Windows for ARM? Maybe, but not in 2011

The Bloomberg news service has suggested that Microsoft will announce a version of Windows for ARM chips at CES 2011 in January. Its sources are people "who asked not to be identified because Microsoft’s plans are confidential.
Written by Jack Schofield, Contributor

The Bloomberg news service has suggested that Microsoft will announce a version of Windows for ARM chips at CES 2011 in January. Its sources are people "who asked not to be identified because Microsoft’s plans are confidential. The software would be tailored for battery-powered devices, such as tablet computers and other handhelds, the people said".

"While other versions of Microsoft software aimed at phones and mobile devices work on ARM chips, this is the first time it will make a full version of Windows available on that technology."

The Wall Street Journal, following up the story, adds the rider: "though it isn't expected to be available for two years".

Of course, Microsoft already has a version of Windows for ARM chips, and it has been used on tablet computers for roughly a decade. That is the Windows CE (Consumer Electronics) version that Microsoft wrote in the 1990s because the chips used in small devices weren't going to run full-blown x86 versions of Windows. CE, a whole new modular real-time operating system, is the basis for PocketPC, Windows Automotive, Windows Mobile, Windows Phone 7 and other varieties. CE has been used in hundreds or perhaps thousands of products from the Dreamcast games console to juke boxes and sewing machines.

Also, the dominant versions of Microsoft Windows in the 1990s -- Windows 3, Windows 95 and Windows 98SE -- were based on MS DOS, which was written for Intel 8086-compatible processors.

However, like CE, the NT (New Technology) version of Windows was not written on or for Intel's x86 platform. When programming started, the prevailing wisdom was that Risc (reduced instruction set computer) chips like Acorn's ARM were going to take over the market, so NT was written with the idea that it would run on multiple Risc chips. To make this simpler, NT was based on a HAL or Hardware Abstraction Layer. At various times, Windows NT has run on Intel i860, MIPS R3000/R4000, DEC Alpha, IBM PowerPC, Intel Itanium and Sun Sparc chips. The pundits (not including me) turned out to be wrong about Risc replacing x86 chips in PC market, but in principle, there's no reason why NT-based versions of Windows (NT, 2000, XP, Vista, Windows 7) shouldn't run on ARM chips.

In fact, because ARM chips have become more powerful while Windows has become more efficient, this could be an attractive combination. The question is whether it's necessary. Intel has tried to make it not necessary by selling off its ARM-based products (StrongARM, XScale) and developing its own successful range of cheap low-power chips: the Intel Atom.

Microsoft did indeed start putting Windows NT on StrongARM chps with its LongARM project, but doing it today might be seen as stabbing Intel in the back.

This is possible.

Only the ignoranti regard Wintel as some sort of joint operation, because the two companies are always trying to insure against the other's failure. For example, Intel puts a lot of money into Linux development and treats Apple as its cutiepie, while Microsoft wrote CE and NT with portability in mind, and goes out of its way to support AMD. Still, both companies know where they make the vast bulk of their profits, and financially, the non-Wintel stuff is just a sideshow.

None the less, there are signs of Microsoft and ARM working together. On July 23, ARM announced Microsoft Licenses ARM Architecture, adding:

"Microsoft is an important member of the ARM ecosystem, and has been for many years," said Mike Muller, CTO ARM. "With this architecture license, Microsoft will be at the forefront of applying and working with ARM technology in concert with a broad range of businesses addressing multiple application areas."

ARM staff have also been telling people (including me) that they would like Windows on ARM, and Warren East, president and CEO of ARM Holdings plc, said as much in a conference call in April 2009. In a story headlined Is Windows 7 launch set to reveal ARM processor support?, EE Times reported:

"East said that 'Microsoft will continue to play an important part in this [netbook] space. If there was Windows support for the ARM processor today clearly it would be a very different marketplace.' He then added: 'Perhaps there will be support in future but that's really for Microsoft to comment on and not for us to comment on, I'm afraid.'"

Will it happen with Windows 8? We'll just have to wait and see....

Real Windows on ARM would certainly be interesting in that it could well expose lots of examples of bad programming. Windows Vista also forced some programmers to improve their shoddy habits, and though Vista suffered for it, it did improve the Windows ecosystem to Windows 7's benefit. The need to run on both x86 and ARM could be a similar wake-up call to many software houses. That would help everyone, not just the people who want ARM-based options.

Editorial standards