Use tiles and queue of them for rendering.

Makefile

@@ -1,6 +1,6 @@
 CXX = g++
 
-CXXFLAGS  = -std=c++14 -Wall -Wextra -Wno-unused-parameter -march=native -Ofast -fopenmp -flto
+CXXFLAGS  = -std=c++20 -Wall -Wextra -Wno-unused-parameter -march=native -Ofast -flto -fopenmp
 
 DEPS = util.h vec3.h color.h ray.h camera.h hittable.h hittable_list.h sphere.h material.h lodepng.h
 OBJ = wtracer.o material.o vec3.o lodepng.o
@@ -20,5 +20,5 @@ $(TARGET): $(OBJ)
 	$(CXX) $(CXXFLAGS) -o $@ $^
 
 clean:
-	$(RM) $(TARGET) $(OBJ) image.ppm
+	$(RM) $(TARGET) $(OBJ) image.png
 

wtracer.cpp

@@ -1,10 +1,18 @@
 #include <iostream>
 #include <iomanip>
+#include <atomic>
 #include <memory>
+#include <cstdio>
 
 #define LODEPNG_NO_COMPILE_DECODER
 #include "lodepng.h"
 
+#include <boost/lockfree/queue.hpp>
+
+#ifdef _OPENMP
+#include <omp.h>
+#endif
+
 #include "color.h"
 #include "vec3.h"
 #include "ray.h"
@@ -31,11 +39,9 @@ Color ray_color(const Ray& r, const Hittable& world, int depth) {
     return (1.0 - t) * Color(1.0, 1.0, 1.0) + t * Color(0.5, 0.7, 1.0);
 }
 
-Hittable_list setup_random_scene() {
+Hittable_list setup_random_scene(const int sph_i) {
     Hittable_list world;
 
-    const int sph_i = 3;
-
     auto ground_material = std::make_shared<Lambertian>(Color(0.5, 0.5, 0.5));
     world.add(std::make_shared<Sphere>(Point3(0, -1000, 0), 1000, ground_material));
     for (int a = -sph_i; a<sph_i; a++) {
@@ -45,7 +51,7 @@ Hittable_list setup_random_scene() {
 
             if ((center - Point3(4, 0.2, 0)).length() > 0.9) {
                 std::shared_ptr<Material> sphere_material;
-                if (choose_mat < 0.8) {
+                if (choose_mat < 0.7) {
                     // diffuse
                     auto albedo = Color::random() * Color::random();
                     sphere_material = std::make_shared<Lambertian>(albedo);
@@ -78,17 +84,78 @@ Hittable_list setup_random_scene() {
     return world;
 }
 
+struct render_params {
+    int image_width;
+    int image_height;
+    int samples_per_pixel;
+    int max_depth;
+    Hittable_list& world;
+    Camera& cam;
+};
+
+struct render_tile {
+    int start_x;
+    int start_y;
+    int end_x;
+    int end_y;
+};
+
+void split_line(const int& line_length, int& start, int& end) {
+#ifdef _OPENMP
+    const int line_length_thread = static_cast<int>(line_length/omp_get_num_threads());
+    start = omp_get_thread_num() * line_length_thread;
+    end = std::min((omp_get_thread_num()+1) * line_length_thread, line_length);
+#else
+    start = 0;
+    end = line_length;
+#endif
+}
+
+void render(boost::lockfree::queue<render_tile>& queue,
+        std::atomic<int>& queue_counter,
+        const render_params& params, std::vector<Color>& image) {
+    render_tile rt;
+    while (queue.pop(rt)) {
+        --queue_counter;
+        int qc = queue_counter;
+        printf("\rTiles remaining: %4d", qc);
+        std::cout << std::flush;
+        for (int i = rt.start_x; i<rt.end_x; ++i) {
+            for (int j = rt.start_y; j<rt.end_y; ++j) {
+                Color pixel_color(0, 0, 0);
+                for (int s = 0; s<params.samples_per_pixel; ++s) {
+                    auto u = double(j + random_double(-0.5, 0.5)) / (params.image_width - 1);
+                    auto v = double(i + random_double(-0.5, 0.5)) / (params.image_height - 1);
+                    Ray r = params.cam.get_ray(u, v);
+                    pixel_color += ray_color(r, params.world, params.max_depth);
+                }
+                image[i*params.image_width+j] = pixel_color;
+            }
+        }
+    }
+}
+
 int main() {
     const auto aspect_ratio = 16.0 / 9.0;
+    //const int image_width = 1920;
     //const int image_width = 1280;
-    const int image_width = 768;
-    //const int image_width = 384;
+    //const int image_width = 768;
+    const int image_width = 384;
     const int image_height = static_cast<int>(image_width / aspect_ratio);
+    //const int samples_per_pixel = 2000;
     //const int samples_per_pixel = 1000;
     const int samples_per_pixel = 400;
     const int max_depth = 50;
 
-    auto world = setup_random_scene();
+    //const int tile_size = 128;
+    const int tile_size = 64;
+
+    const int sph_i = 3;
+    //const int sph_i = 5;
+    //const int sph_i = 7;
+    //const int sph_i = 11;
+
+    auto world = setup_random_scene(sph_i);
 
     Point3 lookfrom(13, 2, 3);
     Point3 lookat(0, 0, 0);
@@ -98,21 +165,36 @@ int main() {
 
     Camera cam(lookfrom, lookat, vup, 20, aspect_ratio, aperture, dist_to_focus);
 
+    render_params params = {
+        image_width,
+        image_height,
+        samples_per_pixel,
+        max_depth,
+        world,
+        cam,
+    };
+
+    std::atomic<int> queue_counter;
+    boost::lockfree::queue<render_tile> queue(0);
+
+    for (int i = 0; i<std::ceil(double(image_height)/double(tile_size)); i++) {
+        for (int j = 0; j<std::ceil(double(image_width)/double(tile_size)); j++) {
+            render_tile rt = {
+                i*tile_size,
+                j*tile_size,
+                std::min((i+1)*tile_size, image_height),
+                std::min((j+1)*tile_size, image_width)
+            };
+            queue.push(rt);
+            ++queue_counter;
+        }
+    }
+
     std::vector<Color> image(image_width * image_height);
 
-    for (int i = image_height-1; i>=0; --i) {
-        std::cerr << "\rScanline remaining: " << std::setw(4) << i << std::flush;
-        #pragma omp parallel for
-        for (int j = 0; j<image_width; ++j) {
-            Color pixel_color(0, 0, 0);
-            for (int s = 0; s<samples_per_pixel; ++s) {
-                auto u = double(j + random_double(-0.5, 0.5)) / (image_width - 1);
-                auto v = double(i + random_double(-0.5, 0.5)) / (image_height - 1);
-                Ray r = cam.get_ray(u, v);
-                pixel_color += ray_color(r, world, max_depth);
-            }
-            image.at(i*image_width+j) = pixel_color;
-        }
+    #pragma omp parallel
+    {
+        render(queue, queue_counter, params, image);
     }
 
     std::cerr << "\nAssembling image.\n";
@@ -120,7 +202,7 @@ int main() {
     std::vector<unsigned char> img_lode(image_width * image_height * 3);
     for (int i = 0; i<image_height; ++i) {
         for (int j = 0; j<image_width; ++j) {
-            write_color_vec(img_lode, i*image_width*3+j*3, image.at((image_height-1-i)*image_width+j), samples_per_pixel);
+            write_color_vec(img_lode, i*image_width*3+j*3, image[(image_height-1-i)*image_width+j], samples_per_pixel);
         }
     }
     std::cerr << "Writing file \"image.png\".\n";