BANGALORE: Chinese legend tells of a monkey king who had a golden stick — a powerful tool that could be scaled up or down, from a tiny needle to a giant pillar to hold up the sky.
“Thanks to highly scalable chip architecture you can all be monkey kings, creating applications for tiny connection devices — and for the world’s largest super computers... from milliwatts to peta flops,” says Intel’s senior vice-president (digital enterprise) Pat Gelsinger.
He was speaking to 5,000 assembled software and hardware designers at the Intel Developer Forum in Shanghai recently — so the analogy went down rather well.
During a halt in Bangalore on his way back to the U.S., Mr. Gelsinger, in a briefing, explained how infinite scalability had helped the microchip leader to create two drastically different processors: The Atom is the world’s smallest computer processor family.
The size of a baby’s fingernail, the chip in its tiniest version packs in 47 million transistors and consumes around 600 milliwatts, that is just over half a watt of power — but it has the muscle to run tomorrow’s Mobile Internet Devices (MIDs), small handheld computers that will access the Net and provide a condensed keyboard to do basic e-mailing, letter writing or calculations.
Slightly larger versions will soon come to India, fuelling the next line of Classmate PCs for home and school that HCL launched in India two months ago.
At the other end of the computing scale is the Tukwila, the new Intel chip which has the highest number of transistors ever put on a slab of silicon — two billion. With four cores on board, the chip will “virtually” do the work of four computers.
The world’s biggest microchip is, unlike the Atom, no milliwatt machine: it gobbles 130 to 170 watts of power. But it will allow scientists to build giant “number crunchers” even more powerful than the Tata-CRL “Eka” supercomputer built in India and ticking away in Pune — the world’s fourth-fastest computing machine.
Fuelled by Tukwila, supercomputers will become peta flop machines, Mr. Gelsinger predicted: a peta flop is a million times billion operations a second.
In less than a decade such machines will create a complete genetic simulation of a human cell, he said.
“Doctors will precisely simulate an unhealthy cell in a human and prescribe the exact medication that is needed,” he predicted.
Holding up a sample of the Tukwila wafer (the chip should reach Indian personal computer-makers by June), Mr. Gelsinger was nostalgic: “When Intel launched the world’s first microchip in 1971 it had just 2,300 transistors. I was part of the team that created the Intel 386 processor in 1985: it had 275,000 transistors... and now in my hand, I can hold this two billion-transistor monster.”
He did not have the Atom chip with him. It was too tiny to hold in its unpackaged form.
But in his briefcase he carries a replica of the Monkey King’s Golden Stick. Like Intel’s newest chips, it can, in your mind’s eye, seem very big — or very small.