Re: [Gcl-devel] Re: Compiling floating point loops to blas calls

[Mike Thomas]

Hi all.

> > Well, if you allow using f2cl to convert Fortran to Lisp, then there's
> > a lot of code out there. :-) Otherwise, I think it's fairly limited
> > and certainly not like the Fortran or even C numerical libraries.
> >
>
> OK, so it seems this project should be on the back burner.  Though it
> really would show off the power of a lisp compiler, I'd think, and
> without all the overhead of writing the optimized blas directly, as
> its already widely available via atlas.

Without altering the current set of priorites for GCL development I would
like to note that two classic application areas for LISP are the modelling
of visual and neurological processes, both of which I would expect to
benefit substantially from optimised floating point loops (eg  neural
networks and the application of mexican hat filters to images).

As usual I feel constrained to point out that I don't do these things myself
but I believe that language implementors need to look to likely application
areas and there seems to have been an historical need to compare LISP to
Fortran for array and numerical processing.