Integrate the MicroProfile profiling library

This brings goodies such as a configurable user interface and
multi-threaded timeline view.

src/video_core/command_processor.cpp

@@ -4,6 +4,7 @@
 
 #include <boost/range/algorithm/fill.hpp>
 
+#include "common/microprofile.h"
 #include "common/profiler.h"
 
 #include "core/hle/service/gsp_gpu.h"
@@ -43,6 +44,8 @@ static const u32 expand_bits_to_bytes[] = {
     0xffff0000, 0xffff00ff, 0xffffff00, 0xffffffff
 };
 
+MICROPROFILE_DEFINE(GPU_Drawing, "GPU", "Drawing", MP_RGB(50, 50, 240));
+
 static void WritePicaReg(u32 id, u32 value, u32 mask) {
     auto& regs = g_state.regs;
 
@@ -126,6 +129,7 @@ static void WritePicaReg(u32 id, u32 value, u32 mask) {
         case PICA_REG_INDEX(trigger_draw_indexed):
         {
             Common::Profiling::ScopeTimer scope_timer(category_drawing);
+            MICROPROFILE_SCOPE(GPU_Drawing);
 
 #if PICA_LOG_TEV
             DebugUtils::DumpTevStageConfig(regs.GetTevStages());

src/video_core/rasterizer.cpp

@@ -7,6 +7,7 @@
 #include "common/color.h"
 #include "common/common_types.h"
 #include "common/math_util.h"
+#include "common/microprofile.h"
 #include "common/profiler.h"
 
 #include "core/hw/gpu.h"
@@ -267,6 +268,7 @@ static int SignedArea (const Math::Vec2<Fix12P4>& vtx1,
 };
 
 static Common::Profiling::TimingCategory rasterization_category("Rasterization");
+MICROPROFILE_DEFINE(GPU_Rasterization, "GPU", "Rasterization", MP_RGB(50, 50, 240));
 
 /**
  * Helper function for ProcessTriangle with the "reversed" flag to allow for implementing
@@ -279,6 +281,7 @@ static void ProcessTriangleInternal(const Shader::OutputVertex& v0,
 {
     const auto& regs = g_state.regs;
     Common::Profiling::ScopeTimer timer(rasterization_category);
+    MICROPROFILE_SCOPE(GPU_Rasterization);
 
     // vertex positions in rasterizer coordinates
     static auto FloatToFix = [](float24 flt) {

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -7,6 +7,7 @@
 
 #include "common/color.h"
 #include "common/math_util.h"
+#include "common/microprofile.h"
 #include "common/profiler.h"
 
 #include "core/hw/gpu.h"
@@ -777,12 +778,16 @@ void RasterizerOpenGL::SyncDrawState() {
     state.Apply();
 }
 
+MICROPROFILE_DEFINE(OpenGL_FramebufferReload, "OpenGL", "FB Reload", MP_RGB(70, 70, 200));
+
 void RasterizerOpenGL::ReloadColorBuffer() {
     u8* color_buffer = Memory::GetPhysicalPointer(Pica::g_state.regs.framebuffer.GetColorBufferPhysicalAddress());
 
     if (color_buffer == nullptr)
         return;
 
+    MICROPROFILE_SCOPE(OpenGL_FramebufferReload);
+
     u32 bytes_per_pixel = Pica::Regs::BytesPerColorPixel(fb_color_texture.format);
 
     std::unique_ptr<u8[]> temp_fb_color_buffer(new u8[fb_color_texture.width * fb_color_texture.height * bytes_per_pixel]);
@@ -822,6 +827,8 @@ void RasterizerOpenGL::ReloadDepthBuffer() {
     if (depth_buffer == nullptr)
         return;
 
+    MICROPROFILE_SCOPE(OpenGL_FramebufferReload);
+
     u32 bytes_per_pixel = Pica::Regs::BytesPerDepthPixel(fb_depth_texture.format);
 
     // OpenGL needs 4 bpp alignment for D24
@@ -868,6 +875,7 @@ void RasterizerOpenGL::ReloadDepthBuffer() {
 }
 
 Common::Profiling::TimingCategory buffer_commit_category("Framebuffer Commit");
+MICROPROFILE_DEFINE(OpenGL_FramebufferCommit, "OpenGL", "FB Commit", MP_RGB(70, 70, 200));
 
 void RasterizerOpenGL::CommitColorBuffer() {
     if (last_fb_color_addr != 0) {
@@ -875,6 +883,7 @@ void RasterizerOpenGL::CommitColorBuffer() {
 
         if (color_buffer != nullptr) {
             Common::Profiling::ScopeTimer timer(buffer_commit_category);
+            MICROPROFILE_SCOPE(OpenGL_FramebufferCommit);
 
             u32 bytes_per_pixel = Pica::Regs::BytesPerColorPixel(fb_color_texture.format);
 
@@ -911,6 +920,7 @@ void RasterizerOpenGL::CommitDepthBuffer() {
 
         if (depth_buffer != nullptr) {
             Common::Profiling::ScopeTimer timer(buffer_commit_category);
+            MICROPROFILE_SCOPE(OpenGL_FramebufferCommit);
 
             u32 bytes_per_pixel = Pica::Regs::BytesPerDepthPixel(fb_depth_texture.format);
 

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -4,6 +4,7 @@
 
 #include "common/make_unique.h"
 #include "common/math_util.h"
+#include "common/microprofile.h"
 #include "common/vector_math.h"
 
 #include "core/memory.h"
@@ -16,6 +17,8 @@ RasterizerCacheOpenGL::~RasterizerCacheOpenGL() {
     FullFlush();
 }
 
+MICROPROFILE_DEFINE(OpenGL_TextureUpload, "OpenGL", "Texture Upload", MP_RGB(128, 64, 192));
+
 void RasterizerCacheOpenGL::LoadAndBindTexture(OpenGLState &state, unsigned texture_unit, const Pica::Regs::FullTextureConfig& config) {
     PAddr texture_addr = config.config.GetPhysicalAddress();
 
@@ -25,6 +28,8 @@ void RasterizerCacheOpenGL::LoadAndBindTexture(OpenGLState &state, unsigned text
         state.texture_units[texture_unit].texture_2d = cached_texture->second->texture.handle;
         state.Apply();
     } else {
+        MICROPROFILE_SCOPE(OpenGL_TextureUpload);
+
         std::unique_ptr<CachedTexture> new_texture = Common::make_unique<CachedTexture>();
 
         new_texture->texture.Create();

src/video_core/shader/shader.cpp

@@ -9,6 +9,7 @@
 
 #include "common/hash.h"
 #include "common/make_unique.h"
+#include "common/microprofile.h"
 #include "common/profiler.h"
 
 #include "video_core/debug_utils/debug_utils.h"
@@ -55,11 +56,13 @@ void Shutdown() {
 }
 
 static Common::Profiling::TimingCategory shader_category("Vertex Shader");
+MICROPROFILE_DEFINE(GPU_VertexShader, "GPU", "Vertex Shader", MP_RGB(50, 50, 240));
 
 OutputVertex Run(UnitState<false>& state, const InputVertex& input, int num_attributes) {
     auto& config = g_state.regs.vs;
 
     Common::Profiling::ScopeTimer timer(shader_category);
+    MICROPROFILE_SCOPE(GPU_VertexShader);
 
     state.program_counter = config.main_offset;
     state.debug.max_offset = 0;