MPI-1 provides datatype objects, which allow users to specify an arbitrary layout of data in memory. The layout information, once put in a datatype, could not be decoded from the datatype. There are several cases, however, where accessing the layout information in opaque datatype objects would be useful.
The two functions in this section are used together to decode datatypes to recreate the calling sequence used in their initial definition. These can be used to allow a user to determine the type map and type signature of a datatype.
| MPI_TYPE_GET_ENVELOPE(datatype, num_integers, num_addresses, num_datatypes, combiner) | |
| IN datatype | datatype to access (handle) |
| OUT num_integers | number of input integers used in the call constructing combiner (nonnegative integer) |
| OUT num_addresses | number of input addresses used in the call constructing combiner (nonnegative integer) |
| OUT num_datatypes | number of input datatypes used in the call constructing combiner (nonnegative integer) |
| OUT combiner | combiner (state) |
int MPI_Type_get_envelope(MPI_Datatype datatype, int *num_integers, int *num_addresses, int *num_datatypes, int *combiner)
MPI_TYPE_GET_ENVELOPE(DATATYPE, NUM_INTEGERS, NUM_ADDRESSES, NUM_DATATYPES, COMBINER, IERROR)
INTEGER DATATYPE, NUM_INTEGERS, NUM_ADDRESSES, NUM_DATATYPES, COMBINER, IERROR
void MPI::Datatype::Get_envelope(int& num_integers, int& num_addresses, int& num_datatypes, int& combiner) const
For the given datatype, MPI_TYPE_GET_ENVELOPE returns information on the number and type of input arguments used in the call that created the datatype. The number-of-arguments values returned can be used to provide sufficiently large arrays in the decoding routine MPI_TYPE_GET_CONTENTS. This call and the meaning of the returned values is described below. The combiner reflects the MPI datatype constructor call that was used in creating datatype.
Rationale.
By requiring that the combiner reflect the constructor used in the creation of the datatype, the decoded information can be used to effectively recreate the calling sequence used in the original creation. One call is effectively the same as another when the information obtained from MPI_TYPE_GET_CONTENTS may be used with either to produce the same outcome. C calls MPI_Type_hindexed and MPI_Type_create_hindexed are always effectively the same while the Fortran call MPI_TYPE_HINDEXED will be different than either of these in some MPI implementations. This is the most useful information and was felt to be reasonable even though it constrains implementations to remember the original constructor sequence even if the internal representation is different.
The decoded information keeps track of datatype duplications. This is
important as one needs to distinguish between a predefined datatype
and a dup of a predefined datatype. The former is a constant object
that cannot be freed, while the latter is a derived datatype that can
be freed.
( End of rationale.)
The list below has the values that can be returned in
combiner on the left and the call associated with them on the
right.
MPI_COMBINER_NAMEDa named predefined datatype
MPI_COMBINER_DUP MPI_TYPE_DUP
MPI_COMBINER_CONTIGUOUS MPI_TYPE_CONTIGUOUS
MPI_COMBINER_VECTOR MPI_TYPE_VECTOR
MPI_COMBINER_HVECTOR_INTEGER MPI_TYPE_HVECTOR from Fortran
MPI_COMBINER_HVECTOR MPI_TYPE_HVECTOR from C or C++
and in some case Fortran
or MPI_TYPE_CREATE_HVECTOR
MPI_COMBINER_INDEXED MPI_TYPE_INDEXED
MPI_COMBINER_HINDEXED_INTEGER MPI_TYPE_HINDEXED from Fortran
MPI_COMBINER_HINDEXED MPI_TYPE_HINDEXED from C or C++
and in some case Fortran
or MPI_TYPE_CREATE_HINDEXED
MPI_COMBINER_INDEXED_BLOCK MPI_TYPE_CREATE_INDEXED_BLOCK
MPI_COMBINER_STRUCT_INTEGER MPI_TYPE_STRUCT from Fortran
MPI_COMBINER_STRUCT MPI_TYPE_STRUCT from C or C++
and in some case Fortran
or MPI_TYPE_CREATE_STRUCT
MPI_COMBINER_SUBARRAY MPI_TYPE_CREATE_SUBARRAY
MPI_COMBINER_DARRAY MPI_TYPE_CREATE_DARRAY
MPI_COMBINER_F90_REAL MPI_TYPE_CREATE_F90_REAL
MPI_COMBINER_F90_COMPLEX MPI_TYPE_CREATE_F90_COMPLEX
MPI_COMBINER_F90_INTEGER MPI_TYPE_CREATE_F90_INTEGER
MPI_COMBINER_RESIZED MPI_TYPE_CREATE_RESIZED
If combiner is MPI_COMBINER_NAMED then datatype is a named predefined datatype.
For calls with address arguments, we sometimes need to differentiate whether the call used an integer or an address size argument. For example, there are two combiners for hvector: MPI_COMBINER_HVECTOR_INTEGER and MPI_COMBINER_HVECTOR. The former is used if it was the MPI-1 call from Fortran, and the latter is used if it was the MPI-1 call from C or C++. However, on systems where MPI_ADDRESS_KIND = MPI_INTEGER_KIND (i.e., where integer arguments and address size arguments are the same), the combiner MPI_COMBINER_HVECTOR may be returned for a datatype constructed by a call to MPI_TYPE_HVECTOR from Fortran. Similarly, MPI_COMBINER_HINDEXED may be returned for a datatype constructed by a call to MPI_TYPE_HINDEXED from Fortran, and MPI_COMBINER_STRUCT may be returned for a datatype constructed by a call to MPI_TYPE_STRUCT from Fortran. On such systems, one need not differentiate constructors that take address size arguments from constructors that take integer arguments, since these are the same. The new MPI-2 calls all use address sized arguments.
Rationale.
For recreating the original call, it is important to know if address
information may have been truncated. The MPI-1 calls from Fortran for a few
routines could be
subject to truncation in the case where the default INTEGER size is
smaller than the size of an address.
( End of rationale.)
The actual arguments used in the creation call for a datatype
can be obtained from the call:
| MPI_TYPE_GET_CONTENTS(datatype, max_integers, max_addresses, max_datatypes, array_of_integers, array_of_addresses, array_of_datatypes) | |
| IN datatype | datatype to access (handle) |
| IN max_integers | number of elements in array_of_integers (non-negative integer) |
| IN max_addresses | number of elements in array_of_addresses (non-negative integer) |
| IN max_datatypes | number of elements in array_of_datatypes (non-negative integer) |
| OUT array_of_integers | contains integer arguments used in constructing datatype (array of integers) |
| OUT array_of_addresses | contains address arguments used in constructing datatype (array of integers) |
| OUT array_of_datatypes | contains datatype arguments used in constructing datatype (array of handles) |
int MPI_Type_get_contents(MPI_Datatype datatype, int max_integers, int max_addresses, int max_datatypes, int array_of_integers[], MPI_Aint array_of_addresses[], MPI_Datatype array_of_datatypes[])
MPI_TYPE_GET_CONTENTS(DATATYPE, MAX_INTEGERS, MAX_ADDRESSES, MAX_DATATYPES, ARRAY_OF_INTEGERS, ARRAY_OF_ADDRESSES, ARRAY_OF_DATATYPES, IERROR)
INTEGER DATATYPE, MAX_INTEGERS, MAX_ADDRESSES, MAX_DATATYPES, ARRAY_OF_INTEGERS(*), ARRAY_OF_DATATYPES(*), IERROR
INTEGER(KIND=MPI_ADDRESS_KIND) ARRAY_OF_ADDRESSES(*)
void MPI::Datatype::Get_contents(int max_integers, int max_addresses, int max_datatypes, int array_of_integers[], MPI::Aint array_of_addresses[], MPI::Datatype array_of_datatypes[]) const
datatype must be a predefined unnamed or a derived datatype; the call is erroneous if datatype is a predefined named datatype.
The values given for max_integers, max_addresses, and max_datatypes must be at least as large as the value returned in num_integers, num_addresses, and num_datatypes, respectively, in the call MPI_TYPE_GET_ENVELOPE for the same datatype argument.
Rationale.
The arguments max_integers, max_addresses, and
max_datatypes allow for error checking in the call. This is
analogous to the topology calls in MPI-1.
( End of rationale.)
The datatypes returned in array_of_datatypes are handles to
datatype objects that are equivalent to the datatypes used in the
original construction call. If these were derived datatypes, then
the returned datatypes are new datatype objects, and the
user is responsible for freeing these datatypes with
MPI_TYPE_FREE.
If these were predefined datatypes, then
the returned datatype is equal to that (constant) predefined datatype
and cannot be freed.
The committed state of returned derived datatypes is undefined, i.e., the datatypes may or may not be committed. Furthermore, the content of attributes of returned datatypes is undefined.
Note that MPI_TYPE_GET_CONTENTS can be invoked with a datatype argument that was constructed using MPI_TYPE_CREATE_F90_REAL, MPI_TYPE_CREATE_F90_INTEGER, or MPI_TYPE_CREATE_F90_COMPLEX (an unnamed predefined datatype). In such a case, an empty array_of_datatypes is returned.
Rationale.
The definition of datatype equivalence implies that equivalent
predefined datatypes are equal.
By requiring the same handle for named predefined datatypes, it is
possible to use the == or .EQ. comparison operator to determine the
datatype involved.
( End of rationale.)
Advice
to implementors.
The datatypes returned in array_of_datatypes must appear to the
user as if each is an equivalent copy of the datatype used in the type
constructor call.
Whether this is done by
creating a new datatype or via another mechanism such as a reference
count mechanism is up to the implementation as long as the semantics
are preserved.
( End of advice to implementors.)
Rationale.
The committed state and attributes of the returned datatype is
deliberately left vague. The datatype used in the original
construction may have been modified since its use in the constructor
call. Attributes can be added, removed, or modified as well as having
the datatype committed. The semantics given allow for a
reference count implementation without having to track these changes.
( End of rationale.)
In the MPI-1 datatype constructor calls, the address arguments in Fortran are of type INTEGER. In the new MPI-2 calls, the address arguments are of type INTEGER(KIND=MPI_ADDRESS_KIND). The call MPI_TYPE_GET_CONTENTS returns all addresses in an argument of type INTEGER(KIND=MPI_ADDRESS_KIND). This is true even if the old MPI-1 calls were used. Thus, the location of values returned can be thought of as being returned by the C bindings. It can also be determined by examining the new MPI-2 calls for datatype constructors for the deprecated MPI-1 calls that involve addresses.
Rationale.
By having all address arguments returned in the
array_of_addresses argument, the result from a C and Fortran
decoding of a datatype gives the result in the same
argument. It is assumed that an integer
of type INTEGER(KIND=MPI_ADDRESS_KIND) will be at least as large as
the INTEGER argument used in datatype construction with the old MPI-1
calls so no loss of information will occur.
( End of rationale.)
The following defines what values are placed in each entry of the
returned arrays depending on the datatype constructor used for
datatype. It also specifies the size of the arrays needed
which is the values returned by MPI_TYPE_GET_ENVELOPE.
In Fortran, the following calls were made:
PARAMETER (LARGE = 1000)
INTEGER TYPE, NI, NA, ND, COMBINER, I(LARGE), D(LARGE), IERROR
INTEGER(KIND=MPI_ADDRESS_KIND) A(LARGE)
! CONSTRUCT DATATYPE TYPE (NOT SHOWN)
CALL MPI_TYPE_GET_ENVELOPE(TYPE, NI, NA, ND, COMBINER, IERROR)
IF ((NI .GT. LARGE) .OR. (NA .GT. LARGE) .OR. (ND .GT. LARGE)) THEN
WRITE (*, *) "NI, NA, OR ND = ", NI, NA, ND, &
" RETURNED BY MPI_TYPE_GET_ENVELOPE IS LARGER THAN LARGE = ", LARGE
CALL MPI_ABORT(MPI_COMM_WORLD, 99)
ENDIF
CALL MPI_TYPE_GET_CONTENTS(TYPE, NI, NA, ND, I, A, D, IERROR)
or in C the analogous calls of:
#define LARGE 1000
int ni, na, nd, combiner, i[LARGE];
MPI_Aint a[LARGE];
MPI_Datatype type, d[LARGE];
/* construct datatype type (not shown) */
MPI_Type_get_envelope(type, &ni, &na, &nd, &combiner);
if ((ni > LARGE) || (na > LARGE) || (nd > LARGE)) {
fprintf(stderr, "ni, na, or nd = %d %d %d returned by ", ni, na, nd);
fprintf(stderr, "MPI_Type_get_envelope is larger than LARGE = %d\n",
LARGE);
MPI_Abort(MPI_COMM_WORLD, 99);
};
MPI_Type_get_contents(type, ni, na, nd, i, a, d);
The C++ code is in analogy to the C code above with the same values returned.
In the descriptions that follow, the lower case name of arguments is used.
If combiner is MPI_COMBINER_NAMED then it is erroneous to call MPI_TYPE_GET_CONTENTS.
If combiner is MPI_COMBINER_DUP then
| Constructor argument | C & C++ location | Fortran location |
| oldtype | d[0] | D(1) |
If combiner is MPI_COMBINER_CONTIGUOUS then
| Constructor argument | C & C++ location | Fortran location |
| count | i[0] | I(1) |
| oldtype | d[0] | D(1) |
If combiner is MPI_COMBINER_VECTOR then
| Constructor argument | C & C++ location | Fortran location |
| count | i[0] | I(1) |
| blocklength | i[1] | I(2) |
| stride | i[2] | I(3) |
| oldtype | d[0] | D(1) |
If combiner is MPI_COMBINER_HVECTOR_INTEGER or MPI_COMBINER_HVECTOR then
| Constructor argument | C & C++ location | Fortran location |
| count | i[0] | I(1) |
| blocklength | i[1] | I(2) |
| stride | a[0] | A(1) |
| oldtype | d[0] | D(1) |
If combiner is MPI_COMBINER_INDEXED then
| Constructor argument | C & C++ location | Fortran location |
| count | i[0] | I(1) |
| array_of_blocklengths | i[1] to i[i[0]] | I(2) to I(I(1)+1) |
| array_of_displacements | i[i[0]+1] to i[2*i[0]] | I(I(1)+2) to I(2*I(1)+1) |
| oldtype | d[0] | D(1) |
If combiner is MPI_COMBINER_HINDEXED_INTEGER or MPI_COMBINER_HINDEXED then
| Constructor argument | C & C++ location | Fortran location |
| count | i[0] | I(1) |
| array_of_blocklengths | i[1] to i[i[0]] | I(2) to I(I(1)+1) |
| array_of_displacements | a[0] to a[i[0]-1] | A(1) to A(I(1)) |
| oldtype | d[0] | D(1) |
If combiner is MPI_COMBINER_INDEXED_BLOCK then
| Constructor argument | C & C++ location | Fortran location |
| count | i[0] | I(1) |
| blocklength | i[1] | I(2) |
| array_of_displacements | i[2] to i[i[0]+1] | I(3) to I(I(1)+2) |
| oldtype | d[0] | D(1) |
If combiner is MPI_COMBINER_STRUCT_INTEGER or MPI_COMBINER_STRUCT then
| Constructor argument | C & C++ location | Fortran location |
| count | i[0] | I(1) |
| array_of_blocklengths | i[1] to i[i[0]] | I(2) to I(I(1)+1) |
| array_of_displacements | a[0] to a[i[0]-1] | A(1) to A(I(1)) |
| array_of_types | d[0] to d[i[0]-1] | D(1) to D(I(1)) |
If combiner is MPI_COMBINER_SUBARRAY then
| Constructor argument | C & C++ location | Fortran location |
| ndims | i[0] | I(1) |
| array_of_sizes | i[1] to i[i[0]] | I(2) to I(I(1)+1) |
| array_of_subsizes | i[i[0]+1] to i[2*i[0]] | I(I(1)+2) to I(2*I(1)+1) |
| array_of_starts | i[2*i[0]+1] to i[3*i[0]] | I(2*I(1)+2) to I(3*I(1)+1) |
| order | i[3*i[0]+1] | I(3*I(1)+2] |
| oldtype | d[0] | D(1) |
If combiner is MPI_COMBINER_DARRAY then
| Constructor argument | C & C++ location | Fortran location |
| size | i[0] | I(1) |
| rank | i[1] | I(2) |
| ndims | i[2] | I(3) |
| array_of_gsizes | i[3] to i[i[2]+2] | I(4) to I(I(3)+3) |
| array_of_distribs | i[i[2]+3] to i[2*i[2]+2] | I(I(3)+4) to I(2*I(3)+3) |
| array_of_dargs | i[2*i[2]+3] to i[3*i[2]+2] | I(2*I(3)+4) to I(3*I(3)+3) |
| array_of_psizes | i[3*i[2]+3] to i[4*i[2]+2] | I(3*I(3)+4) to I(4*I(3)+3) |
| order | i[4*i[2]+3] | I(4*I(3)+4) |
| oldtype | d[0] | D(1) |
If combiner is MPI_COMBINER_F90_REAL then
| Constructor argument | C & C++ location | Fortran location |
| p | i[0] | I(1) |
| r | i[1] | I(2) |
If combiner is MPI_COMBINER_F90_COMPLEX then
| Constructor argument | C & C++ location | Fortran location |
| p | i[0] | I(1) |
| r | i[1] | I(2) |
If combiner is MPI_COMBINER_F90_INTEGER then
| Constructor argument | C & C++ location | Fortran location |
| r | i[0] | I(1) |
If combiner is MPI_COMBINER_RESIZED then
| Constructor argument | C & C++ location | Fortran location |
| lb | a[0] | A(1) |
| extent | a[1] | A(2) |
| oldtype | d[0] | D(1) |
Example
This example shows how a datatype can be decoded. The routine
printdatatype prints out the elements of the datatype. Note the use
of MPI_Type_free for datatypes that are not predefined.
/*
Example of decoding a datatype.
Returns 0 if the datatype is predefined, 1 otherwise
*/
#include <stdio.h>
#include <stdlib.h>
#include "mpi.h"
int printdatatype( MPI_Datatype datatype )
{
int *array_of_ints;
MPI_Aint *array_of_adds;
MPI_Datatype *array_of_dtypes;
int num_ints, num_adds, num_dtypes, combiner;
int i;
MPI_Type_get_envelope( datatype,
&num_ints, &num_adds, &num_dtypes, &combiner );
switch (combiner) {
case MPI_COMBINER_NAMED:
printf( "Datatype is named:" );
/* To print the specific type, we can match against the
predefined forms. We can NOT use a switch statement here
We could also use MPI_TYPE_GET_NAME if we prefered to use
names that the user may have changed.
*/
if (datatype == MPI_INT) printf( "MPI_INT\n" );
else if (datatype == MPI_DOUBLE) printf( "MPI_DOUBLE\n" );
... else test for other types ...
return 0;
break;
case MPI_COMBINER_STRUCT:
case MPI_COMBINER_STRUCT_INTEGER:
printf( "Datatype is struct containing" );
array_of_ints = (int *)malloc( num_ints * sizeof(int) );
array_of_adds =
(MPI_Aint *) malloc( num_adds * sizeof(MPI_Aint) );
array_of_dtypes = (MPI_Datatype *)
malloc( num_dtypes * sizeof(MPI_Datatype) );
MPI_Type_get_contents( datatype, num_ints, num_adds, num_dtypes,
array_of_ints, array_of_adds, array_of_dtypes );
printf( " %d datatypes:\n", array_of_ints[0] );
for (i=0; i<array_of_ints[0]; i++) {
printf( "blocklength %d, displacement %ld, type:\n",
array_of_ints[i+1], array_of_adds[i] );
if (printdatatype( array_of_dtypes[i] )) {
/* Note that we free the type ONLY if it
is not predefined */
MPI_Type_free( &array_of_dtypes[i] );
}
}
free( array_of_ints );
free( array_of_adds );
free( array_of_dtypes );
break;
... other combiner values ...
default:
printf( "Unrecognized combiner type\n" );
}
return 1;
}