L’ordi de demain sera-t-il biologique?
Pour ses opérations, le bio-ordi, l’ordi du futur
passera par des labyrinthes cellulaires, des voies internes à des cellules. La
biologie réduira son volume à celui d’un portable et, grace à la nanobiochimie,
son activité coûtera peu d’énergie.
Ci-dessous un article tire de http://futurism.com/researchers-found-way-shrink-supercomputer-size-laptop/
Scientists at the University of Lund in
Sweden have found a way to use "biological motors" for parallel
computing. The findings could mean vastly more powerful and energy efficient
computers in a decade's time.
Parallel
Computing
Nanotechnologists at Lund University in
Sweden have discovered a way to miniaturize the processing power that is found
today only in the largest and most unwieldy of supercomputers. Their findings,
which were published in the Proceedings
of the National Academy of Sciences, point the way to a future when our
laptops and other personal, handheld computing devices pack the computational heft
of a Cray Titan or
IBM Blue Gene/Q.
But the solution may be a little surprising.
These computers will not be like our
familiar, electronic variety—they’ll be “biological computers,” machines that
rely on the molecular activities of nature’s true nanomachines: cellular
organelles.
The computing problem that needed to be
solved is an old one, and involves the fact that today’s electronic computers
are woefully inadequate when faced with the task of performing multiple
computations. In other words, calculations can be tackled sequentially, one
after another—a painfully sluggish process that, when faced with mathematical
problems of a high order (as computing more frequently is nowadays), is simply
not up to it.
The solution is “parallel computing,”
essentially slaving processors together to perform multiple calculations
simultaneously, thus cracking the problem through sheer computational brute
force. This way, these so-called combinatorial problems—those that require many
different solutions to be tested—can be resolved at much higher speeds than
today’s electronic computers.
Of course, quantum
computers, with their bizarre, non-binary “qubits,” are favored contenders
to tackle this new frontier of computing; but they’ve proven devilishly hard to
build, require technology that’s still in many ways embryonic, and are probably
many years, if not decades, away from being brought online.
Not so with biological computing. They are,
as the Lund team has shown, easier and less expensive to build (after all, they
utilize a technology evolution has already conveniently provided us), and are
likely only a decade away from production.
One of the researchers, Heiner Linke,
director of NanoLund and a participant in the study, summed it up nicely, “The fact that molecules are very
cheap and that we have now shown the biocomputer’s calculations work leads me
to believe that biocomputers have the prerequisites for practical use within
ten years.”
Schematic
diagram of actual biological computational device. The green channels represent
microtubules used in calculations; insets show scanning electron micrographs of
computational “split-pass” junctions. Credit: PNAS.
Molecular
Motors and Protein Filaments
The problem with using a conventional
computer in a parallel configuration is obvious: scalability and energy use.
Simply put, in order to perform multiple calculations, you’ve got to have
several large, bulky, energy-sucking computers working at the problem. And
there’s nothing miniature and efficient about that.
But the Lund team ingeniously circumvented
this problem by using molecular motors, large intracellular molecules capable
of performing mechanical work within our bodies—in fact, you’re using them
right now to move your eyes and read this story. They’re called myosin, and
they’re found in muscle cells; they operate by guiding protein filaments along
artificial paths.
It was the insight of the nanotechnologists
that this process could be harnessed to execute calculations. “In simple terms,
it involves the building of a labyrinth of nano-based channels that have
specific traffic regulations for protein filaments. The solution in the
labyrinth corresponds to the answer of a mathematical question, and many
molecules can find their way through the labyrinth at the same time,” Linke
explains.
So rather than bulky computers working in
tandem, performing multiple simultaneous computations, you have nano-scale
molecular motors doing the same thing. Which means much smaller, and much more
powerful, computers.
And these biocomputers are extremely
energy-efficient. So efficient that it requires less than one percent of the
energy used by an electronic transistor to execute a single computational step.
So while quantum computing may furnish us
with the hefty supercomputers of the future, used to solve phenomenal
scientific and mathematical problems with almost godlike computing power, our
own personal computers may someday squeeze the might of today’s supercomputers
into a more modest package.
And all this using nothing more extraordinary
than muscle power.
