fast inversion

Torbjörn Granlund tg at
Mon May 18 07:07:21 UTC 2015

bodrato at writes:

  The new code is faster for n==1, slower for 2 <= n <= 4, and faster (more
  than twice) for n >= 16.
Nice speedup!  In mpn/x86_64/fastsse/com.asm we have an mpn_com which
will speed things up another 2x.  It is not enabled on any platforms now
as it needs tweaking for small operands in order to avoid slowdown in a
range.  There is a special code for small operands there already, but it
is not as fast as the current code.

The copyi and copyd code in the same directory is actually being used
for many platforms.  I am afraid it too should get better small-operands

Please encrypt, key id 0xC8601622

More information about the gmp-devel mailing list