MPI_DATA_TYPE(3)MPI_DATA_TYPE(3)NAME
MPI_Data_type - Defines C and Fortran data types
STANDARDS
This release implements the MPI 1.2 standard, as documented by the MPI
Forum in the spring 1997 release of MPI: A Message Passing Interface
Standard.
DESCRIPTION
The MPI_Data_type routine defines special data types for both C and
Fortran and data types of type MPI_Datatype for C programs and of type
integer for Fortran programs. The C types should be used only in C
programs and the Fortran types should be used only in Fortran programs.
For example, it is an error to use MPI_INT for a Fortran integer.
Special Data Types
The following data types can be used in C or Fortran programs:
MPI_PACKED Used for the MPI_Pack(3) and MPI_Unpack(3)
routines
MPI_UB Used for the MPI_Type_struct(3) routine (an
upper-bound indicator)
MPI_LB Used for the MPI_Type_struct(3) routine (a
lower-bound indicator)
C Data Types
The following data types can be used in all C programs:
MPI_CHAR Specifies char
MPI_BYTE Specifies unsigned char (see the standard)
MPI_SHORT Specifies short
MPI_INT Specifies int
MPI_LONG Specifies long
MPI_FLOAT Specifies float
MPI_DOUBLE Specifies double
MPI_UNSIGNED_CHAR Specifies unsigned char
MPI_UNSIGNED_SHORT Specifies unsigned short
MPI_UNSIGNED Specifies unsigned int
Page 1
MPI_DATA_TYPE(3)MPI_DATA_TYPE(3)
MPI_UNSIGNED_LONG Specifies unsigned long
MPI_LONG_DOUBLE Specifies long double (some systems may not
implement)
The following data types can be used for the MPI functions MPI_MAXLOC and
MPI_MINLOC:
MPI_FLOAT_INT Specifies struct { float, int }
MPI_LONG_INT Specifies struct { long, int }
MPI_DOUBLE_INT Specifies struct { double, int }
MPI_SHORT_INT Specifies struct { short, int }
MPI_2INT Specifies struct { int, int }
MPI_LONG_DOUBLE_INT Specifies struct { long double, int } (this is
an optional type and can be set to NULL)
MPI_LONG_LONG_INT Specifies struct { long long, int } (this is an
optional type and can be set to NULL)
Fortran Data Types
The following data types can be used in all Fortran programs:
MPI_REAL Specifies REAL
MPI_INTEGER Specifies INTEGER
MPI_LOGICAL Specifies LOGICAL
MPI_DOUBLE_PRECISION Specifies DOUBLE PRECISION
MPI_COMPLEX Specifies COMPLEX
MPI_DOUBLE_COMPLEX Specifies COMPLEX*16 (or COMPLEX*32) where
supported
The following data types are optional:
MPI_INTEGER1 Specifies INTEGER*1 if supported
MPI_INTEGER2 Specifies INTEGER*2 if supported
MPI_INTEGER4 Specifies INTEGER*4 if supported
MPI_REAL4 Specifies REAL*4 if supported
Page 2
MPI_DATA_TYPE(3)MPI_DATA_TYPE(3)
MPI_REAL8 Specifies REAL*8 if supported
The following data types are for the MPI functions MPI_MAXLOC and
MPI_MINLOC. In Fortran, these data types always consist of two elements
of the same Fortran type.
MPI_2INTEGER Specifies INTEGER, INTEGER
MPI_2REAL Specifies REAL, REAL
MPI_2DOUBLE_PRECISION Specifies DOUBLE PRECISION, DOUBLE
PRECISION
MPI_2COMPLEX Specifies COMPLEX, COMPLEX
MPI_2DOUBLE_COMPLEX Specifies COMPLEX*16, COMPLEX*16
Page 3
