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

#define WYMIAR 13

main ()
{
  double a[WYMIAR][WYMIAR];
  int n,i,j;

  for(i=0;i<WYMIAR;i++) for(j=0;j<WYMIAR;j++) a[i][j]=1.02*i+1.01*j;

  n=WYMIAR;

  double suma=0.0;
  for(i=0;i<WYMIAR;i++) {
    for(j=0;j<WYMIAR;j++) {
      suma += a[i][j];
    }
  }

  printf("Suma wyrazów tablicy: %lf\n", suma);

  omp_set_nested(1);
  int number_of_threads = 3;
  omp_set_num_threads(number_of_threads);
  double suma_p;
  int id;

  double suma_parallel=0.0;
  printf("\n Wierszowo zewnetrznie \n");

 //#pragma omp parallel for ordered schedule(static, 3) private(j) reduction(+ : suma_p)
 //#pragma omp parallel for ordered schedule(static) private(j) reduction(+ : suma_p)
 //#pragma omp parallel for ordered schedule(dynamic, 2) private(j) reduction(+ : suma_p)
  #pragma omp parallel for ordered schedule(dynamic) private(j) reduction(+ : suma_p)

  for(i=0;i<WYMIAR;i++) {
    int id_w = omp_get_thread_num();
    // ...
    for(j=0;j<WYMIAR;j++) {
      suma_parallel += a[i][j];
      #pragma omp ordered
      printf("(%2d,%2d)-W(%1d,%1d) ",i,j,id_w,omp_get_thread_num());
    }
    // ...
    printf("\n");
  }

  printf("Suma: %lf\n", suma_parallel);

  printf( "\nWierszowo wewnetrznie\n" );
	suma_p = 0.0;
	double *tab_sum = malloc(sizeof *tab_sum * number_of_threads);
	for ( i = 0; i < number_of_threads; i++ )//tablica do sumowania
		tab_sum[i] = 0.0;

	int k;

	for ( i = 0; i < WYMIAR; i++ )
	{
		id = omp_get_thread_num();

		//#pragma omp parallel for ordered schedule(static, 3) firstprivate(i) num_threads(3)
		//#pragma omp parallel for ordered schedule(static) firstprivate(i) num_threads(3)
		//#pragma omp parallel for ordered schedule(dynamic, 2) firstprivate(i) num_threads(3)
		#pragma omp parallel for ordered schedule(dynamic) firstprivate(i) num_threads(3)
		for ( j = 0; j < WYMIAR; j++ )
		{
			tab_sum[omp_get_thread_num()] += a[i][j];
			#pragma omp ordered
			printf("(%2d,%2d)-W(%1d,%1d)", i, j, id, omp_get_thread_num() );
		}
		printf("\n");
	}

	for ( k = 0; k < number_of_threads; k++ )
			suma_p += tab_sum[k];

	printf( "SUMA_P = %lf\n", suma_p );

	printf( "\nKolumnowo wewnetrznie\n" );
	suma_p = 0.0;
	for ( j = 0; j < WYMIAR; j++ )
	{
		id = omp_get_thread_num();
		//#pragma omp parallel for ordered schedule(static, 3) firstprivate(j) reduction(+ : suma_p)
		//#pragma omp parallel for ordered schedule(static) firstprivate(j) reduction(+ : suma_p)
		//#pragma omp parallel for ordered schedule(dynamic, 2) firstprivate(j) reduction(+ : suma_p)
		#pragma omp parallel for ordered schedule(dynamic) firstprivate(j) reduction(+ : suma_p)
		for ( i = 0; i < WYMIAR; i++ )
		{
			suma_p += a[i][j];
			#pragma omp ordered
			printf("(%2d,%2d)-W(%1d,%1d)", i, j, id, omp_get_thread_num() );
		}
		printf("\n");
	}
	printf( "SUMA_P = %lf\n", suma_p );



	printf( "\nKolumnowo zewnetrznie\n" );
	suma_p = 0.0;
	double suma_k = 0.0;

	//#pragma omp parallel for ordered schedule(static, 3) private(i, suma_p)
	//#pragma omp parallel for ordered schedule(static) private(i, suma_p)
	//#pragma omp parallel for ordered schedule(dynamic, 2) private(i, suma_p)
	#pragma omp parallel for ordered schedule(dynamic) private(i, suma_p)
	for ( j = 0; j < WYMIAR; j++ )
	{
		suma_p = 0.0;
		id = omp_get_thread_num();
		for ( i = 0; i < WYMIAR; i++ )
		{
			suma_p += a[i][j];
			#pragma omp ordered
			printf("(%2d,%2d)-W(%1d,%1d)", i, j, id, omp_get_thread_num() );
		}

		#pragma omp critical
		{
			suma_k += suma_p;
		}
		printf("\n");
	}
	printf( "SUMA_P = %lf\n", suma_k );

}