WestKelowna.com - West Kelowna. Right Now!

Canwest News Service

Four guys are standing on a street corner… an American, a Russian, a Chinese man, and an Israeli…. A reporter comes up to the group and says to them: "Excuse me … What's your opinion on the meat shortage?"

The American says: What's a shortage? The Russian says: What's meat? The Chinese man says: What's an opinion? The Israeli says: What's "Excuse me"?

— Mike Leigh, Two Thousand Years

'We did it the Israeli way; we argued our case to death." That's how Shmuel "Mooly" Eden sums up a months-long showdown between senior executives of the high-tech firm that gave Silicon Valley its name and an upstart group of the firm's employees working in an outpost in Haifa. As it turns out, the survival and future prosperity of Intel, the computer-chip manufacturer, would turn on the outcome. But the fierce internecine dispute was about more than just Intel; it would determine whether the ubiquitous laptop computer — so much taken for granted today — would ever exist.

Eden was then a leader of Intel's Israeli operation who had helped build it into the largest private-sector employer in the country, responsible for $1.53-billion in annual exports. The story he tells about Intel in Israel, and Intel's battles with Israel, helps explain the nature of Israel's emergence in the past decade as perhaps the economy with the most potent combination of innovation and entrepreneurship on Earth.

Throughout most of the history of modern computing, the speed of data processing — the amount of time it takes your computer to do anything — was determined by the speed of a chip's transistors. The transistors flipped on and off, and the order in which they did so produced a code, much like letters are used to make words. Together, millions of flips could record and manipulate data in endless ways. The faster the transistors could be made to flip on and off (the transistors' "clock speed"), the more powerful the software they could run, transforming computers from glorified calculators to multimedia entertainment and enterprise machines.

But until the 1970s, computers were used predominantly by rocket scientists and big universities. All that began to change in 1980, when Intel's Haifa team designed the 8088 chip, whose transistors could flip almost 5 million times per second (4.77 megahertz) and were small enough to allow for the creation of computers that would fit in homes and offices. IBM chose Israel's 8088 chip as the brains for its first "personal computer," or PC, launching a new era of computing. It was also a major breakthrough for Intel. According to journalist Michael Malone, "With the IBM contract, Intel won the microprocessor wars."

From then on, computing technology continued to get smaller and faster. By 1986, Intel's only foreign chip factory was producing the 386 chip. Built in Jerusalem, its processing speed was 33 megahertz. Though only a small fraction of today's chip speeds, Intel called it "blazing" — it was almost seven times faster than the 8088. The company was solidly on the path imagined by one of its founders, Gordon Moore, who had predicted that the industry would shrink transistors to half their size every 18 to 24 months, roughly doubling a chip's processing speed. This constant halving was dubbed Moore's Law, and the chip industry was built around this challenge to deliver faster and faster chips. IBM, Wall Street and the business press all caught on, too. Clock speed and size were how they measured the value of new chips.

This was proceeding well until about 2000, when another factor came into the mix: power. Chips were getting smaller and faster, just as Moore had predicted. But as they did, they also used more power and generated more heat. Chips overheating would soon become a critical problem. The obvious solution was a fan, but in the case of laptops, the fan needed to cool the chips would be much too big to fit inside. Industry experts dubbed this dead end the "power wall."

Intel's Israeli team was the first group within the company to see this coming. Many late nights at Intel's Haifa facility were dedicated to hot coffee, cold takeout and ad hoc brainstorming sessions about how to get around the power wall. The Israeli team was more focused than any other on what the industry called "mobility" — designing chips for laptop computers and, eventually, for all sorts of mobile devices. Noticing this tendency, Intel put the Israeli branch in charge of building mobility chips for the whole company.

Even given this responsibility, Israelis still resisted fitting into the Intel mainstream. "The development group in Israel, even before it was tasked as the mobility group, pushed ideas for mobility that went against the common wisdom at Intel," explained another former chief of Intel Israel, David "Dadi" Perlmutter, who had begun designing chips at Intel Israel in 1980. One of these unconventional ideas was a way to get around the power wall, and it came from Rony Friedman, one of Intel Israel's top engineers at the time. Just for fun, he had been tinkering with a way to produce low-power chips, which went blatantly against the prevailing orthodoxy that the only way to make chips faster was to pack in more transistors, making them more "powerful." This, he thought, was a bit like making cars go faster by revving their engines harder. There was definitely a connection between the speed of the engine and the speed of the car, but at some point the engine would go too fast, get too hot and the car would have to slow down.

Friedman and the Israeli team realized that the solution to the problem was something like a gear system in a car: If you could change gears, you could run the engine more slowly while still making the car go faster. In a chip, this was accomplished differently, by splitting the instructions fed into the chip. But the effect was similar: The transistors in Intel Israel's low-power chips did not need to flip on and off as fast, yet in a process analogous to shifting a car into high gear, they were able to run software faster.

When Intel's Israel team euphorically introduced its innovation to headquarters in Santa Clara, the engineers thought their bosses would be thrilled. What could be better than a car that goes faster without overheating? Yet what the Israeli team saw as an asset — that the engine turned more slowly — headquarters saw as a big problem. After all, the entire industry measured the power of chips by how fast the engine turned: clock speed.

It did not matter that Israeli chips ran software faster. The computer's engine — composed of its chip's transistors — wasn't turning on and off fast enough. Wall Street analysts would opine on the attractiveness (or unattractiveness)

of Intel's stock based on performance along a parameter that said "Faster clock speed: Buy. Slower clock speed: Sell." Trying to persuade the industry and the press that this concept was obsolete was impossible. This was especially the case in this instance because Intel had itself created — through Moore's Law — the industry's Pavlovian attachment to clock speed. It was tantamount to trying to convince Ford to abandon its quest for more horsepower or telling Tiffany's that carat size does not matter.

"We weren't in the mainstream–clock speed was king, and we were on the outside," Israel's Rony Friedman said.

The head of Intel's chip division, Paul Otellini, tried to mothball the project. The clock-speed doctrine was enshrined among Intel's brass, and its members weren't about to hold a seminar to decide whether to change it. But a seminar, it turned out, was exactly what Intel needed.

— – This article is adapted from Dan Senor and Paul Singer's new book, Start-Up Nation: The Story of Israel's Economic Miracle, just published by McClelland & Stewart.

———

Monday

Changing Intel's mind

This entry was posted on Saturday, November 28th, 2009 at 6:50 am and is filed under National Business. You can follow any responses to this entry through the RSS 2.0 feed. You can skip to the end and leave a response. Pinging is currently not allowed.