I re-wrote this once, and it's probably still too convoluted, and I'm not even sure I'm making a point of any kind). What I'm far more interested in discussing is figuring out who's gonna build a Gamecube controller adapter that let's me plug that thing into a Wiimote, in advance of the next Smash Bros. game. Failing that, I'll gladly buy at least one Classic controller merely designed like the Gamecube controller (hear that, Nintendo? That's like new money for old rope! Make it happen!)