[Axiom-developer] Re: [Axiom-mail] A slow summation

[Waldek Hebisch]

Bill Page wrote:
> On 15 Jun 2007 07:49:11 +0200, Martin Rubey wrote:
> >
> > > > reduce(+,[1.0/i for i in 1..20000])
> > > >
> > > > This works, but is (I think) unreasonably slow; it takes over 21 
> > > > seconds on
> > > > my computer.  The equivalent command in Maxima takes less than 1 second.
> >
> > Bill, could you file that as a bug on MathAction, please.
> 
> Well, I would rather that somebody new do that. Alasdair? :-)
> 
> The "issue" (I am not quite sure to call it a "bug") is probably some
> very inefficient implementation of list comprehension in the Axiom
> interpreter. E.g.
> 
> (1) -> )set message time on
> (1) -> [i for i in 1..20000];
> 
>                                                Type: List PositiveInteger
>       Time: 14.73 (EV) + 0.02 (OT) + 0.05 (GC) = 14.80 sec
> 
> (2) -> expand(1..20000);
> 
>                                                            Type: List Integer
>                              Time: 0.07 (IN) + 0.18 (OT) = 0.25 sec
> 
> -------
> 
> Both (1) and (2) do essentially the same thing.
> 
> >
> > (If you have an sbcl compiled axiom available, you could check
> > whether it is gcl's fault, or rather a bug in axiom.)
> >
> 
> I am quite sure that this is not a lisp implementation issue,
> but no I don't have a sbcl version of Axiom available. Do you?
> Still, maybe this is a job for someone who knows how to use
> the sbcl profiler to find out where the Axiom interpreter is
> spending all it's time - Waldek?
> 

On my machine, I get the following (on the second run, to
exclude time for loading):

                                      gcl      sbcl          sbcl
                                             interpreted  compiled
 reduce(+,[1.0/i for i in 1..20000])   8.70      1.76        0.17
 [i for i in 1..20000];                6.23      0.78        0.01
 expand(1..20000);                     0         0.004       0.01

Comment: sbcl evaluator has two modes of operation: interpreted
and compiled.  In compiled mode the code is first compiled and
then the resulting machine code is run.  One can switch mode
setting variable sb-ext:*evaluator-mode*:

)lisp (setf sb-ext:*evaluator-mode* :compile)

or

)lisp (setf sb-ext:*evaluator-mode* :interpret)

sbcl profiler showed that 98% of time were spent in Lisp evaluator,
and that agrees very well with getting much higher speed using
compiled mode.


-- 
                              Waldek Hebisch
address@hidden