Last month I laid out a prognosis for setting up a million-digit Leyland prime search. That endeavour has now started its run!
I sieved my L(999999,10) - L(1000099,10) candidates to 2*10^11 resulting in a 59536-term file. Running the sieve from 10^11 to 2*10^11 was not really necessary. The 12.5 days that it took (on a 10-core machine) netted 1632 composites but a direct primality test would have netted ~50 composites per core in the same amount of time and, at 100 cores, would have resulted in three times the yield. At any rate, the effort was not wasted since nine of my Mac minis are still working on their previous project and are therefore not yet search-ready.
I have now initialized 54 cores on nine different computers to begin the search. In a week I will have added the 54 cores on those nine Mac minis finishing their assignments. So 108 cores on eighteen computers dedicated to the task! I am hoping for completion some time in September. Of course, prime finds (should I be so lucky) could happen at any time.
No comments:
Post a Comment