IBM Has World's Coolest Chip
Page 1 of 1
IBM today announced a breakthrough development in semiconductor performance today.
Working with the Georgia Institute of Technology's Cryogenic Lab, IBM said it was able to achieve a blazing 500GHz performance, with one caveat; the chip had to be cooled to near-absolute zero degrees temperatures.
"We realize 4.2 degrees Kelvin is not a practical point of operation for a transistor," David Ahlgren, an IBM researcher at the company's Fishkill, N.Y. manufacturing facility, told internetnews.com.
4.2 degrees Kelvin translates to 452.11 Fahrenheit, just a few degrees shy of absolute zero.
But Ahlgren said the point of the research is to test the limits of how fast they could go.
He said IBM actually has run simulations that suggest future versions of the Silicon-Germanium (SiGe) chip could run twice the 500GHz speed, already considered a world record for a Silicon-based processor.
There are faster transistors based on non-Silicon materials such as Gallium Arsenide.
Nearer term, Ahlgren said the current test processor can run at a speedy 350GHz at room temperature. In the digital communications market, where SiGe chips are common, cell phone chips typically operate at a "mere" 2GHz.
Sageza Group analyst Clay Ryder is impressed with the implications of the technology.
"The fact that IBM and Georgia Tech have created such a screamer in the labs with today's technology shows a lot of promise," Ryder told internetnews.com.
"This could be a very important development when you look at the increasing focus on energy savings."
Practical applications for the superfast chip include "automotive radar," 60 GHz wireless LANs and broadband applications.
Automotive radar encompasses such features as adaptive cruise control, where the car senses, for example, how far the car in front of the driver is and adjusts the speed accordingly. Another is crash avoidance, where the car would be able to sense an impending crash and either swerve away, slow down or brake.
"You can't do this kind of technology cheaply enough with conventional semiconductors," said Ahlgren. "The is the first time we think we can make it inexpensive enough for the car makers to implement, down to a few dollars a chip."
Ahlgren said we're a good five years away from high production that would get costs down that low, however.
He also said 60GHz wireless has the most near-term potential, helping companies implement, for example, wireless connections within a building.
"These are fairly short-term signals that only go a few hundred feet so it's not like a cell tower. But that can also be an advantage in a building where you don't want the signal going too far for security purposes."