MPI_Scatterv example

scatterv.c

#include <mpi.h>
#include <stdio.h>
#include <stdlib.h>
#include <math.h>

#define SIZE 4

int main(int argc, char *argv[])
{
    int rank, size; // for storing this process' rank, and the number of processes

    MPI_Init(&argc, &argv);
    MPI_Comm_rank(MPI_COMM_WORLD, &rank);
    MPI_Comm_size(MPI_COMM_WORLD, &size);

    int *sendcounts;            // array describing how many elements to send to each process
    int *displs;                // array describing the displacements where each segment begins
    int rem = (SIZE*SIZE)%size; // elements remaining after division among processes
    int sum = 0;                // Sum of counts. Used to calculate displacements
    char rec_buf[100];          // buffer where the received data should be stored

    // the data to be distributed
    char data[SIZE][SIZE] = {
        {'a', 'b', 'c', 'd'},
        {'e', 'f', 'g', 'h'},
        {'i', 'j', 'k', 'l'},
        {'m', 'n', 'o', 'p'}
    };

    sendcounts = malloc(sizeof(int)*size);
    displs = malloc(sizeof(int)*size);

    // calculate send counts and displacements
    for (int i = 0; i < size; i++) {
        sendcounts[i] = (SIZE*SIZE)/size;
        if (rem > 0) {
            sendcounts[i]++;
            rem--;
        }

        displs[i] = sum;
        sum += sendcounts[i];
    }

    // print calculated send counts and displacements for each process
    if (0 == rank) {
        for (int i = 0; i < size; i++) {
            printf("sendcounts[%d] = %d\tdispls[%d] = %d\n", i, sendcounts[i], i, displs[i]);
        }
    }

    // divide the data among processes as described by sendcounts and displs
    MPI_Scatterv(&data, sendcounts, displs, MPI_CHAR, &rec_buf, 100, MPI_CHAR, 0, MPI_COMM_WORLD);

    // print what each process received
    printf("%d: ", rank);
    for (int i = 0; i < sendcounts[rank]; i++) {
        printf("%c\t", rec_buf[i]);
    }
    printf("\n");

    MPI_Finalize();

    free(sendcounts);
    free(displs);

    return 0;
}

Uninitialized variable size is used at line 14 to calculate the remainder. But, size gets the number of processes later, at line 28. It can fixed moving lines 26-28 at the beginning of main(), between lines 10 and 11.

For those who want to know what it prints out without compiling:

1: e    f       g       h
sendcounts[0] = 4       displs[0] = 0
sendcounts[1] = 4       displs[1] = 4
sendcounts[2] = 4       displs[2] = 8
sendcounts[3] = 4       displs[3] = 12
0: a    b       c       d
3: m    n       o       p
2: i    j       k       l

Great, thanks

LO DE ABAJO ES EL CODIGO CORRECTO:
.
.
.
.
.
.
.
.
.

#include <mpi.h>
#include <stdio.h>
#include <stdlib.h>
#include <math.h>

#define N 4

int main(int argc, char *argv[])
{
int rank, numProcs; // para almacenar rango número de procesos

MPI_Init(&argc, &argv);
MPI_Comm_rank(MPI_COMM_WORLD, &rank);
MPI_Comm_size(MPI_COMM_WORLD, &numProcs);

int *sendcounts;    // matriz que describe cuántos elementos enviar a cada proceso
int *displs;        // array que describe los desplazamientos donde comienza cada segmento

int rem = (N*N)%numProcs; // Elementos que quedan después de la división entre procesos
int sum = 0;                // Suma de cuentas. Se utiliza para calcular desplazamientos
char rec_buf[100];          // buffer donde se deben almacenar los datos recibidos

// los datos a distribuir
char data[N][N] = {
    {'a', 'b', 'c', 'd'},
    {'e', 'f', 'g', 'h'},
    {'i', 'j', 'k', 'l'},
    {'m', 'n', 'o', 'p'}
};


sendcounts = malloc(sizeof(int)*numProcs);
displs = malloc(sizeof(int)*numProcs);

// calcular conteos y desplazamientos de envío
for (int i = 0; i < numProcs; i++) {
    sendcounts[i] = (N*N)/numProcs;

    if (rem > 0) {
        sendcounts[i]++;
        rem--;
    }

    displs[i] = sum;
    sum += sendcounts[i];
}

// imprimir los conteos y desplazamientos de envío calculados para cada proceso
if (0 == rank) {
    for (int i = 0; i < numProcs; i++) {
        printf("sendcounts[%d] = %d\tdispls[%d] = %d\n", i, sendcounts[i], i, displs[i]);
    }
}

// dividir los datos entre procesos según lo descrito por sendcounts y displs
MPI_Scatterv(&data, sendcounts, displs, MPI_CHAR, &rec_buf, 100, MPI_CHAR, 0, MPI_COMM_WORLD);

// imprime lo que recibió cada proceso
printf("%d: ", rank);
for (int i = 0; i < sendcounts[rank]; i++) {
    printf("%c\t", rec_buf[i]);
}
printf("\n");

MPI_Finalize();

free(sendcounts);
free(displs);

return 0;

}

Thanks a lot, very helpful ❤️

Thank you for the code, very helpfull