Torbjörn Granlund tg at gmplib.org
Sun Aug 25 00:28:08 UTC 2019

Now we have a nice set of x86_64 gcd_22.  The code is not as well tuned
as the gcd_11 code, but it runs somewhat fast.

I haven't explored the table based variant which gives 3 bits of
progress per iteration.  It might make the new code obsolete for
machines with fast multiply.

Now what?  Should we have gcd_33, gcd_44, and gcd_55 also?  :-) (It is
clear that these could improve speed greatly, and with the gcd_22 code
they would not be hard to write these.  Well, gcd_44 and above would not
be able to keep things in the 15 usable registers of x86_64.)

Please encrypt, key id 0xC8601622

More information about the gmp-devel mailing list