// MPI Initialization

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

	numWorkProcs = CalcNumBlocks(size, num_proc, blockSize);
	numBlocksWidth=size/blockSize;

// Buffer Initialization 
	int buffer_size = 2*size*size+size;
	int *buffer = new int[buffer_size];
	MPI_Buffer_attach(buffer,buffer_size);

	if (rank < numWorkProcs)
	{
// Memory Allocation
	Matrix blockC = MatrixInitialization(blockSize,0); 
	Matrix blockA = MatrixInitialization(blockSize,0); 
	Matrix blockB = MatrixInitialization(blockSize,0); 
	
	if (rank==0)
		{
			A=MatrixInitialization(size,1); 
			B=MatrixInitialization(size,1);
			C=MatrixInitialization(size,0);

			starttime = MPI_Wtime(); 
			A=Repack(A,size,blockSize);
			B=Repack(B,size,blockSize);			
		}
	
	MPI_Barrier(MPI_Comm_Work);

// Delivery blocks
	MPI_Scatter(A, blockSize*blockSize,MPI_INT,blockA,blockSize*blockSize,MPI_INT,0,MPI_Comm_Work);
	MPI_Scatter(B, blockSize*blockSize,MPI_INT,blockB,blockSize*blockSize,MPI_INT,0,MPI_Comm_Work);

	int row_num = (int)rank/numBlocksWidth; 
	int col_num = rank%numBlocksWidth; 

	if (row_num!=0)
	{
		if (col_num < row_num)
			MPI_Bsend (blockA,blockSize*blockSize,MPI_INT,rank+numBlocksWidth-row_num,0,MPI_Comm_Work);
		else
			MPI_Bsend (blockA,blockSize*blockSize,MPI_INT,rank-row_num,0,MPI_Comm_Work);
	}

	
	if (col_num!=0)
		if (row_num < col_num)
			MPI_Bsend (blockB,blockSize*blockSize,MPI_INT,rank+(numBlocksWidth-col_num)*numBlocksWidth,1,MPI_Comm_Work);
		else
			MPI_Bsend (blockB,blockSize*blockSize,MPI_INT,rank-numBlocksWidth*col_num,1,MPI_Comm_Work);

	if ((row_num!=0)&&(col_num!=0)) 
	{
		MPI_Recv(blockA,blockSize*blockSize,MPI_INT,MPI_ANY_SOURCE,0,MPI_Comm_Work,&status);
		MPI_Recv(blockB,blockSize*blockSize,MPI_INT,MPI_ANY_SOURCE,1,MPI_Comm_Work,&status);
	}
	else
		if((row_num==0)&&(col_num!=0)) 
			MPI_Recv(blockB,blockSize*blockSize,MPI_INT,MPI_ANY_SOURCE,1,MPI_Comm_Work,&status);
		else
		if ((row_num!=0)&&(col_num==0))
			MPI_Recv(blockA,blockSize*blockSize,MPI_INT,MPI_ANY_SOURCE,0,MPI_Comm_Work,&status);

	MPI_Barrier(MPI_Comm_Work);

// Multiplication
	MultMatrix (blockA,blockB,blockC,blockSize);


	for (int i=0; i < numBlocksWidth-1; i++)
	{
		if (col_num==0)
			MPI_Bsend (blockA,blockSize*blockSize,MPI_INT,(row_num+1)*numBlocksWidth-1,0,MPI_Comm_Work);
		else
			MPI_Bsend (blockA,blockSize*blockSize,MPI_INT,rank-1,0,MPI_Comm_Work);
	
		if (rank < numBlocksWidth)
			MPI_Bsend (blockB,blockSize*blockSize,MPI_INT,rank+(numBlocksWidth-1)*numBlocksWidth,1,MPI_Comm_Work);
		else
			MPI_Bsend (blockB,blockSize*blockSize,MPI_INT,rank-numBlocksWidth,1,MPI_Comm_Work);

		
// Receive blocks
	MPI_Recv(blockA,blockSize*blockSize,MPI_INT,MPI_ANY_SOURCE,0,MPI_Comm_Work,&status);
	MPI_Recv(blockB,blockSize*blockSize,MPI_INT,MPI_ANY_SOURCE,1,MPI_Comm_Work,&status);
	MultMatrix (blockA,blockB,blockC,blockSize);
	}

// Assembly of the resulting matrix	
	MPI_Gather (blockC,blockSize*blockSize,MPI_INT,C,blockSize*blockSize,MPI_INT,0,MPI_Comm_Work);

	MPI_Barrier(MPI_Comm_Work);
		
	if (rank==0)
		{
			C=BackRepack(C,size,blockSize);
			endtime = MPI_Wtime();
			time1=endtime-starttime;
		}