octuple-precision and arbitrary-precision numbers

[Tie-Feng_F_]

Hi,

I am using Intel Visual Fortran Compiler for windows.

It only provides  the single-, double-, and quadruple-precision numbers.

But I now need  to perform very accurate  calculations.

Is there any way to obtain, within the  Intel Visual Fortran for windows, 

the octuple- and even arbitrary-precision floating-point formats?

Thank you very much for your help!

Tie-Feng Fang

[Les_Neilson]

Hi,

(from Google - Fortran arbitrary precision)

You could try this from Professor David Smith of LMU Los Angeles

Also the Fortran wiki here gives a list of various precision libraries.

HTH

Les

[rase]

Thanks, Les, for the link to the Fortran wiki which I was not aware of until now.

[Jeff_Arnold]

If you are willing to work with a library which only has a C language interface, investigate MPFR.  See http://www.mpfr.org/  It's a high quality arbitrary-precision math library which produces correctly rounded results.  It is used by many of the compilers in the GNU compiler collection during compilation; it is not required at run-time.

There was a discussion in 2011 on comp.lang.fortran about using MPFR from Fortran.  Seach the newsgroup for the thread titled "arbitrary precision".

I'm not aware that anyone has implemented a Fortran interface to MPFR.