Using C/C++ and Fortran together:

Free Information Technology Software and Development Magazine Subscriptions and Document Downloads

Order of multi dimensional arrays in C/C++ is the opposite of FORTRAN.

• Equivalent multidimension arrays:
  • FORTRAN 77: INTEGER I(2,3,4)
  • FORTRAN 90: INTEGER, DIMENSION(2,3,4) :: I
  • C: int i[4][3][2];
• Accessing a FORTRAN array in C:
  Native FORTRAN:

     INTEGER   I,J,NI,NJ
     PARAMETER(NI=2,NJ=3)
     INTEGER   B(NI,NJ)                same as B(2,3)
  
     DO J = 1, NJ                      loop to 3
         DO I = 1, NI                  loop to 2
             PRINT 10, I, J, B(I,J)
  10         FORMAT(‘B(', I1, ',', I1, ') =', I2)
         END DO
     END DO
                

  
  Native C:

      int a[3][2])
      int i, j;
   
      for (i = 0; i < 3; i++)
          for (j = 0; j < 2; j++)
              printf(“a[%d][%d] = %d\n", i, j, a[i][j]);
                

  C array a[3][2] memory layout:

  a[0][0]

  a[0][1]

  a[1][0]

  a[1][1]

  a[2][0]

  a[2][1]

• In FORTRAN 90, also check out the "RESHAPE" directive.
• It is best not to re-dimension multi dimensional arrays within a function. Pass array size "n" and declare array as x[n][];

Fortran subroutines are the equivalent of "C" functions returning "(void)".

Note: The entry point names for some FORTRAN compilers have an underscore appended to the name. This is also true for common block/structure names as shown above.

The GNU gfortran comiler flags "-fno-underscoring" (older f77: "-fno-underscore") and "-fno-second-underscore" will alter the default naming in the object code and thus affect linking. One may view the object file with the command nm (i.e.: nm file.o).

Note: The case in FORTRAN is NOT preserved and is represented in lower case in the object file. The gfortran compiler option "-fcase-lower is default. The older g77 compiler option "-fsource-case-lower" is also default. GNU g77 FORTRAN can be case sensitive with the compile option "-fsource-case-preserve".

NOTE: When debugging with GDB, the Fortran subroutines must be referenced with names as they appear in the symbol table. This is the same as the "C" representation. Thus when setting a break point at the Fortran subroutine subra(), issue the comand "break subra_".

Man pages:

• nm - list symbols from object files
• gfortran - GNU FORTRAN compiler man page

All arguments in FORTRAN are passed by reference and not by value. Thus C must pass FORTRAN arguments as a pointer.

Character variables:

• Linux and GNU compilers: When passing character strings, the length must follow as separate arguments which are passed by value.

  FORTRAN	C
  call subra( string_a, string_b)	len_a = strlen(string_a); len_b = strlen(string_a); subra_( char *string_a, len_a, char *string_b, len_b)

  I have also seen the passing of a data structure containing two elements, the character string and an integer storing the length. This is common with databases such as Oracle.

• Classic AT&T UNIX:
  When passing character strings, the length must be appended as separate arguments which are passed by value.

  FORTRAN	C
  call subra( string_a, string_b)	subra_( char *string_a, char *string_b, len_a, len_b)

C expects strings to be null-terminated. Thus character strings from FORTRAN which are passed back to "C" should be null terminated with CHAR(0)

CHARACTER(LEN=32) :: sample_string = "This is a sample"//CHAR(0)

or

CALL SUBRA ('A string'//CHAR(0))
      

call sub(a,b,c,*,*)
return 1
end

int sub_(int *a,int *b,int *c)
{
   return(1)
}

goto(1, 2, 3), sub()

if( sub() ) goto ERRS;

Note: When using alternate returns to turn on/off an intlevel by returning a 1/0 you must use a FORTRAN wrapper to perform this function on the CSC Norwich mainframe. This is because the C compiler is old.

          #include <stdio.h>

          void ersetb(void){
              setbuf(stdout,NULL); /* set output to unbuffered */
          }

Fortran common block and global C/C++ extern structs of same name are equivalent. Never use un-named common blocks! Reference variables in same order, same type and with the same name for both C and FORTRAN. Character data is aligned on word boundaries.

          FORTRAN:
               DOUBLE PRECISION X
               INTEGER A, B, C
               COMMON/ABC/ X, A, B, C
 
          C:
               extern struct{
                   double x;
                   int a, b, c;
               } abc_;
 
          C++:
             extern "C" {
               extern struct{
                   double x;
                   int a, b, c;
               } abc_;
             }
 
        

Note: use of extern requires that the common block be referenced first by FORTRAN. If referenced first by C then drop the extern. The extern statement states that it is trying to reference memory which has already been set aside elsewhere.

[Potential Pitfall]: Byte alignment can be a source of data corruption if memory boundaries between FORTRAN and C/C++ are different. Each language may also align structure data differently. One must preserver the alignment of memory between the C/C++ "struct" and FORTRAN "common block" by ordering the variables in the exact same order and exactly matching the size of each variable. It is best to order the variables from the largest word size down to the smallest. Start with "double" followed by "float" and "int". Bool and byte aligned data should be listed last.

          FORTRAN:
               INTEGER A, B, C
               DOUBLE PRECISION D, E, F
               LOGICAL*1  FLAG
               COMMON/ABC/ A, D, FLAG, B, E
 
          C:
               extern struct{
                   int a;
                   double d;
                   bool flag;
                   int b;
                   double e;
               } abc_;
 
          C++:
             extern "C" {
               extern struct{
                   int a;
                   double d;
                   bool flag;
                   int b;
                   double e;
               } abc_;
             }
        

Using GDB to examine alignment:

• Set a breakpoint in the C/C++ section of code which has visibility to the struct.
• While in a C/C++ section of code: (gdb) print &abc_.b
  $3 = (int *) 0x5013e8
• Set a breakpoint in the FORTRAN section of code which has visibility to the common block.
• While in a FORTRAN section of code: (gdb) print &b
  $2 = (PTR TO -> ( integer )) 0x5013e8

Forcing alignment with compiler arguments: Mixing Intel FORTRAN compiler and GNU g++: use the following compiler flags to force a common memory alignment and padding to achieve a common double word alignment of variables:

• Intel FORTRAN: -warn alignments -align all -align rec8byte
• Intel C/C++: -Zp8
• GNU g++: -Wpadded -Wpacked -malign-double -mpreferred-stack-boundary=8
  Example warning: warning: padding struct size to alignment boundary
• GNU g77: -malign-double

FORTRAN program calling a C function:

      program test

      integer ii, jj, kk
      common/ijk/ ii, jj, kk
      real*8  ff
      character*32 cc

      ii = 2
      jj = 3
      kk = 4
      ff = 9.0567
      cc = 'Example of a character string'

      write(6,10) ii, ff
10    format('ii= ',i2,' ff= ',f10.4)

      call abc(ii)

      write(6,20) ii
20    format('ii= ',i2)

      write(6,30) ii, jj, kk

      call doubleIJK(cc)

      write(6,30) ii, jj, kk
30    format('ii= ',i2,' jj= ', i2, ' kk= ', i2)

      write(6, 40) cc
40    format(a32)

      stop
      end

      subroutine abc(jj)
      jj = jj * 2
      return
      end

#include <stdio.h>

extern struct
{
   int ii, jj, kk;
} ijk_;

int doubleijk_(char *cc, int ll)
{
   cc[ll--] = '\0';  // NULL terminate the string

   printf("From doubleIJK: %s\n",cc);

   ijk_.ii *=2;
   ijk_.jj *=2;
   ijk_.kk *=2;

   return(1);
}

Compile:

• gfortran -c testF.f
• gcc -c testC.c
• gfortran -o test testF.o testC.o

Note: If there is use of C/C++ standard libraries you may have to include the following linking arguments: -lc or -lstdc++

Run: ./test

ii=  2 ff=     9.0567
ii=  4
ii=  4 jj=  3 kk=  4
From doubleIJK: Example of a character string
ii=  8 jj=  6 kk=  8
Example of a character string

[Potential Pitfall]: If one does not terminate the string with NULL, the default with the new compilers (fort77 1.15 and gcc 4.5.2) and linker is to terminate the string with '\004' which is EOT (end of transmission). Previously this was not required but it is now. This code (cc[ll--] = '\0';) replaces the string terminating EOT character with a NULL.

C++ calling a FORTRAN function:

#include <iostream>

using namespace std;

extern"C" {
void fortfunc_(int *ii, float *ff);
}

main()
{

   int ii=5;
   float ff=5.5;

   fortfunc_(&ii, &ff);

   return 0;
}
                  

      subroutine fortfunc(ii,ff)
      integer ii
      real*4  ff

      write(6,100) ii, ff
 100  format('ii=',i2,' ff=',f6.3)

      return
      end
                  

Compile:

• gfortran -c testF.f
• g++ -c testC.cpp
• g++ -o test testF.o testC.o -lg2c

Run: ./test

ii= 5 ff= 5.500

FORTRAN unit numbers can not be shared with "C" and "C" file pointers can not be used by FORTRAN. Pass data to a function (pick one language) which does the I/O for you. Use this function by both languages.

The GNU FORTRAN compilers have only ported a small subset of VAX extensions. The vast majority will require a clever re-write.

VAX Variable Format Expressions:

      integer*4 ivar(3), nfor
      nfor=3
      ...
 
      write(6,100) (ivar(i),  i=1,nfor)
  100 format(<nfor>I3)
              

      do 20 i=1,nfor
         write(6,200) ivar(i)
  200    format(I3,$)  !! Supress carriage return
   20 continue

      write(6,400)     !! Write carriage return
  400 format()
              

VAX intrinsic functions:

Example: inint.c

short int inint_(float *rval)
{
   if(*rval < 0.0)
      return (*rval - 0.5);
   else
      return (*rval + 0.5);
}
        

• gcc -c inint.c
• gcc -c idnint.c
• ...
• Create library: ar -cvq libvax.a inint.o idnint.o ...

When mixing Fortran with C++, name mangling must be prevented.

          #ifdef __cplusplus
             extern"C" {
          #endif
          .
          .
          place declarations here
          .
          .
          #ifdef __cplusplus
          }
          #endif
        

The GNU FORTRAN compilers have supported the various standards over the years. The current (RHEL6 and Ubuntu 14.04) GNU gfortran compiler supports f95, f2003 and f2008 standards. By default, gfortran supports a superset of f95. Use the "-std=" flag to specify other levels of support. The older (RHEL4) GNU FORTRAN compiler supported the f77 standard.

Installation:

• Red Hat/CentOS: yum install gcc-gfortran
• Ubuntu/Debian: apt-get install gfortran

The Intel FORTRAN compiler has become a popular FORTRAN compiler for Linux as it supports many of the FORTRAN extensions supported by the Compaq (old DEC vax) and SGI FORTRAN compilers. (i.e. "structure", "record", "external", "encode", "decode", "find", "virtual", "pointer", "union", various intr8insic functions, I/O directives, ...) It links with object code generated by GNU gcc/g++ compilers as well as their own Intel C/C++ compilers.

Installation:

• mkdir /opt/intel
• cd /opt/intel
• move Intel Fortram compiler tar ball to this directory.
• tar xzf 1_fc_c_9.0.033_ia32.tar.gz
• cd 1_fc_c_9.0.033/
• ./install.sh
  1 : Install
  2 : provide name of an existing license file.
  License file path :
  /path-to-license-file/commercial_for_1_F2WC-5FDZV87R.lic
  (file copied to /opt/intel/licenses)
  1 (typical installation)
  Type "accept" to agree to license terms.
  Accept default location: /opt/intel/fc/9.0
  Accept default location: /opt/intel/idb/9.0
  x : Installation done

Compiler use and user configurations:

File: $HOME/.bashrc

..
...

#
# Intel Compiler
#

# FlexLM license server
export INTEL_LICENSE_FILE=28518@license-server

# Support for Intel FORTRAN compiler
if [ -f /opt/intel/fc/9.0/bin/ifortvars.sh ];
then
   source /opt/intel/fc/9.0/bin/ifortvars.sh
fi

# Support for Intel C/C++ compiler
if [ -f /opt/intel/cc/9.0/bin/iccvars.sh ];
then
   source /opt/intel/cc/9.0/bin/iccvars.sh
fi

# Support for Intel Debugger
if [ -f /opt/intel/cc/9.0/bin/idbvars.sh ];
then
   source /opt/intel/cc/9.0/bin/idbvars.sh
fi

...
...

export LD_LIBRARY_PATH
export PATH
           

F77=/opt/intel/fc/9.0/bin/ifort
F77FLAGS= -extend_source -fpp -f77rtl -intconstant -ftz -pad-source -sox \
          -lowercase -warn alignments -cxxlibgcc
CXX=g++
CPPFLAGS=-Wpadded -Wpacked
LDFLAGS=-L/opt/intel/fc/9.0/lib -lifport -lifcore -limf -Wabi -Wcast-align
DEBUG=-g

OBJS=file1.o file2.o

cpp-exe: $(OBJS)
	$(CC) $(LDFLAGS) -o name-of-exe $(OBJS)

.cpp.o:
     $(CXX) -c $(DEBUG) $(CPPFLAGS) $<	

.f.o:
     $(F77) -c $(DEBUG) $(F77FLAGS) $<	
        

Use GNU g++ or Intel C++ compiler to compile and link with main():
g++ -o name-of-exe main_prog.cpp file1.o file2.o -L/opt/intel/fc/9.0/lib -lifport -lifcore -limf -Wabi -Wcast-align

Using ddd as a front-end for the Intel debugger:

[Potential Pitfall]: I found that I could not install the Intel FORTRAN compiler from an NFS mounted drive. I had to copy the Intel installation files to a local drive and install from there.

Documentation: /opt/intel/cc/9.0/doc/main_for/mergedProjects/bldaps_for/

• Mixed language programming using C++ and FORTRAN 77
• FORTRAN GL API: (SGI GL library emulation for Linux)
  • YGL: Written in X11 (2D) and OpenGL (3D) for C and FORTRAN
  • repGL

Books:

	"Introduction to Programming with Fortran" with coverage of Fortran 90, 95, 2003 and 77 by Ian Chivers, Jane Sleightholme Springer; 1st edition, ISBN# 1846280532
	"FORTRAN 90 for Scientists and Engineers" by Brian Hahn Butterworth-Heinemann, ISBN# 0340600349
	"Introduction to FORTRAN 90 for Engineers and Scientists" by Larry R. Nyhoff, Sanford Leestma Prentice Hall; 1st edition, ISBN# 0135052157

   

Advertisements