sendbuf - send buffer (array)
sendcount - count of elements to send to each rank (integer)
sendtype - type of elements in send buffer (handle)
recvcount - count of elements in receive buffer (integer)
recvtype - type of elements in receive buffer (handle)
root - root rank (integer)
comm - communicator for messages (handle)
recvbuf - receive buffer (array)
The MPI_Scatter() function is used to send messages from a root process to every other process in the communicator comm . Given a vector described by sendcount elements of type sendtype for each rank, in an array specified by sendbuf , each process is sent the data that corresponds its location given by its rank within the communicator. This is the reverse process of the MPI_Gather() function.
The sends may (or may not) block depending on the amount of internal buffering done in the MPI implementation. At the root, if recvbuf is MPI_IN_PLACE then the destination of the message is taken from root's corresponding offset in sendbuf . This corresponds to no self-send or copy. Otherwise, the root will perform a copy from its requisite offset in sendbuf to recvbuf .
Note that each process should receive the same amount of data from root . If that is not the case, then the messages will be silently truncated. The recvtype and recvcount determine exactly how the messages are interpreted at each rank. Note that for performance reasons this MPI library does not check that the sent and received datatypes match.
If differing amounts of data need to be transferred from root , use the MPI_Scatterv() function.
The communicator must be a valid one (not MPI_COMM_NULL). PMPI_Scatter() is the profiling version of this function.
All MPI routines except for MPI_Wtime and MPI_Wtick return an error code. The the current MPI error handler is invoked if the return value is not MPI_SUCCESS. The default error handler aborts, but this may be changed with by using the MPI_Errhandler_set() function. The predefined error handler MPI_ERRORS_RETURN may be used to cause error values to be returned instead. Note that MPI does not guarentee that an MPI program can continue past an error. In this implementation, all errors except MPI_ERR_INTERN or MPI_ERR_OTHER should always be recoverable.
In C, the error code is passed as the return value. In FORTRAN, all functions have an parameter ierr which returns the error code. MPI C++ functions do not directly return an error code. However, C++ users may want to use the MPI::ERRORS_THROW_EXCEPTIONS handler. This will throw an MPI::Exception with the corresponding error code. To prevent exceptions from being raised from within C and Fortran code, they will see all error return values as MPI_ERR_PENDING when this handler is chosen. In this implementation, call MPI::throw_exception() to throw the correct exception if this occurs.
MPI_SUCCESS - No error;
MPI_ERR_PENDING - Pending exception;
MPI_ERR_COMM - Invalid communicator;
MPI_ERR_COUNT - Invalid element count;
MPI_ERR_BUFFER - Invalid buffer;
MPI_ERR_TYPE - Invalid data type;
MPI_ERR_ROOT - Invalid root;
MPI_ERR_INTERN - Out of Memory. This may be fatal.
Company Info |
Product Index |
Category Index |
Copyright © Lockless Inc All Rights Reserved.