In Theory: How Microsoft could build a new 360 for 2010

So the cat's out of the bag or thereabouts. Xbox bosses will wish he hadn't, but Microsoft CEO Steve Ballmer says that a "new Xbox" bundled with, or even integrating, the Project Natal 3D camera will launch in 2010, just days after the company distanced itself from talk of a new machine. Bearing in mind this is the same firm that refused to confirm the existence of the Xbox 360 Arcade SKU even when photos showing the packaging in actual shops appeared online, this is clearly something of an unplanned turn of events at best, a marketing disaster at worst.

The news first emerged when 1UP carried a story claiming the existence of new Xbox hardware in its most recent podcast, tantalising its audience by promising that along with bundled Project Natal, the console itself would be an enhanced, more powerful but still backwards-compatible version of the current 360 architecture, something I was quick to dismiss in a follow-up blog post on the Digital Foundry channel.

But does the 1UP story have more substance than we give it credit for, bearing in mind the timing of Ballmer's bombshell just days later? Is there really no smoke without fire? Bearing in mind that the Xbox 360 will be a veteran in console terms by the time Natal comes to market, is there any argument or indeed precedent for boosting the specs of the machine? Traditionally, Sony has introduced the replacement to its current PlayStation five years after launch, so why not Microsoft too? More to the point, is there any evidence in development circles that hints that the new Xbox will be anything other than a mid-life revision of the existing hardware?

Such revisions are, of course, nothing new. The Atari VCS was remodelled into the Atari 2600 primarily for marketing reasons, and a 2700 version with wireless controllers was even developed (but never released owing to its comedic garage-opening side effects). Moving forward a generation or two, the SEGA Mega Drive went through a massive array of changes, both for cost-cutting reasons (Mega Drive II) and for additional flexibility and functionality, culminating in its ultimate evolution as both a handheld (Nomad) and a gloriously gonzo console-cum-portable CD player (Multimega).

More recently, with the PlayStation and PlayStation 2, we've seen a bewildering range of product revisions, culminating with slim versions made possible by integrating custom silicon spread across multiple chips into single dies, resulting in smaller, cooler machines - but all retaining the basic functionality. Moving into the here and now, the PS3 has been subject to cost-cutting within months of its launch - its integrated PS2 chipset cut down then excised completely for the sake of the bottom line, its hard disk capacity regularly changing over the months depending on the best deal Sony could get on the components.

Most of these precedents suggest that our original thinking with regards the 'new' Xbox 360 are right on the money. Industry pundits' predictions on the evolution of the Xbox 360 hardware have been unerringly accurate, from the Xenon shift to the HDMI-equipped Zephyr, through to the Falcon/Opus and the current, more reliable Jasper. Expected next is the so-called Valhalla chip, to be inserted into a so-far unnamed motherboard. Similar to the slim PS2's merging of Graphics Synthesiser and Emotion Engine, Valhalla sees the GPU and CPU integrated into a single, more power-efficient processor. Originally slated for 2009, Valhalla may well have been pushed back to 2010, thus making its debut in a Natal-equipped "mini" console a distinct possibility. However, bearing in mind that 360 has been revised every year so far, sometimes more than once, it may well be the case that the integrated chip will still launch this year, while a smaller, cooler version based on the 45nm fabrication process will debut in the Natal box in 2010.

But standing out like a sore thumb in terms of established precedents is the mighty Sony PSP. In the transition from the original PSP "Phat" to the slimline PSP-2000, the machine doubled its internal RAM from 32MB to 64MB and added a TV output, while still being cheaper overall to manufacture. Around the same time, Sony also unlocked the full power of the PSP CPU, allowing developers to run it at 333MHz, up from the basic 222MHz stock setting. Indeed, the internal PS1 emulator, running entirely in software, dynamically switches between the speed settings as and when needed. Through a combination of both unlocking the existing hardware and indeed adding to it, today's PSP is a clear step beyond the launch configuration. However, unless you like to use Skype on your handheld, or you're using hacked apps and emulators, it's unlikely you'll notice.

Can we expect anything similar from the Natal revision of the Xbox 360? 1UP seems to think so. It makes the case that Nintendo Wii is essentially a hardware revision of the GameCube: an upgraded, overclocked piece of existing tech designed offer a leap over what has come before while maintaining complete compatibility with the existing library. So, why not 360? It is an intriguing thought, and one that is almost certain to factor into the design of the next Xbox and PlayStation whenever they do actually make an appearance, simply based on the success of the Wii model - a triumph of innovation over raw gaming power.

And there is a shred of evidence that some kind of expansion may be in the offing. It's a matter of fact that Microsoft will soon be shipping 1GB-equipped development kits to studios which will replace the existing 512MB model. While it has hitherto been accepted that this is to aid the development process, it means that the motherboard chipsets (presumably based on the existing Jasper model) do now have compatibility for higher-density RAM chips.

The original 1UP report also suggests that Xbox Natal will be a relatively modest increase in available power, and short of completely engineering and fabricating new silicon, a PSP-style overclock is the best way - probably the only way - to achieve this. Smaller manufacturing processes on the key chips have been used by the PC enthusiast community for years in ekeing out more performance from their gaming hardware. It could work here. To maintain compatibility, the chips could simply be set to the old clockspeed or the new one in software, similar to the way PSP works. In fact, even without changing the speed of the hardware, just a small increase in the size of the onboard 10MB of eDRAM available to the Xenos GPU could also see a big performance boost in new games. Developers would no longer need to "tile" graphics data in and out of the GPU, speeding up the process of effects like anti-aliasing immeasurably. Games that lag due to excessive use of alpha textures (explosions, smoke and the like) would gain a noticeable performance advantage.