All MPI implementations are required to support the data representation defined in this section. Support of optional datatypes (e.g., MPI_INTEGER2) is not required.
All floating point values are in big-endian IEEE format [43] of the appropriate size. Floating point values are represented by one of three IEEE formats. These are the IEEE ``Single (binary32),'' ``Double (binary64),'' and ``Double Extended (binary128)'' formats, requiring 4, 8, and 16 bytes of storage, respectively. For the IEEE ``Double Extended (binary128)'' formats, MPI specifies a format width of 16 bytes, with 15 exponent bits, bias = +16383, 112 fraction bits, and an encoding analogous to the ``Double (binary64)'' format. All integral values are in two's complement big-endian format. Big-endian means most significant byte at lowest address byte. For C _Bool, Fortran LOGICAL, and C++ bool, 0 implies false and nonzero implies true. C float _Complex, double _Complex, and long double _Complex, Fortran COMPLEX and DOUBLE COMPLEX, and other complex types are represented by a pair of floating point format values for the real and imaginary components. Characters are in ISO 8859-1 format [44]. Wide characters (of type MPI_WCHAR) are in Unicode format [69].
All signed numerals (e.g., MPI_INT, MPI_REAL) have the sign bit at the most significant bit. MPI_COMPLEX and MPI_DOUBLE_COMPLEX have the sign bit of the real and imaginary parts at the most significant bit of each part.
According to IEEE specifications [43], the ``NaN'' (not a number) is system dependent. It should not be interpreted within MPI as anything other than ``NaN.''
Advice
to implementors.
The MPI treatment of ``NaN'' is similar to the approach used in XDR [66].
( End of advice to implementors.)
All data is byte aligned, regardless of type.
All data items are stored contiguously in the file
(if the file view is contiguous).
Advice
to implementors.
All bytes of LOGICAL and bool must be checked
to determine the value.
( End of advice to implementors.)
Advice to users.
The type MPI_PACKED is treated as bytes and is not converted.
The user should be aware that MPI_PACK has the option of placing
a header in the beginning of the pack buffer.
( End of advice to users.)
Table 13: "external32" sizes of predefined datatypes
| Predefined Type | Length |
| MPI_PACKED | 1 |
| MPI_BYTE | 1 |
| MPI_CHAR | 1 |
| MPI_UNSIGNED_CHAR | 1 |
| MPI_SIGNED_CHAR | 1 |
| MPI_WCHAR | 2 |
| MPI_SHORT | 2 |
| MPI_UNSIGNED_SHORT | 2 |
| MPI_INT | 4 |
| MPI_LONG | 4 |
| MPI_UNSIGNED | 4 |
| MPI_UNSIGNED_LONG | 4 |
| MPI_LONG_LONG_INT | 8 |
| MPI_UNSIGNED_LONG_LONG | 8 |
| MPI_FLOAT | 4 |
| MPI_DOUBLE | 8 |
| MPI_LONG_DOUBLE | 16 |
| MPI_C_BOOL | 1 |
| MPI_INT8_T | 1 |
| MPI_INT16_T | 2 |
| MPI_INT32_T | 4 |
| MPI_INT64_T | 8 |
| MPI_UINT8_T | 1 |
| MPI_UINT16_T | 2 |
| MPI_UINT32_T | 4 |
| MPI_UINT64_T | 8 |
| MPI_AINT | 8 |
| MPI_COUNT | 8 |
| MPI_OFFSET | 8 |
| MPI_C_COMPLEX | 2*4 |
| MPI_C_FLOAT_COMPLEX | 2*4 |
| MPI_C_DOUBLE_COMPLEX | 2*8 |
| MPI_C_LONG_DOUBLE_COMPLEX | 2*16 |
| MPI_CHARACTER | 1 |
| MPI_LOGICAL | 4 |
| MPI_INTEGER | 4 |
| MPI_REAL | 4 |
| MPI_DOUBLE_PRECISION | 8 |
| MPI_COMPLEX | 2*4 |
| MPI_DOUBLE_COMPLEX | 2*8 |
| MPI_CXX_BOOL | 1 |
| MPI_CXX_FLOAT_COMPLEX | 2*4 |
| MPI_CXX_DOUBLE_COMPLEX | 2*8 |
| MPI_CXX_LONG_DOUBLE_COMPLEX | 2*16 |
| Predefined Type | Length |
| MPI_INTEGER1 | 1 |
| MPI_INTEGER2 | 2 |
| MPI_INTEGER4 | 4 |
| MPI_INTEGER8 | 8 |
| MPI_INTEGER16 | 16 |
| MPI_REAL2 | 2 |
| MPI_REAL4 | 4 |
| MPI_REAL8 | 8 |
| MPI_REAL16 | 16 |
| MPI_COMPLEX4 | 2*2 |
| MPI_COMPLEX8 | 2*4 |
| MPI_COMPLEX16 | 2*8 |
| MPI_COMPLEX32 | 2*16 |
| C++ Types | Length |
| MPI_CXX_BOOL | 1 |
| MPI_CXX_FLOAT_COMPLEX | 2*4 |
| MPI_CXX_DOUBLE_COMPLEX | 2*8 |
| MPI_CXX_LONG_DOUBLE_COMPLEX | 2*16 |
Advice
to implementors.
When converting a larger size integer to a smaller size integer,
only the least significant bytes are moved.
Care must be taken to preserve the sign bit value.
This allows no conversion errors
if the data range is within the range of the smaller size integer.
( End of advice to implementors.)
Table 13, 14,
and 15 specify the sizes of
predefined, optional, and C++ datatypes in "external32" format,
respectively.