// Adrien Bourmault 3677850
// Volodymyr Patuta
// Groupe K

#include "easypap.h"

#include <omp.h>

///////////////////////////// Sequential version (tiled)
// Suggested cmdline(s):
// ./run -l images/1024.png -k blur -v seq -si
//
// resultat -O3 : 3295.680
// resultat -O2 : 3240.540
// resultat -O1 : 3289.684
// resultat -O0 : 9681.364
//
int blur_do_tile_default (int x, int y, int width, int height)
{
	for (int i = y; i < y + height; i++)
		for (int j = x; j < x + width; j++) {
			unsigned r = 0, g = 0, b = 0, a = 0, n = 0;

			int i_d = (i > 0) ? i - 1 : i;
			int i_f = (i < DIM - 1) ? i + 1 : i;
			int j_d = (j > 0) ? j - 1 : j;
			int j_f = (j < DIM - 1) ? j + 1 : j;

			for (int yloc = i_d; yloc <= i_f; yloc++)
				for (int xloc = j_d; xloc <= j_f; xloc++) {
					unsigned c = cur_img (yloc, xloc);
					r += extract_red (c);
					g += extract_green (c);
					b += extract_blue (c);
					a += extract_alpha (c);
					n += 1;
				}

			r /= n;
			g /= n;
			b /= n;
			a /= n;

			next_img (i, j) = rgba (r, g, b, a);
		}

	return 0;
}

int blur_do_tile_default_nb (int x, int y, int width, int height)
{
	for (int i = y + 1; i < y + height - 1; i++) {
		for (int j = x + 1; j < x + width - 1; j++) {
			unsigned r = 0, g = 0, b = 0, a = 0, n = 0;
			for (int yloc = i - 1; yloc <= i + 1; yloc++) {
				for (int xloc = j - 1; xloc <= j + 1; xloc++) {
					unsigned c = cur_img (yloc, xloc);
					r += extract_red (c);
					g += extract_green (c);
					b += extract_blue (c);
					a += extract_alpha (c);
					n += 1;
				}
			}

			r /= n;
			g /= n;
			b /= n;
			a /= n;

			next_img (i, j) = rgba (r, g, b, a);
		}
	}

	for (int i = 0; i < width - 1; i++) {
		next_img (i, 0) = cur_img(i, 0);
		next_img (i, DIM - 1) = cur_img(i, DIM - 1);
		next_img (0, i) = cur_img(0, i);
		next_img (DIM - 1, i) = cur_img(DIM - 1, i);
	}

	return 0;
}

// resultat optim1 Denver: 7349.435
// resultat optim1 Cortex: 23348.009
// resultat nb Denver: 13512.337
// resultat nb Cortex: 27133.248
int blur_do_tile_default_optim1 (int x, int y, int width, int height)
{
	for (int i = y + 1; i < y + height - 1; i++) {
		for (int j = x + 1; j < x + width - 1; j++) {
			unsigned r = 0, g = 0, b = 0, a = 0, n = 0;
			for (int yloc = i - 1; yloc <= i + 1; yloc++) {
				unsigned c = cur_img (yloc, j - 1);
				r += extract_red (c);
				g += extract_green (c);
				b += extract_blue (c);
				a += extract_alpha (c);
				c = cur_img (yloc, j);
				r += extract_red (c);
				g += extract_green (c);
				b += extract_blue (c);
				a += extract_alpha (c);
				c = cur_img (yloc, j + 1);
				r += extract_red (c);
				g += extract_green (c);
				b += extract_blue (c);
				a += extract_alpha (c);
				n += 3;
			}

			r /= n;
			g /= n;
			b /= n;
			a /= n;

			next_img (i, j) = rgba (r, g, b, a);
		}
	}

	for (int i = 0; i < width - 1; i++) {
		next_img (i, 0) = cur_img(i, 0);
		next_img (i, DIM - 1) = cur_img(i, DIM - 1);
		next_img (0, i) = cur_img(0, i);
		next_img (DIM - 1, i) = cur_img(DIM - 1, i);
	}

	return 0;
}

// resultat Denver: 6355.280
// resultat Cortex: 23297.872
int blur_do_tile_default_optim2 (int x, int y, int width, int height)
{
	for (int i = y + 1; i < y + height - 1; i++) {
		for (int j = x + 1; j < x + width - 1; j++) {
			unsigned r = 0, g = 0, b = 0, a = 0, n = 0;

			unsigned c = cur_img (i - 1, j - 1);
			r += extract_red (c);
			g += extract_green (c);
			b += extract_blue (c);
			a += extract_alpha (c);
			c = cur_img (i - 1, j);
			r += extract_red (c);
			g += extract_green (c);
			b += extract_blue (c);
			a += extract_alpha (c);
			c = cur_img (i - 1, j + 1);
			r += extract_red (c);
			g += extract_green (c);
			b += extract_blue (c);
			a += extract_alpha (c);

			c = cur_img (i, j - 1);
			r += extract_red (c);
			g += extract_green (c);
			b += extract_blue (c);
			a += extract_alpha (c);
			c = cur_img (i, j);
			r += extract_red (c);
			g += extract_green (c);
			b += extract_blue (c);
			a += extract_alpha (c);
			c = cur_img (i, j + 1);
			r += extract_red (c);
			g += extract_green (c);
			b += extract_blue (c);
			a += extract_alpha (c);

			c = cur_img (i + 1, j - 1);
			r += extract_red (c);
			g += extract_green (c);
			b += extract_blue (c);
			a += extract_alpha (c);
			c = cur_img (i + 1, j);
			r += extract_red (c);
			g += extract_green (c);
			b += extract_blue (c);
			a += extract_alpha (c);
			c = cur_img (i + 1, j + 1);
			r += extract_red (c);
			g += extract_green (c);
			b += extract_blue (c);
			a += extract_alpha (c);

			n += 9;

			r /= n;
			g /= n;
			b /= n;
			a /= n;

			next_img (i, j) = rgba (r, g, b, a);
		}
	}

	for (int i = 0; i < width - 1; i++) {
		next_img (i, 0) = cur_img(i, 0);
		next_img (i, DIM - 1) = cur_img(i, DIM - 1);
		next_img (0, i) = cur_img(0, i);
		next_img (DIM - 1, i) = cur_img(DIM - 1, i);
	}

	return 0;
}

// resultat Denver: 5189.197
// resultat Cortex: 16848.161
int blur_do_tile_default_optim3 (int x, int y, int width, int height)
{
	for (int i = y + 1; i < y + height - 1; i++) {
		for (int j = x + 1; j < x + width - 1; j++) {
			unsigned r = 0, g = 0, b = 0, a = 0, n = 0;

			unsigned c = cur_img (i - 1, j - 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i - 1, j);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i - 1, j + 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;

			c = cur_img (i, j - 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i, j);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i, j + 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;

			c = cur_img (i + 1, j - 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i + 1, j);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i + 1, j + 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;

			n += 9;

			r /= n;
			g /= n;
			b /= n;
			a /= n;

			next_img (i, j) = rgba (r, g, b, a);
		}
	}

	for (int i = 0; i < width - 1; i++) {
		next_img (i, 0) = cur_img(i, 0);
		next_img (i, DIM - 1) = cur_img(i, DIM - 1);
		next_img (0, i) = cur_img(0, i);
		next_img (DIM - 1, i) = cur_img(DIM - 1, i);
	}

	return 0;
}

int blur_do_tile_default_optim4 (int x, int y, int width, int height)
{
	for (int i = y + 1; i < y + height - 1; i++) {
		for (int j = x + 1; j < x + width - 1; j++) {
			unsigned r = 0, g = 0, b = 0, a = 0, n = 0;

			unsigned c = cur_img (i - 1, j - 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i - 1, j);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i - 1, j + 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;

			c = cur_img (i, j - 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i, j);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i, j + 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;

			c = cur_img (i + 1, j - 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i + 1, j);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;
			c = cur_img (i + 1, j + 1);
			r += c >> 24;
			g += (c >> 16) & 255;
			b += (c >> 8) & 255;
			a += c & 255;

			n += 9;

			r /= n;
			g /= n;
			b /= n;
			a /= n;

			next_img (i, j) = rgba (r, g, b, a);
		}
	}

	for (int i = 0; i < width - 1; i++) {
		next_img (i, 0) = cur_img(i, 0);
		next_img (i, DIM - 1) = cur_img(i, DIM - 1);
		next_img (0, i) = cur_img(0, i);
		next_img (DIM - 1, i) = cur_img(DIM - 1, i);
	}

	return 0;
}

///////////////////////////// Sequential version (tiled)
// Suggested cmdline(s):
// ./run -l images/1024.png -k blur -v seq
//
unsigned blur_compute_seq (unsigned nb_iter)
{
	for (unsigned it = 1; it <= nb_iter; it++) {

		do_tile (0, 0, DIM, DIM, 0);

		swap_images ();
	}

	return 0;
}

///////////////////////////// Tiled sequential version (tiled)
// Suggested cmdline(s):
// ./run -l images/1024.png -k blur -v tiled -ts 32 -m si
//
unsigned blur_compute_tiled (unsigned nb_iter)
{
	for (unsigned it = 1; it <= nb_iter; it++) {

		for (int y = 0; y < DIM; y += TILE_H)
			for (int x = 0; x < DIM; x += TILE_W)
				do_tile (x, y, TILE_W, TILE_H, 0);

		swap_images ();
	}

	return 0;
}