Windows Mobile in the (Intel) Atom(ic) Age
By Charles Wilde, CTO and Founder of Aton International, Inc.
I think Intel has introduced a real game changer in the ever evolving world of mobile devices.
Even before the stampede to buy the iPhone and the buzz surrounding the Android mobile Linux platform, corporate IT had begun to acknowledge that mobile devices are here to stay. The clamor for better communications began in the executive offices and now reaches down the organization to front line staff.
Enterprise mobility initiatives have, in turn, driven changes by the cell phone carriers to become the chosen player in the enterprise arena. The walled gardens, where the customer could only select the carriers’ proprietary phones and applications, are opening to include outside devices and applications. For example, Verizon Wireless proudly announced its “any phone, any application” initiative last November.
For Microsoft Windows Mobile, the question today is how to retain its lead in the enterprise business market for high end smart phones. To dramatically boost its count of licenses sold, Microsoft also needs to figure out how to compete with the likes of iPhone in the consumer marketplace.
In some ways, Microsoft Windows Mobile is in the position of many smaller companies that have pioneered a market, only to see the fruits of their labor taken by larger latecomers to the marketplace. The latecomers can build something new and flashy to draw customers from the pioneers, but not be weighed down by compatibility concerns for existing customers.
Reinvent Windows Mobile?
If I were the manager responsible for Windows Mobile, I would be looking for a way to reinvent the brand about now. I think that Intel has offered one of the key items for reinvention with its Atom chip that is both extremely low power and x86 architecture compatible. Intel has also used a strategy that Microsoft could do well to adapt to its own products.
Intel inherited what was to become the XScale ARM CPU series in 1997 through a joint venture with DEC (Digital Equipment Corporation). After acquisition, Intel invested upwards of a billion dollars into the XScale ARM CPU and supporting chip sets.
Although this investment produced a very technically capable and popular CPU for Pocket PC and Smartphone devices, this series was not profitable and was not expected to become profitable for Intel.
In a strategic move, Intel terminated most of its support of the ARM architecture by selling the entire XScale processor line and supporting products to Marvell Technology Group in June, 2006. That seemed like the end of Intel in the mobile device field. It turns out that Intel had other plans.
Intel has now introduced a new CPU line called the Atom, which combines the x86 architecture into a very low power device. It is a brilliant strategic move! The complexity and design costs of the XScale ARM processors were rapidly approaching that of the Pentium. Why support two architectures when you can support just one?
What This Means to Microsoft Windows Mobile
Although Windows Mobile has always been based upon Windows CE, there is no inherent reason for that to be true forever. In the ideal world, close cooperation between the operating system team and the product software team is highly desirable. This works best when both teams are trying to achieve the same goal.
Microsoft brought Windows Mobile and Windows Embedded into the same organization a few short years ago. But each group continued to pursue different goals. Windows Mobile was only one of many Windows Embedded customers.
It appears that proximity did not result in a happy marriage. Now the two groups have diverged again in the Microsoft corporate structure.
The latest versions of Windows Mobile 5.0, 6.0, and 6.1 are all based on the Windows CE 5.0 operating system. The current Windows Mobile has pretty much tapped out the capacity of Windows CE 5.0 in terms of processes and memory management.
The next version of Windows Mobile is expected to leverage Windows CE 6.0 which dramatically expands upon the current limits of Windows CE 5.0. Transiting Windows Mobile from Windows CE 5.0 to Windows CE 6.0 will cause some pain for both the Windows Mobile team and for the third party vendors who have developed the many application programs that are an essential part of the Windows Mobile world.
To get the features the market is now demanding for Windows Mobile requires migrating significant amounts of code from the desktop Windows to Windows CE. As the market demands more and more features currently available on the desktop OS, more of Microsoft’s energy will be consumed in maintaining two parallel operating system universes.
Meanwhile, back on the desktop, we see that the Windows OS group is doing a major hedge trimming operation in preparation for the post Vista world. It would seem that there are at least two initiatives in motion, projects named MinWin and Singularity that seek to dramatically reduce the size of the Windows kernel. One of the goals of these initiatives is to make the OS kernel more granular, and allow it to be tailored to purpose along the lines of Windows XP Embedded.
A possible strategic move for Microsoft would be to migrate Windows Mobile off of Windows CE and onto the new generation desktop Windows kernel. Moving Windows Mobile off the Windows CE base and onto a minimalist subset of the new desktop OS kernel could provide some cost and time to market advantages over incorporating the desktop features into Windows CE.
To compete with the likes of iPhone, Windows Mobile needs to add a huge set of features now available or in development for the desktop OS. The experiments within Microsoft to produce a highly granular OS with a smaller footprint are timely. And, the Intel Atom is ready to offer its low power use advantage.
One Issue Could Derail This Vision
The cost of the Atom chip is significantly higher than the equivalent ARM chip. Intel is currently targeting Mobile Internet Devices (MID) running Vista that command higher selling prices. But, how many people will pay $1000 for the MID?
The real volume opportunity is in the smart phone devices powered by Windows Mobile, the Apple iPhone and others. The pricing, and therefore, success of these devices is highly dependent on component costs, including the CPU. Given the right price point, one can also see Apple shifting the iPhone from the ARM CPU to the Intel Atom to utilize the same architecture as the desktop Mac systems.
The next strategic decision for Intel would be licensing the core Atom IP to other silicon vendors in the same way ARM does with its IP. This would create the eco-system of customized CPU’s that has made the ARM concept so popular. The licensing fees would help drive the costs of the Atom chip ever lower. Ultimately the Intel Atom could displace ARM as the low power architecture of choice for mobile devices.
The excitement and action has moved from the desktop to the mobile device. The possibility is that the Intel / Microsoft duopoly that has ruled the desktop for years, can now be recreated in the mobile space with the Intel Atom and Microsoft Windows Mobile. But, both companies still need to take major steps to get there.
Charles Wilde, CTO and Founder of Aton International, Inc., is a mobile/embedded system veteran with 25+ years of successful application development experience. As time permits, he’ll answer your tough questions about mobile app development or Windows Mobile to help you complete a project successfully.
To contact Charles Wilde, go to the Aton International, Inc website.
© 2008 Charles A. Wilde All Rights Reserved
Charles,
Absolutely agree - the concept of moving Windows Mobile to use the Windows kernel - not Windows CE, hasn’t been discussed nearly enough - surely well worth considering to streamline application development and integration.
I think phone volume would drive down costs on x86 compatible processors like the Atom.
Good article.
cheers,
Peter
Charles:
I am looking for system on a chip capability for my device.
Products from freescale, STMicroelectronics and Toshiba fit the bill and I realized that I could add other functionality to the basic board design at minimal cost.
The ATOM is just a CPU and for this reason, was eliminated from my CPU selection process.
While I agree that an X86 instruction set is desirable from a Windows or Mac OS perspective, Intel’s ATOM is to late to market. As you point out, there is no compelling Windows operating system to run on handhelds. I will stick with proven Wind River Linux and ARM Jazelle model as the OS and development environment.
With billions of ARM chips shipped and featured for about every process controller or consumer electronic device you can think off, Intel will never gain ARM’s economies of scale and diversity of on chip functionality nor the multitude of competing vendors to force the price down.
Apple’s acquisition of PA Semiconductor - an ARM design shop is a strong sign of future of Apple’s platform architectures. The issue for Apple is to decide if it should migrate to ARM Cortex for all products and drop Intel or to continue support legacy Motorola, Intel and Samsung ARM architectures.
The name of the game now is microelectronics whose ever falling cost will make them more and more ubiquitous in the home and offices of the near future. ARM will continue to be richly rewarded for being the chip design of choice.
Regards
Jim Collier
Toronto