I think the breakingoffornot depends on the timescale. Splitting the range will speed up the sieving at the moment. However, in the long run overall sieving time would be shorter when sieving longer ranges. At least that's what I think.
Anyway, one core freed up and I ran a sr1sieve for 2.7e6 < n < 4.1e6 and 314e12 < P < 325e12. Estimation was 24 factors in about 3 days or about 11000 s per candidate removed. To confirm, I ran it for 7 hours, about 10%, and found 4 factors, i.e. 6300 s per candidate removed.
LLR2 takes 2500 s for n = 2.7e6, so I stopped the sieve and continued LLR. It seems for this nrange sieving doesn't make sense anymore. Of course I could extend the sieve, but for now I want to finish n < 4.1e6 and then see how and if I'll continue.
Last fiddled with by bur on 20210218 at 06:59
