LibSoftGPU: Add option to render a debug overlay

This displays statistics regarding frame timings and number of pixels
rendered.

Timings are based on the time between draw_debug_overlay() invocations.
This measures actual number of frames presented to the user vs. wall
clock time so this also includes everything the app might do besides
rendering.

Triangles are counted after clipping. This number might actually be
higher than the number of triangles coming from LibGL.

Pixels are counted after the initial scissor and coverage test. Pixels
rejected here are not counted. Shaded pixels is the percentage of all
pixels that made it to the shading stage. Blended pixels is the
percentage of shaded pixels that were alpha blended to the color buffer.

Overdraw measures how many pixels were shaded vs. how many pixels the
render target has. e.g. a 640x480 render target has 307200 pixels. If
exactly that many pixels are shaded the overdraw number will read 0%.
614400 shaded pixels will read as an overdraw of 100%.

Sampler calls is simply the number of times sampler.sample_2d() was
called.

3 files changed, 86 insertions, 1 deletions

diff --git a/Userland/Libraries/LibSoftGPU/Config.h b/Userland/Libraries/LibSoftGPU/Config.h
index 8a551ca40b..0292a153cd 100644
--- a/Userland/Libraries/LibSoftGPU/Config.h
+++ b/Userland/Libraries/LibSoftGPU/Config.h
@@ -6,8 +6,15 @@
 
 #pragma once
 
+#define INCREASE_STATISTICS_COUNTER(stat, n)     \
+    do {                                         \
+        if constexpr (ENABLE_STATISTICS_OVERLAY) \
+            stat += (n);                         \
+    } while (0)
+
 namespace SoftGPU {
 
+static constexpr bool ENABLE_STATISTICS_OVERLAY = false;
 static constexpr int RASTERIZER_BLOCK_SIZE = 8;
 static constexpr int NUM_SAMPLERS = 32;
 
diff --git a/Userland/Libraries/LibSoftGPU/Device.cpp b/Userland/Libraries/LibSoftGPU/Device.cpp
index f97d95b18b..da18894ec1 100644
--- a/Userland/Libraries/LibSoftGPU/Device.cpp
+++ b/Userland/Libraries/LibSoftGPU/Device.cpp
@@ -6,6 +6,7 @@
  */
 
 #include <AK/Function.h>
+#include <LibCore/ElapsedTimer.h>
 #include <LibGfx/Painter.h>
 #include <LibGfx/Vector2.h>
 #include <LibGfx/Vector3.h>
@@ -14,6 +15,12 @@
 
 namespace SoftGPU {
 
+static long long g_num_rasterized_triangles;
+static long long g_num_pixels;
+static long long g_num_pixels_shaded;
+static long long g_num_pixels_blended;
+static long long g_num_sampler_calls;
+
 using IntVector2 = Gfx::Vector2<int>;
 using IntVector3 = Gfx::Vector3<int>;
 
@@ -113,6 +120,8 @@ static constexpr void setup_blend_factors(BlendFactor mode, FloatVector4& consta
 template<typename PS>
 static void rasterize_triangle(const RasterizerOptions& options, Gfx::Bitmap& render_target, DepthBuffer& depth_buffer, const Triangle& triangle, PS pixel_shader)
 {
+    INCREASE_STATISTICS_COUNTER(g_num_rasterized_triangles, 1);
+
     // Since the algorithm is based on blocks of uniform size, we need
     // to ensure that our render_target size is actually a multiple of the block size
     VERIFY((render_target.width() % RASTERIZER_BLOCK_SIZE) == 0);
@@ -257,6 +266,7 @@ static void rasterize_triangle(const RasterizerOptions& options, Gfx::Bitmap& re
 
             // Generate the coverage mask
             if (!options.scissor_enabled && test_point(b0) && test_point(b1) && test_point(b2) && test_point(b3)) {
+                INCREASE_STATISTICS_COUNTER(g_num_pixels, RASTERIZER_BLOCK_SIZE * RASTERIZER_BLOCK_SIZE);
                 // The block is fully contained within the triangle. Fill the mask with all 1s
                 for (int y = 0; y < RASTERIZER_BLOCK_SIZE; y++)
                     pixel_mask[y] = -1;
@@ -268,8 +278,10 @@ static void rasterize_triangle(const RasterizerOptions& options, Gfx::Bitmap& re
                     pixel_mask[y] = 0;
 
                     for (int x = 0; x < RASTERIZER_BLOCK_SIZE; x++, coords += dbdx) {
-                        if (test_point(coords) && (!options.scissor_enabled || render_bounds.contains(x0 + x, y0 + y)))
+                        if (test_point(coords) && (!options.scissor_enabled || render_bounds.contains(x0 + x, y0 + y))) {
+                            INCREASE_STATISTICS_COUNTER(g_num_pixels, 1);
                             pixel_mask[y] |= 1 << x;
+                        }
                     }
                 }
             }
@@ -401,6 +413,7 @@ static void rasterize_triangle(const RasterizerOptions& options, Gfx::Bitmap& re
                     float fog_fragment_depth = interpolate(vertex0_eye_absz, vertex1_eye_absz, vertex2_eye_absz, barycentric);
 
                     *pixel = pixel_shader(uv, vertex_color, fog_fragment_depth);
+                    INCREASE_STATISTICS_COUNTER(g_num_pixels_shaded, 1);
                 }
             }
 
@@ -490,6 +503,7 @@ static void rasterize_triangle(const RasterizerOptions& options, Gfx::Bitmap& re
                             + FloatVector4(float_dst.w(), float_dst.w(), float_dst.w(), float_dst.w()) * dst_factor_dst_alpha;
 
                         *dst = (*dst & ~options.color_mask) | (to_rgba32(*src * src_factor + float_dst * dst_factor) & options.color_mask);
+                        INCREASE_STATISTICS_COUNTER(g_num_pixels_blended, 1);
                     }
                 }
             } else {
@@ -795,6 +809,7 @@ void Device::submit_triangle(const Triangle& triangle, Vector<size_t> const& ena
             auto const& sampler = m_samplers[i];
 
             FloatVector4 texel = sampler.sample_2d({ uv.x(), uv.y() });
+            INCREASE_STATISTICS_COUNTER(g_num_sampler_calls, 1);
 
             // FIXME: Implement more blend modes
             switch (sampler.config().fixed_function_texture_env_mode) {
@@ -890,6 +905,9 @@ void Device::blit(Gfx::Bitmap const& source, int x, int y)
 {
     wait_for_all_threads();
 
+    INCREASE_STATISTICS_COUNTER(g_num_pixels, source.width() * source.height());
+    INCREASE_STATISTICS_COUNTER(g_num_pixels_shaded, source.width() * source.height());
+
     Gfx::Painter painter { *m_render_target };
     painter.blit({ x, y }, source, source.rect(), 1.0f, true);
 }
@@ -900,6 +918,65 @@ void Device::blit_to(Gfx::Bitmap& target)
 
     Gfx::Painter painter { target };
     painter.blit({ 0, 0 }, *m_render_target, m_render_target->rect(), 1.0f, false);
+
+    if constexpr (ENABLE_STATISTICS_OVERLAY)
+        draw_statistics_overlay(target);
+}
+
+void Device::draw_statistics_overlay(Gfx::Bitmap& target)
+{
+    static Core::ElapsedTimer timer;
+    static String debug_string;
+    static int frame_counter;
+
+    frame_counter++;
+    int milliseconds = 0;
+    if (timer.is_valid())
+        milliseconds = timer.elapsed();
+    else
+        timer.start();
+
+    Gfx::Painter painter { target };
+
+    if (milliseconds > 500) {
+
+        if (g_num_pixels == 0)
+            g_num_pixels = 1;
+
+        int num_rendertarget_pixels = m_render_target->width() * m_render_target->height();
+
+        StringBuilder builder;
+        builder.append(String::formatted("Timings      : {:.1}ms {:.1}FPS\n",
+            static_cast<double>(milliseconds) / frame_counter,
+            (milliseconds > 0) ? 1000.0 * frame_counter / milliseconds : 9999.0));
+        builder.append(String::formatted("Triangles    : {}\n", g_num_rasterized_triangles));
+        builder.append(String::formatted("Pixels       : {}, Shaded: {}%, Blended: {}%, Overdraw: {}%\n",
+            g_num_pixels,
+            g_num_pixels_shaded * 100 / g_num_pixels,
+            g_num_pixels_blended * 100 / g_num_pixels_shaded,
+            g_num_pixels_shaded * 100 / num_rendertarget_pixels - 100));
+        builder.append(String::formatted("Sampler calls: {}\n", g_num_sampler_calls));
+
+        debug_string = builder.to_string();
+
+        frame_counter = 0;
+        timer.start();
+    }
+
+    g_num_rasterized_triangles = 0;
+    g_num_pixels = 0;
+    g_num_pixels_shaded = 0;
+    g_num_pixels_blended = 0;
+    g_num_sampler_calls = 0;
+
+    auto& font = Gfx::FontDatabase::default_fixed_width_font();
+
+    for (int y = -1; y < 2; y++)
+        for (int x = -1; x < 2; x++)
+            if (x != 0 && y != 0)
+                painter.draw_text(target.rect().translated(x + 2, y + 2), debug_string, font, Gfx::TextAlignment::TopLeft, Gfx::Color::Black);
+
+    painter.draw_text(target.rect().translated(2, 2), debug_string, font, Gfx::TextAlignment::TopLeft, Gfx::Color::White);
 }
 
 void Device::wait_for_all_threads() const
diff --git a/Userland/Libraries/LibSoftGPU/Device.h b/Userland/Libraries/LibSoftGPU/Device.h
index 9fdd2803ee..f8e2c1dcb6 100644
--- a/Userland/Libraries/LibSoftGPU/Device.h
+++ b/Userland/Libraries/LibSoftGPU/Device.h
@@ -91,6 +91,7 @@ public:
 
 private:
     void submit_triangle(Triangle const& triangle, Vector<size_t> const& enabled_texture_units);
+    void draw_statistics_overlay(Gfx::Bitmap&);
 
 private:
     RefPtr<Gfx::Bitmap> m_render_target;