Tech Analysis: Nintendo Wii U

Another positive element is that while Nintendo has clearly concentrated on its tablet/pad concept, it recognises that it needs to be competitive with Microsoft and Sony in attracting third party developers to the platform. In partnering with AMD and IBM for the core components in the new machine, it has chosen the best possible collaborators, providing instant parallels to developers who are very familiar with the make-up of the Xbox 360. In going with this approach, game-makers should be able to hit the ground running and won't need extended periods of time to get to grips with a whole new architecture, as was the case with PlayStation 3.

Details are starting to emerge about the Wii U's hardware technology, limited as they are. IBM has confirmed a bespoke, energy efficient CPU design with a significant amount of embedded DRAM, operating on a 45nm process and based on the POWER architecture. The official IBM Watson Twitter account even goes so far as to confirm that the company's latest POWER7 tech forms the basis of the new CPU, but really it could be nothing else since IBM's prior CPU architectures don't actually have onboard DRAM.

IBM has confirmed a bespoke, energy efficient CPU design for Wii U, based on its cutting edge POWER7 architecture.

Current POWER7 chips operate with anything between four to eight cores, each with eight threads and featuring access to 4MB of eDRAM. The more accurate end of the Project Cafe rumour mill suggested a tri-core IBM CPU, which would require a significant redesign of the existing architecture, but this doesn't seem too unlikely: POWER7 is a server-level CPU that's monstrously large compared to the typical console CPU, and with IBM confirming a 45nm fabrication process, a basic quad would simply be too large and too hot for the Wii U's diminutive form factor. Indeed, by our reckoning, it would be larger than the 90nm Xenon core in the launch version of the Xbox 360. We'd expect to see the CPU significantly cut-down from its server-targeted sibling, but even so, we'd still hope for a useful performance bump when compared directly against the 360's processor.

Even less is known about the make-up of the AMD graphics core in the Wii U. The notion of it being sourced from the R700 line of AMD GPUs (a two-year-old design) seems likely, but very little evidence of the tech's capabilities could be discerned from the E3 demos.

The concept demos were effectively Wii-era visuals running at 720p with no anti-aliasing - not exactly a taxing graphical workout. The excellent Zelda HD Experience dazzled with a number of accomplished effects, lots of light sources and some phenomenal art, but was once again running at 720p with no AA. The most impressive tech demo, the Japanese garden, again seemed to be running at the same resolution and for all its grandeur, it suffered from frame-rate issues on some sections and appeared to have some notable texture filtering problems.

Few conclusions can be drawn from these demos of course, but it is notable that elements such as anisotropic filtering and multi-sampling anti-aliasing are hardware features that will have been built into the AMD core - quite why they were not being used is a bit of a mystery. Anti-aliasing in particular would have really made a difference to the clean aesthetic that was on display in the Wii-style demos.

There's been plenty of speculation of late about the raw processing capabilities of the Wii U, with the 50 per cent processing boost claim from the pre-E3 rumours once again getting an airing courtesy of Sterne Agee analyst Arvind Bhatia. Aside from the odd mention of 1080p (not reflected in the demos), Nintendo has made no effort to distance itself from the 3D capabilities of the current generation consoles to the point where it was happy to showcase PS3 and 360 footage of upcoming cross-platform titles to illustrate forthcoming Wii U releases.

Perhaps more telling are the recent comments from id software technical mastermind John Carmack. He pegs Wii U at the same level as the PS3 and 360, and believes there's plenty of mileage left in the current platforms.

"The technology level... brings it up to parity with the other consoles, which is nice for us," he told Gamespot.

"The current platforms are so powerful and so sophisticated. I don't think there's any person anywhere that can really honestly say they know everything about one of these platforms."

Raw specifics on the Wii U GPU remain unclear, but even AMD's entry level enthusiast GPUs - the RV740-based Radeon HD 4750 and 4770 have an embarrassing amount of graphical power compared to the current Microsoft and Sony consoles. However, the cut-down HD 4650/4670 could be repurposed fairly easily into a tight, compact, efficient console GPU that would be a good match for the current generation consoles. By using modern GPU tech, Nintendo also gains other useful, non-rendering advantages. The Avivo video processing technology should give a significantly superior quality level compared to the first generation implementation seen in the 360's Xenos chip, resulting in superior upscaling and downscaling.

The compact form factor of the Wii U also suggests that Nintendo isn't pushing out the boat on an expensive, power hungry architecture. The machine itself isn't very large, and with a 45nm CPU and what we would expect to be a 40nm graphics chip, Nintendo could be in danger of precipitating another RROD disaster if it went for higher-end, performance-centric components. The pre-production unit at E3 looked to be smaller than the current Xbox 360 Slim - and it contained an internal power supply (something Microsoft continues to externalise), another source of additional heat.

Overall impressions of Wii U are positive if not spectacularly overwhelming once you've had the controller in your hands and you've got some idea of the approach Nintendo is going for with its new design. Rendering performance on a par with Xbox 360 and PlayStation 3 is absolutely fine so long as the concepts behind the games produce new and fresh experiences. That's the key challenge going forward, as John Carmack realises.

"I'm kind of excited about the touch-screen aspect on there," he says.

"I think that probably has broader general utility for games than most of the motion control stuff, where you really have to design a game around motion control and you can't just tack it onto a finely crafted FPS. But I think the DS has really shown what the extra little touchscreen can do - almost any game can do something useful with that."