Re: [Gretl-devel] Collection of gretl speed examples
Am 09.09.2017 um 01:53 schrieb Summers, Peter:
I like Sven's idea, and I'd be glad to contribute a Bayesian
mcmc
example. One thought occurs to me though. I've ported a bunch of
matlab code to hansl, but my goal has always been getting it to run,
not optimization. My point is that an octave guru could look at my
code and sped it up by x%, while Jack speeds my hansl code up by y%.
Maybe I'm anticipating the referee's report, but making an "apples to
apples " comparison in all the cases might not be easy.
Yes that's true and is a well-known problem of language comparisons. And
we would want to avoid to include obviously silly implementations.
However, I think a widespread perception now is: "Nobody is doing
research in this interpreted scripting language hansl, it is probably
quite slow." If we manage to move that perception to: "Hansl's speed
advantage is partly an artefact." Then I think that would already be a
big achievement.
cheers,
sven