sieve completed... 0.000000 user, 0.000000 system 2**3 - 1 is prime 0.000000 user, 0.000000 system 2**5 - 1 is prime 0.000000 user, 0.000000 system 2**7 - 1 is prime 0.000000 user, 0.000000 system 2**13 - 1 is prime 0.000000 user, 0.000000 system 2**17 - 1 is prime 0.000000 user, 0.000000 system 2**19 - 1 is prime 0.000000 user, 0.000000 system 2**31 - 1 is prime 0.000000 user, 0.000000 system 2**61 - 1 is prime 0.000000 user, 0.000000 system 2**89 - 1 is prime 0.000000 user, 0.000000 system 2**107 - 1 is prime 0.000000 user, 0.000000 system 2**127 - 1 is prime 0.000000 user, 0.000000 system 2**521 - 1 is prime 0.012000 user, 0.000000 system 2**607 - 1 is prime 0.020001 user, 0.000000 system 2**1279 - 1 is prime 0.192012 user, 0.000000 system 2**2203 - 1 is prime 1.092068 user, 0.000000 system 2**2281 - 1 is prime 1.240077 user, 0.000000 system 2**3217 - 1 is prime 3.972248 user, 0.000000 system 2**4253 - 1 is prime 10.808675 user, 0.044002 system 2**4423 - 1 is prime 12.324770 user, 0.044002 system 2**9689 - 1 is prime 204.052752 user, 0.712044 system 2**9941 - 1 is prime 225.430088 user, 0.744046 system 2**11213 - 1 is prime 339.633225 user, 1.136071 system 2**19937 - 1 is prime 2574.836917 user, 8.732545 system 2611.599u 8.828s 1:27:19.73 50.0% 0+0k 0+0io 0pf+0w

Addendum: Sorry, I was too tired by the time I finished this last night to really comment. I wrote up a simple C program using the GNU multiprecision arithmetic library that used a simple sieve to find all numbers < 20000 which are prime, and then tested each one using the Lucas-Lehmer test to determine if it was, in fact, actually prime. It spews out the total runtime needed each time it finds one. You can see that it took 2611 seconds (43 minutes or so? my math might be off) of runtime (I was running another number cruncher at the same time, so the elapsed time was twice as long). The highest Mersenne prime I checked (2^19937 -1) was discovered in 1971. That means that in 43 minutes I was able to duplicate every computation on Mersenne primes done throughout human history up to 1971 myself. If I get some time later, I'm going to try to use the numbers generated to provide projections on how long it would take me to compute larger, more absurd primes using only my machine.