gather large sub Matrices in MPI in c - c

Question

i wanna make a program that calculate matrix multiplication "large matrices 1000*1000 for example"
in master-worker paradigm and select to each worker number of rows it will work on
for example if i have 1000*1000 matrix and i got 4 processes
process 0 is master wont work on it
process 1 will get from 0 to 333
process 2 will get from 333 to 666
process 3 will get from 666 to 1000
simply here is my code
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "mpi.h"
int **alloc(int rows, int cols)
{
int *data = (int *)malloc(rows*cols*sizeof(int));
int **array= (int **)malloc(rows*sizeof(int*));
int i;
for (i=0; i<rows; i++)
array[i] = &(data[cols*i]);
return array;
}
void free2D(int **mat)
{
free(mat[0]);
free(mat);
}
int main(int argc, char * argv[])
{
int my_rank; /* rank of process */
int p; /* number of process */
int source; /* rank of sender */
int dest; /* rank of reciever */
int tag = 0; /* tag for messages */
MPI_Status status; /* return status for */
MPI_Init( &argc, &argv );
MPI_Comm_rank(MPI_COMM_WORLD, &my_rank);
MPI_Comm_size(MPI_COMM_WORLD, &p);
int **mat1;
int **mat2;
int **mat3;
int n,r,m;
int i,j,k,sum;
int start[100];
int end[100];
int istart,iend;
int avg,extra,taken,temp;
if (my_rank==0)
{
n=1000;
r=1000;
m=1000;
mat1=alloc(n,r);
mat2=alloc(r,m);
mat3=alloc(n,m);
//found = (int *)malloc(n*m*sizeof(int));
MPI_Bcast(&n, 1, MPI_INT, 0, MPI_COMM_WORLD);
MPI_Bcast(&r, 1, MPI_INT, 0, MPI_COMM_WORLD);
MPI_Bcast(&m, 1, MPI_INT, 0, MPI_COMM_WORLD);
avg=n/(p-1);
extra=n%(p-1);
temp=0;
for(i=0; i<(p-1); i++)
{
taken=(extra--<=0)? avg: avg+1;
start[i]=temp;
temp+=taken;
end[i]=temp;
}
for (source=1; source<p; source++)
{
MPI_Send(&start[source-1],1,MPI_INT,source,tag,MPI_COMM_WORLD);
MPI_Send(&end[source-1],1,MPI_INT,source,tag,MPI_COMM_WORLD);
}
for (i=0; i<n; i++)
{
for (j=0; j<r; j++)
{
mat1[i][j]=1;
MPI_Bcast(&(mat1[i][j]), 1, MPI_INT, 0, MPI_COMM_WORLD);
}
}
for (i=0; i<r; i++)
{
for (j=0; j<m; j++)
{
mat2[i][j]=5;
MPI_Bcast(&(mat2[i][j]), 1, MPI_INT, 0, MPI_COMM_WORLD);
}
}
///here i supposed to receive my sub matrices from worker
free2D(mat1);
free2D(mat2);
free2D(mat3);
}
else{
MPI_Bcast(&n, 1, MPI_INT, 0, MPI_COMM_WORLD);
MPI_Bcast(&r, 1, MPI_INT, 0, MPI_COMM_WORLD);
MPI_Bcast(&m, 1, MPI_INT, 0, MPI_COMM_WORLD);
MPI_Recv(&istart,1,MPI_INT,0,tag, MPI_COMM_WORLD,&status);
MPI_Recv(&iend,1,MPI_INT,0,tag, MPI_COMM_WORLD,&status);
mat1=alloc(n,r);
mat2=alloc(r,m);
mat3=alloc(iend-istart,m);
for (i=0; i<n; i++)
{
for (j=0; j<r; j++)
{
MPI_Bcast(&(mat1[i][j]), 1, MPI_INT, 0, MPI_COMM_WORLD);
}
}
for (i=0; i<r; i++)
{
for (j=0; j<m; j++)
{
MPI_Bcast(&(mat2[i][j]), 1, MPI_INT, 0, MPI_COMM_WORLD);
}
}
sum=0;
for (i=istart; i<iend; i++)
{
for (j=0; j<m; j++)
{
for(k=0; k<r; k++)
{
sum += mat1[i][k]*mat2[k][j];
}
// printf("%d %d\n",i,j);
mat3[i-istart][j]=sum;
sum = 0;
}
}
/// here i supposed to send sub matrices
//printMat(mat3,iend-istart,m);
free2D(mat1);
free2D(mat2);
free2D(mat3);
}
/* shutdown MPI */
MPI_Finalize();
return 0;
}
it works fine every worker calculate its sub matrix correctly but i don't know how to send them back to master and it takes alot of time if i used MPI_Send and MPI_Recv "as its point to point messaging"
Any idea how to gather them ? i'm new to MPI btw.