Centrino's Dadi discusses offspring

INTEL'S HAIFA FACILITY site sits barely 200 meters from the glittering Mediterranean Sea in the shadow of Mount Carmel, on the outskirts of the northern Israeli town. The three-hour drive from Jerusalem has been cut dramatically by the new north-south toll highway that briefly skirts the controversial separation barrier, designed, Israeli premier Ariel Sharon insists, to reduce terror attacks from within the Palestinian West Bank.

The Development Centre in Haifa is the largest such Intel facility outside the United States. Some 1,800 people are employed here, many of whom are highly-trained scientists and engineers beavering away on Intel's next-generation "mobile technology".

In charge of the facility is David 'Dadi' Perlmutter, a youthful 50-year old, recently promoted to an international role as corporate vice-president, the first employee based outside the US to attain such a high rank. He now reports directly to CEO Craig Barrett and president and chief operating officer Paul Otellini.

Permutter's promotion is widely seen as recognition of the success of the Centrino mobile platform which was developed by his team in Haifa.

At the heart of Centrino is the radical low-power chip, the Pentium M, designed in Haifa under the code-name Banias. Banias is a remarkable chip in a number of respects. The engineers threw out the rule-book in developing the processor, taking power consumption as the key determinant at every step of the design and engineering process. In Intel, such a radical departure from the norm is termed 'a right-hand turn'.

Dadi, shaven-headed and wearing Levis, tells us he and his team recognised that the growing market for mobile PCs would demand something new. Processing power as expressed by clock-frequency, he says, was "ultimately easy to understand. It's an easy way to market a complex product: higher frequency means higher power." But in a notebook there are other considerations, battery life being primary amongst these.

"We approached it from a different way. There are multiple ways of getting to a target. We chose a new way." As he suggests, in such a highly-charged, creative environment "ideas here aren't a problem." The problem, from a management point of view, is not generating ideas, "it's filtering them".

Dadi describes how one particular engineer had an idea he wouldn't let go of. He declines to name the individual, keeping him under wraps "for later", but it seems his role was key to the whole Banias project. In the end, it seems the engineer was given his head of steam and the whole corporation eventually went with him.

"Within the chip there's the critical path," Dadi explains. "You can tune the critical path but you can't neglect it. But what is not in the critical path you can leave. It is drawing power that is not needed.

Extending such thinking across the whole of the design eventually allowed Intel to deliver a chip capable of doubling the sort of battery life you can expect to demand from a non-Centrino notebook.

But, as an engineer - Perlmutter led the team that developed the xi387 maths co-processor for the 80386 chip and himself holds patents on that design, as well as a prize awarded for his efforts by the Israeli president in 1987 - Dadi brings, he says, a "holistic approach" to product development. "You develop a product, not a technology," he says. "We didn't deliver just a microprocessor, we delivered a whole platform solution."

Indeed, what was notable about the very loud noise Intel made when delivering the Centrino platform was the way the chipmaker shyed away from the announcement of 'just another processor'.

Banias was a whole different story. Despite the sexy technology and its wholly innovative design, it was given the decidedly unsexy label of Pentium M and slotted quietly into a pre-existing processor family. Fundamental to the marchitectural spin Intel placed on the "platform" enabled by the processor, was the emphasis on wirelessness - portability, freedom from battery constraints, freedom from wires. A whole solution. A holistic approach, in fact.

Whether the Perlmutter magic persuaded Intel's heavyweights to follow his approach or whether the approach was dictated by the bosses in Santa Clara is unclear. Still, the combination of the two management teams succeeded in delivering the Centrino-branded concept. The concept, of course, is backed by solid technology, at the heart of which is the Banias Pentium M chip.

Dadi has been rewarded with a promotion and has been lauded, in some insider circles, as a future Intel CEO. He greets the suggestion with a pleased, faux humility: there are still "hurdles to pass along the way," he says.

Banias' successor, codenamed Dothan, appeared on Intel roadmaps scheduled for a launch last November. The launch date came and went with Dothan still under wraps, and Intel remained mute on the subject.

Tomorrow, we'll discuss what went wrong and what you can expect from this update to the wonder chip once it appears - as Dadi promised us - in "Q2". µ