|
The Arecibo telescope in
Puerto Rico gathers data consisting of radio signals almost every day. The data
it receives in one day is 35 gigabytes — too large to be sent over the Internet.
Instead, the data is shipped as regular mail to UC Berkeley, where it is divided
up and sent out to everyone with SETI@home client software. Across the internet,
people receive "work-units" — .25 megabyte chunks of data.
A computer will process the
data in its idle time, checking the radio frequencies. Instead of checking a
broadband signal, which would be spread over a large number of frequencies,
the scientists focus on specific frequencies. The scientists are especially
interested in the frequencies with a lot of strength within a narrow bandwidth
— which could suggest an extraterrestrial message.
|
|
Fig. 3: The Arecibo Telescope's control room, where servers are housed and radio signals are monitored. (©2000 SETI@Home)
|
 |
It should take most computers
between 10 and 50 hours to complete one work unit. In that time, the computer
will have done more than 175 billion calculations on the 107-second piece of
data. (Hipschman) Then the work-unit will be sent back to SETI's servers and
the computer will receive a new work-unit.
SETI@home is the most widely-known
distributed computing project, with more than two million users searching for
extraterrestrial life. All of these CPUs have done more than 300,000 years of
work so far, averaging almost a millennium a day. Although there have been a
few false alarms — it turned out that the Arecibo telescope had been dysfunctional
— nothing has yet been found. The SETI researchers realize that the odds of
actually finding life is slim —they believe there is perhaps a 1% chance —
but the users don't seem to mind. In fact, people continue to sign up; every
day, at least a few hundred users register, making this not only the largest
distributed computing project ever, but also, in essence, the world's largest
supercomputer.
continued...
©2000 Team DC (Thinkquest Team C007645). Hosted by ThinkQuest.
|