> Date: Thu, 30 Sep 2004 15:43:55 +0000
> From: [log in to unmask]
> James Giles wrote:
>
> > robin wrote:
> > >> Date: Tue, 28 Sep 2004 22:58:09 +0000
> > >> From: [log in to unmask]
> > >> A simple experiment is to try something like
> > >> Do I = 1,50
> > >> print *, I, selected_real_kind(I)
> > >> Enddo
> > >
> > > This won't necessarily do it.
> > >
> [snip]
> > There's no rule against all the results from Hendrickson's
> > program being the same KIND number and for that correspond to
> > single precision.
> >
> > --
> > J. Giles
> >
> True, but there is a rule requiring two distinct kinds, "single"
> and "double". A person really interested in safety would
> probably do something like
> I = 0
> DO
> I = I + 1
> print *, I, selected_real_kind(I)
> if (selected_real_kind(I) < 0) exit
> ENDDO
I = 0
DO
I = I + 1
if (selected_real_kind(i) /= selected_real_kind(i+1)) print *, I, selected_real_kind(I)
if (selected_real_kind(I) < 0) exit
END DO
would give a briefer report.
> but that seems like overkill on most of today's popular
> machines ;) .
>
> Dick Hendrickson
|