Geoff,
A quick question for you. If you consider 4*5^n+1. All n's left after sieving till 3 are multiples of 2 ie. all numbers left are x^2+1. Note that numbers of the form x^(2^y)+1 are generalized fermats and have factors of the form K*2^(Y+1)+1
So for 4*5^n+1 you only have to consider p=4*K+1. This makes sieving twice as fast compared to sieving for all primes.
I am currently working on 625. only numbers of the form 625*2^(4*n) are left. So all factors are of the form p=8K+1. I was wondering is it possible to make the srsieve only consider these special factors and hence make the sieve faster than newpgen.
Thank you in advance for the answer.
|