Magma, GMP, and the FFT

Torbjorn Granlund tege at
Tue Dec 6 11:58:32 CET 2005

Ismail Donmez <ismail at> writes:

  Salı 6 Aralık 2005 10:25 tarihinde, Paul Zimmermann şunları yazmıştı: 
  > The Magma developers kindly gave me the opportunity to look at the source
  > code of the Magma FFT. I can therefore say it is a completely independent
  > implementation of Scho"nhage-Strassen's algorithm, and in particular it is
  > not based on my implementation within GMP, nor my implementation is based
  > on the Magma one. :-)

Good and bad.

Good that their code is independent.

Bad that you of all people have now seen "proprietary" FFT code,
since that could open for claims that your future work on FFT
belongs to Magma.

  > I hope the Magma developers will agree to work together with me to
  > understand why their implementation of the same algorithm is faster than
  > mine, which would help to improve GMP, and hopefully Magma too.
  Still their way of advertising this ( "We are based on GMP and faster than 
  it!" ) is not ethical given that GMP is free software.
If they at least said that, it would be better...
Now they're shouting just "we are faster than GMP" from the rooftops,
pretending they are not GMP-based.


More information about the gmp-discuss mailing list