Maybe your factor base is just too big. More primes -> more trial division.
My siqs implementation wants 913 relations normally. In this case, it uses 0.436 sec sieving and 0.19 sec doing trial division.
If I force it to use a factor base of size 1826 (x2 bigger), this becomes 0.27 sec sieving and 0.61 sec doing trial division, for a net slowdown.
- ben
|