nisse at lysator.liu.se
Fri Dec 21 18:13:13 UTC 2018
tg at gmplib.org (Torbjörn Granlund) writes:
> I think we should consider carefully whether mini-gmp should be
> generalised to do small limbs. To me, it makes sense to keep to
> mini-gmp's main goal of being small en simple.
Agree, we should not compromise simplicity. To me it might make sense to
1. Add some more complexity in umul_ppmm, which is critical for
performance. Besides small-limb support, maybe use __uint128 where
available (#ifdef __SIZEOF_INT128__)
2. Reduce the number of places that depend on mp_limb_t == unsigned
long, where that also makes code simpler. E.g, using init_set_ui or
roinit_ui in more of the _ui functions.
3. Make it easy to omit remaining the functions/macros depending on
mp_limb_t == unsigned long, to enable some separate wrapping which
defines substitutes and then includes mini-gmp.c.
Niels Möller. PGP-encrypted email is preferred. Keyid 368C6677.
Internet email is subject to wholesale government surveillance.
More information about the gmp-devel