> Date: Wed, 15 Nov 2000 09:00:35 +0100
> From: Pierre Hugonnet <[log in to unmask]>
> robin wrote:
> >
> > >
> > > A solution should be to use :
> > >
> > > integer, parameter :: digits = 10 ! my regular precision
> > > integer, parameter :: R1 = selected_real_kind(digits)
> > > integer, parameter :: R2 = selected_real-kind(2*digits)
> > >
> > > for portability, i think we should always use kind.
> >
> > This won't be portable. On many machines, R1 will be double precison
> > and R2 will specify 20 digits, which will exceed the available precision.
> >
>
> Hello,
>
> you should say: "on a few machines". For, most of compilers on classical
> scientific Unix platforms (SGI, Sun, IBM,...) handle what is often named
> a "quadruple precision" kind (128bit), with typically around 30 digit precision.
There are some 50 million PCs that don't have it. That makes it
"many".
> Regards,
>
> Pierre
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|