Microsoft Can Soon Take Your ARM

And finally, some news worth blogging about. On the same day, Nvidia announced a new ARM based processor and Microsoft announced that the next version of Windows will support the ARM architecture. What an amazing coincidence?

It's fair to assume that there has been some collusion here between Microsoft and Nvidia. Nvidia has been selling the Tesla supercomputer systems for a long time and it's reasonable to assume that this latest ARM processor is a natural extension of that direction. So it seems less than likely that the two companies have been working together for long and more likely that this is the direction which Nvidia was heading, and Microsoft has jumped on board later into the game.

There's been much speculation about the purpose of the new chip but it does seem (re: CES announcement) to be targeted more at the desktop / small office server market. There's no question mark as to whether Nvidia are capable of building a graphics chip nor computer chip set. The only question mark over this project is whether Nvidia are capable of effectively integrating ARM cores into their GPU's? Given Nvidia's track record, it seems highly unlikely that this announcement would have been made if the product wasn't close to market.

So how effective will a multi-core ARM CPU integrated with multiple Nvidia GPU's be? That remains to be seen, but if the performance / power consumption of the Tegra CPU doesn't resemble the Tesla supercomputer, I'll eat my shorts.

Intel only just released it's own decent integrated GPU solution (after a decade of dragging their heals) and it makes one wonder if the decision was based on some insider knowledge somehow leaking from Nvidia's offices? Once again, the timing appears a little too co-incidental.. Intel managing to make it's Sandy Bridge announcement a mere 2 days before Nvidia pulls it's new processor out of the bag and more notably, 2 days before CES 2011.

No, there's no doubt in my mind that the Nvidia processor is going to knock people's socks off. A lot of writers are talking about this chip like it's going to be the chip to integrate into people's television sets or integrated media box solutions. This may well be the case, but my gut is telling me that this historical view of how ARM chips are used is looking like poorly thought out reactivity. Tegra is clearly the successor to Tesla, of that there's little doubt. Microsoft working to support the Tegra platform is a clear indication that there's a wider market here, certainly for small business servers but more than likely desktop and portable systems.

And this is a particular clincher for Microsoft and may well turn out to be the savior of the monopoly. Once these 'Tegra Inside' systems start selling through storefront providers, software manufacturers will need to port their software to a processor agnostic architecture. There are many, many packages built using the .NET platform and similar to Microsoft Office, these will be fairly simple to port to this new architecture. If the price / performance entry point of the Tegra is competitive, then some of the business market will also be interested in using the platform. This combined with the HPC market and enthusiast / power user market will likely see a largely commoditized processor / platform / chipset available at increasingly competitive pricing levels.. hence, used widely. How does this save Microsoft? I'd imagine that the ARM version of Windows won't support legacy API's, hence Microsoft can finally secure their operating environment by ditching the age old legacy support chain dangling around their neck. And if Windows 8 does support the legacy API's then adapting applications to the architecture will mostly require a simple re-compilation.

It looks like Nvidia is seriously in a position to challenge Intel and they certainly seem to be making a run at it. Even a total failure of Nvidia to enter the desktop / SME market will leave the new integrated core as (most likely) the undisputed HPC king and likely champion of integrated media chip deployment. The move is far more risky for Microsoft as porting and supporting ARM Windows is undoubtedly no small task. Microsoft does however have strong relationships with computer suppliers and if committed to the challenge, then the architecture migration is likely to succeed.

Most importantly for us people out in user land, we're going to have much faster, more capable, more powerful commodity equipment available to us and improved competition in the processor / board market. If computer hardware gets much cheaper, computers will start getting thrown in for free when you purchase your copy of Windows.