General mpn_gcd_basecase

Marco Bodrato bodrato at
Sat Aug 31 13:40:57 UTC 2019


Il Ven, 30 Agosto 2019 12:33 pm, Niels Möller ha scritto:
> tg at (Torbjörn Granlund) writes:

>> 2. The idea with m = (CNST_LIMB(1) << cnt - 1) - m is to chop of a bit
>>    also in the top end of U, not just get >= 5 bits of zeros in the low
>>    I actually believe this is a great idea,
>>    but I haven't checked if it is as effective as I think it is...

> Sounds like the opposite of binary euclid.

Sounds more interesting than binary euclid, to me. Because looking at the
lowest bits and just detecting the bit size seems easier than extracting
the highest bits each loop.



More information about the gmp-devel mailing list