#include <mpi.h>
#include <sys/timeb.h>
#include <stdio.h>
#include <unistd.h>
#include <stdlib.h>

#define SZ 5000

int main(int argc, char** argv) {
    MPI_Status status;
    int i, j, np, id;
    float *b, *c;
    float **A;
    
    time_t sec;
    struct timeb t1, t2;
    unsigned short usec;

    MPI_Init(&argc, &argv);
    MPI_Comm_size(MPI_COMM_WORLD, &np);
    MPI_Comm_rank(MPI_COMM_WORLD, &id);

    b = (float *) malloc(sizeof(float) * SZ);
    c = (float *) malloc(sizeof(float) * SZ);    
    A = (float **) malloc(sizeof(float) * SZ);
    for(i = 0; i < SZ; i++) {
    	b[i] = i;
    	A[i] = (float *) malloc(sizeof(float) * SZ);
	for(j = 0; j < SZ; j++)
	    A[i][j] = i*j;
    }
    
    ftime(&t1);
    for(i = id; i < SZ; i += np) {
	c[i] = 0;
	for(j = 0; j < SZ; j++)
	    c[i] += A[i][j] * b[j];
    }
    if(np > 1) {
        if(id == 0) {
	    for(i = 0; i < SZ; i++)
		if(i != i%np)
    		    MPI_Recv(&c[i], 1, MPI_INT, MPI_ANY_SOURCE, i, MPI_COMM_WORLD, &status);
	}
	else
	    for(i = id; i < SZ; i++)
		MPI_Send(&c[i], 1, MPI_INT, 0, i, MPI_COMM_WORLD);
    }
    ftime(&t2);

    //Calculate time
    sec = t2.time - t1.time;
    usec = t2.millitm - t1.millitm;
    if(t2.millitm - t1.millitm < 0)
	sec--;
    printf("Thread %i/%i. Calculation time = %ld.%d.\n", id, np, sec, usec);
    
    MPI_Finalize();
    return 0;
}