17.2.292. MPI_Reduce

MPI_Reduce, MPI_Ireduce, MPI_Reduce_init - Reduces values on all processes within a group.

17.2.292.1. SYNTAX

17.2.292.1.1. C Syntax

#include <mpi.h>

int MPI_Reduce(const void *sendbuf, void *recvbuf, int count,
               MPI_Datatype datatype, MPI_Op op, int root,
               MPI_Comm comm)

int MPI_Ireduce(const void *sendbuf, void *recvbuf, int count,
                MPI_Datatype datatype, MPI_Op op, int root,
                MPI_Comm comm, MPI_Request *request)


int MPI_Reduce_init(const void *sendbuf, void *recvbuf, int count,
                MPI_Datatype datatype, MPI_Op op, int root,
                MPI_Comm comm, MPI_Info info, MPI_Request *request)

17.2.292.1.2. Fortran Syntax

USE MPI
! or the older form: INCLUDE 'mpif.h'
MPI_REDUCE(SENDBUF, RECVBUF, COUNT, DATATYPE, OP, ROOT, COMM,
             IERROR)
     <type>  SENDBUF(*), RECVBUF(*)
     INTEGER COUNT, DATATYPE, OP, ROOT, COMM, IERROR

MPI_IREDUCE(SENDBUF, RECVBUF, COUNT, DATATYPE, OP, ROOT, COMM,
            REQUEST, IERROR)
     <type>  SENDBUF(*), RECVBUF(*)
     INTEGER COUNT, DATATYPE, OP, ROOT, COMM, REQUEST, IERROR

MPI_REDUCE_INIT(SENDBUF, RECVBUF, COUNT, DATATYPE, OP, ROOT, COMM,
            INFO, REQUEST, IERROR)
     <type>  SENDBUF(*), RECVBUF(*)
     INTEGER COUNT, DATATYPE, OP, ROOT, COMM, INFO, REQUEST, IERROR

17.2.292.1.3. Fortran 2008 Syntax

USE mpi_f08
MPI_Reduce(sendbuf, recvbuf, count, datatype, op, root, comm, ierror)
     TYPE(*), DIMENSION(..), INTENT(IN) :: sendbuf
     TYPE(*), DIMENSION(..) :: recvbuf
     INTEGER, INTENT(IN) :: count, root
     TYPE(MPI_Datatype), INTENT(IN) :: datatype
     TYPE(MPI_Op), INTENT(IN) :: op
     TYPE(MPI_Comm), INTENT(IN) :: comm
     INTEGER, OPTIONAL, INTENT(OUT) :: ierror

MPI_Ireduce(sendbuf, recvbuf, count, datatype, op, root, comm, request,
             ierror)
     TYPE(*), DIMENSION(..), INTENT(IN), ASYNCHRONOUS :: sendbuf
     TYPE(*), DIMENSION(..), ASYNCHRONOUS :: recvbuf
     INTEGER, INTENT(IN) :: count, root
     TYPE(MPI_Datatype), INTENT(IN) :: datatype
     TYPE(MPI_Op), INTENT(IN) :: op
     TYPE(MPI_Comm), INTENT(IN) :: comm
     TYPE(MPI_Request), INTENT(OUT) :: request
     INTEGER, OPTIONAL, INTENT(OUT) :: ierror


MPI_Reduce_init(sendbuf, recvbuf, count, datatype, op, root, comm, info, request,
             ierror)
     TYPE(*), DIMENSION(..), INTENT(IN), ASYNCHRONOUS :: sendbuf
     TYPE(*), DIMENSION(..), ASYNCHRONOUS :: recvbuf
     INTEGER, INTENT(IN) :: count, root
     TYPE(MPI_Datatype), INTENT(IN) :: datatype
     TYPE(MPI_Op), INTENT(IN) :: op
     TYPE(MPI_Comm), INTENT(IN) :: comm
     TYPE(MPI_Info), INTENT(IN) :: info
     TYPE(MPI_Request), INTENT(OUT) :: request
     INTEGER, OPTIONAL, INTENT(OUT) :: ierror

17.2.292.2. INPUT PARAMETERS

  • sendbuf: Address of send buffer (choice).

  • count: Number of elements in send buffer (integer).

  • datatype: Data type of elements of send buffer (handle).

  • op: Reduce operation (handle).

  • root: Rank of root process (integer).

  • comm: Communicator (handle).

  • info: Info (handle, persistent only).

17.2.292.3. OUTPUT PARAMETERS

  • recvbuf: Address of receive buffer (choice, significant only at root).

  • request: Request (handle, non-blocking and persistent only).

  • ierror: Fortran only: Error status (integer).

17.2.292.4. DESCRIPTION

The global reduce functions (MPI_Reduce, MPI_Op_create, MPI_Op_free, MPI_Allreduce, MPI_Reduce_scatter, MPI_Scan) perform a global reduce operation (such as sum, max, logical AND, etc.) across all the members of a group. The reduction operation can be either one of a predefined list of operations, or a user-defined operation. The global reduction functions come in several flavors: a reduce that returns the result of the reduction at one node, an all-reduce that returns this result at all nodes, and a scan (parallel prefix) operation. In addition, a reduce-scatter operation combines the functionality of a reduce and a scatter operation.

MPI_Reduce combines the elements provided in the input buffer of each process in the group, using the operation op, and returns the combined value in the output buffer of the process with rank root. The input buffer is defined by the arguments sendbuf, count, and datatype; the output buffer is defined by the arguments recvbuf, count, and datatype; both have the same number of elements, with the same type. The routine is called by all group members using the same arguments for count, datatype, op, root, and comm. Thus, all processes provide input buffers and output buffers of the same length, with elements of the same type. Each process can provide one element, or a sequence of elements, in which case the combine operation is executed element-wise on each entry of the sequence. For example, if the operation is MPI_MAX and the send buffer contains two elements that are floating-point numbers (count = 2 and datatype = MPI_FLOAT), then recvbuf(1) = global max (sendbuf(1)) and recvbuf(2) = global max(sendbuf(2)).

17.2.292.5. USE OF IN-PLACE OPTION

When the communicator is an intracommunicator, you can perform a reduce operation in-place (the output buffer is used as the input buffer). Use the variable MPI_IN_PLACE as the value of the root process sendbuf. In this case, the input data is taken at the root from the receive buffer, where it will be replaced by the output data.

Note that MPI_IN_PLACE is a special kind of value; it has the same restrictions on its use as MPI_BOTTOM.

Because the in-place option converts the receive buffer into a send-and-receive buffer, a Fortran binding that includes INTENT must mark these as INOUT, not OUT.

17.2.292.6. WHEN COMMUNICATOR IS AN INTER-COMMUNICATOR

When the communicator is an inter-communicator, the root process in the first group combines data from all the processes in the second group and then performs the op operation. The first group defines the root process. That process uses MPI_ROOT as the value of its root argument. The remaining processes use MPI_PROC_NULL as the value of their root argument. All processes in the second group use the rank of that root process in the first group as the value of their root argument. Only the send buffer arguments are significant in the second group, and only the receive buffer arguments are significant in the root process of the first group.

17.2.292.7. PREDEFINED REDUCE OPERATIONS

The set of predefined operations provided by MPI is listed below (Predefined Reduce Operations). That section also enumerates the datatypes each operation can be applied to. In addition, users may define their own operations that can be overloaded to operate on several datatypes, either basic or derived. This is further explained in the description of the user-defined operations (see the man pages for MPI_Op_create and MPI_Op_free).

The operation op is always assumed to be associative. All predefined operations are also assumed to be commutative. Users may define operations that are assumed to be associative, but not commutative. The ``canonical’’ evaluation order of a reduction is determined by the ranks of the processes in the group. However, the implementation can take advantage of associativity, or associativity and commutativity, in order to change the order of evaluation. This may change the result of the reduction for operations that are not strictly associative and commutative, such as floating point addition.

Predefined operators work only with the MPI types listed below (Predefined Reduce Operations, and the section MINLOC and MAXLOC, below). User-defined operators may operate on general, derived datatypes. In this case, each argument that the reduce operation is applied to is one element described by such a datatype, which may contain several basic values. This is further explained in Section 4.9.4 of the MPI Standard, “User-Defined Operations.”

The following predefined operations are supplied for MPI_Reduce and related functions MPI_Allreduce, MPI_Reduce_scatter, and MPI_Scan. These operations are invoked by placing the following in op:

Name                Meaning
---------           --------------------
MPI_MAX             maximum
MPI_MIN             minimum
MPI_SUM             sum
MPI_PROD            product
MPI_LAND            logical and
MPI_BAND            bit-wise and
MPI_LOR             logical or
MPI_BOR             bit-wise or
MPI_LXOR            logical xor
MPI_BXOR            bit-wise xor
MPI_MAXLOC          max value and location
MPI_MINLOC          min value and location

The two operations MPI_MINLOC and MPI_MAXLOC are discussed separately below (MINLOC and MAXLOC). For the other predefined operations, we enumerate below the allowed combinations of op and datatype arguments. First, define groups of MPI basic datatypes in the following way:

C integer:            MPI_INT, MPI_LONG, MPI_SHORT,
                      MPI_UNSIGNED_SHORT, MPI_UNSIGNED,
                      MPI_UNSIGNED_LONG
Fortran integer:      MPI_INTEGER
Floating-point:       MPI_FLOAT, MPI_DOUBLE, MPI_REAL,
                      MPI_DOUBLE_PRECISION, MPI_LONG_DOUBLE
Logical:              MPI_LOGICAL
Complex:              MPI_COMPLEX
Byte:                 MPI_BYTE

Now, the valid datatypes for each option is specified below.

Op                              Allowed Types
----------------         ---------------------------
MPI_MAX, MPI_MIN                C integer, Fortran integer,
                                        floating-point

MPI_SUM, MPI_PROD               C integer, Fortran integer,
                                        floating-point, complex

MPI_LAND, MPI_LOR,              C integer, logical
MPI_LXOR

MPI_BAND, MPI_BOR,              C integer, Fortran integer, byte
MPI_BXOR

Example 1: A routine that computes the dot product of two vectors that are distributed across a group of processes and returns the answer at process zero.

SUBROUTINE PAR_BLAS1(m, a, b, c, comm)
REAL a(m), b(m)       ! local slice of array
REAL c                ! result (at process zero)
REAL sum
INTEGER m, comm, i, ierr

! local sum
sum = 0.0
DO i = 1, m
   sum = sum + a(i)*b(i)
END DO

! global sum
CALL MPI_REDUCE(sum, c, 1, MPI_REAL, MPI_SUM, 0, comm, ierr)
RETURN

Example 2: A routine that computes the product of a vector and an array that are distributed across a group of processes and returns the answer at process zero.

SUBROUTINE PAR_BLAS2(m, n, a, b, c, comm)
REAL a(m), b(m,n)    ! local slice of array
REAL c(n)            ! result
REAL sum(n)
INTEGER n, comm, i, j, ierr

! local sum
DO j= 1, n
  sum(j) = 0.0
  DO i = 1, m
    sum(j) = sum(j) + a(i)*b(i,j)
  END DO
END DO

! global sum
CALL MPI_REDUCE(sum, c, n, MPI_REAL, MPI_SUM, 0, comm, ierr)

! return result at process zero (and garbage at the other nodes)
RETURN

17.2.292.8. MINLOC AND MAXLOC

The operator MPI_MINLOC is used to compute a global minimum and also an index attached to the minimum value. MPI_MAXLOC similarly computes a global maximum and index. One application of these is to compute a global minimum (maximum) and the rank of the process containing this value.

The operation that defines MPI_MAXLOC is

         ( u )    (  v )      ( w )
         (   )  o (    )   =  (   )
         ( i )    (  j )      ( k )

where

    w = max(u, v)

and

         ( i            if u > v
         (
   k   = ( min(i, j)    if u = v
         (
         (  j           if u < v)

MPI_MINLOC is defined similarly:

         ( u )    (  v )      ( w )
         (   )  o (    )   =  (   )
         ( i )    (  j )      ( k )

where

    w = min(u, v)

and

         ( i            if u < v
         (
   k   = ( min(i, j)    if u = v
         (
         (  j           if u > v)

Both operations are associative and commutative. Note that if MPI_MAXLOC is applied to reduce a sequence of pairs (u(0), 0), (u(1), 1), …, (u(n-1), n-1), then the value returned is (u , r), where u= max(i) u(i) and r is the index of the first global maximum in the sequence. Thus, if each process supplies a value and its rank within the group, then a reduce operation with op = MPI_MAXLOC will return the maximum value and the rank of the first process with that value. Similarly, MPI_MINLOC can be used to return a minimum and its index. More generally, MPI_MINLOC computes a lexicographic minimum, where elements are ordered according to the first component of each pair, and ties are resolved according to the second component.

The reduce operation is defined to operate on arguments that consist of a pair: value and index. For both Fortran and C, types are provided to describe the pair. The potentially mixed-type nature of such arguments is a problem in Fortran. The problem is circumvented, for Fortran, by having the MPI-provided type consist of a pair of the same type as value, and coercing the index to this type also. In C, the MPI-provided pair type has distinct types and the index is an int.

In order to use MPI_MINLOC and MPI_MAXLOC in a reduce operation, one must provide a datatype argument that represents a pair (value and index). MPI provides nine such predefined datatypes. The operations MPI_MAXLOC and MPI_MINLOC can be used with each of the following datatypes:

Fortran:
Name                     Description
MPI_2REAL                pair of REALs
MPI_2DOUBLE_PRECISION    pair of DOUBLE-PRECISION variables
MPI_2INTEGER             pair of INTEGERs

C:
Name                     Description
MPI_FLOAT_INT            float and int
MPI_DOUBLE_INT           double and int
MPI_LONG_INT             long and int
MPI_2INT                 pair of ints
MPI_SHORT_INT            short and int
MPI_LONG_DOUBLE_INT      long double and int

The data type MPI_2REAL is equivalent to:

MPI_TYPE_CONTIGUOUS(2, MPI_REAL, MPI_2REAL)

Similar statements apply for MPI_2INTEGER, MPI_2DOUBLE_PRECISION, and MPI_2INT.

The datatype MPI_FLOAT_INT is as if defined by the following sequence of instructions.

type[0] = MPI_FLOAT
type[1] = MPI_INT
disp[0] = 0
disp[1] = sizeof(float)
block[0] = 1
block[1] = 1
MPI_TYPE_STRUCT(2, block, disp, type, MPI_FLOAT_INT)

Similar statements apply for MPI_LONG_INT and MPI_DOUBLE_INT.

Example 3: Each process has an array of 30 doubles, in C. For each of the 30 locations, compute the value and rank of the process containing the largest value.

...
/* each process has an array of 30 double: ain[30]
 */
double ain[30], aout[30];
int  ind[30];
struct {
    double val;
    int   rank;
} in[30], out[30];
int i, myrank, root;

MPI_Comm_rank(MPI_COMM_WORLD, &myrank);
for (i=0; i<30; ++i) {
    in[i].val = ain[i];
    in[i].rank = myrank;
}
MPI_Reduce( in, out, 30, MPI_DOUBLE_INT, MPI_MAXLOC, root, comm );
/* At this point, the answer resides on process root
 */
if (myrank == root) {
    /* read ranks out
     */
    for (i=0; i<30; ++i) {
        aout[i] = out[i].val;
        ind[i] = out[i].rank;
    }
}

Example 4: Same example, in Fortran.

...
! each process has an array of 30 double: ain(30)

DOUBLE PRECISION :: ain(30), aout(30)
INTEGER :: ind(30)
DOUBLE PRECISION :: in(2,30), out(2,30)
INTEGER :: i, myrank, root, ierr

call MPI_COMM_RANK(MPI_COMM_WORLD, myrank)
DO I=1, 30
    in(1,i) = ain(i)
    in(2,i) = myrank    ! myrank is coerced to a double
END DO

call MPI_REDUCE( in, out, 30, MPI_2DOUBLE_PRECISION, MPI_MAXLOC, root, &
                                                          comm, ierr )
! At this point, the answer resides on process root

IF (myrank == root) THEN
    ! read ranks out
    DO I= 1, 30
        aout(i) = out(1,i)
        ind(i) = out(2,i)  ! rank is coerced back to an integer
    END DO
END IF

Example 5: Each process has a nonempty array of values. Find the minimum global value, the rank of the process that holds it, and its index on this process.

#define  LEN   1000

float val[LEN];        /* local array of values */
int count;             /* local number of values */
int myrank, minrank, minindex;
float minval;

struct {
    float value;
    int   index;
} in, out;

/* local minloc */
in.value = val[0];
in.index = 0;
for (i=1; i < count; i++)
    if (in.value > val[i]) {
        in.value = val[i];
        in.index = i;
    }

/* global minloc */
MPI_Comm_rank(MPI_COMM_WORLD, &myrank);
in.index = myrank*LEN + in.index;
MPI_Reduce( in, out, 1, MPI_FLOAT_INT, MPI_MINLOC, root, comm );
    /* At this point, the answer resides on process root
     */
if (myrank == root) {
    /* read answer out
     */
    minval = out.value;
    minrank = out.index / LEN;
    minindex = out.index % LEN;

All MPI objects (e.g., MPI_Datatype, MPI_Comm) are of type INTEGER in Fortran.

17.2.292.9. NOTES ON COLLECTIVE OPERATIONS

The reduction functions ( MPI_Op ) do not return an error value. As a result, if the functions detect an error, all they can do is either call MPI_Abort or silently skip the problem. Thus, if you change the error handler from MPI_ERRORS_ARE_FATAL to something else, for example, MPI_ERRORS_RETURN , then no error may be indicated.

The reason for this is the performance problems in ensuring that all collective routines return the same error value.

17.2.292.10. ERRORS

Almost all MPI routines return an error value; C routines as the return result of the function and Fortran routines in the last argument.

Before the error value is returned, the current MPI error handler associated with the communication object (e.g., communicator, window, file) is called. If no communication object is associated with the MPI call, then the call is considered attached to MPI_COMM_SELF and will call the associated MPI error handler. When MPI_COMM_SELF is not initialized (i.e., before MPI_Init/MPI_Init_thread, after MPI_Finalize, or when using the Sessions Model exclusively) the error raises the initial error handler. The initial error handler can be changed by calling MPI_Comm_set_errhandler on MPI_COMM_SELF when using the World model, or the mpi_initial_errhandler CLI argument to mpiexec or info key to MPI_Comm_spawn/MPI_Comm_spawn_multiple. If no other appropriate error handler has been set, then the MPI_ERRORS_RETURN error handler is called for MPI I/O functions and the MPI_ERRORS_ABORT error handler is called for all other MPI functions.

Open MPI includes three predefined error handlers that can be used:

  • MPI_ERRORS_ARE_FATAL Causes the program to abort all connected MPI processes.

  • MPI_ERRORS_ABORT An error handler that can be invoked on a communicator, window, file, or session. When called on a communicator, it acts as if MPI_Abort was called on that communicator. If called on a window or file, acts as if MPI_Abort was called on a communicator containing the group of processes in the corresponding window or file. If called on a session, aborts only the local process.

  • MPI_ERRORS_RETURN Returns an error code to the application.

MPI applications can also implement their own error handlers by calling:

Note that MPI does not guarantee that an MPI program can continue past an error.

See the MPI man page for a full list of MPI error codes.

See the Error Handling section of the MPI-3.1 standard for more information.