Re: [Caml-list] Question on writing efficient Ocaml.

["Ian Oversby" <oversby@hotmail.com>]

Hi Olivier,

Thanks for the response.

> > Hi,
>
> > I've written some Ocaml code to solve the problem of placing 8 queens on 
>a
> > board so that none of them attack each-other.  I've also written a C++
> > equivalent which is running much more quickly than the Ocaml.  I assume
> > I've
> > made some basic errors with the Ocaml - does anyone have any suggestions
> > as
> > to what?
>
>there is room for improvement: for instance the type definitions of posn
>and board at the very beginning of the program introduce some unneeded
>boxing of values.

Does this mean that unboxing is inefficient in OCaml?  I've written an 
alternative version of the C++ that returns NULL instead of out of bound 
values which was close to the same speed so it would be a little 
disappointing if I couldn't achieve something similar in OCaml with Some / 
None.  Let me try to convert the OCaml to use out of bounds board values 
instead to see if that solves the speed problem.

>You might want to compare with this solution of the queens problem in 
>ocaml:
>   http://caml.inria.fr/pub/old_caml_site/Examples/oc/basics/queens.ml

I've written a queens solver along the same lines which is much faster than 
my other example as it makes many fewer calls and constructs fewer (and 
simpler) boards.

> > I compiled the Ocaml with the following command:
> >
> > ocamlopt -noassert -unsafe -ccopt -O3 -ccopt -fomit-frame-pointer q.ml 
>-o
> > q.exe
>
>the "-ccopt -O3 -ccopt -fomit-frame-pointer" are completely pointless: the
>ocaml compiler does not generate C code, it generates asm !

Well, that is certainly good to know.  Thanks very much :)

>--
>   Olivier

Ian

_________________________________________________________________
MSN Hotmail is evolving – check out the new Windows Live Mail 
http://ideas.live.com