3 ms·

Given that the finder of a new prime gets a $3000 award, can someone do the math and see if running Prime95 is actually better than playing the lottery?

by speps 8y ago

schoen 8y ago
I think there's a simple calculation to estimate that the expected financial value of running Prime95 (just for this particular prize) is negative. Suppose GIMPS as a whole discovers one new Mersenne prime per year and pays its discoverer $3000. According to https://www.electricchoice.com/electricity-prices-by-state/ https://www.electricchoice.com/electricity-prices-by-state/, the average U.S. residental customer pays 13.11 cents per kilowatt-hour for electricity. Suppose that you wanted to run a desktop computer with Prime95, hypothetically consuming 80 watts (which is probably low). You have: 13.11 cent/kWh You want: dollar/(80 watt*year) * 91.935844 So it's going to cost about $91 to run that computer for a year, which means that your net earnings are negative in expectation if your computational contribution to GIMPS from that machine is less than than 3000/91 or about 1/33 of the network's total computational power. There are lots of ways to make this estimate better, for example by finding the actual fraction of computational capacity that an individual machine contributes, or the actual power consumption of that machine, or the actual cost of electricity for that machine's owner, or a better estimate than "one per year" of the GIMPS discovery rate (for example, their current press release says that recently they've found about three times as many primes as they expected).
- mhh__ 8y ago
  So you're saying I need to use someone else's computer?
  - schoen 8y ago
    This is a real phenomenon that people used to get in trouble for back before the days of cryptocurrency mining. For the EFF cooperative computing awards, we have a provision about this: > EFF reserves the right to require a more detailed disclosure. For example, EFF may require evidence that the hardware was used legally, with the owner's permission and support. https://www.eff.org/awards/coop/rules https://www.eff.org/awards/coop/rules A particularly notable case where someone was accused of unauthorized computer use for running a distributed computing client back in 2001: https://www.extremetech.com/extreme/49898-georgia-cracks-down-on-distributed-computing https://www.extremetech.com/extreme/49898-georgia-cracks-dow...

Given that the finder of a new prime gets a $3000 award, can someone do the math and see if running Prime95 is actually better than playing the lottery?

by speps 8y ago

Given that the finder of a new prime gets a $3000 award, can someone do the math and see if running Prime95 is actually better than playing the lottery?

schoen 8y ago

I think there's a simple calculation to estimate that the expected financial value of running Prime95 (just for this particular prize) is negative. Suppose GIMPS as a whole discovers one new Mersenne prime per year and pays its discoverer $3000. According to https://www.electricchoice.com/electricity-prices-by-state/ https://www.electricchoice.com/electricity-prices-by-state/, the average U.S. residental customer pays 13.11 cents per kilowatt-hour for electricity. Suppose that you wanted to run a desktop computer with Prime95, hypothetically consuming 80 watts (which is probably low). You have: 13.11 cent/kWh You want: dollar/(80 watt*year) * 91.935844 So it's going to cost about $91 to run that computer for a year, which means that your net earnings are negative in expectation if your computational contribution to GIMPS from that machine is less than than 3000/91 or about 1/33 of the network's total computational power. There are lots of ways to make this estimate better, for example by finding the actual fraction of computational capacity that an individual machine contributes, or the actual power consumption of that machine, or the actual cost of electricity for that machine's owner, or a better estimate than "one per year" of the GIMPS discovery rate (for example, their current press release says that recently they've found about three times as many primes as they expected).

mhh__ 8y ago
So you're saying I need to use someone else's computer?
- schoen 8y ago
  This is a real phenomenon that people used to get in trouble for back before the days of cryptocurrency mining. For the EFF cooperative computing awards, we have a provision about this: > EFF reserves the right to require a more detailed disclosure. For example, EFF may require evidence that the hardware was used legally, with the owner's permission and support. https://www.eff.org/awards/coop/rules https://www.eff.org/awards/coop/rules A particularly notable case where someone was accused of unauthorized computer use for running a distributed computing client back in 2001: https://www.extremetech.com/extreme/49898-georgia-cracks-down-on-distributed-computing https://www.extremetech.com/extreme/49898-georgia-cracks-dow...