OFF: SETI at home

[Paul Mather]

On Mon, 25 Oct 1999, M Holmes wrote:

=> > > > The quickest one I've seen is some guy at Silicon Graphics who's averaging
=> > > > less than three hours!!!  Eeek!
=> > >
=> > > I just checked the top 100 users, and the fastest is 1hr 54min!!
=> > oops! made a mistake it actually is 1hr 35min!
=> > > I do not know what kind of a machine that's supposed to be.
=>
=> There was a rumour recently that the Israelis had built a quantum
=> machine for code cracking (most doubt it, but it's just about possible
=> with the sort of money spooks supply and a technical breakthrough or
=> two).  I guess that with such a machine, and the right algorithm, they
=> could do this sort of stuff in microseconds.

Quantum computing is useful for certain applications, but is not a
panacea.  From my limited understanding (i.e., some researcher gave a
talk about it here:), it is good for, e.g., problems in NP, or of
exponential complexity---rendering them polynomial time solvable.
Some quantum computing algorithms of polynomial time algorithms have
been proposed, too, e.g., a linear search [O(N)] solvable in
O(\sqrt{N}).  But, even the massive parallelism hinted at by quantum
computing doesn't necessarily mean "microsecond" solutions.  The devil
is in the details (and those scaling constants).

Because the SETI at Home people keep their algorithm secret (to "prevent
spoofing"), it should be fairly easy to determine the fastest
hardware---just look at the (small-ish) set of available binaries, and
find out the most humongous hardware that can run it (e.g., big SGI,
AlphaServer, etc.)

Do they have a parallelised version of the client?  We have an IBM SP2
supercomputer here at VT... >;-)

Cheers,

Paul.

NP: Fairport Convention, _Unhalfbricking_

e-mail: paul at gromit.dlib.vt.edu

"I don't live today; maybe tomorrow..."
        --- James Marshall Hendrix