Engineers at a startup in Austin, Texas, say the ultrafast processor they've developed is really cool. Cool as in running at much lower temperatures than today's fastest chips. And cool as in being manufactured with essentially the same conventional methods that until now have resulted in chips only a quarter as fast as the prototype produced by Intrinsity Inc.
The new chip, boasting a clock speed of 2.2 GHz, is made through the automatic generation of dynamic logic circuits that are as powerful as conventional static circuits. The process, called Fast14 (the atomic number of silicon) Technology, enables Intrinsity to come up with new chips quickly, because dynamic circuits allow easier data flow. But it still takes intensive circuit design and layout to create the processors, and regardless of the circuit choice, there's difficulty lining up logic with gates.
The Intrinsity processor uses a clocking scheme to reduce the heat it generates. High-speed chips frequently burn more power than the batteries embedded in devices can handle. Intrinsity says its chip won't burn more power, because it uses between 5 and 15 watts. "The other guys will burn 80 to 120 watts of power," CEO Paul Nixon says.
The "other guys" include Intel, which last month announced its own 2.2-GHz chip. But the manufacturing process Intel uses is very expensive and specialized.
Intrinsity processors could start showing up in products within 18 months, Nixon says. He expects them to be used in systems embedded in cell-phone towers, digital video, and medical-imaging machines. One day, the chips could even let laptops stay cool while on laps.
Intrinsity has a breakthrough on paper, says industry analyst Will Strauss at Forward Concepts--but he warns that the company must focus on ease of use and universal acceptability. "Other movements in chip technology took five years," Strauss says. "Even if it's universal in scope, it could be five years out before the market embraces Fast14."