Quote:
Originally Posted by VBCurtis
I'm not sure why LaurV suggests splitting the range that gives up a bunch of efficiency.

Nope. I mean, yes, if you think about the same sieving depth. But you will sieve different chunks to different depths, and that is where you GET speed. Say you work a single k, and you sieve very large large n to N, with some depth p to P (srXsieve notation). You sieve 100k primes per second and eliminate 5 candidates per hour, but this is misleading because you also eliminate large candidates, for which the LLR test would take hours, but also eliminate small candidates, which would be eliminated faster by LL test. That is why you have to test how the things are for your system. I will try to give a numerical example soon, let me make one first.