GMP 4.4 remaining tasks

Torbjorn Granlund tg at
Mon Dec 7 18:32:40 CET 2009

David Harvey <dmharvey at> writes:

  About a month ago I wrote some approximate reciprocal code using the
  middle product. It's well-tested, and it *should* be pretty fast (at
  least on K8 where good middle product basecase assembly code is
  available), but I have not profiled it yet. I was planning to post it
  once I had some performance figures, but since this has come up now
  on gmp-devel I thought I would post it now:
  Unfortunately I do not have time right now to work on this.
This is too late for GMP 4.4, I am afraid, in particular considering
paperwork issues.

I hope that you will have time to take care of paperwork and that
somebody has time to integrate it before GMP 4.5.

For GMP 4.4 we need mpn_invert code that uses FFT wraparound.  If I
understand it correctly, your mulmid code will help in a (large) operand
size interval before FFT kicks in.


More information about the gmp-devel mailing list