Xavier,
Using inline PDL::PP gets around the problem with makefiles, etc... but
adds a small compile time to the first run of the program.
The thought is that this article will lead to readers thinking "I need
to start a project in one of these languages, which one should I use?".
In that case, it would make sense for the benchmark to reflect what the
finished project would be if someone who knew PDL fairly well had
written it. And I think most people who know PDL fairly well would end
up using at least inline PDL::PP for this particular problem.
I think the same goes for the other languages as well, they should all
be using their own bag of tricks. When all said and done, the real
"money" comparison is the time it takes to get it coded at an optimal
speed (if possible in that language). Just my 2 cents, though...
-Judd
On Thu, 2007-03-01 at 22:46 +0100, Xavier Calbet wrote:
> Hi Judd,
>
> Thanks for your response.
> I do not quite understand everything you say. I think that
> if you go for PDL::PP you end up compiling, makefiles, etc.
> This is NOT user friendly, so I think this solution
> does not compare well with other equivalent languages
> where you do NOT need to compile.
>
> Xavier
>
>
>
>
> On 3/1/07, Judd Taylor <[EMAIL PROTECTED]> wrote:
> > I think it should not. I don't think anyone that uses PDL seriously
> > would consider using a perl-level for() loop through a big dataset like
> > that.
> >
> > We all use PDL::PP and we mostly use Inline as well. If the benchmark is
> > to represent what we actually experience, then we should go ahead and
> > get rid of that for() loop and put in some PDL::PP code.
> >
> > If it _is_ considered cheating, then we should advise using pdl
> > threading to precompute several of the * calcs before getting into the
> > for loop.
> >
> > -Judd
> >
> >
> > On Thu, 2007-03-01 at 15:39 +0200, Kaj Wiik wrote:
> > > On Thu, 2007-03-01 at 07:32 +0000, Xavier Calbet wrote:
> > >
> > > > The results are the following:
> > > >
> > > > gcc 201 sec
> > > > g77 201 sec
> > > > PDL 651 sec
> > > > IDL 694 sec
> > > > MatLab 2738 sec
> > > > Octave 2031 sec
> > >
> > > Would Inline Pdlpp considered cheating ;-)?
> > > It would be interesting though..
> > >
> > > Kaj
> > >
> > --
> > ____________________________
> > Judd Taylor
> > Software Engineer
> >
> > Orbital Systems, Ltd.
> > 8304 Esters Blvd, Suite 870
> > Irving, TX 75063-2209
> >
> > [EMAIL PROTECTED]
> > (469) 442-1767 x127
> >
> >
>
--
____________________________
Judd Taylor
Software Engineer
Orbital Systems, Ltd.
8304 Esters Blvd, Suite 870
Irving, TX 75063-2209
[EMAIL PROTECTED]
(469) 442-1767 x127
_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl
