Intel in the palm of your hand

The launch of the Orange San Diego handset represents a major milestone for Intel. Despite ruling the server, desktop and laptop markets for years, if not decades, Intel hasn't made an impact into the handheld device arena - until now. But even with the San Diego out the door, many are asking whether it's already too late for Intel to make an impact on such a fast moving market.

There's no denying that ARM based hardware solutions account for the majority of smartphone and tablet devices on out there, but does that really mean that there isn't space for an Intel solution?

It's worth remembering that Intel offered an ARM based solution itself, having purchased StrongARM from DEC in 1997 and evolving it into the XScale processor line. Intel saw decent uptake of its XScale hardware during the PDA boom of the early 2000s, but it sold XScale to Marvel in 2006, essentially pulling out of the handheld hardware market.

In reality though, the idea that Intel was pulling out of handheld devices was a simplistic view of the company's decision to sell its XScale technology. The real decision that Intel made was to focus entirely on IA technology from top to bottom - or from the server down to the handheld.

In theory, having the same underlying technology in every chip that Intel manufacturers makes perfect sense, but the challenges, as always, were power draw, heat and physical form factor. To address these problems, Intel has been applying the same power efficiency mantra to all its chips, no matter what their application may be. After all, it's just as important to reduce power draw and heat in a server chip, especially when you have data centres filled with thousands of racks stuffed with servers.

And ever since the introduction of Centrino in 2003, Intel has improved the power/performance ratio of its chips, making laptops more capable, while imbuing them with greater battery life. But it was the launch of the Atom processor back in 2008 that proved to be the foundation for Intel's eventual move into the smartphone and tablet markets.

I can still remember being at the Intel Developer Forum in 2007 when Anand Chandrasekher - then Senior Vice President Ultra Mobility Group - was talking about a Mobile Internet Device (MID) revolution based on the Menlow platform, which would incorporate the original 45nm Silverthorne Atom core. But despite Samsung's best efforts, with its Q1 Ultra-Mobile PC (UMPC), that MID revolution never happened, and even Samsung dropped the form factor before Menlow ever appeared.

Of course the real problem with the MID and UMPC concepts was Windows. The real weak point of Samsung's Q1 was that it ran Windows Vista - an operating system that could bring a fully featured laptop to its knees! The result was a plethora of Intel prototype devices wheeled out by Chandrasekher each IDF, with no sign of a production unit in sight.

At 2009's IDF I got my paws on an MID running Moblin on a Moorestown platform, and although it was far from a finished article, it definitely showed some promise. But even while I was playing with that Moorestown device, I couldn't help wondering if I would ever see an Intel smartphone on sale.

I've been waiting for the San Diego to arrive for a very long time, and while it has taken far longer than it should have, it's nonetheless good to see it now.

To get some deeper insight into Intel's mobile platform strategy I met with Intel's Robby Swinnen and Mike Bell yesterday, both of whom were extremely confident of Intel's foray into the mobile arena - as you'd expect them to be.

Swinnen was keen to highlight that the San Diego is just the first step on a long road for Intel. The device itself is a reference design, which Intel developed to speed up the roll out, but handset manufacturers are free to use all, some or none of that reference design in their devices. He said that handsets from both Motorola and ZTE, due in the latter half of the year, will likely be very different from the San Diego.

Neither Swinnen or Bell saw the extensive ARM app support as a problem. With every phone manufacturer developing their own CPU based on ARM's licensed core, there's no guarantee that every app will run on every ARM-based handset anyway - a fact that I can vouch for first hand.

Both agreed that software optimisation is key to getting the best from the Medfield platform, and from future Intel mobile platforms. "To create great devices, it's not just about hardware," commented Bell. Intel is investing a significant amount of resource in software optimisation and development for its mobile platforms.

The software part of the equation goes some way to explain why Intel chose to launch a single core solution into a market stuffed full of dual and quad-core processor phones - "Android doesn't make as effective use of multi-core as it could" noted Bell. Bell said that Intel's significant experience optimising Windows for multi-core environments is helping to optimise Android.

But if Intel is already working to optimise Android to behave more efficiently in a multi-core environment, why is Medfield a single-core solution? Both Bell and Swinnen said that there's a way to go before Android and the apps it runs will make good use of multi-core platforms, but that the optimisation Intel has already done will benefit Medfield due to its Hyper Threading functionality.

So, were all those years of rolling out Moblin based devices a waste of time? Did Intel just switch to an Android platform so it could actually bring a product to market? Bell answered those questions, saying "Android is the incumbent OS, we'd be crazy not to support it." It's hard to argue with that, but Intel is still working with Samsung on Tizen based devices - Tizen being the evolution of MeeGo, which itself evolved from Moblin. No matter how you slice it up though, if Intel hadn't jumped on the Android bandwagon, it would probably still be showcasing prototype devices, albeit on Tizen by now.

What about other operating systems? Bell refused to be drawn on whether we'd see Intel hardware in Apple mobile devices, and even though both he and Swinnen confirmed that Clovertrail is already running in several Windows 8 tablet designs, there was little going on in the Windows Phone sector right now. Bell said that Intel's customers are asking for Android phone support now, but if customers start to ask for Windows Phone 8 support, his team will deliver it.

Although the San Diego might not be the kind of flagship handset that steals the headlines, like the Galaxy S3 or iPhone 4S, it's a great proof of concept for Intel. Performance is impressive, proving Intel's claims about multi-core support, to a degree at least. And the bargain price that Orange has attached to the new handset should see it grabbing a respectable market share.

But the best thing about the San Diego and the Medfield platform is that it merely represents the beginning. Intel has already announced that we'll be seeing an SoC Atom - rolling the compute and communications silicon together - while multi-core versions are as inevitable as a rainy British Summer.

One thing Intel won't be doing is offering a licensing solution like ARM. Both Swinnen and Bell confirmed that Intel will focus on an end-to-end solution, and why not? After all, as Swinnen succinctly put it, "we have Fabs".

With IDF only a few months away, you can expect Intel to be talking up its mobile platform strategy, and confirming its commitment to handheld platforms at its San Francisco expo. That commitment was made pretty clear by one of Bell's closing statements - "we want to own mobile" - and given Intel's history in the server, desktop and laptop spaces, I wouldn't bet against him.

404

Sorry! Page not found.

The article you requested has either been moved or removed from the site.