ni...@lysator.liu.se (Niels Möller) writes:
No immediate plans. To me, it seems stable enough, if documented
together with mpn_mulmod_bnm1_itch and mpn_mulmod_bnm1_next_size.
We should integrate the small primes FFT code,
Torbjorn Granlund t...@gmplib.org writes:
We should integrate the small primes FFT code,
https://gmplib.org/repo/gcd-nisse/file/0d591aa7e02c/mpn/generic/mul_spfft.c
in the next major GMP release. Perhaps this would make mpn_mulmod_bnm1
trickier to maintain (at least as a cutting-edge
ok I see we are far from a public mulmod_bnm1 interface. May I ask that in the
meantime, you don't change the internal functions mpn_mul_fft, mpn_fft_best_k
and mpn_fft_next_size, so that we can use them for the wrap-around trick
(for example in GMP-ECM)?
Paul
David Harvey d.har...@unsw.edu.au writes:
The main problem with this approach is that there is not much
granularity in the choice of r.
In my code, I used some huristics like this: Start with r = 5, and
choose the appropriate size k for the splitting into pieces, and
transform size 2^n. Then