From mboxrd@z Thu Jan  1 00:00:00 1970
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on polar.synack.me
X-Spam-Level: 
X-Spam-Status: No, score=-0.3 required=5.0 tests=BAYES_00,
	REPLYTO_WITHOUT_TO_CC autolearn=no autolearn_force=no version=3.4.4
Path: 
 eternal-september.org!reader01.eternal-september.org!reader02.eternal-september.org!news.eternal-september.org!mx02.eternal-september.org!feeder.eternal-september.org!aioe.org!.POSTED!not-for-mail
From: "Nasser M. Abbasi" <nma@12000.org>
Newsgroups: comp.lang.ada
Subject: Re: getting same output as gfortran, long_float
Date: Fri, 01 May 2015 02:32:27 -0500
Organization: Aioe.org NNTP Server
Message-ID: <mhva6b$t35$1@speranza.aioe.org>
References: <mhu65r$ntj$1@speranza.aioe.org>
 <1kxou0nloqg9c$.1x0itzgdrlosm$.dlg@40tude.net>
 <mhujuf$jbs$1@speranza.aioe.org> <1i8x3r1feyzkt$.j85il7e3wpv9.dlg@40tude.net>
Reply-To: nma@12000.org
NNTP-Posting-Host: CV72NQ0GT7rQd8cP1ZYi/A.user.speranza.aioe.org
Mime-Version: 1.0
Content-Type: text/plain; charset=windows-1252; format=flowed
Content-Transfer-Encoding: 7bit
X-Complaints-To: abuse@aioe.org
User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64;
 rv:31.0) Gecko/20100101 Thunderbird/31.6.0
X-Notice: Filtered by postfilter v. 0.8.2
Xref: news.eternal-september.org comp.lang.ada:25682
Date: 2015-05-01T02:32:27-05:00
List-Id: <comp.lang.ada>

On 5/1/2015 1:52 AM, Dmitry A. Kazakov wrote:
> On Thu, 30 Apr 2015 20:12:47 -0500, Nasser M. Abbasi wrote:
>
>> I found that gfortran can do 128 bit floating point without
>> the use of the compiler switch -fdefault-real-8. Which will
>> map to similar thing as the above Ada construct:
>>
>> -------------------
>> PROGRAM foo
>> IMPLICIT NONE
>> REAL(KIND = 16) :: x  !-- kind=16 tells it is double quad
>> x = 12.0D0 * 0.0001D0/(1.0D0 * (1.0D0 - 0.1D0)**4 )
>> PRINT *, x
>> END PROGRAM
>> ------------------
>
> I didn't use FORTRAN for decades, but in good old FORTRAN-IV you declare
> specific lengths using T*n, e.g.
>
> INTEGER*2
> REAL*4
> REAL*8
>
> so, I would suggest, maybe naively,
>
> REAL*16
>

I am no Fortran expert by any means, but yes, REAL*16
will also work. However, modern Fortran suggests to use KIND.

Actually, it is much more complicated. The way I wrote it
above is not the optimal way. One is supposed to call
SELECTED_REAL_KIND(n,e) requesting n significant digits
and e number of digits in the exponent (e is optional).

i.e. one is supposed to write

SELECTED_REAL_KIND(8,3)  specifies real type of
(+-) 0.xxxxxxxx * 10 ^(+-)xxx

If the compiler does not support this, then the compile
will fail. This is in a way similar to Ada's

    type my_type is digits n;

And it is supposed to be portable way of doing things, vs.
using Real*16.  Here is an example:

---------------------
PROGRAM foo
IMPLICIT NONE
INTEGER, PARAMETER :: quad = SELECTED_REAL_KIND(30,10)
REAL(KIND = quad) :: x
x = 12.0D0 * 0.0001D0/(1.0D0 * (1.0D0 - 0.1D0)**4 )
PRINT *, x
END PROGRAM
------------------------
>gfortran -Wall  foo2.f90
>./a.out
    1.82898948331047112025871115292829927E-0003

If one would to ask for more digits than the compiler can support,
one gets an -1 from SELECTED_REAL_KIND() call, and this generates
a compile error in the next statement, since KIND=-1 is compile error.

There is much more to this, and it can get really complicated
to understand it all. The bottom line is that using KIND is the
recommended way to define real variables in modern Fortran to obtain
same precision  on different systems.

Here is one link on the subject

http://fortranwiki.org/fortran/show/Real+precision

--Nasser