Re: fannkuch

[Christophe TROESTLER-3]

On Thu, 19 May 2005, Isaac Gouy <igouy2@...> wrote:

>
> --- Christophe TROESTLER <del-con@...> wrote:
> > I believe than fannkuch should be run for at least N=10 (possibly
> > N=11).  I just checked in the Caml code with aesthetic modifications
> > (no code change) and the running time went from 0.09 to 0.12 -- quite
> > a few places down.  That means the running times are too short not to
> > be seriously influenced by random effects.
>
> It may mean that changes made to the scripts, over the last week or so,
> have had an effect on the measured times. (Let's not jump to conclusions.)

I doubt it since I am talking from one day to the next.

ChriS

_______________________________________________
Shootout-list mailing list
Shootout-list@...
http://lists.alioth.debian.org/mailman/listinfo/shootout-list

[Christophe TROESTLER-3]

On Thu, 19 May 2005, Isaac Gouy <igouy2@...> wrote:
>
> > > > That means the running times are too short
> > > > not to be seriously influenced by random effects.
>
> Let's be clear

The cosmetic change I made was only to add a newline at the end (NO
code change)!  I replaced the old code with the new one (so, yes,
there is only OCaml #3) and I wrote down the running time (which was
computed from a run some time ago).  As you remarked, the new time
vastly differs...

You need not to believe me, you can see for yourself:

- diff(1.228, 2005/05/18 07:55:32     1.231, 2005/05/19 07:06:10)
  http://cvs.alioth.debian.org/cgi-bin/cvsweb.cgi/shootout/website/data/data.csv.diff?cvsroot=shootout&r1=text&tr1=1.228&r2=text&tr2=1.231&f=h

- select the diff on http://cvs.alioth.debian.org/cgi-bin/cvsweb.cgi/shootout/bench/fannkuch/fannkuch.ocaml-3.ocaml?cvsroot=shootout

Do you have a better explanation? (maybe something changed on the
bench machine but then all programs should be rerun).

Regards,
ChriS

_______________________________________________
Shootout-list mailing list
Shootout-list@...
http://lists.alioth.debian.org/mailman/listinfo/shootout-list