RE: Octave speed up (Was: execution speed in *oct files)

[John W. Eaton]

On  8-Jun-1999, Van den Eynde Gert <address@hidden> wrote:

| By the way, do you think it is feasible to use Blitz++ or MTL as a
| Matrix/Vector base for Octave ? Do you think it is useful ?

[This question was sent directly to me, but I'm responding to the list
since it is probably of somewhat general interest.  --jwe]

Yes, in the long term, I think it would probably be useful to use
something that is better designed than the Octave matrix classes.  It
would help if the solution we adopt is defined by some sort of
standard, which neither Blitz++ nor MTL has become (yet).

Blitz++ (and perhaps MTL too, I'm not as familiar with it) is quite
complex and that could add some maintenance trouble.  How fast are
these packages changing?  Are they going to be actively maintained for
the forseeable future?

Some of the neat things that Blitz++ can do may provide some
benefit to the compiled code in Octave, but will not directly benefit
the interpreted Octave code (well, at least not without a lot of work).
For example, eliding the temporary variables and extra loops in a
matrix expression like `R = A + B + C + D' is nice, but Octave's
interpreter would have to become a lot smarter to take advantage of
that.

jwe



---------------------------------------------------------------------
Octave is freely available under the terms of the GNU GPL.  To ensure
that development continues, see www.che.wisc.edu/octave/giftform.html
Instructions for unsubscribing: www.che.wisc.edu/octave/archive.html
---------------------------------------------------------------------