Open MPI logo

Open MPI User's Mailing List Archives

  |   Home   |   Support   |   FAQ   |   all Open MPI User's mailing list

Subject: Re: [OMPI users] mpi.h:: OMPI_HAVE_FORTRAN_LOGICAL / INTEGER / REAL are set to 0 (zero)
From: Shiqing Fan (fan_at_[hidden])
Date: 2011-06-11 13:03:11


Hi,

I did a few tests to reproduce the problem, and found out that it is
because of using a single cmake cache to handle the fortran settings, so
in some cases the part of the cache can be refreshed. I'm thinking to
have a second cache to make it more convenient for users.

So the current solution is just to run "configure" only once after
enabling OMPI_WANT_F77_BINDINGS. I will implement the update for the
CMake files soon.

Thanks,
Shiqing

On 2011-05-31 2:32 PM, hi wrote:
> Greetings once again !!!
>
> Environment:
> OS: Windows 7 64-bit
> Compilers: ifort and cl.exe (both 64-bit)
> CMAKE: version-2.8.4
> OpenMPI: version 1.5.3
>
> As described in the README.WINDOWS.txt, when performing 'First
> approach:" I am observing following...
> 1) via cmake-gui.exe clicked 'Configure' and it shown all config
> options in pink color
> see output of this step in attached cmake_gui_output1.txt
> 2) then I enabled OMPI_WANT_F77_BINDING and OMPI_WANT_F90_BINDING to
> true and then clicked 'Configure'.
> see output of this step in attached cmake_gui_output2.txt
>
> As you can observe in cmake_gui_output2.txt file, it succeeded on all
> Fortran data types (except INTEGER*16 and REAL*2).
> Now what I am observing is that eventho' fortran compiler supports
> most of the datatypes, but still mpi.h file reports '0' for
> corresponding datatype.
> <from_mpi.h>
> /* Whether we have FORTRAN LOGICAL*1 or not */
> #define OMPI_HAVE_FORTRAN_LOGICAL1 0
>
> /* Whether we have FORTRAN LOGICAL*2 or not */
> #define OMPI_HAVE_FORTRAN_LOGICAL2 0
>
> /* Whether we have FORTRAN LOGICAL*4 or not */
> #define OMPI_HAVE_FORTRAN_LOGICAL4 0
>
> /* Whether we have FORTRAN LOGICAL*8 or not */
> #define OMPI_HAVE_FORTRAN_LOGICAL8 0
>
> /* Whether we have FORTRAN INTEGER*1 or not */
> #define OMPI_HAVE_FORTRAN_INTEGER1 0
>
> /* Whether we have FORTRAN INTEGER*16 or not */
> #define OMPI_HAVE_FORTRAN_INTEGER16 0
>
> /* Whether we have FORTRAN INTEGER*2 or not */
> #define OMPI_HAVE_FORTRAN_INTEGER2 0
>
> /* Whether we have FORTRAN INTEGER*4 or not */
> #define OMPI_HAVE_FORTRAN_INTEGER4 0
>
> /* Whether we have FORTRAN INTEGER*8 or not */
> #define OMPI_HAVE_FORTRAN_INTEGER8 0
>
> /* Whether we have FORTRAN REAL*16 or not */
> #define OMPI_HAVE_FORTRAN_REAL16 0
>
> /* Whether we have FORTRAN REAL*2 or not */
> #define OMPI_HAVE_FORTRAN_REAL2 0
>
> /* Whether we have FORTRAN REAL*4 or not */
> #define OMPI_HAVE_FORTRAN_REAL4 0
>
> /* Whether we have FORTRAN REAL*8 or not */
> #define OMPI_HAVE_FORTRAN_REAL8 0
> ======
>
> Any idea on why mpi.h is setting zero for all above fortran datatypes
> despite fortran compiler supports most of the datatypes (see
> cmake_gui_output2.txt)???
>
> Thank you in advance.
> -Hiral
>
>
> _______________________________________________
> users mailing list
> users_at_[hidden]
> http://www.open-mpi.org/mailman/listinfo.cgi/users

-- 
---------------------------------------------------------------
Shiqing Fan
High Performance Computing Center Stuttgart (HLRS)
Tel: ++49(0)711-685-87234      Nobelstrasse 19
Fax: ++49(0)711-685-65832      70569 Stuttgart
http://www.hlrs.de/organization/people/shiqing-fan/
email:fan_at_[hidden]