Atoms
Offer Giant Leap In Computing Power
By Matthew
Herper
08.29.00
The strange science of quantum
mechanics promises a revolution in computer power as
radical as that following the invention of the
transistor in the 1950s.
That invention sent computer
performance on a rocket-like trajectory upward, its
ascent governed by what became popularly known as
Moore's Law, after Intel founder Gordon Moore who
coined it. It states that computers tend to double
their speed every 18 months.
Computers get faster as the
transistors on microchips get smaller and closer
together. In about 20 years, transistors will shrink
down to the size of atoms. When that happens, the laws
of physics will demand another revolution. This is
when the laws of quantum mechanics, a branch of
physics that describes the interaction of objects
smaller than molecules, wreak havoc on the way chips
are designed.
But as quantum mechanics shuts the
door on digital computers, it may also lead us to the
next revolution. Devices known as quantum computers
show amazing promise. Although they are decades from
being practical, according to researchers, quantum
computers may one day add previously unimagined
computing power to such complex tasks as
database-searching and breaking secret codes, tasks
that now require powerful supercomputers.
Peter Shor, a scientist at AT&T
(nyse: T),
ignited interest in quantum computers in 1994 when he
proved they could be used to crack the codes that make
computer connections secure. They could also allow the
creation of new encryptions methods that would be, in
theory, unbreakable.
In
one step a quantum computer can do what a
conventional computer needs several to accomplish. A
digital computer is really a series of switches, or
"bits," that can be set to two positions, on
or off, 0 or 1, true or false. By lining up thousands
of these bits, a computer can perform complicated
mathematical calculations, and even use that ability
to do tasks like downloading a Web page and displaying
it on a screen.
A quantum computer uses quantum
bits, known as "qubits," which are also
on-and-off switches. But qubits can exist in several
states at once, rather than one at a time. This allows
a quantum computer to perform a single mathematical
function on many numbers at once, answering many
different versions of one question at the same time.
"In a very naive way, the
advantage of quantum computing is that by being a
little bit of zero and a little bit of one, you can do
two calculations at once," says Raymond Laflamme,
a researcher at the Los Alamos National Laboratory in
New Mexico.
This ability would allow a quantum
computer to do the mathematical heavy-lifting required
for, say, cracking secret codes, while even the
fastest and most advanced supercomputers plod along on
the same problem. A quantum computer might need a
month to break the code that would take a
supercomputer billions of years.
But quantum computing has a long way
to go before that happens. Two weeks ago a team led by
Isaac Chuang at IBM (nyse: IBM)
disclosed that it had built a 5-qubit quantum computer
and used it to solve a relatively simple mathematical
problem. Five qubits is less computing power than a
typical handheld calculator. But with more qubits
comes more power. Just simulating the abilities of a
40-qubit quantum computer would use all the resources
of any of the most advanced supercomputers in the
world.
"It is true that any task a
classical computer can do, a quantum computer can
do," says Jeff Kimble, a professor of physics at
the California Institute of Technology in Pasadena.
"But there's a whole lot of difficult technical
overhead for quantum computers."
Quantum computers will not be able
to supplant conventional computers until they reach a
size of at least 100 qubits, and it would take at
least several thousand qubits for them to be really
useful for such heavy-duty tasks as code-breaking.
Getting from five qubits to 1,000
isn't going to be easy. Quantum computers are nothing
like the boxes of electronics we're used to. The IBM
team's quantum computer was a straw-thin vial filled
with a rust-colored liquid packed with millions of
copies of a single organic molecule. Five fluorine
atoms on the molecule were used as qubits. Each atom
could be in one of two quantum states, which served as
the 1 and 0. The qubits were "programmed"
using radio pulses, and the results of the
calculations read using nuclear magnetic resonance
imaging equipment, the same kind of machine a medical
doctor might use to look inside at an injured head.
Adding another qubit to such a
computer requires designing a completely new molecule
to meet very rigorous criteria. Essentially it
requires building a whole new computer from the ground
up.
And quantum computing is so new that
scientists can only begin to imagine what it might one
day be able to do, says John Preskill, a physicist at
Caltech. "One of the things we don't understand
very well from a theoretical standpoint is what they
are good at," he says.
What is understood is that there is
enormous potential behind quantum computers, he says.
"This computational power is woven into the
fabric of nature."
Yet despite its immense
capabilities, practical quantum computing is still a
long way from reality, says Steven Koonin, provost at
Caltech and a quantum-computing researcher.
"I'm not about to sell my stock
in the semiconductor manufacturers on this," he
says.
™ © 2000
Forbes.com Terms,
Conditions and Notices
Chart: Quantum
Speed
Related Links:
Forbes.com
Future Technology
