Challenges include figuring out who's going to need a supercomputer in their living rooms.

Antone Gonsalves, Contributor

February 16, 2007

2 Min Read

Intel's experimental 80-core microprocessor is a marvel in processing power and low energy consumption, but a number of hurdles stand between it and everyday use.

The Teraflop Research Chip, unveiled last week, is the size of a fingertip. It can crunch a trillion calculations a second, a feat that 11 years ago required a 2,500-square-foot supercomputer. The chip's size and the fact that it uses about the same power as a two-core processor mean that it could provide unprecedented processing power.


Teraflop Research Chip -- Eighty cores on the tip of your finger


Eighty cores on the tip of your finger

The explosive growth of multicore chips with low power consumption will drive the mass adoption of virtualized computing environments. Massive scale multicore processors will get people thinking more broadly about what servers and operating systems can do to drive business applications.

Closer to home, such muscle would let video games look like TV shows and computers recognize and catalog photos. But beyond gaming and specialty applications, the usefulness of a supercomputer in the home isn't clear.

It will take Intel at least five years to move to production, time enough to identify other applications and overcome several challenges. The biggest of those is divvying up applications to make efficient use of all 80 processors, something that's much easier with today's two- and four-core designs. The process is similar to what's currently done with multicore chips, but for 80 cores, it requires a dramatic change in architecture and more intelligence in the operating system and the software that runs on top of the operating system.

New application development tools will be needed, too. With a processor this complex, code won't be written, but more likely generated from models "It's a whole different ball game," says In-Stat analyst Jim McGregor. "Think of the human brain. We can't think of anything simultaneously, and yet we're building a chip that can."

There probably won't be 80 cores within five years, McGregor adds. "It's a very feasible timetable, but not at 80 cores. You could, however, see chips jump from four to 20 or 40."

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like


More Insights