The company has disclosed little more than its basic concept. But at least one major telecommunications vendor plans to provide seed money and a dedicated market for the company's idea of creating a chip that morphs more than 1,000 times every second into the exact set of gates needed to process every new instruction passing through it.

"We are changing the face of silicon," said Paul Master, vice president of technology at the San Jose, Calif.-based company. Other companies are working on reconfigurable devices that change infrequently and generally allow one chip design to work in several different types of systems. Master said Quicksilver's vision is to create a chip that is constantly in flux. He said that in order to eke out the most performance from the least power, the concept depends on a core within a system-on-chip design that is dedicated to processing instructions but that changes its structure to process each instruction most efficiently.

With new commands coming through the chip at all times, the core reinvents itself 1,000 or even a million times every second. While this novel approach may seem like science fiction, co-founder, president and chief executive officer Jaime Cummins said the company will have working silicon samples in its office this year.

BellSouth calls
Quicksilver has released few details of the chip's design and has multiple patent applications pending. However, it has received initial funding from BellSouth, which has already placed an order for more than a million units as soon as they are available, Master said. The company already has its second and third customer agreements lined up, he added.

Initially the company will target consumer applications and hopes to provide chips that will allow a single system to perform several tasks. This could include a device that merges the various functions of a cell phone, a pager and a PDA. The chip could allow a single IC to function as the main driver for all those various applications, and because it is designed to process every instruction with the most efficient approach, Master stressed that the chip will be very power-sensitive.

Jim Feldham, president of Semico Research Corp. (Phoenix), was not certain such an approach could succeed, primarily because the Quicksilver chip may turn out to be very difficult to implement. "In theory it sounds like a great idea, and if they can do it, it will be an awesome machine," he said. "But they have a very difficult road ahead of them."