[SciPy-user] fftw3 wrappers

David Cournapeau david@ar.media.kyoto-u.ac...
Wed Aug 8 01:42:35 CDT 2007

Lars Friedrich wrote:
> Hi David,
> thank you for your answer.
> David Cournapeau wrote:
>> Mmh, sorry, I missed that you were interested in multi dimensional fft, 
>> this may explain the result. For 1d, fftw3 wrappers are almost always 
>> faster than numpy, but not for multi dimensional
> Maybe I just did not tell before... I thought this would not make any 
> difference.
It shouldn't, but the current implementation is suboptimal for multi 
dimensional fft (please note that the original implementation of fft in 
scipy makes it possible to use not less than 5 different fft backend, 
which is already quite an achievement; I don't want to sound like I am 
bashing anyone work, just that you can expect better performances in a 
near future).
>> What does scipy.show_config() tells you ?
> ...
> fftw2_info:
> fftw3_info:
Well, that means that fftw3 was not found, or not used when it was 
packaged :)

Which platform are you on ? Did you install fftw3 (as well as 
development package on linux) ?

More information about the SciPy-user mailing list