Toom-8 testing (mpz level)

[Torbjorn Granlund]

bodrato at mail.dm.unipi.it writes:

  Dear Paul, Torbjorn and others...
  
  I know that the answer can be much less easy than the question, but..
  
  > please could you update your graphics with the new FFT code at
  > http://www.loria.fr/~kruppaal/mul_fft-4.3.1.1.tgz?
  
  Will this code ever be integrated in GMP?
  
There are legal problems with that code wrt GNU.  Paul, Pierrick, Alex and
myself have code there.  My code is already assigned to the FSF.  Paul
used to have assignments, but revoked them when a few years ago (stating
that it was a reaction to that I released development code under the GPL
(as opposed to the LGPL)).

I have not been able to get assignments for Pierrick's and Alex's work,
despite a number of attempts.

(This is not the right list for discussions about FSF's policy of asking
for legal papers from authors.  If people feel the urge to discuss that
in the context of GMP, please use gmp-discuss at gmplib.org.)

  PS: Toom4.5 is ready(mpn), then I'll work on T6.5; next one will be T8.5.

Toom4.5, that'd be toom54, toom63, and toom72 in GMP speak, i.e, 8 eval
points?  The former two would be awesome!  T6.5 would use 12 and T8.5 16
eval points?

I am curious about your preference of # of evaluation points that are
divisible by 4?  Does that have something to do with interpolation
symmetries...?

-- 
Torbjörn