Centrino consists largely of three parts: a new microprocessor called the Pentium-M, which is the heart of the project; a companion chipset; and a Wi-Fi module. Software and specialised packaging also are included. The Pentium-M, which comes out of Intel's labs in Israel, seeks to solve one of the historical problems with Pentium notebooks: power consumption. Despite new chip technologies and industrywide efforts in the 1990s to increase overall notebook energy efficiency, growing screen sizes and faster chips wiped out many of the conservation gains. Intel-based notebooks still typically conked out after two to four hours. In late 1998, details began to emerge about Transmeta, a then hypersecretive company that was working on an energy-efficient processor that could run Windows. Although company executives have said that the Pentium-M was independently conceived, Transmeta fuelled the urgency, a number of Intel executives have said. Rather than retrofit the Pentium, the Israeli team scrapped it. The Pentium-M was designed from the ground up to fit into notebooks. Although it runs the same Windows software as Pentiums, the chip is different from an architectural point of view. Among the new architectural features, Micro Ops Fusion will combine routine instructions and tasks and thereby save time and energy. Anand Chandrasekher, general manager of Intel's mobile product division, likened the process to a bunch of people at the airport sharing a cab, rather than taking separate taxis. Advanced Branch Prediction will let the processor better schedule tasks, and different parts of the chip such as the system bus and even the Wi-Fi chips will shut down when not in use to conserve power. Energy-efficiency, though, means lower megahertz. The chip will initially top out at 1.6GHz, far slower than the Pentium 4, which runs at 2.4GHz. The imbalance, some analysts say, could muddle the marketing part for Intel. Demand, finally
Despite all the complaints about battery power over the years, energy-efficient chips such as Transmeta's Crusoe and Intel's low-voltage Pentium III chips haven't sold in large volumes. Interest in these products in many ways has started to accelerate only with the advent of Wi-Fi in the past 18 months. This new interest, though, seems earnest. "In the corporate space it should be a good sell. Corporate buyers care less about frequency and more about cost of ownership and longevity," said Kevin Krewell, senior editor of The Microprocessor Report. "It's also going to be a good sell to people who are second-time notebook users and savvy buyers." Energy-efficiency and Wi-Fi are necessary partners, Krewell emphasised. At a recent Advanced Micro Devices briefing, he noticed that nearly all of the attendees checked their email via Wi-Fi during the break, but they had to huddle around power strips to keep plugged in. "So they weren't quite wireless," he said. The Pentium-M so far is succeeding better than expectations, according to Don MacDonald, director of mobile platforms at Intel. During the development process, Intel expected the chip to mostly go into "thin and light notebooks," which cost anywhere from $1,500 (about £900) to $2,500 and get sold mostly to business customers. Now the company says the chip will become its predominant notebook product by the end of the year and will be found in all types of notebooks and tablet PCs except the budget "desknote" machines. There are more than four times as many Pentium-M designs at this point in the cycle than there are notebooks for the Pentium 4 mobile chip, according to Intel's Chandrasekher. In turn, this could help Intel's bottom line. The price difference between Intel's mobile and desktop chips, usually hundreds of dollars, has been steadily decreasing in the past year. That will likely continue, but increased volumes will more than compensate for the slide. "There is going to be some uptick," Gartner's Margevicius said. "Mobile is something like 20 percent of the market. We believe it will go up to something like 35 percent" over the next few years. A bumpy ride
Building the wireless environment, though, will cost money. According to estimates, Intel will spend around $300m promoting Centrino, close to the cost of the original Pentium campaign. In addition, the company is working with telecommunications carriers, hotel chains and cellular providers to develop networks for "hot spots" -- public places that give people wireless access. Sean Maloney, general manager of Intel's communication group, wouldn't say that the company is partly subsidising these networks, but admitted there is a "thorough and lively" campaign afoot. "You'd be amazed at how open the service providers have been. None of them are in denial," Maloney said. Once the first networks are up, the next phase will involve "getting the owner of 1,000 hot spots in the United Kingdom to sit down with a hot spot provider in France and work out a revenue-sharing agreement" for roaming, he said. But competitors aren't standing still. Atheros, Broadcom, Intersil and other wireless chip companies are fiercely seeking to get their Wi-Fi chips outfitted into notebooks and hot spots. The delay of Intel's own Wi-Fi chip has helped these companies maintain their existing lead: Hewlett-Packard, Dell and IBM will all offer notebooks on Wednesday with Intel and non-Intel solutions. To attract customers, Intel is touting its engineering department. The company has spent thousands of hours testing how the Pentium-M and its approved wireless solutions work with hot spots and third-party software. The company has not tested other wireless products. "We have developed [test and verification] chips whose only aim in life was to torture the rest of the system," said Mooly Eden, general manager of Intel's Israel design center. Whether this succeeds has analysts split. Some say PC makers will want to work with the communications specialists. Others say that a Wi-Fi chip is just a fancy modem, and the only thing that is important after a while is price. Transmeta, meanwhile, is gearing up for Astro, a new energy-efficient chip slated for the second half of the year, while AMD will come out with new notebook chips next week and in September. Intel's recent history of monumental changes to PCs is also uneven. The company tried to make Rambus the standard PC memory. It flopped. Timna, another chip from the Israeli group, never came out. Still, even with the hurdles, the wireless genie seems to be out of the bottle. Like the first PCs, it has become a technology driven by ordinary customers rather than marketing departments. "It is one of those things that sort of happened spontaneously," Krewell said.
See Chips Central for the latest headlines on processors and semiconductors.
To find out more about the computers and hardware that these chips are being used in, see ZDNet UK's Hardware News Section.
Let the Chips Central editor know what you think by email. And sign up for the weekly Chips Central newsletter.
