Reduce_scatter

int MPI_Reduce_scatter(void *sbuf, void *rbuf, int *rcounts, 
                       MPI_Datatype dtype, MPI_Op op, MPI_Comm comm)
{
  int *disps = 0;
  int i, myrank, nprocs;
  void *tmpbuf;

  MPI_Comm_rank(comm, &myrank);
  MPI_Comm_size(comm, &nprocs);

  if (myrank == 0) {
    create a temporary buffer tmpbuf large enough for count dtypes;
    disps = malloc(nprocs * sizeof(int));

    disps[0] = 0;
    for (i = 0; i < (nprocs - 1); i++) {
      disps[i + 1] = disps[i] + rcounts[i];
    }
  }

  MPI_Reduce(sbuf, tmpbuf, count, dtype, op, 0, comm);

  MPI_Scatterv(tmpbuf, rcounts, disps, dtype, 
               rbuf, rcounts[myrank], dtype, 0, comm);

  if (disps) free(disps);

  return(MPI_SUCCESS);
}