Your browser is not supported. Please update it.

14 September 2005

Intel power per watt campaign to result in handtop breakthrough

In the past two years, Intel’s Developer Forums have become primarily showcases for its wireless and mobile developments, and the event two weeks ago in San Francisco was no exception. Initially, wireless was the means to encourage increased uptake of conventional laptop PCs incorporating Intel’s premium chips. Now, the challenge is to prolong the life of the Intel architecture by evolving it into new devices.

Eventually, there will be a single handheld device that combines wireless IP communications for voice, data and video; mass storage; and acts as a credit card, home automation controller, music and video player and telephone directory. The big question is whether the dominant converged devices spring from a cellphone or PC heritage and, while Intel and Microsoft are both making a play for the former market, their real opportunities lie in creating a PC-oriented IP platform that can incorporate telephony capabilities too.

One step on this road for Intel was detailed by Paul Otellini, leading his first Developer Forum as CEO, and is dubbed the ‘handtop’. This is superficially a PDA, but promises to deliver the full PC experience ‘ including running the full upcoming version of Windows, Vista (formerly Longhorn) in a truly mobile device, something that will only be enabled by significant advances in battery technology that Intel claims it will achieve within the next two years.

Such a product would boast the performance of a laptop and would feature Wi-Fi and WiMAX connections, multi-gigabyte hard drive and optional cellular links.

It would act as a hub for communications and multimedia activity and a personal data store, and would use the forthcoming Wireless USB short range wireless standard to link at high speed to peripherals such as PC screens, televisions and larger storage drives to improve the user experience.

This is how the PC concept will survive, and Intel has pursued the idea for years, though with less technical detail than it provided this time. One of the key barriers to such ideas in the past has always been battery technology, with handheld devices needing to operate at very low power and therefore having problems supporting fully fledged operating environments. Important in the advances Intel claims to be making in this respect is a new alliance with Matsushita Battery Industrial to develop technology that will provide a 30% gain in Centrino battery life in 2006, with the goal to achieve eight-hour battery lifespan for a mobile device running full Windows in 2008.

It was interesting that Otellini was still focusing heavily on Windows. While some commentators took the talk about Vista as a snub to Microsoft, because it effectively makes Windows Mobile redundant, we believe the software giant should be encouraged by the continuing public loyalty to the Wintel combination. After all, the ‘handtop’ or micro-PC concept relies heavily on Intel technologies, but does not require Windows. As Nokia’s recent launch of an internet tablet, a Wi-Fi (but not cellular) enabled handheld running Linux and Java, it showed there is now a choice of operating systems for the new breed of miniaturized notebooks.

Nokia’s product can be seen as an early step towards a handtop-type future, and the Finnish giant is also promising major battery breakthroughs ‘ though with far more reservations than Intel ‘ that could enable it to pack a full PC functionality on to its future devices.

Such Linux/Java products, heavily backed by the likes of IBM, are threatening to Microsoft. The loss of Windows Mobile is not, since it was only created to get round the limitations of the mobile platform and would be easily sacrificed should it become possible to put the whole Vista on to a tiny device.

This raises a further issue ‘ in a world where battery and chipsize advances allow a fully fledged PC to be shrunk down into a palm-sized device with none of the compromises normally associated with mobiles, specific mobile operating systems become redundant and, as Otellini pointed out, a single OS can be run on all types of platform (this would also be a bonus for Java, whose desktop implementation has many functions that are currently denied to the mobile variant for power reasons). Extending that argument, what becomes of architectures that were specifically designed for the mobile market, notably cellphones?

This is the real daring at the heart of Intel’s handtop roadmap. If it can shrink Centrino and either desktop Windows or desktop Linux into a cellphone-sized package with an adaptive radio to allow connections to many wireless options and a large disk drive, and can make that device run on battery for five hours or more, what is left for the architectures of Texas Instruments/Nokia and Qualcomm, especially when the specialized cellular networks are giving way to all-IP systems?

Like Microsoft, Intel has had limited success in attacking the cellphone market itself and breaking into the territory of TI and others. But a company of this scale has to play a long game and we have always regarded the XScale cellular architecture primarily as a means to gain contacts and expertise in the mobile market in preparation for unleashing a fully converged chip platform for the multi-radio, all-IP world ‘ a battle where Intel believes it can outperform TI by effectively bringing the PC on to the wireless network and into the pocket.

Otellini touted “performance per watt” as one of the core slogans for the Intel strategy for mid-decade, with the ultimate goal a 0.5W processor with sufficient horsepower to run Vista by 2009.

To get there, Intel will first merge its NetBurst and Pentium M architectures into the ‘next generation power optimized microarchitecture’, explained Otellini, creating a single development platform on all devices including handhelds, centered on the key Intel technologies – x86, 64-bit, virtualization, active management and trusted computing.

As baby steps towards the full low power vision, the second half of 2006 will see the Conroe processor for desktops and the Merom for notebooks, both 65 nanometer dual-core parts with the latter rated at 5W. Merom promises three times the performance per watt of the current Centrino. Raw performance gains will rely on multiple cores, an area where Intel is only a recent convert and where IBM has amassed a huge lead. Otellini said processors with 10 cores or more are currently in development so that, by 2009, a tenfold reduction in power consumption will be balanced by a tenfold increase in performance from multiple processing cores.

An initial handtop is expected from OEMs in the first half of 2006. This will not be equipped to run Vista, at least not until 2008, but will offer significantly better performance and power ratings than current handheld platforms, pledges Intel, and will support WiMAX in 2007.