CODEWITHSUNDEEP
cmpi
Sean Wayland
Sean Wayland

Reputation: 39

difficulty understanding MPI scatter and gather in C

I am trying to learn to use MPI. Below is my simple program to test MPI scatter and gather. I don't understand how it works and why it produces the result

1 2 3 4 4 5 6 7 8 9 10 11

instead of expected

1 2 3 4 5 6 7 8 9 10 11 12

The documentation and all the examples I can find are too complicated/poorly worded for me to understand. I just want to scatter an array across 3 processes and add one to each value in each process. Alternatively I would be happy to see how a 2D array was sent row by row to each process and each row was processed simply.

int main(int argc, char **argv) {
    int rank; // my process ID
    int size = 3; // number of processes/nodes
    MPI_Status status;
    MPI_Init(&argc, &argv); // start MPI
    MPI_Comm_size(MPI_COMM_WORLD, &size); // initialize MPI
    MPI_Comm_rank(MPI_COMM_WORLD, &rank);

    unsigned char inData[12]; // data returned after being "processed"
    unsigned char outData[12]; // buffer for receiving data
    unsigned long datasize = 12; // size of data to process
    unsigned char testData[12]; // data to be processed

    if (rank == 0) {
        // initialize data
        for (int i = 0; i < datasize; i++) {
            testData[i] = i;
            outData[i] = 0;
            inData[i] = 0;
        }
    }

    // scatter the data to the processes
    // I am not clear about the numbers sent in and out
    MPI_Scatter(&testData, 12, MPI_UNSIGNED_CHAR, &outData, 
        12, MPI_UNSIGNED_CHAR, 0, MPI_COMM_WORLD);
    MPI_Barrier(MPI_COMM_WORLD);

    // process data
    for (int i = 0; i < 4; i++) { outData[i] = outData[i] + 1; }

    MPI_Barrier(MPI_COMM_WORLD);

    // gather processed data
    MPI_Gather(&outData, 12, MPI_UNSIGNED_CHAR, &inData, 
        12, MPI_UNSIGNED_CHAR, 0, MPI_COMM_WORLD);

    //print processed data from root 
    if (rank == 0) {
        for (int i = 0; i < 12; i++) {
            printf("\n%d", inData[i]);
        }

        MPI_Finalize();
    }

    return 0;
}

Upvotes: 0

Views: 1760

Answers (1)

Evg
Evg

Reputation: 26362

Though your main error is using 12 instead of 4, let's do it step-by-step.

// int size = 3; // number of processes/nodes
int size;
...
MPI_Comm_size(MPI_COMM_WORLD, &size); // initialize MPI
assert(size == 3);

There is no point in setting size to 3. This value will be overwritten by MPI_Comm_size with the actual number of processes. This number is determined by how you run your MPI application (e.g. mpirun -np 3).

//unsigned char outData[12]; // buffer for receiving data
unsigned char outData[4];

We have 12 elements and 3 processes, 4 elements per processes. So, 4 elements are enough for outData.

outData[i] = 0;
inData[i] = 0;

There is no point in zeroing these buffers, they will be overwritten.

// scatter the data to the processes
// I am not clear about the numbers sent in and out
MPI_Scatter(&testData, 4 /*12*/, MPI_UNSIGNED_CHAR, &outData,
    4 /*12*/, MPI_UNSIGNED_CHAR, 0, MPI_COMM_WORLD);

We have 4 elements per processes, so the number should be 4, not 12.

MPI_Barrier(MPI_COMM_WORLD);

You don't need barriers here.

MPI_Gather(&outData, 4 /*12*/, MPI_UNSIGNED_CHAR, &inData, 
    4 /*12*/, MPI_UNSIGNED_CHAR, 0, MPI_COMM_WORLD);

Same story, 4 instead of 12.

MPI_Finalize();

This should be called by all processes.

Upvotes: 2

Related Questions