Mar 17, 2008 4:36 pm US/Pacific
Don't Waste That Downtime!
John DeFore
We all know computers consume a lot of electricity (particularly older ones,
which can rival electric heaters as juice-hogs) and should be shut down when not
in use. After years of pressure, manufacturers are competing to work better
power management systems into PCs, and a small market has developed for devices,
like smartly designed power strips, that help users make sure gear they think
they've shut down has actually stopped drawing power.
But what about those hours of the day when computers need to be running but
aren't actively in use? As we all learn in school when we write our first term
papers, much "work time" is spent staring at a blank screen. Isn't it a shame we
waste so much energy keeping our processors humming while we wait for
inspiration?
Actually, all those billions of wasted processor cycles could be put to good
use. While your computer waits for you to formulate an eloquent thesis sentence
or to receive an important email, it could be busy solving some of the world's
great problems, from mathematical imponderables to global epidemics and the
diseases that afflict us.
That's thanks to a brilliant breed of screensavers built around the principle
of "distributed computing," also called "volunteer computing."
With distributed computing, projects involving enormously complex sets of
calculations are taken away from the room-filling supercomputers that once
worked on them, broken into manageable little chunks, and sent to thousands or
millions of personal computers around the world. When a participant stops using
her computer for a while, instead of seeing the swimming fish or psychedelic
fractals of a traditional screensaver, she watches as her processor becomes one
cog in a vast, amorphous machine — a "cloud" of computing power, as some
describe it.
The granddaddy of these applications is
SETI@Home, which
uses participants' computing time to look for aliens. Data gathered by banks of
radio telescopes is parceled out to individual users, whose CPUs sift through it
in search of patterns that might indicate transmissions from intelligent life
forms on other planets.
A bit of the program's documentation explains the vastness of its computing
needs: "More computing power enables searches to cover greater frequency ranges
with more sensitivity. Radio SETI, therefore, has an insatiable appetite for
computing power." It also describes the inadequacy of once-cutting-edge
supercomputers to this task: "The most powerful computer, IBM's ASCI White, is
rated at 12 TeraFLOPS [a measurement of mathematical operations per second] and
costs $110 million. SETI@home currently gets about 15 TeraFLOPs and has cost
$500K so far."
SETI@home was born in the late Nineties, and was quickly adopted by young
techies for whom finding extraterrestrial life sounded impossibly cool. A decade
later, those who find this quest far-fetched or frivolous have options. A wide
array of distributed computing projects have been launched (see some lists
here and
here), tackling everything from
chess problems to
scientific issues of immense real-world importance.
Of particular interest to Green Right Now readers is a
collaborative effort
directed at "ensemble forecasting" of the Earth's climate, which runs through
endless models visualizing climate change, each based on slightly different
parameters. The goal is to sort useful models from those that don't hold up
against known data, offering policy makers the best possible guess on issues
that are inherently speculative.
An umbrella site called
AFRICA@home was established to focus on various issues facing
that continent. Their current project,
MalariaControl.net studies the various ways malaria can be
transmitted in order to devise the best strategies to control it.
Fighting disease, in fact, appears to be the most popular use of distributed
computing. A surprising number of projects (like
Predictor and
Rosetta) have been
launched to analyze the way proteins within the body "fold," a process that,
when it goes awry, causes diseases from cancer to Mad Cow Disease.
A newcomer to this arena might ask why so many protein-folding projects
exist. If the power of distributed computing lies in its ability to cobble
near-infinite resources toward a single academic goal, why don't the teams join
forces? Prof. Vijay S. Pande, director of the
Folding@home
project, explains. Of the three leading protein-centered projects, he says,
"Rosetta and Predictor are the most similar" in that "they try to predict the
final structure" of a protein once it has been folded.
"In my opinion," Pande continues, "that's still something best done
experimentally" — in the physical world, that is, outside of computer models.
"Folding@home is different in that it's studying the
process of folding
and misfolding" — not trying to model the end result — "especially as it's
relevant to diseases such as Alzheimer's." Since different kinds of mis-folding
lead to different diseases, Pande's team has narrowed its focus to the mis-folds
that cause Alzheimer's, interested not simply in understanding the disease "but
even in coming up with small-molecule drugs to fight it."
That explains the multiplicity of protein-studying teams. But once computer
owners have decided which project they want to support, a nagging environmental
question lingers: Won't making your computer do more work use more
electricity?
Well, yes. David Anderson, director of
BOINC — the
software platform, built at the University of California, Berkeley, upon which
most distributed computing projects rely — acknowledges that "relative to being
idle, it's an increase of about 50 watts." While Anderson offers one
circumstance in which there's no environmental impact ("For people living in
cold places who use electric heat, volunteer computing can be energy-neutral,
since their PC is acting as a heater."), the bottom line is that there is a
cost. The Folding team estimates that for a participant paying California
utility rates "the difference between an idled computer and one running F@H
would be close to $4 a month - and if the computer was already being used 8
hours a day, it would be closer to $3 a month."
Individuals will have to decide for themselves which makes more sense for the
planet: minimizing the energy used by their computer down-time but accepting
that it is 100% waste, or allowing their computers to use a little more power in
pursuit of answers to the most pressing questions facing us today. One thing is
certain: Using a computer's downtime to draw geometric eye candy or to present
slideshows of cute animals isn't the answer.
- For more information on how to volunteer your computer for humanitarian or
science-related projects see the listings of current projects on the BOINC website
which doesn't endorse particular programs, but maintains links to ongoing work
by academic and industry-based groups such as IBM's World Community
Grid. which supports research on AIDS, cancer, Muscular Dystrophy and Dengue
Fever.
(Copyright © 2009 | Distributed by Noofangle Media)
Click to enlarge
Stumble It!
Comments