I've been using next_prime (as part of my work on nth_prime) so I added support for next_prime to tune/speed.
next_prime is pretty slow so I changed the -s to set number of bits instead of number of limbs for this function. This means that the graph is much less sparse, still the results are pretty noisey. Currently It takes ~30 seconds to generate the results for time ./speed -s 32-2000 -t 1 -f 1.04 mpz_nextprime -P nextprime If people don't mind it taking 2-5x longer I'd love to chain several next_prime calls so that the amortized behavior shows through better. Two graphs and code here: https://github.com/sethtroisi/libgmp/pull/6 [image: image.png]
_______________________________________________ gmp-devel mailing list gmp-devel@gmplib.org https://gmplib.org/mailman/listinfo/gmp-devel
