Quote:
Originally Posted by rogue
IIRC, it sieves to 1e6, but also eliminates numbers with algebraic factorizations. The difficulty is a rough guess to the number of tests needed at 1e12. The reason it sieves to 1e6 is because many of the k are larger than 1e5. It isn't perfect, but hopefully more accurate.

That's sieving Pdepth. I think he wants to know what nrange that it sieves. In other words, does it sieve n=100001110000 like for Nash weight? I'm not sure about that. Do you remember?
On the Pdepth, it does sieve to P=1e6 and then uses a multiplier to estimate a sieve to P=1e12...and to me it's more than a "rough guess" on the number of tests at P=1e12. I've found it to be very accurate.