[SciPy-Dev] Using MKL for FFTs (was: Accuracy of single-precision FFT)

Pauli Virtanen pav at iki.fi
Fri Jun 25 11:17:33 EDT 2010


Fri, 25 Jun 2010 16:59:40 +0200, Sturla Molden wrote:
[clip]
> All I am saying is this: Why not allow MKL as an alternative to FFTPACK
> as well?

I don't think there are valid objections to this. The issues are just 
practical: (i) can we find someone who wants to do it, and (ii) the multi-
backend stuff should be done well enough to be maintainable in the future.

-- 
Pauli Virtanen




More information about the SciPy-Dev mailing list