In Theory: What's Inside Project Cafe?
Digital Foundry analyses the rumours and puts forward informed suggestions
Six weeks away from E3 and we now know that Nintendo will announce its high-definition successor to the Wii at the Los Angeles event. It's a new piece of hardware that the rumour mill tells us will not only offer a superior spec to the current crop of HD consoles, but will also change the way we play our games. Based on comments in Nintendo's recent announcement, the new hardware could be with us within a year.
We've been aware of various "Wii HD" rumours for a while now, and have some perspective on the various leaks that have appeared just recently. Photos of what appear to be Project Cafe presentation slides, stealthily snapped using a mobile telephone, have been hotly debated on the internet. Are they genuine? Are they fake? What of the increasing torrent of rumours describing power significantly in excess of the Xbox 360 and PlayStation 3?
One thing that appears to be self-evident is that, once again, the controller is the core of the next Nintendo experience. The Cafe hardware itself appears to operate either as a conventional console tethered to a TV, or else as a lounge-based hub, beaming out gameplay to gamers effectively playing on handheld units with short-range links to the unit. Almost like an entirely home-based OnLive, if you will.
No longer will there be arguments over who's using the HDTV. If someone wants to watch TV, conceivably the player could simply switch over to gaming on Cafe's controller, with rumours suggesting a six-inch widescreen display with resolution in the EDTV-esque 800x480 area - the kind of display utilised by larger sat-navs. It's been described as "an iPad with buttons", but a large joypad with integrated touch-screen is more in line with what we've been hearing from our sources.
Obviously there are other benefits to this configuration. Streaming could be used to send player-specific information, or used to de-clutter the main display by moving HUD information across to the satellite screen. However, applications in this scenario would probably be quite limited. Most HUD info typically needs to be easily accessible to the player, and shifting focus between two screens is not a user-friendly experience.
Cafe's video streaming is a potentially very cool technology. The question is, how could it work and what transmission format could be used?
On the one hand, we already have a working example: PS3's Remote Play, where the framebuffer is shrunk down, encoded and beamed over WiFi to PSP. In theory, then, gameplay could be transmitted to anywhere in the house, and a Wireless-N connection could conceivably handle the bandwidth.
However, two issues here suggest this isn't such a good idea: latency would be introduced by encoding and decoding the video signal, plus there would be a significant degradation in image quality brought about by image compression. Bearing in mind how many Nintendo titles operate at 60Hz, gameplay would suffer a drop in quality were the platform holder to adopt this strategy. Nintendo wants technology to enhance the way we play games, not to detract from it.
The other solution would be something along the lines of WirelessHD, which offers a spec level essentially identical to HDMI beamed out from the console with enough bandwidth to support 1080p at 60Hz (and presumably, by extension, four remote screens on Cafe controllers). We've already seen this deployed on the latest iteration of Alienware's M17x laptop and it's said to offer zero latency and a ten-metre range. Nintendo is all about refactoring existing technology in an innovative fashion, and here we have a template that could be relatively easily adapted.
This is pretty advanced tech though and in concert with integrated touch-screens, we can expect additional Cafe controllers to be significantly more expensive than what we're used to - and we'd be hugely surprised if the unit shipped with more than one pad/tablet.
However, bearing in mind that at one point Orange was selling an Android-powered mobile phone with an 800x480 AMOLED touch-screen display for less than £100 (the Orange San Francisco/ZTE Blade), you can imagine that, with additional economies of scale, Nintendo should be able to make these far less complex controllers fairly cheaply, and still factor in a healthy profit margin.
Of course, a third option is that Cafe controllers could be wired, tethered to the console. Unwieldy and a retrograde step in many ways, but it would resolve the streaming issue in a cost-effective manner and it would also take care of another issue: running a touch-screen display on a wireless controller is going to require battery power way in excess of the requirements of the current-generation joypads and wands.
So what of the core hardware make-up of the machine? Rumours we have not been able to independently confirm are suggesting a triple-core PowerPC processor similar to the Xbox 360's, though we would like to believe that, in the five years since Xenon first shipped, refinements have been made to the design that would increase processing efficiency even if clock speed remains in the 3.2GHz territory.
Going for higher clockspeeds is an option, but would increase cooling costs and could impact on the unit's reliability. A smarter CPU design simply makes more sense, especially bearing in mind how key elements of Xenon's design are inefficient or lacking (onboard cache utilisation for example).
There's also talk of an AMD GPU from the R700 family, which unfortunately tells us very little about what we can expect from Cafe's graphics. However, it does fit in with Nintendo's modus operandi in utilising cheap, existing hardware, most likely with console-specific refinements.
In terms of available power, R700 scales up from anything from a low-power Radeon HD 4350 with 80 streaming processing units all the way to the HD 4890 that utilises 10 times as many. Optimistic speculation zeroes in on the HD 4770 as the most likely candidate for inclusion in Cafe, but our sources suggest that the overall profile of the GPU is much closer to the Xbox 360 - and we've even heard that in some applications it may even operate at a deficit compared to Xenos.
While this may come as a disappointment to many, it's important to remember that Nintendo has achieved incredible success in recent years through a combination of factors: new concepts that appeal to a mainstream audience, a price point that's "right" straight from launch, per-unit profitability from day one and not getting involved in spec wars with rival manufacturers. Wii - and perhaps Project Cafe - aren't consoles with 10-year lifecycles in the way that Xbox 360 and PS3 need to be in order to recoup their costs.
Over and above that, we need to understand that Nintendo's primary focus isn't the core gamer audience. Let's assume for the moment that the Cafe GPU is indeed 50 per cent faster than the Xenos chip in the 360 and identical in every other way, perhaps with an eDRAM boost. In a typical third-party game, what would that 50 per cent actually translate into? Would a mainstream audience be impressed with a higher frame-rate, less aggressive LODs, higher texture detail or even 1080p resolution? If you can't achieve a true generational leap in graphical fidelity that would be noticeable and immediately appealing to a mainstream audience, targeting a new controller concept makes a lot more sense.
That leads us on quite nicely to the business perspective. There's a set unit cost Nintendo needs to stick to and what is obvious is that the inclusion of just one joypad/tablet-style hybrid controller is going to take a significant slice of the available budget, cutting down the available cash for the CPU and graphics tech.
While the platform holder could gamble on game and peripheral sales to offset a loss made on each console, it's just not the way that Nintendo has historically operated and it's difficult to imagine anything different here, especially when you factor in an economic climate unlikely to improve that much in the next couple of years.
So what of the rumours of backwards compatibility? Assuming provision for buttons, d-pad and analogue sticks, the new controllers would effectively negate the need for supporting the old GameCube pad socket in the way that Wii did. There's nothing to stop Nintendo using Bluetooth for controller input/output traffic (in concert with WirelessHD or something similar for the visuals) so that would allow for Wii peripherals to work just fine. Nintendo could simply provide a socket on the Cafe unit itself for the Wii sensor bar, but for all we know the Cafe unit itself could ship with its own sensor bar that interacts with the new controllers as well as the old.
Technologically, if Cafe is indeed using an AMD graphics chip and PowerPC CPU, there should be enough commonality with the old Wii and GameCube designs to ensure decent backwards compatibility, though some of the more bizarro elements of the original hardware spec may introduce issues. With over 86 million Wiis shipped, we should assume that backwards compatibility will be of concern to Nintendo as a cost-effective means of giving the new system a headstart of sorts.
What's almost as interesting as the rumours themselves right now are the topics not being addressed by the speculation. For example, how much RAM does Project Cafe have? With the 3DS packing 512MB, we would expect more: at least 1GB. And what about onboard storage: iOS-style flash RAM, SD card support or onboard HDD? Assuming an optical drive, is it DVD or Blu-ray based? Can movies be streamed across to the controllers too?
Then there's the issue of support services - none of which have factored into the current rumour-mongering. From an infrastructure point of view, Nintendo is woefully equipped in terms of a viable online networking presence and is generations behind Xbox Live. Elements in 3DS such as StreetPass suggest that the platform holder has been giving some serious thought to innovative uses of networking, but concepts such as friends codes and the lack of voice chat severely limit the possibilities of the service and make Nintendo look limited and backward in comparison.
We would be surprised and delighted by a technological leap ahead of the current generation, but our gut feeling is that the true innovation surrounds the controller and how the platform holder will make use of it. Interesting times ahead for Nintendo, then. Roll on E3.