The power and the glory: scene set for processor power-play

So here's a question. If you were a low-powered device vendor, which processor supplier would you go with today: ARM or Intel? And why?

On one level it's straight-forward. ARM has clout right now in the mobile device business because its processors can run sophisticated programs on sleek devices for 10 hours or more on a single charge. The Apple iPad, Lenovo Skylight smartbook and various Nvidia Tegra-enabled tablets, among other recently announced products, show that global vendors are backing ARM for new product types aimed at the ultra-mobile market.

In contrast, incumbent vendor Intel’s newer Atom processors have yet to show a sensational win even though its technology allows use of x86 programs such as Microsoft Windows. Intel has always assumed that its x86 heritage would give it advantage in the emerging connected device market. Admittedly, it dominates the netbook category that it created, but that is Intel’s only game right now.

The Apple iPad debut, in fact, contrasted sharply Intel’s debut of a forthcoming LG smartphone, called the GW990, which Intel showed off at CES. Remember that device? It was nice looking and worked well but drew immediate criticism as big and clunky and awkward. As the first device to run on Intel’s forthcoming Atom platform for MIDs called Moorestown, this debut failed to inspire. And the GW990’s power consumption and battery life during data usage have not been publicised. You have to wonder about that.

So while power consumption is a fundamental factor companies must consider when selecting a processor for mobile Internet devices, it's the business matters associated with an ARM versus Intel decision that might be crucial - given they underscore several points of concern throughout the industry today.

For example, control is a big issue. ARM is well-known for giving its licensees the freedom to use its architecture to design and manufacture chips for any type of product and manufacturers can work with any licensed chip supplier to develop customised solutions for specific platforms or markets.

“OEMs, carriers and service providers want to be able to differentiate their products and services.

That’s a huge concern,” says Jim McGregor, chief technology strategist at In-Stat, a research firm in the U.S.

Intel does not let licensees build their own chips and requires its OEMs to build products according to certain requirements that Intel defines in order to standardise use of the technology. A manufacturer building a netbook to run on the Intel Atom, for example, must keep the netbook within certain screen size and other configuration parameters that Intel specifies.

Intel’s move last year to let Taiwan Semiconductor Manufacturing Corp. develop customised parts based on the Atom will give manufacturers access to more silicon design options while expanding Intel’s reach into new consumer electronics markets. This is a new business model for Intel and a way of loosening up. Businesses should welcome this.

Processor choice influences innovation. While ARM’s licensing model offers more innovation in silicon, the pervasive use of Intel architecture gives Intel a software innovation advantage. Intel will lose some of that advantage, however, as cloud-based services and business models gain traction and as open development models like the Android OS attract more participants. ARM’s architecture is ideal for many of these newer usage models. While many Internet services are based on x86, which has favoured Intel, the Internet is basically agnostic and doesn’t care what architecture or OS is used.

Ecosystems matter. Nobody can build or sell anything in the computing and communications industry without an ecosystem and this is one of Intel’s strengths. The company is supported by the largest software ecosystem in the world and the most common standards are associated with its technology. Its processors are backwards compatible and have specific instruction sets. These are all reasons to use Intel.

ARM’s ecosystem is broad but more varied because it includes so many different IP vendors, silicon and software firms, each with their own relationships. ARM’s use in smartphones has unleashed an ecosystem of professional and amateur developers for a host of developer environments, from Linux to the iPhone to the Android OS and proprietary ones like BlackBerry and Palm.

Fragmentation can be an issue with ARM-based products. One vendor’s products may not work with another’s and the use of proprietary OSes adds hassles for chip suppliers and developers. Android may lessen this problem a bit as it consolidates efforts previously expended on many OSes into a unified ecosystem.

So it's more than a straight 'my silicon uses less power than yours' tussle here. The next year should show us how important those 'business matters' are.