Primes a^8*2^8m+1 = (a*2^m)^8+1 for odd prime powers a < 72, m < 42,500: (9*2^m)^8+1 : m=5,22,224,281,2492,3239,10940,13833,21060,45120* (17*2^m)^8+1 : m=8,12,15,25,105,191,260,393,466,682,743,990,8219,13450 (19*2^m)^8+1 : m=3,54,736,2574,10440,15597 (25*2^m)^8+1 : m=5,7,36,158,912,1893,2456,7377,20061,47004 (43*2^m)^8+1 : m=19,22,28,36,37,43,99,175,303,427,536,592,4352,6268,13021,24615 (47*2^m)^8+1 : m=8,13,16,94,1750,5035,6457,13273,19855,28284 (49*2^m)^8+1 : m=3,13,17,35,93,365,675,1031,2025,2207,2991,3657,17295 (53*2^m)^8+1 : m=10,45,71,124,244,255,429,469,2004,2979,14914,37463 (59*2^m)^8+1 : m=1,10,52,87,211,345,762,2337,18244 (67*2^m)^8+1 : m=32,174,292,344,1957,2919,15900,19816,21560,31672,38633 In standard base 2 format (n=8m, n < 340,000): 9^8*2^n+1 : n=40,176,1792,2248,19936,25912,87520,110664,168480,360960* 17^8*2^n+1 : n=64,96,120,200,840,1528,2080,3144,3728,5456,5944,7920,65752, 107600 19^8*2^n+1 : n=24,432,5888,20592,83520,124776 25^8*2^n+1 : n=40,56,288,1264,7296,15144,19648,59016,160488,376032 43^8*2^n+1 : n=152,176,224,288,296,344,792,1400,2424,3416,4288,4736,34816, 50144,104168,196920 47^8*2^n+1 : n=64,104,128,752,14000,40280,51656,106184,158840,226272 49^8*2^n+1 : n=24,104,136,280,744,2920,5400,8248,16200,17656,23928,29256, 138360 53^8*2^n+1 : n=80,360,568,992,1952,2040,3432,3752,16032,23832,119312,299704 59^8*2^n+1 : n=8,80,416,696,1688,2760,6096,18696,145952 67^8*2^n+1 : n=256,1392,2336,2752,15656,23352,127200,158528,172480,253376, 309064 *This prime is from the subsequence (3*2^m)^16+1 = 3^16*2^n+1 which has been tested to higher levels. (n > 1,000,000) The siever needs a^8 written out in full: 9^8 = 43046721 17^8 = 6975757441 19^8 = 16983563041 25^8 = 152587890625 43^8 = 11688200277601 47^8 = 23811286661761 49^8 = 33232930569601 53^8 = 62259690411361 59^8 = 146830437604321 67^8 = 406067677556641 Recent primes: 8 Jan 2007 62259690411361*2^299704+1 is prime! 11 Feb 2007 406067677556641*2^309064+1 is prime! 24 Mar 2007 152587890625*2^376032+1 is prime! The choice of a < 72, exponent=8 allows efficient sieving (all factors are of the form 16k+1), but a^8 is not so large that LLR has to use a slow generic FFT. (Using exponent=16 would reduce the choices for a to 3 or 5). n=1,000,000 takes about 10,000 P4 GHz seconds to test with LLR 3.7.1. Sieving speed for all 10 sequences 200,000 < n < 1,000,000 is about 1.35 million p per P3 GHz second. (At p=2.5 trillion with srsieve 0.5.2).