[Numpy-discussion] VSIPL?

Konrad Hinsen hinsen at cnrs-orleans.fr
Mon Aug 21 10:01:20 CDT 2000


> Just by chance I found the link to an ANSI-C library which looks
> interesting in the light of the reimplementation of NumPy. The link is:

Is anyone working on a reimplementation?

> data space, gather/scatter, interfacing to Fortran. Are there people
> on the list, which do know this library and the possible problems or
> merits of it? This library was used as an positive design example in a
> discussion about the GSL.

Not me, but at first sight it does look reasonable. The main benefit
for using this in NumPy would be the possibility of substituting
optimized implementations for the reference implementation in C.

> definitions, so the question is, if the new NumPy should have more
> characteristics of an interface, so that the underlying numeric engine
> could be changed. Would it help in such a case to wrap a library as a

To be useful in C modules, at least the data layout must be documented
and stable. Given that there are at least two popular layouts (C style
and Fortran style), it is difficult to accomodate all existing array
libraries.

Konrad.
-- 
-------------------------------------------------------------------------------
Konrad Hinsen                            | E-Mail: hinsen at cnrs-orleans.fr
Centre de Biophysique Moleculaire (CNRS) | Tel.: +33-2.38.25.56.24
Rue Charles Sadron                       | Fax:  +33-2.38.63.15.17
45071 Orleans Cedex 2                    | Deutsch/Esperanto/English/
France                                   | Nederlands/Francais
-------------------------------------------------------------------------------




More information about the Numpy-discussion mailing list