[Numpy-discussion] Enhancements for NumPy's FFTs

[David Cournapeau]

Sturla Molden wrote:
>
> There is a version of fftpack_litemodule.c, fftpack.c and fftpack.h that
> does this attached to ticket #1055. The two important changes are
> releasing the GIL and using npy_intp for 64 bit support.
>   

Would it be possible to make the changes as a patch (svn diff) - this
makes things easier to review.

> Minor changes:
>
> There is a restrict qualifier in fftpack.c. If it is not compiled with
> C99, it tries to use similar GNU or MS extensions.
>
> There is some OpenMP pragmas in ftpack_litemodule.c. If you don't compile
> with OpenMP support, they do nothing. If you do compile with OpenMP, they
> will make certain FFTs run in parallel.  I can comment them out if you
> prefer.
>   

Yes, I would be more comfortable without them (for 1.3). This is
typically the kind of small changes which can be a PITA to deal with
just before a release because it breaks some platforms in non obvious ways.

For the restrict keyword support, I will add a distutils check to avoid
the compiler-specifics (again after 1.3).

cheers,

David