archiv~1.txt: Re: SETI-L: Everybody may be afraid & a Answer

Re: SETI-L: Everybody may be afraid & a Answer

David Woolley ( david@djwhome.demon.co.uk )
Sat, 27 Mar 1999 09:33:53 +0000 (GMT)

> accumulate 15 terabytes of data. I don't know how many of you are
> computer professionals, or even know that much about computers, but that
> is an extremely ludicrous number. They said all processing is done in

You are assuming general purpose computers, whereas this will be a special
purpose FFT machine.

> realtime. If you can show me a computer that can process 15 terabytes of
> information into complex search algorithms in the short time they spend to

The bulk processing will use simple algorithms and any complex algorithms
will be applied to interesting signals. This is a standard tactic for
efficient searching - reduce the problem by using a simple filter first.

> btw, I've used beowulf clusters on 1terabyte raid systems, and it has
> taken WEEKS to do certain things. I'm sure the seti algorithms are
> equally complex.

I think you are thinking of disk limited systems; the bulk processing will
only be performed on a small window of the data, and as said above,
the bulk of the data will only have simple algorithms applied.

5TB really isn't that difficult to capture. If you are scanning a 500MHz
window, at 4 bits per sample, it only takes about 3 hours of raw data.

Incidentally, SETI League members don't seem to publish raw data either;
you simply get the result of applying an FFT and integrating.

I think it is fairly safe to assume that a respectable organisation
may be economical with the truth (not mention the competition, for example,
or rely on people thinking in terms of general purpose computers) but
would never deliberately lie. Tera Bytes is not an everyday term, so it
also unlikely to have been used by mistake.