<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
</head>
<body>
<div class="BodyFragment"><font size="2"><span style="font-size:10pt;">
<div class="PlainText"><br>
I am confused by Blaise's statement. "selected_real_kind(5)" is well defined across<br>
compilers; it returns an integer value that can be used in the kind statements and<br>
selected real numbers with at least 5 digits of precision.<br>
<br>
With the exception of -1, -2, and -3 (plus -4 and -5 in Fortran 2008) indicating <br>
no support. The actual integer values returned are compiler specific.<br>
<br>
If we wanted to have a better chance of getting ieee single, double, and quad, <br>
we should probably use something like:<br>
<br>
selected_real_kind(6,37)<br>
selected_real_kind(15,307)<br>
selected_real_kind(33,4931)<br>
<br>
in PETSc. You would still need to do some testing during configuration to <br>
really ensure that<br>
<br>
real(kind=selected_real_kind(7,38))<br>
real(kind=selected_real_kind(15,307))<br>
real(kind=selected_real_kind(34,4931))<br>
<br>
is the same as a C float, double, and __float128 respectively? <br>
<br>
=====<br>
<br>
Its not until Fortran 2008 that the standard has constants for REAL32, REAL64, <br>
and REAL128 defined in the ISO_FORTRAN_ENV module. However, if we are using<br>
Fortran 2008 features, go all out and use the ISO_C_BINDING module with <br>
the C_FLOAT and C_DOUBLE constants! Some compilers, such as the gnu<br>
compilers. extend the constants and include C_FLOAT128.<br>
<br>
I have attached a simple program to output the kinds. Try it on your machine.<br>
On my machine with gfortran from gcc version 4.8.5 (MacPorts gcc48 4.8.5_0), I<br>
get the following:<br>
<br>
C_FLOAT : 4<br>
C_DOUBLE : 8<br>
C_FLOAT128: 16<br>
REAL32 : 4<br>
REAL64 : 8<br>
REAL128 : 10<br>
sel(6) : 4<br>
sel(15) : 8<br>
sel(16) : 10<br>
sel(33) : 16<br>
<br>
The integers provide the number of bytes on my machine. Note REAL128 is <br>
apparently an 80-bit long double and not a 128-bit __float128!! The <br>
type using selected_real_kind appears correct though. The same <br>
happens for REAL128 with gfortran from gcc 4.9.3, gcc 5.4.0, <br>
and gcc 6.1.0 on my machine.<br>
<br>
Todd.<br>
<br>
</div>
</span></font></div>
<div class="BodyFragment"><font size="2"><span style="font-size:10pt;">
<div class="PlainText"><br>
<br>
> On Sep 2, 2016, at 12:51 AM, William Gropp <wgropp@illinois.edu> wrote:<br>
> <br>
> This is true. The meaning of integer values of kind is up to the implementation, and at least two choices are in use.<br>
> <br>
> Bill<br>
> <br>
> William Gropp<br>
> Acting Director and Chief Scientist, NCSA<br>
> Director, Parallel Computing Institute<br>
> Thomas M. Siebel Chair in Computer Science<br>
> University of Illinois Urbana-Champaign<br>
> <br>
> <br>
> <br>
> <br>
> <br>
> On Sep 1, 2016, at 6:15 PM, Blaise A Bourdin <bourdin@lsu.edu> wrote:<br>
> <br>
>> If I recall correctly, fortran does not mandate that "selected_real_kind(5)" means the same across compilers, so that hardcoding kind values may not be portable.<br>
> <br>
<br>
</div>
</span></font></div>
</body>
</html>