Actually Berkeley can't handle that kind of bandwidth. They will
probably make us all use caches or increase the Science so the
processing is slowed down again. Now if Berkeley were to get an influx
of cash they could spend it on bandwidth and making units available
for us to process. But at under an hour for each unit that is almost
3 1/2 times as fast as you are doing now! That means that Berkeley has
to be able to provide 3 1/2 times as many units to process as they do
now, and that is just for you! The rest of us should speed up our
processing rate too, so Berkeley will probably make us do more
analysis on each unit in the future. Meaning that we could be
re-analyzing these same units over and over for years to come.
Also meaning that they won't have to provide "more" units just recycle
the old ones so they get processed with the new Science"!

 On Wed, 16 Jul 2003 13:57:49 GMT, "Barbi Satin" <jk11@pa.net> wrote:

"Wally AngleseaT" <wanglese@spbigpondammersareparasites.net.au> wrote in
message news:b2c2hvg5coq2p1micmtqaoseou3d1tgobk@4ax.com...

seti@home has been going for some time now,

I'm just wondering if anyone knows what the average processing time
per unit was back when we began, and if there's a graph of how this
has decreased over time. Purely from an intellectual exercise.

I guess the numbers of users per month has either remained static,
grown or decreased over time, so I'm just interested in if any stat
analysis is worthwhile (outside of the work unit processing of course)
:-))

All I know is that on the first computer I ran Seti on, my average time per

unit was 8 days. It was a Packard Bell with Win 98 and an AMD 333. On my
next computer also with Win 98 and a Pentium III 800 I averaged about 12
hours per unit. Now I run Win XP pro with an AMD Athlon 3000 with the
Bartion Core and average about 3:30 per unit. If chip speeds keep climbing
at this rate I can imagine that in five years a unit may take under 1 hour
to crunch.

-- Have you checked your Smoke Detector...LATELY?