At Extro4 at Berkeley in 1999, Gregory Stock did a pitch on
the effect of cash prizes on technology. I mentioned the Electronic
Frontier Foundation prize of 100Kbucks for the first prime
number over 10 million digits. (The 50k$ prize has since been
claimed for the first prime over 1 million digits.)
I just did a calculation that shows that using a state of the art
P4-2GHz, doing Lucas-Lehmer tests, each exponent would
take about 6.5 days and is worth about 37 cents, assuming you
take the (probability of prime) * (100K$). The power cost
at 11 cents a kWh would run you about 2 bucks to check
that exponent. So we still aren't at breakeven for chasing
the EFF prize. But we are getting close. So in another 3 to 5
years, computers will be able to run 24/7 for free.
Perhaps other cash prizes for distributed computing will
be offered that will get us to breakeven before we get
another factor of 5 in computing power. spike
This archive was generated by hypermail 2b30 : Sat May 11 2002 - 17:44:23 MDT