aurora: Reimplement textures

CMakeLists.txt

@@ -472,7 +472,7 @@ function(shaderc out)
 endfunction()
 
 include(hecl/ApplicationTools.cmake)
-#add_subdirectory(Shaders)
+add_subdirectory(extern/xxhash EXCLUDE_FROM_ALL)
 add_subdirectory(imgui)
 add_subdirectory(extern/libSquish EXCLUDE_FROM_ALL)
 add_subdirectory(extern/libpng EXCLUDE_FROM_ALL)

DataSpec/CMakeLists.txt

@@ -74,7 +74,7 @@ add_library(AssetNameMapNull
 
 get_target_property(HECL_INCLUDES hecl-full INCLUDE_DIRECTORIES)
 target_include_directories(RetroDataSpec PUBLIC ${HECL_INCLUDES} ${CMAKE_CURRENT_SOURCE_DIR} ${CMAKE_SOURCE_DIR})
-target_link_libraries(RetroDataSpec PUBLIC amuse zeus nod squish ${PNG_LIBRARIES} ${ZLIB_LIBRARIES} lzokay logvisor aurora)
+target_link_libraries(RetroDataSpec PUBLIC amuse zeus nod squish ${PNG_LIBRARIES} ${ZLIB_LIBRARIES} lzokay logvisor aurora xxhash)
 if (COMMAND add_sanitizers)
     add_sanitizers(RetroDataSpec)
 endif ()

NESEmulator/CNESEmulator.hpp

@@ -35,7 +35,7 @@ private:
     zeus::CColor m_color;
   };
 
-  std::shared_ptr<aurora::gfx::TextureHandle> m_texture;
+  aurora::gfx::TextureHandle m_texture;
 //  boo::ObjToken<boo::IGraphicsBufferD> m_uniBuf;
 //  boo::ObjToken<boo::IGraphicsBufferS> m_vbo;
 //  boo::ObjToken<boo::IShaderDataBinding> m_shadBind;

Runtime/AutoMapper/CMappableObject.hpp

@@ -64,7 +64,7 @@ private:
     CLineRenderer m_outline;
     explicit DoorSurface()
     : m_surface(skDoorVerts, skDoorIndices)
-    , m_outline(CLineRenderer::EPrimitiveMode::LineLoop, 5, nullptr, false, false, true) {}
+    , m_outline(CLineRenderer::EPrimitiveMode::LineLoop, 5, {}, false, false, true) {}
   };
   std::optional<DoorSurface> m_doorSurface;
   std::optional<CTexturedQuadFilter> m_texQuadFilter;

Runtime/Graphics/CBooRenderer.cpp

@@ -644,7 +644,7 @@ void CBooRenderer::ReallyRenderFogVolume(const zeus::CColor& color, const zeus::
 //  m_scanLinesOddVBO = ctx.newStaticBuffer(boo::BufferUse::Vertex, verts.data(), sizeof(zeus::CVector3f), verts.size());
 //}
 
-std::shared_ptr<aurora::gfx::TextureHandle> CBooRenderer::GetColorTexture(const zeus::CColor& color) {
+aurora::gfx::TextureHandle CBooRenderer::GetColorTexture(const zeus::CColor& color) {
   const auto search = m_colorTextures.find(color);
   if (search != m_colorTextures.end()) {
     return search->second;
@@ -652,7 +652,7 @@ std::shared_ptr<aurora::gfx::TextureHandle> CBooRenderer::GetColorTexture(const
 
   std::array<u8, 4> pixel{};
   color.toRGBA8(pixel[0], pixel[1], pixel[2], pixel[3]);
-  auto tex = aurora::gfx::new_static_texture_2d(1, 1, 1, aurora::gfx::TextureFormat::RGBA8, pixel, "Color Texture"sv);
+  auto tex = aurora::gfx::new_static_texture_2d(1, 1, 1, aurora::gfx::TextureFormat::RGBA8, pixel, "Color Texture");
   m_colorTextures.emplace(color, tex);
   return tex;
 }
@@ -681,13 +681,13 @@ CBooRenderer::CBooRenderer(IObjectStore& store, IFactory& resFac)
 
   constexpr std::array<u8, 4> clearPixel{0, 0, 0, 0};
   m_clearTexture =
-      aurora::gfx::new_static_texture_2d(1, 1, 1, aurora::gfx::TextureFormat::RGBA8, clearPixel, "Clear Texture"sv);
+      aurora::gfx::new_static_texture_2d(1, 1, 1, aurora::gfx::TextureFormat::RGBA8, clearPixel, "Clear Texture");
   constexpr std::array<u8, 4> blackPixel{0, 0, 0, 255};
   m_blackTexture =
-      aurora::gfx::new_static_texture_2d(1, 1, 1, aurora::gfx::TextureFormat::RGBA8, blackPixel, "Black Texture"sv);
+      aurora::gfx::new_static_texture_2d(1, 1, 1, aurora::gfx::TextureFormat::RGBA8, blackPixel, "Black Texture");
   constexpr std::array<u8, 4> whitePixel{255, 255, 255, 255};
   m_whiteTexture =
-      aurora::gfx::new_static_texture_2d(1, 1, 1, aurora::gfx::TextureFormat::RGBA8, whitePixel, "White Texture"sv);
+      aurora::gfx::new_static_texture_2d(1, 1, 1, aurora::gfx::TextureFormat::RGBA8, whitePixel, "White Texture");
 
   //  GenerateFogVolumeRampTex();
 //  GenerateSphereRampTex();

Runtime/Graphics/CBooRenderer.hpp

@@ -101,20 +101,20 @@ class CBooRenderer final : public IRenderer {
   // boo::ITextureS* xe4_blackTex = nullptr;
   bool xee_24_ : 1 = true;
 
-  std::shared_ptr<aurora::gfx::TextureHandle> m_clearTexture;
+  aurora::gfx::TextureHandle m_clearTexture;
-  std::shared_ptr<aurora::gfx::TextureHandle> m_blackTexture;
+  aurora::gfx::TextureHandle m_blackTexture;
-  std::shared_ptr<aurora::gfx::TextureHandle> m_whiteTexture;
+  aurora::gfx::TextureHandle m_whiteTexture;
-  std::unordered_map<zeus::CColor, std::shared_ptr<aurora::gfx::TextureHandle>> m_colorTextures;
+  std::unordered_map<zeus::CColor, aurora::gfx::TextureHandle> m_colorTextures;
 
-  std::shared_ptr<aurora::gfx::TextureHandle> x14c_reflectionTex;
+  aurora::gfx::TextureHandle x14c_reflectionTex;
   // boo::ITextureS* x150_mirrorRamp = nullptr;
-//  std::shared_ptr<aurora::gfx::TextureHandle> x1b8_fogVolumeRamp;
+//  aurora::gfx::TextureHandle x1b8_fogVolumeRamp;
-  std::shared_ptr<aurora::gfx::TextureHandle> x220_sphereRamp;
+  aurora::gfx::TextureHandle x220_sphereRamp;
 //  TLockedToken<CTexture> m_thermoPaletteTex;
-//  std::shared_ptr<aurora::gfx::TextureHandle> x288_thermoPalette;
+//  aurora::gfx::TextureHandle x288_thermoPalette;
 //  TLockedToken<CTexture> m_ballFadeTex;
-  std::shared_ptr<aurora::gfx::TextureHandle> m_ballFade;
+  aurora::gfx::TextureHandle m_ballFade;
-  std::shared_ptr<aurora::gfx::TextureHandle> m_ballShadowId;
+  aurora::gfx::TextureHandle m_ballShadowId;
 //  boo::ObjToken<boo::IGraphicsBufferS> m_scanLinesEvenVBO;
 //  boo::ObjToken<boo::IGraphicsBufferS> m_scanLinesOddVBO;
   int m_ballShadowIdW = 64;
@@ -269,11 +269,11 @@ public:
 //  const boo::ObjToken<boo::IGraphicsBufferS>& GetScanLinesEvenVBO() const { return m_scanLinesEvenVBO; }
 //  const boo::ObjToken<boo::IGraphicsBufferS>& GetScanLinesOddVBO() const { return m_scanLinesOddVBO; }
 
-  const std::shared_ptr<aurora::gfx::TextureHandle>& GetClearTexture() const { return m_clearTexture; }
+  const aurora::gfx::TextureHandle& GetClearTexture() const { return m_clearTexture; }
-  const std::shared_ptr<aurora::gfx::TextureHandle>& GetBlackTexture() const { return m_blackTexture; }
+  const aurora::gfx::TextureHandle& GetBlackTexture() const { return m_blackTexture; }
-  const std::shared_ptr<aurora::gfx::TextureHandle>& GetWhiteTexture() const { return m_whiteTexture; }
+  const aurora::gfx::TextureHandle& GetWhiteTexture() const { return m_whiteTexture; }
 
-  std::shared_ptr<aurora::gfx::TextureHandle> GetColorTexture(const zeus::CColor& color);
+  aurora::gfx::TextureHandle GetColorTexture(const zeus::CColor& color);
 
   static void BindMainDrawTarget() {
 //    CGraphics::g_BooMainCommandQueue->setRenderTarget(CGraphics::g_SpareTexture);

Runtime/Graphics/CLineRenderer.cpp

@@ -20,7 +20,7 @@ void CLineRenderer::Shutdown() {
 //hecl::VertexBufferPool<CLineRenderer::SDrawVertNoTex> CLineRenderer::s_vertPoolNoTex = {};
 //hecl::UniformBufferPool<CLineRenderer::SDrawUniform> CLineRenderer::s_uniformPool = {};
 
-CLineRenderer::CLineRenderer(EPrimitiveMode mode, u32 maxVerts, const std::shared_ptr<aurora::gfx::TextureHandle>& texture,
+CLineRenderer::CLineRenderer(EPrimitiveMode mode, u32 maxVerts, const aurora::gfx::TextureHandle& texture,
                              bool additive, bool zTest, bool zGEqual)
 : m_mode(mode), m_maxVerts(maxVerts) {
   OPTICK_EVENT();

Runtime/Graphics/CLineRenderer.hpp

@@ -72,7 +72,7 @@ public:
 //  hecl::UniformBufferPool<SDrawUniform>::Token m_uniformBuf;
 //  std::array<boo::ObjToken<boo::IShaderDataBinding>, 2> m_shaderBind;
 
-  CLineRenderer(EPrimitiveMode mode, u32 maxVerts, const std::shared_ptr<aurora::gfx::TextureHandle>& texture, bool additive,
+  CLineRenderer(EPrimitiveMode mode, u32 maxVerts, const aurora::gfx::TextureHandle& texture, bool additive,
                 bool zTest = false, bool zGEqual = false);
   CLineRenderer(CLineRenderer&&) = default;
 

Runtime/Graphics/CModel.hpp

@@ -239,16 +239,16 @@ public:
   static void EnsureViewDepStateCached(const CBooModel& model, const CBooSurface* surf, zeus::CMatrix4f* mtxsOut,
                                        float& alphaOut);
 
-  static inline std::shared_ptr<aurora::gfx::TextureHandle> g_shadowMap;
+  static inline aurora::gfx::TextureHandle g_shadowMap;
   static inline zeus::CTransform g_shadowTexXf;
-  static void EnableShadowMaps(const std::shared_ptr<aurora::gfx::TextureHandle>& map, const zeus::CTransform& texXf);
+  static void EnableShadowMaps(const aurora::gfx::TextureHandle& map, const zeus::CTransform& texXf);
   static void DisableShadowMaps();
 
-  static inline std::shared_ptr<aurora::gfx::TextureHandle> g_disintegrateTexture;
+  static inline aurora::gfx::TextureHandle g_disintegrateTexture;
-  static void SetDisintegrateTexture(const std::shared_ptr<aurora::gfx::TextureHandle>& map) { g_disintegrateTexture = map; }
+  static void SetDisintegrateTexture(const aurora::gfx::TextureHandle& map) { g_disintegrateTexture = map; }
 
-  static inline std::shared_ptr<aurora::gfx::TextureHandle> g_reflectionCube;
+  static inline aurora::gfx::TextureHandle g_reflectionCube;
-  static void SetReflectionCube(const std::shared_ptr<aurora::gfx::TextureHandle>& map) { g_reflectionCube = map; }
+  static void SetReflectionCube(const aurora::gfx::TextureHandle& map) { g_reflectionCube = map; }
 
   static void SetDummyTextures(bool b) { g_DummyTextures = b; }
   static void SetRenderModelBlack(bool b) { g_RenderModelBlack = b; }

Runtime/Graphics/CModelBoo.cpp

@@ -135,11 +135,11 @@ void CBooModel::EnsureViewDepStateCached(const CBooModel& model, const CBooSurfa
   }
 }
 
-void CBooModel::EnableShadowMaps(const std::shared_ptr<aurora::gfx::TextureHandle>& map, const zeus::CTransform& texXf) {
+void CBooModel::EnableShadowMaps(const aurora::gfx::TextureHandle& map, const zeus::CTransform& texXf) {
   g_shadowMap = map;
   g_shadowTexXf = texXf;
 }
-void CBooModel::DisableShadowMaps() { g_shadowMap = nullptr; }
+void CBooModel::DisableShadowMaps() { g_shadowMap.reset(); }
 
 CBooModel::~CBooModel() {
   if (m_prev)

Runtime/Graphics/CMoviePlayer.cpp

@@ -373,8 +373,8 @@ void CMoviePlayer::DrawFrame() {
 
   /* draw appropriate field */
   CTHPTextureSet& tex = x80_textures[xd0_drawTexSlot];
-  aurora::gfx::queue_movie_player(tex.Y[m_deinterlace ? (xfc_fieldIndex != 0) : 0]->ref, tex.U->ref, tex.V->ref,
+  aurora::gfx::queue_movie_player(tex.Y[m_deinterlace ? (xfc_fieldIndex != 0) : 0], tex.U, tex.V, zeus::skWhite, m_hpad,
-                                  zeus::skWhite, m_hpad, m_vpad);
+                                  m_vpad);
 
   /* ensure second field is being displayed by VI to signal advance
    * (faked in metaforce with continuous xor) */
@@ -495,18 +495,18 @@ void CMoviePlayer::DecodeFromRead(const void* data) {
           memcpy(buffer.get() + x6c_videoInfo.width * y, m_yuvBuf.get() + x6c_videoInfo.width * (y * 2),
                  x6c_videoInfo.width);
         }
-        aurora::gfx::write_texture(tex.Y[0]->ref, {buffer.get(), planeSizeHalf});
+        aurora::gfx::write_texture(tex.Y[0], {buffer.get(), planeSizeHalf});
         for (unsigned y = 0; y < x6c_videoInfo.height / 2; ++y) {
           memcpy(buffer.get() + x6c_videoInfo.width * y, m_yuvBuf.get() + x6c_videoInfo.width * (y * 2 + 1),
                  x6c_videoInfo.width);
         }
-        aurora::gfx::write_texture(tex.Y[1]->ref, {buffer.get(), planeSizeHalf});
+        aurora::gfx::write_texture(tex.Y[1], {buffer.get(), planeSizeHalf});
       } else {
         /* Direct planar load */
-        aurora::gfx::write_texture(tex.Y[0]->ref, {m_yuvBuf.get(), planeSize});
+        aurora::gfx::write_texture(tex.Y[0], {m_yuvBuf.get(), planeSize});
       }
-      aurora::gfx::write_texture(tex.U->ref, {m_yuvBuf.get() + planeSize, planeSizeQuarter});
+      aurora::gfx::write_texture(tex.U, {m_yuvBuf.get() + planeSize, planeSizeQuarter});
-      aurora::gfx::write_texture(tex.V->ref, {m_yuvBuf.get() + planeSize + planeSizeQuarter, planeSizeQuarter});
+      aurora::gfx::write_texture(tex.V, {m_yuvBuf.get() + planeSize + planeSizeQuarter, planeSizeQuarter});
 
       break;
     }

Runtime/Graphics/CMoviePlayer.hpp

@@ -69,9 +69,9 @@ private:
   };
 
   struct CTHPTextureSet {
-    std::shared_ptr<aurora::gfx::TextureHandle> Y[2];
+    std::array<aurora::gfx::TextureHandle, 2> Y;
-    std::shared_ptr<aurora::gfx::TextureHandle> U;
+    aurora::gfx::TextureHandle U;
-    std::shared_ptr<aurora::gfx::TextureHandle> V;
+    aurora::gfx::TextureHandle V;
     u32 playedSamples = 0;
     u32 audioSamples = 0;
     std::unique_ptr<s16[]> audioBuf;

Runtime/Graphics/CRainSplashGenerator.cpp

@@ -71,7 +71,7 @@ void CRainSplashGenerator::Draw(const zeus::CTransform& xf) {
 }
 
 CRainSplashGenerator::SSplashLine::SSplashLine()
-: m_renderer(CLineRenderer::EPrimitiveMode::LineStrip, 3, nullptr, false) {}
+: m_renderer(CLineRenderer::EPrimitiveMode::LineStrip, 3, {}, false) {}
 
 CRainSplashGenerator::SRainSplash::SRainSplash() {
   for (size_t i = 0; i < x0_lines.capacity(); ++i) {

Runtime/Graphics/CTexture.hpp

@@ -33,30 +33,30 @@ private:
   u16 x6_h;
   u32 x8_mips;
 //  boo::ObjToken<boo::ITexture> m_booTex;
-  std::shared_ptr<aurora::gfx::TextureHandle> m_tex;
+  aurora::gfx::TextureHandle m_tex;
-  std::shared_ptr<aurora::gfx::TextureHandle> m_paletteTex;
+  aurora::gfx::TextureHandle m_paletteTex;
   std::unique_ptr<u8[]> m_otex;
   EFontType m_ftype = EFontType::None;
   const CTextureInfo* m_textureInfo;
 
   size_t ComputeMippedTexelCount() const;
   size_t ComputeMippedBlockCountDXT1() const;
-  void BuildI4FromGCN(CInputStream& in, std::string_view label);
+  void BuildI4FromGCN(CInputStream& in, aurora::zstring_view label);
-  void BuildI8FromGCN(CInputStream& in, std::string_view label);
+  void BuildI8FromGCN(CInputStream& in, aurora::zstring_view label);
-  void BuildIA4FromGCN(CInputStream& in, std::string_view label);
+  void BuildIA4FromGCN(CInputStream& in, aurora::zstring_view label);
-  void BuildIA8FromGCN(CInputStream& in, std::string_view label);
+  void BuildIA8FromGCN(CInputStream& in, aurora::zstring_view label);
-  void BuildC4FromGCN(CInputStream& in, std::string_view label);
+  void BuildC4FromGCN(CInputStream& in, aurora::zstring_view label);
-  void BuildC8FromGCN(CInputStream& in, std::string_view label);
+  void BuildC8FromGCN(CInputStream& in, aurora::zstring_view label);
-  void BuildC14X2FromGCN(CInputStream& in, std::string_view label);
+  void BuildC14X2FromGCN(CInputStream& in, aurora::zstring_view label);
-  void BuildRGB565FromGCN(CInputStream& in, std::string_view label);
+  void BuildRGB565FromGCN(CInputStream& in, aurora::zstring_view label);
-  void BuildRGB5A3FromGCN(CInputStream& in, std::string_view label);
+  void BuildRGB5A3FromGCN(CInputStream& in, aurora::zstring_view label);
-  void BuildRGBA8FromGCN(CInputStream& in, std::string_view label);
+  void BuildRGBA8FromGCN(CInputStream& in, aurora::zstring_view label);
-  void BuildDXT1FromGCN(CInputStream& in, std::string_view label);
+  void BuildDXT1FromGCN(CInputStream& in, aurora::zstring_view label);
-  void BuildRGBA8(const void* data, size_t length, std::string_view label);
+  void BuildRGBA8(const void* data, size_t length, aurora::zstring_view label);
-  void BuildC8(const void* data, size_t length, std::string_view label);
+  void BuildC8(const void* data, size_t length, aurora::zstring_view label);
-  void BuildC8Font(const void* data, EFontType ftype, std::string_view label);
+  void BuildC8Font(const void* data, EFontType ftype, aurora::zstring_view label);
-  void BuildDXT1(const void* data, size_t length, std::string_view label);
+  void BuildDXT1(const void* data, size_t length, aurora::zstring_view label);
-  void BuildDXT3(const void* data, size_t length, std::string_view label);
+  void BuildDXT3(const void* data, size_t length, aurora::zstring_view label);
 
 public:
   CTexture(ETexelFormat, s16, s16, s32);
@@ -70,11 +70,11 @@ public:
   u16 GetHeight() const { return x6_h; }
   u32 GetNumMips() const { return x8_mips; }
   void Load(int slot, EClampMode clamp) const;
-  const std::shared_ptr<aurora::gfx::TextureHandle>& GetTexture() const { return m_tex; }
+  const aurora::gfx::TextureHandle& GetTexture() const { return m_tex; }
 //  const boo::ObjToken<boo::ITexture>& GetBooTexture() const { return m_booTex; }
-  const std::shared_ptr<aurora::gfx::TextureHandle>& GetPaletteTexture() const { return m_paletteTex; }
+  const aurora::gfx::TextureHandle& GetPaletteTexture() const { return m_paletteTex; }
   std::unique_ptr<u8[]> BuildMemoryCardTex(u32& sizeOut, ETexelFormat& fmtOut, std::unique_ptr<u8[]>& paletteOut) const;
-  const std::shared_ptr<aurora::gfx::TextureHandle>& GetFontTexture(EFontType tp);
+  const aurora::gfx::TextureHandle& GetFontTexture(EFontType tp);
 
   const CTextureInfo* GetTextureInfo() const { return m_textureInfo; }
 };

Runtime/Graphics/CTextureBoo.cpp

@@ -75,7 +75,7 @@ size_t CTexture::ComputeMippedBlockCountDXT1() const {
   return ret;
 }
 
-void CTexture::BuildI4FromGCN(CInputStream& in, std::string_view label) {
+void CTexture::BuildI4FromGCN(CInputStream& in, aurora::zstring_view label) {
   const size_t texelCount = ComputeMippedTexelCount();
   std::unique_ptr<RGBA8[]> buf(new RGBA8[texelCount]);
 
@@ -115,7 +115,7 @@ void CTexture::BuildI4FromGCN(CInputStream& in, std::string_view label) {
                                              {reinterpret_cast<const uint8_t*>(buf.get()), texelCount * 4}, label);
 }
 
-void CTexture::BuildI8FromGCN(CInputStream& in, std::string_view label) {
+void CTexture::BuildI8FromGCN(CInputStream& in, aurora::zstring_view label) {
   const size_t texelCount = ComputeMippedTexelCount();
   std::unique_ptr<RGBA8[]> buf(new RGBA8[texelCount]);
 
@@ -155,7 +155,7 @@ void CTexture::BuildI8FromGCN(CInputStream& in, std::string_view label) {
                                              {reinterpret_cast<const uint8_t*>(buf.get()), texelCount * 4}, label);
 }
 
-void CTexture::BuildIA4FromGCN(CInputStream& in, std::string_view label) {
+void CTexture::BuildIA4FromGCN(CInputStream& in, aurora::zstring_view label) {
   const size_t texelCount = ComputeMippedTexelCount();
   std::unique_ptr<RGBA8[]> buf(new RGBA8[texelCount]);
 
@@ -196,7 +196,7 @@ void CTexture::BuildIA4FromGCN(CInputStream& in, std::string_view label) {
                                              {reinterpret_cast<const uint8_t*>(buf.get()), texelCount * 4}, label);
 }
 
-void CTexture::BuildIA8FromGCN(CInputStream& in, std::string_view label) {
+void CTexture::BuildIA8FromGCN(CInputStream& in, aurora::zstring_view label) {
   const size_t texelCount = ComputeMippedTexelCount();
   std::unique_ptr<RGBA8[]> buf(new RGBA8[texelCount]);
 
@@ -278,7 +278,7 @@ static std::vector<RGBA8> DecodePalette(int numEntries, CInputStream& in) {
   return ret;
 }
 
-void CTexture::BuildC4FromGCN(CInputStream& in, std::string_view label) {
+void CTexture::BuildC4FromGCN(CInputStream& in, aurora::zstring_view label) {
   const size_t texelCount = ComputeMippedTexelCount();
   std::unique_ptr<RGBA8[]> buf(new RGBA8[texelCount]);
   std::vector<RGBA8> palette = DecodePalette(16, in);
@@ -316,7 +316,7 @@ void CTexture::BuildC4FromGCN(CInputStream& in, std::string_view label) {
                                              {reinterpret_cast<const uint8_t*>(buf.get()), texelCount * 4}, label);
 }
 
-void CTexture::BuildC8FromGCN(CInputStream& in, std::string_view label) {
+void CTexture::BuildC8FromGCN(CInputStream& in, aurora::zstring_view label) {
   const size_t texelCount = ComputeMippedTexelCount();
   std::unique_ptr<RGBA8[]> buf(new RGBA8[texelCount]);
   std::vector<RGBA8> palette = DecodePalette(256, in);
@@ -354,11 +354,11 @@ void CTexture::BuildC8FromGCN(CInputStream& in, std::string_view label) {
                                              {reinterpret_cast<const uint8_t*>(buf.get()), texelCount * 4}, label);
 }
 
-void CTexture::BuildC14X2FromGCN(CInputStream& in, std::string_view label) {
+void CTexture::BuildC14X2FromGCN(CInputStream& in, aurora::zstring_view label) {
   Log.report(logvisor::Fatal, FMT_STRING("C14X2 not implemented"));
 }
 
-void CTexture::BuildRGB565FromGCN(CInputStream& in, std::string_view label) {
+void CTexture::BuildRGB565FromGCN(CInputStream& in, aurora::zstring_view label) {
   const size_t texelCount = ComputeMippedTexelCount();
   std::unique_ptr<RGBA8[]> buf(new RGBA8[texelCount]);
 
@@ -397,7 +397,7 @@ void CTexture::BuildRGB565FromGCN(CInputStream& in, std::string_view label) {
                                              {reinterpret_cast<const uint8_t*>(buf.get()), texelCount * 4}, label);
 }
 
-void CTexture::BuildRGB5A3FromGCN(CInputStream& in, std::string_view label) {
+void CTexture::BuildRGB5A3FromGCN(CInputStream& in, aurora::zstring_view label) {
   size_t texelCount = ComputeMippedTexelCount();
   std::unique_ptr<RGBA8[]> buf(new RGBA8[texelCount]);
 
@@ -443,7 +443,7 @@ void CTexture::BuildRGB5A3FromGCN(CInputStream& in, std::string_view label) {
                                              {reinterpret_cast<const uint8_t*>(buf.get()), texelCount * 4}, label);
 }
 
-void CTexture::BuildRGBA8FromGCN(CInputStream& in, std::string_view label) {
+void CTexture::BuildRGBA8FromGCN(CInputStream& in, aurora::zstring_view label) {
   const size_t texelCount = ComputeMippedTexelCount();
   std::unique_ptr<RGBA8[]> buf(new RGBA8[texelCount]);
 
@@ -488,7 +488,7 @@ void CTexture::BuildRGBA8FromGCN(CInputStream& in, std::string_view label) {
                                              {reinterpret_cast<const uint8_t*>(buf.get()), texelCount * 4}, label);
 }
 
-void CTexture::BuildDXT1FromGCN(CInputStream& in, std::string_view label) {
+void CTexture::BuildDXT1FromGCN(CInputStream& in, aurora::zstring_view label) {
   const size_t blockCount = ComputeMippedBlockCountDXT1();
   std::unique_ptr<DXT1Block[]> buf(new DXT1Block[blockCount]);
 
@@ -536,7 +536,7 @@ void CTexture::BuildDXT1FromGCN(CInputStream& in, std::string_view label) {
                                              {reinterpret_cast<const uint8_t*>(buf.get()), blockCount * 8}, label);
 }
 
-void CTexture::BuildRGBA8(const void* data, size_t length, std::string_view label) {
+void CTexture::BuildRGBA8(const void* data, size_t length, aurora::zstring_view label) {
   size_t texelCount = ComputeMippedTexelCount();
   size_t expectedSize = texelCount * 4;
   if (expectedSize > length)
@@ -546,7 +546,7 @@ void CTexture::BuildRGBA8(const void* data, size_t length, std::string_view labe
                                              {reinterpret_cast<const uint8_t*>(data), expectedSize}, label);
 }
 
-void CTexture::BuildC8(const void* data, size_t length, std::string_view label) {
+void CTexture::BuildC8(const void* data, size_t length, aurora::zstring_view label) {
   size_t texelCount = ComputeMippedTexelCount();
   if (texelCount > length)
     Log.report(logvisor::Fatal, FMT_STRING("insufficient TXTR length ({}/{})"), length, texelCount);
@@ -560,7 +560,7 @@ void CTexture::BuildC8(const void* data, size_t length, std::string_view label)
                                              label);
 }
 
-void CTexture::BuildC8Font(const void* data, EFontType ftype, std::string_view label) {
+void CTexture::BuildC8Font(const void* data, EFontType ftype, aurora::zstring_view label) {
   size_t texelCount = ComputeMippedTexelCount();
 
   size_t layerCount = 1;
@@ -659,12 +659,12 @@ void CTexture::BuildC8Font(const void* data, EFontType ftype, std::string_view l
   //  } BooTrace);
 }
 
-void CTexture::BuildDXT1(const void* data, size_t length, std::string_view label) {
+void CTexture::BuildDXT1(const void* data, size_t length, aurora::zstring_view label) {
   m_tex = aurora::gfx::new_static_texture_2d(x4_w, x6_h, x8_mips, aurora::gfx::TextureFormat::DXT1,
                                              {reinterpret_cast<const uint8_t*>(data), length}, label);
 }
 
-void CTexture::BuildDXT3(const void* data, size_t length, std::string_view label) {
+void CTexture::BuildDXT3(const void* data, size_t length, aurora::zstring_view label) {
   m_tex = aurora::gfx::new_static_texture_2d(x4_w, x6_h, x8_mips, aurora::gfx::TextureFormat::DXT3,
                                              {reinterpret_cast<const uint8_t*>(data), length}, label);
 }
@@ -881,10 +881,10 @@ std::unique_ptr<u8[]> CTexture::BuildMemoryCardTex(u32& sizeOut, ETexelFormat& f
   return ret;
 }
 
-const std::shared_ptr<aurora::gfx::TextureHandle>& CTexture::GetFontTexture(EFontType tp) {
+const aurora::gfx::TextureHandle& CTexture::GetFontTexture(EFontType tp) {
   if (m_ftype != tp && x0_fmt == ETexelFormat::C8PC) {
     m_ftype = tp;
-    BuildC8Font(m_otex.get() + 12, m_ftype, "Font (TODO)"sv);
+    BuildC8Font(m_otex.get() + 12, m_ftype, "Font (TODO)");
   }
   return m_tex;
 }

Runtime/Graphics/Shaders/CColoredStripShader.cpp

@@ -61,7 +61,7 @@ void CColoredStripShader::Shutdown() {
 //                                        stages.data(), nullptr, nullptr, texs.size(), texs.data(), nullptr, nullptr);
 //}
 
-CColoredStripShader::CColoredStripShader(size_t maxVerts, Mode mode, std::shared_ptr<aurora::gfx::TextureHandle> tex) {
+CColoredStripShader::CColoredStripShader(size_t maxVerts, Mode mode, aurora::gfx::TextureHandle tex) {
 //  CGraphics::CommitResources([this, maxVerts, mode, tex](boo::IGraphicsDataFactory::Context& ctx) {
 //    BuildResources(ctx, maxVerts, mode, tex);
 //    return true;

Runtime/Graphics/Shaders/CColoredStripShader.hpp

@@ -32,7 +32,7 @@ public:
   };
   static void Initialize();
   static void Shutdown();
-  CColoredStripShader(size_t maxVerts, Mode mode, std::shared_ptr<aurora::gfx::TextureHandle> tex);
+  CColoredStripShader(size_t maxVerts, Mode mode, aurora::gfx::TextureHandle tex);
   void draw(const zeus::CColor& color, size_t numVerts, const Vert* verts);
 };
 

Runtime/Graphics/Shaders/CFluidPlaneShader.cpp

@@ -198,7 +198,7 @@ CFluidPlaneShader::CFluidPlaneShader(EFluidType type, const TLockedToken<CTextur
                                      const TLockedToken<CTexture>& patternTex2, const TLockedToken<CTexture>& colorTex,
                                      const TLockedToken<CTexture>& bumpMap, const TLockedToken<CTexture>& envMap,
                                      const TLockedToken<CTexture>& envBumpMap, const TLockedToken<CTexture>& lightmap,
-                                     const std::shared_ptr<aurora::gfx::TextureHandle>& rippleMap, bool doubleLightmapBlend,
+                                     const aurora::gfx::TextureHandle& rippleMap, bool doubleLightmapBlend,
                                      bool additive, u32 maxVertCount)
 : m_patternTex1(patternTex1)
 , m_patternTex2(patternTex2)

Runtime/Graphics/Shaders/CFluidPlaneShader.hpp

@@ -106,7 +106,7 @@ private:
   TLockedToken<CTexture> m_envMap;
   TLockedToken<CTexture> m_envBumpMap;
   TLockedToken<CTexture> m_lightmap;
-  std::shared_ptr<aurora::gfx::TextureHandle> m_rippleMap;
+  aurora::gfx::TextureHandle m_rippleMap;
 //  boo::ObjToken<boo::IGraphicsBufferD> m_vbo;
 //  boo::ObjToken<boo::IGraphicsBufferD> m_pvbo;
 //  boo::ObjToken<boo::IGraphicsBufferD> m_uniBuf;
@@ -145,7 +145,7 @@ public:
                     const TLockedToken<CTexture>& patternTex2, const TLockedToken<CTexture>& colorTex,
                     const TLockedToken<CTexture>& bumpMap, const TLockedToken<CTexture>& envMap,
                     const TLockedToken<CTexture>& envBumpMap, const TLockedToken<CTexture>& lightmap,
-                    const std::shared_ptr<aurora::gfx::TextureHandle>& rippleMap, bool doubleLightmapBlend, bool additive,
+                    const aurora::gfx::TextureHandle& rippleMap, bool doubleLightmapBlend, bool additive,
                     u32 maxVertCount);
   CFluidPlaneShader(const TLockedToken<CTexture>& patternTex1, const TLockedToken<CTexture>& patternTex2,
                     const TLockedToken<CTexture>& colorTex, u32 maxVertCount);

Runtime/Graphics/Shaders/CLineRendererShaders.cpp

@@ -54,7 +54,7 @@ void CLineRendererShaders::Shutdown() {
 }
 
 void CLineRendererShaders::BuildShaderDataBinding(CLineRenderer& renderer,
-                                                  const std::shared_ptr<aurora::gfx::TextureHandle>& texture, bool additive,
+                                                  const aurora::gfx::TextureHandle& texture, bool additive,
                                                   bool zTest, bool zGEqual) {
 //  std::array<boo::ObjToken<boo::IShaderPipeline>, 2>* pipeline = nullptr;
 //

Runtime/Graphics/Shaders/CLineRendererShaders.hpp

@@ -26,7 +26,7 @@ public:
   static void Initialize();
   static void Shutdown();
   static void BuildShaderDataBinding(CLineRenderer& renderer,
-                                     const std::shared_ptr<aurora::gfx::TextureHandle>& texture, bool additive, bool zTest,
+                                     const aurora::gfx::TextureHandle& texture, bool additive, bool zTest,
                                      bool zGEqual);
 };
 

Runtime/Graphics/Shaders/CSpaceWarpFilter.hpp

@@ -17,7 +17,7 @@ class CSpaceWarpFilter {
     zeus::CVector3f m_strength;
   };
   std::array<std::array<std::array<u8, 4>, 8>, 4> m_shiftTexture{};
-  std::shared_ptr<aurora::gfx::TextureHandle> m_warpTex;
+  aurora::gfx::TextureHandle m_warpTex;
 //  boo::ObjToken<boo::IGraphicsBufferS> m_vbo;
 //  boo::ObjToken<boo::IGraphicsBufferD> m_uniBuf;
 //  boo::ObjToken<boo::IShaderDataBinding> m_dataBind;

Runtime/Graphics/Shaders/CTexturedQuadFilter.cpp

@@ -178,11 +178,11 @@ void CTexturedQuadFilterAlpha::Shutdown() {
 //  }
 //}
 
-CTexturedQuadFilter::CTexturedQuadFilter(const std::shared_ptr<aurora::gfx::TextureHandle>& tex) : m_booTex(tex) {
+CTexturedQuadFilter::CTexturedQuadFilter(const aurora::gfx::TextureHandle& tex) : m_booTex(tex) {
   m_flipRect = true; // TODO?
 }
 
-CTexturedQuadFilter::CTexturedQuadFilter(EFilterType type, const std::shared_ptr<aurora::gfx::TextureHandle>& tex, ERglEnum ztest)
+CTexturedQuadFilter::CTexturedQuadFilter(EFilterType type, const aurora::gfx::TextureHandle& tex, ERglEnum ztest)
 : m_booTex(tex), m_zTest(ztest) {
   m_flipRect = true; // TODO?
 //  tex->setClampMode(boo::TextureClampMode::ClampToEdge);
@@ -201,7 +201,7 @@ CTexturedQuadFilter::CTexturedQuadFilter(EFilterType type, const std::shared_ptr
 }
 
 CTexturedQuadFilter::CTexturedQuadFilter(EFilterType type, TLockedToken<CTexture> tex, ERglEnum ztest)
-: CTexturedQuadFilter(type, (tex ? tex->GetTexture() : nullptr), ztest) {
+: CTexturedQuadFilter(type, (tex ? tex->GetTexture() : aurora::gfx::TextureHandle{}), ztest) {
   m_flipRect = true; // TODO?
   m_tex = tex;
 }
@@ -322,7 +322,7 @@ void CTexturedQuadFilter::DrawFilter(EFilterShape shape, const zeus::CColor& col
   }
 }
 
-CTexturedQuadFilterAlpha::CTexturedQuadFilterAlpha(EFilterType type, const std::shared_ptr<aurora::gfx::TextureHandle>& tex)
+CTexturedQuadFilterAlpha::CTexturedQuadFilterAlpha(EFilterType type, const aurora::gfx::TextureHandle& tex)
 : CTexturedQuadFilter(tex) {
 //  CGraphics::CommitResources([&](boo::IGraphicsDataFactory::Context& ctx) {
 //    m_vbo = ctx.newDynamicBuffer(boo::BufferUse::Vertex, 32, 4);
@@ -339,7 +339,7 @@ CTexturedQuadFilterAlpha::CTexturedQuadFilterAlpha(EFilterType type, const std::
 }
 
 CTexturedQuadFilterAlpha::CTexturedQuadFilterAlpha(EFilterType type, TLockedToken<CTexture> tex)
-: CTexturedQuadFilterAlpha(type, (tex ? tex->GetTexture() : nullptr)) {
+: CTexturedQuadFilterAlpha(type, (tex ? tex->GetTexture() : aurora::gfx::TextureHandle{})) {
   m_tex = tex;
 }
 

Runtime/Graphics/Shaders/CTexturedQuadFilter.hpp

@@ -26,7 +26,7 @@ protected:
     float m_lod = 0.f;
   };
   TLockedToken<CTexture> m_tex;
-  std::shared_ptr<aurora::gfx::TextureHandle> m_booTex;
+  aurora::gfx::TextureHandle m_booTex;
   //  boo::ObjToken<boo::IGraphicsBufferD> m_vbo;
   //  boo::ObjToken<boo::IGraphicsBufferD> m_uniBuf;
   //  boo::ObjToken<boo::IShaderDataBinding> m_dataBind;
@@ -34,7 +34,7 @@ protected:
   ERglEnum m_zTest;
   bool m_flipRect = false;
 
-  explicit CTexturedQuadFilter(const std::shared_ptr<aurora::gfx::TextureHandle>& tex);
+  explicit CTexturedQuadFilter(const aurora::gfx::TextureHandle& tex);
 
 public:
   struct Vert {
@@ -45,7 +45,7 @@ public:
   static void Shutdown();
   static constexpr zeus::CRectangle DefaultRect{0.f, 0.f, 1.f, 1.f};
   explicit CTexturedQuadFilter(EFilterType type, TLockedToken<CTexture> tex, ERglEnum zTest = ERglEnum::Never);
-  explicit CTexturedQuadFilter(EFilterType type, const std::shared_ptr<aurora::gfx::TextureHandle>& tex,
+  explicit CTexturedQuadFilter(EFilterType type, const aurora::gfx::TextureHandle& tex,
                                ERglEnum zTest = ERglEnum::Never);
   CTexturedQuadFilter(const CTexturedQuadFilter&) = delete;
   CTexturedQuadFilter& operator=(const CTexturedQuadFilter&) = delete;
@@ -56,7 +56,7 @@ public:
   void drawVerts(const zeus::CColor& color, std::array<Vert, 4> verts, float lod = 0.f);
   void DrawFilter(EFilterShape shape, const zeus::CColor& color, float t);
   const TLockedToken<CTexture>& GetTex() const { return m_tex; }
-  const std::shared_ptr<aurora::gfx::TextureHandle>& GetBooTex() const { return m_booTex; }
+  const aurora::gfx::TextureHandle& GetBooTex() const { return m_booTex; }
 };
 
 class CTexturedQuadFilterAlpha : public CTexturedQuadFilter {
@@ -64,7 +64,7 @@ public:
   static void Initialize();
   static void Shutdown();
   explicit CTexturedQuadFilterAlpha(EFilterType type, TLockedToken<CTexture> tex);
-  explicit CTexturedQuadFilterAlpha(EFilterType type, const std::shared_ptr<aurora::gfx::TextureHandle>& tex);
+  explicit CTexturedQuadFilterAlpha(EFilterType type, const aurora::gfx::TextureHandle& tex);
 };
 
 } // namespace metaforce

Runtime/Graphics/Shaders/CWorldShadowShader.hpp

@@ -13,7 +13,7 @@
 namespace metaforce {
 
 class CWorldShadowShader {
-  std::shared_ptr<aurora::gfx::TextureHandle> m_tex;
+  aurora::gfx::TextureHandle m_tex;
   std::optional<CTexturedQuadFilter> m_prevQuad;
   u32 m_w, m_h;
 
@@ -40,7 +40,7 @@ public:
   u32 GetWidth() const { return m_w; }
   u32 GetHeight() const { return m_h; }
 
-  const std::shared_ptr<aurora::gfx::TextureHandle>& GetTexture() const { return m_tex; }
+  const aurora::gfx::TextureHandle& GetTexture() const { return m_tex; }
 };
 
 } // namespace metaforce

Runtime/Graphics/Shaders/CXRayBlurFilter.hpp

@@ -15,7 +15,7 @@ class CXRayBlurFilter {
     std::array<zeus::CMatrix4f, 8> m_uv;
   };
   TLockedToken<CTexture> m_paletteTex;
-  std::shared_ptr<aurora::gfx::TextureHandle> m_booTex;
+  aurora::gfx::TextureHandle m_booTex;
 //  boo::ObjToken<boo::IGraphicsBufferS> m_vbo;
 //  boo::ObjToken<boo::IGraphicsBufferD> m_uniBuf;
 //  boo::ObjToken<boo::IShaderDataBinding> m_dataBind;

Runtime/GuiSys/CAuiImagePane.cpp

@@ -83,17 +83,17 @@ void CAuiImagePane::DoDrawImagePane(const zeus::CColor& color, const CTexture& t
 
   bool zTest = xac_drawFlags == EGuiModelDrawFlags::Shadeless || xac_drawFlags == EGuiModelDrawFlags::Opaque;
   if (noBlur) {
-    aurora::gfx::queue_textured_quad_verts(aurora::gfx::CameraFilterType(filter), tex.GetTexture()->ref,
+    aurora::gfx::queue_textured_quad_verts(aurora::gfx::CameraFilterType(filter), tex.GetTexture(),
                                            aurora::gfx::ZTest::LEqual, zTest, useColor, xe0_coords, realUseUvs, 0);
     // quad.drawVerts(useColor, verts);
   } else if ((x14c_deResFactor == 0.f && alpha == 1.f) || tex.GetNumMips() == 1) {
-    aurora::gfx::queue_textured_quad_verts(aurora::gfx::CameraFilterType(filter), tex.GetTexture()->ref,
+    aurora::gfx::queue_textured_quad_verts(aurora::gfx::CameraFilterType(filter), tex.GetTexture(),
                                            aurora::gfx::ZTest::LEqual, zTest, useColor, xe0_coords, realUseUvs, 0);
   } else {
     const float tmp = (1.f - x14c_deResFactor) * alpha;
     const float tmp3 = 1.f - tmp * tmp * tmp;
     const float mip = tmp3 * static_cast<float>(tex.GetNumMips() - 1);
-    aurora::gfx::queue_textured_quad_verts(aurora::gfx::CameraFilterType(filter), tex.GetTexture()->ref,
+    aurora::gfx::queue_textured_quad_verts(aurora::gfx::CameraFilterType(filter), tex.GetTexture(),
                                            aurora::gfx::ZTest::LEqual, zTest, useColor, xe0_coords, realUseUvs, mip);
   }
 }

Runtime/GuiSys/CCompoundTargetReticle.cpp

@@ -72,9 +72,9 @@ CCompoundTargetReticle::CCompoundTargetReticle(const CStateManager& mgr)
 CCompoundTargetReticle::SScanReticuleRenderer::SScanReticuleRenderer() {
 //  CGraphics::CommitResources([this](boo::IGraphicsDataFactory::Context& ctx) {
     for (size_t i = 0; i < m_lineRenderers.size(); ++i) {
-      m_lineRenderers[i].emplace(CLineRenderer::EPrimitiveMode::Lines, 8, nullptr, true, true);
+      m_lineRenderers[i].emplace(CLineRenderer::EPrimitiveMode::Lines, 8, aurora::gfx::TextureHandle{}, true, true);
       for (auto& stripRenderer : m_stripRenderers[i]) {
-        stripRenderer.emplace(CLineRenderer::EPrimitiveMode::LineStrip, 4, nullptr, true, true);
+        stripRenderer.emplace(CLineRenderer::EPrimitiveMode::LineStrip, 4, aurora::gfx::TextureHandle{}, true, true);
       }
     }
 //    return true;

Runtime/GuiSys/CRasterFont.hpp

@@ -148,7 +148,7 @@ public:
                   const char16_t* str, int len) const;
   const CGlyph* GetGlyph(char16_t chr) const { return InternalGetGlyph(chr); }
   void GetSize(const CDrawStringOptions& opts, int& width, int& height, const char16_t* str, int len) const;
-  const std::shared_ptr<aurora::gfx::TextureHandle>& GetTexture() { return x80_texture->GetFontTexture(x2c_mode); }
+  const aurora::gfx::TextureHandle& GetTexture() { return x80_texture->GetFontTexture(x2c_mode); }
 
   bool IsFinishedLoading() const;
 };

Runtime/GuiSys/CSplashScreen.cpp

@@ -75,7 +75,7 @@ void CSplashScreen::Draw() {
 
   aurora::gfx::queue_textured_quad(
       aurora::gfx::CameraFilterType::Blend,
-      m_texture->GetTexture()->ref,
+      m_texture->GetTexture(),
       aurora::gfx::ZTest::Always,
       false,
       color,

Runtime/MP1/CFrontEndUI.cpp

@@ -2045,7 +2045,7 @@ void CFrontEndUI::Draw() {
                                   x38_pressStart->GetHeight() / 480.f * vPad);
       zeus::CColor color = zeus::skWhite;
       color.a() = x64_pressStartAlpha;
-      aurora::gfx::queue_textured_quad(aurora::gfx::CameraFilterType::Add, x38_pressStart->GetTexture()->ref,
+      aurora::gfx::queue_textured_quad(aurora::gfx::CameraFilterType::Add, x38_pressStart->GetTexture(),
                                        aurora::gfx::ZTest::Always, false, color, 1.f, rect, 0.f);
     }
 

Runtime/Particle/CElementGen.cpp

@@ -209,7 +209,7 @@ CElementGen::CElementGen(TToken<CGenDescription> gen, EModelOrientationType orie
 
   if (x26c_31_LINE) {
     CUVElement* texr = desc->x54_x40_TEXR.get();
-    std::shared_ptr<aurora::gfx::TextureHandle> tex;
+    aurora::gfx::TextureHandle tex;
     if (texr)
       tex = texr->GetValueTexture(0).GetObj()->GetTexture();
     int maxVerts = x90_MAXP;

Runtime/Particle/CParticleElectric.cpp

@@ -98,8 +98,8 @@ void CParticleElectric::DrawLineStrip(const std::vector<zeus::CVector3f>& verts,
     m_lineRenderers.resize(m_nextLineRenderer);
   }
   if (!m_lineRenderers[useIdx]) {
-    m_lineRenderers[useIdx] =
+    m_lineRenderers[useIdx] = std::make_unique<CLineRenderer>(CLineRenderer::EPrimitiveMode::LineStrip, x150_SSEG,
-        std::make_unique<CLineRenderer>(CLineRenderer::EPrimitiveMode::LineStrip, x150_SSEG, nullptr, true, true);
+                                                              aurora::gfx::TextureHandle{}, true, true);
   }
   CLineRenderer& renderer = *m_lineRenderers[useIdx];
   const zeus::CColor useColor = x1b8_moduColor * color;

Runtime/Particle/CParticleSwoosh.cpp

@@ -67,7 +67,7 @@ CParticleSwoosh::CParticleSwoosh(const TToken<CSwooshDescription>& desc, int len
     if (x1c_desc->x44_29_WIRE) {
       const int maxVerts = x1b4_LENG * (x1b0_SPLN + 1) * x1b8_SIDE * 12;
       m_lineRenderer.reset(
-          new CLineRenderer(CLineRenderer::EPrimitiveMode::Lines, maxVerts * 2, nullptr, x1d0_25_AALP));
+          new CLineRenderer(CLineRenderer::EPrimitiveMode::Lines, maxVerts * 2, {}, x1d0_25_AALP));
     } else {
       const auto maxVerts = size_t(x1b4_LENG * (x1b0_SPLN + 1) * x1b8_SIDE * 4);
       m_cachedVerts.reserve(maxVerts);

Runtime/Weapon/CPlasmaProjectile.cpp

@@ -12,17 +12,16 @@
 
 namespace metaforce {
 
-CPlasmaProjectile::RenderObjects::RenderObjects(std::shared_ptr<aurora::gfx::TextureHandle> tex,
+CPlasmaProjectile::RenderObjects::RenderObjects(aurora::gfx::TextureHandle tex, aurora::gfx::TextureHandle glowTex)
-                                                std::shared_ptr<aurora::gfx::TextureHandle> glowTex)
 : m_beamStrip1(8, CColoredStripShader::Mode::Additive, {})
-, m_beamStrip2( 10, CColoredStripShader::Mode::FullAdditive, tex)
+, m_beamStrip2(10, CColoredStripShader::Mode::FullAdditive, tex)
-, m_beamStrip3( 18, CColoredStripShader::Mode::FullAdditive, tex)
+, m_beamStrip3(18, CColoredStripShader::Mode::FullAdditive, tex)
-, m_beamStrip4( 14, CColoredStripShader::Mode::Additive, glowTex)
+, m_beamStrip4(14, CColoredStripShader::Mode::Additive, glowTex)
-, m_beamStrip1Sub( 8, CColoredStripShader::Mode::Subtractive, {})
+, m_beamStrip1Sub(8, CColoredStripShader::Mode::Subtractive, {})
-, m_beamStrip2Sub( 10, CColoredStripShader::Mode::Subtractive, tex)
+, m_beamStrip2Sub(10, CColoredStripShader::Mode::Subtractive, tex)
-, m_beamStrip3Sub( 18, CColoredStripShader::Mode::Subtractive, tex)
+, m_beamStrip3Sub(18, CColoredStripShader::Mode::Subtractive, tex)
-, m_beamStrip4Sub( 14, CColoredStripShader::Mode::Subtractive, glowTex)
+, m_beamStrip4Sub(14, CColoredStripShader::Mode::Subtractive, glowTex)
-, m_motionBlurStrip( 16, CColoredStripShader::Mode::Alpha, {}) {}
+, m_motionBlurStrip(16, CColoredStripShader::Mode::Alpha, {}) {}
 
 CPlasmaProjectile::CPlasmaProjectile(const TToken<CWeaponDescription>& wDesc, std::string_view name, EWeaponType wType,
                                      const CBeamInfo& bInfo, const zeus::CTransform& xf, EMaterialTypes matType,

Runtime/Weapon/CPlasmaProjectile.hpp

@@ -91,8 +91,7 @@ private:
     CColoredStripShader m_beamStrip3Sub;
     CColoredStripShader m_beamStrip4Sub;
     CColoredStripShader m_motionBlurStrip;
-    RenderObjects(std::shared_ptr<aurora::gfx::TextureHandle> tex,
+    RenderObjects(aurora::gfx::TextureHandle tex, aurora::gfx::TextureHandle glowTex);
-                  std::shared_ptr<aurora::gfx::TextureHandle> glowTex);
   };
   std::optional<RenderObjects> m_renderObjs;
 

Runtime/Weapon/CWaveBuster.cpp

@@ -25,8 +25,8 @@ CWaveBuster::CWaveBuster(const TToken<CWeaponDescription>& desc, EWeaponType typ
 , x360_busterSwoosh2(g_SimplePool->GetObj("BusterSwoosh2"))
 , x36c_busterSparks(g_SimplePool->GetObj("BusterSparks"))
 , x378_busterLight(g_SimplePool->GetObj("BusterLight"))
-, m_lineRenderer1(CLineRenderer::EPrimitiveMode::LineStrip, 36 * 6, nullptr, true)
+, m_lineRenderer1(CLineRenderer::EPrimitiveMode::LineStrip, 36 * 6, {}, true)
-, m_lineRenderer2(CLineRenderer::EPrimitiveMode::LineStrip, 36 * 6, nullptr, true) {
+, m_lineRenderer2(CLineRenderer::EPrimitiveMode::LineStrip, 36 * 6, {}, true) {
   x354_busterSwoosh1.GetObj();
   x360_busterSwoosh2.GetObj();
   x36c_busterSparks.GetObj();
@@ -302,7 +302,7 @@ CRayCastResult CWaveBuster::SeekDamageTarget(TUniqueId& uid, const zeus::CVector
     uid = physId;
     return physRes;
   }
-  
+
   if (actRes.IsValid() && ApplyDamageToTarget(physId, actRes, physRes, res, mgr, dt)) {
     uid = actId;
     return actRes;

Runtime/World/CActor.hpp

@@ -75,7 +75,7 @@ protected:
   bool xe7_30_doTargetDistanceTest : 1 = true;
   bool xe7_31_targetable : 1 = true;
 
-  std::shared_ptr<aurora::gfx::TextureHandle> m_reflectionCube;
+  aurora::gfx::TextureHandle m_reflectionCube;
   zeus::CColor m_debugAddColor = zeus::skClear;
   float m_debugAddColorTime = 0.f;
 

Runtime/World/CFluidPlaneCPU.cpp

@@ -271,7 +271,7 @@ CFluidPlaneShader::RenderSetupInfo CFluidPlaneCPU::RenderSetup(const CStateManag
     m_cachedAdditive = mgr.GetThermalDrawFlag() == EThermalDrawFlag::Hot;
     m_shader.emplace(x44_fluidType, x10_texPattern1, x20_texPattern2, x30_texColor, xb0_bumpMap, xc0_envMap,
                      xd0_envBumpMap, xe0_lightmap,
-                     m_tessellation ? CFluidPlaneManager::RippleMapTex : std::shared_ptr<aurora::gfx::TextureHandle>{},
+                     m_tessellation ? CFluidPlaneManager::RippleMapTex : aurora::gfx::TextureHandle{},
                      m_cachedDoubleLightmapBlend, m_cachedAdditive, m_maxVertCount);
   }
 

Runtime/World/CFluidPlaneManager.cpp

@@ -81,7 +81,7 @@ static bool g_RippleMapSetup = false;
 std::array<std::array<u8, 64>, 64> CFluidPlaneManager::RippleValues{};
 std::array<u8, 64> CFluidPlaneManager::RippleMins{};
 std::array<u8, 64> CFluidPlaneManager::RippleMaxs{};
-std::shared_ptr<aurora::gfx::TextureHandle> CFluidPlaneManager::RippleMapTex;
+aurora::gfx::TextureHandle CFluidPlaneManager::RippleMapTex;
 
 void CFluidPlaneManager::SetupRippleMap() {
   if (g_RippleMapSetup) {

Runtime/World/CFluidPlaneManager.hpp

@@ -46,7 +46,7 @@ public:
   static std::array<std::array<u8, 64>, 64> RippleValues;
   static std::array<u8, 64> RippleMins;
   static std::array<u8, 64> RippleMaxs;
-  static std::shared_ptr<aurora::gfx::TextureHandle> RippleMapTex;
+  static aurora::gfx::TextureHandle RippleMapTex;
 
   CFluidPlaneManager();
   void StartFrame(bool);

aurora/CMakeLists.txt

@@ -14,13 +14,16 @@ add_library(aurora STATIC
     lib/imgui.cpp
     lib/dawn/BackendBinding.cpp
     lib/gfx/common.cpp
+    lib/gfx/texture.cpp
+    lib/gfx/movie_player/shader.cpp
+    # TODO move to imgui? or move imgui impl to here?
     ../extern/imgui/backends/imgui_impl_sdl.cpp
     ../extern/imgui/backends/imgui_impl_wgpu.cpp
     )
 target_compile_definitions(aurora PRIVATE IMGUI_USER_CONFIG="imconfig_user.h") # IMGUI_USE_WCHAR32
 target_include_directories(aurora PUBLIC include ../Runtime)
 target_include_directories(aurora PRIVATE ../imgui ../extern/imgui)
-target_link_libraries(aurora PRIVATE dawn_native dawncpp webgpu_dawn zeus logvisor SDL2-static)
+target_link_libraries(aurora PRIVATE dawn_native dawncpp webgpu_dawn zeus logvisor SDL2-static xxhash)
 if (APPLE)
   target_compile_definitions(aurora PRIVATE DAWN_ENABLE_BACKEND_METAL)
   target_sources(aurora PRIVATE lib/dawn/MetalBinding.mm)

aurora/include/aurora/gfx.hpp

@@ -3,6 +3,7 @@
 #include "common.hpp"
 
 #include <cstdint>
+#include <utility>
 
 #include <zeus/CAABox.hpp>
 #include <zeus/CColor.hpp>
@@ -39,16 +40,13 @@ struct CFogState {
 } // namespace metaforce
 
 namespace aurora::gfx {
-struct TextureRef {
+struct TextureRef;
-  uint32_t id;
-  bool render;
-};
 struct TextureHandle {
-  TextureRef ref;
+  std::shared_ptr<TextureRef> ref;
-  explicit TextureHandle(TextureRef ref) : ref(ref) {}
+  TextureHandle() = default;
-  ~TextureHandle() noexcept;
+  TextureHandle(std::shared_ptr<TextureRef>&& ref) : ref(std::move(ref)) {}
-  TextureHandle(const TextureHandle&) = delete;
+  operator bool() const { return ref.operator bool(); }
-  TextureHandle& operator=(const TextureHandle&) = delete;
+  void reset() { ref.reset(); }
 };
 enum class TextureFormat : uint8_t {
   RGBA8,
@@ -106,24 +104,23 @@ void add_model(/* TODO */) noexcept;
 void queue_aabb(const zeus::CAABox& aabb, const zeus::CColor& color, bool z_only) noexcept;
 void queue_fog_volume_plane(const ArrayRef<zeus::CVector4f>& verts, uint8_t pass);
 void queue_fog_volume_filter(const zeus::CColor& color, bool two_way) noexcept;
-void queue_textured_quad_verts(CameraFilterType filter_type, TextureRef texture, ZTest z_comparison, bool z_test,
+void queue_textured_quad_verts(CameraFilterType filter_type, const TextureHandle& texture, ZTest z_comparison,
-                               const zeus::CColor& color, const ArrayRef<zeus::CVector3f>& pos,
+                               bool z_test, const zeus::CColor& color, const ArrayRef<zeus::CVector3f>& pos,
                                const ArrayRef<zeus::CVector2f>& uvs, float lod) noexcept;
-void queue_textured_quad(CameraFilterType filter_type, TextureRef texture, ZTest z_comparison, bool z_test,
+void queue_textured_quad(CameraFilterType filter_type, const TextureHandle& texture, ZTest z_comparison, bool z_test,
                          const zeus::CColor& color, float uv_scale, const zeus::CRectangle& rect, float z) noexcept;
 void queue_colored_quad_verts(CameraFilterType filter_type, ZTest z_comparison, bool z_test, const zeus::CColor& color,
                               const ArrayRef<zeus::CVector3f>& pos) noexcept;
 void queue_colored_quad(CameraFilterType filter_type, ZTest z_comparison, bool z_test, const zeus::CColor& color,
                         const zeus::CRectangle& rect, float z) noexcept;
-void queue_movie_player(TextureRef tex_y, TextureRef tex_u, TextureRef tex_v, const zeus::CColor& color, float h_pad,
+void queue_movie_player(const TextureHandle& tex_y, const TextureHandle& tex_u, const TextureHandle& tex_v,
-                        float v_pad) noexcept;
+                        const zeus::CColor& color, float h_pad, float v_pad) noexcept;
 
-std::shared_ptr<TextureHandle> new_static_texture_2d(uint32_t width, uint32_t height, uint32_t mips,
+TextureHandle new_static_texture_2d(uint32_t width, uint32_t height, uint32_t mips, TextureFormat format,
-                                                     TextureFormat format, ArrayRef<uint8_t> data,
+                                    ArrayRef<uint8_t> data, zstring_view label) noexcept;
-                                                     std::string_view label) noexcept;
+TextureHandle new_dynamic_texture_2d(uint32_t width, uint32_t height, uint32_t mips, TextureFormat format,
-std::shared_ptr<TextureHandle> new_dynamic_texture_2d(uint32_t width, uint32_t height, uint32_t mips,
+                                     zstring_view label) noexcept;
-                                                      TextureFormat format, std::string_view label) noexcept;
+TextureHandle new_render_texture(uint32_t width, uint32_t height, uint32_t color_bind_count, uint32_t depth_bind_count,
-std::shared_ptr<TextureHandle> new_render_texture(uint32_t width, uint32_t height, uint32_t color_bind_count,
+                                 zstring_view label) noexcept;
-                                                  uint32_t depth_bind_count, std::string_view label) noexcept;
+void write_texture(const TextureHandle& handle, ArrayRef<uint8_t> data) noexcept;
-void write_texture(TextureRef ref, ArrayRef<uint8_t> data) noexcept;
 } // namespace aurora::gfx

aurora/lib/aurora.cpp

@@ -2,6 +2,8 @@
 
 #include <SDL.h>
 #include <dawn/native/DawnNative.h>
+// TODO HACK: dawn doesn't expose device toggles
+#include "../extern/dawn/src/dawn/native/Toggles.h"
 #include <dawn/webgpu_cpp.h>
 #include <logvisor/logvisor.hpp>
 #include <magic_enum.hpp>
@@ -25,6 +27,14 @@
 #include <backends/imgui_impl_sdl.h>
 #include <backends/imgui_impl_wgpu.h>
 
+// TODO HACK: dawn doesn't expose device toggles
+namespace dawn::native {
+class DeviceBase {
+public:
+  void SetToggle(Toggle toggle, bool isEnabled);
+};
+} // namespace dawn::native
+
 namespace aurora {
 // TODO: Move global state to a class/struct?
 static logvisor::Module Log("aurora");
@@ -184,13 +194,7 @@ static bool poll_events() noexcept {
       Log.report(logvisor::Info, FMT_STRING("Received quit request"));
       return false;
     }
-    // TODO why doesn't this work?
+    // Log.report(logvisor::Info, FMT_STRING("Received SDL event: {}"), event.type);
-//    const auto typedEvent = magic_enum::enum_cast<SDL_EventType>(event.type);
-//    if (typedEvent) {
-//      Log.report(logvisor::Info, FMT_STRING("Received SDL event: {}"), magic_enum::enum_name(typedEvent.value()));
-//    } else {
-//      Log.report(logvisor::Info, FMT_STRING("Received SDL event: {}"), event.type);
-//    }
   }
   return true;
 }
@@ -252,7 +256,20 @@ void app_run(std::unique_ptr<AppDelegate> app, Icon icon, int argc, char** argv)
   const auto backendName = magic_enum::enum_name(g_AdapterProperties.backendType);
   Log.report(logvisor::Info, FMT_STRING("Using {} graphics backend"), backendName);
 
-  g_Device = wgpu::Device::Acquire(g_Adapter.CreateDevice());
+  {
+    const std::array<wgpu::FeatureName, 1> requiredFeatures{
+        wgpu::FeatureName::TextureCompressionBC,
+    };
+    const auto deviceDescriptor = wgpu::DeviceDescriptor{
+        .requiredFeaturesCount = requiredFeatures.size(),
+        .requiredFeatures = requiredFeatures.data(),
+    };
+    g_Device = wgpu::Device::Acquire(g_Adapter.CreateDevice(&deviceDescriptor));
+    // TODO HACK: dawn doesn't expose device toggles
+    static_cast<dawn::native::DeviceBase*>(static_cast<void*>(g_Device.Get()))
+        ->SetToggle(dawn::native::Toggle::UseUserDefinedLabelsInBackend, true);
+  }
+
   g_BackendBinding = std::unique_ptr<utils::BackendBinding>(
       utils::CreateBinding(g_AdapterProperties.backendType, g_Window, g_Device.Get()));
   if (!g_BackendBinding) {
@@ -268,6 +285,7 @@ void app_run(std::unique_ptr<AppDelegate> app, Icon icon, int argc, char** argv)
   {
     auto size = get_window_size();
     auto format = static_cast<wgpu::TextureFormat>(g_BackendBinding->GetPreferredSwapChainTextureFormat());
+    Log.report(logvisor::Info, FMT_STRING("Using swapchain format {}"), magic_enum::enum_name(format));
     g_SwapChain.Configure(format, wgpu::TextureUsage::RenderAttachment, size.width, size.height);
   }
 
@@ -283,7 +301,6 @@ void app_run(std::unique_ptr<AppDelegate> app, Icon icon, int argc, char** argv)
   g_AppDelegate->onAppLaunched();
   g_AppDelegate->onAppWindowResized(get_window_size());
 
-  bool showDemo = true;
   while (poll_events()) {
     ImGui_ImplWGPU_NewFrame();
     ImGui_ImplSDL2_NewFrame();
@@ -293,9 +310,6 @@ void app_run(std::unique_ptr<AppDelegate> app, Icon icon, int argc, char** argv)
 
     const wgpu::TextureView view = g_SwapChain.GetCurrentTextureView();
     g_AppDelegate->onAppDraw();
-    if (showDemo) {
-      ImGui::ShowDemoWindow(&showDemo);
-    }
     ImGui::Render();
 
     auto encoder = g_Device.CreateCommandEncoder();

aurora/lib/gfx/common.cpp

@@ -1,24 +1,65 @@
-#include <aurora/gfx.hpp>
+#include "common.hpp"
+
+#include <magic_enum.hpp>
+
+#include "movie_player/shader.hpp"
 
 namespace aurora::gfx {
-bool get_dxt_compression_supported() noexcept {
+struct ShaderDrawCommand {
-  return true; // TODO
+  ShaderType type;
-}
+  union {
+    movie_player::DrawData moviePlayer;
+  };
+};
+struct PipelineCreateCommand {
+  ShaderType type;
+  union {
+    movie_player::PipelineConfig moviePlayer;
+  };
+};
+enum class CommandType {
+  SetViewport,
+  SetScissor,
+  Draw,
+};
+struct Command {
+  CommandType type;
+  union Data {
+    struct SetViewportCommand {
+      zeus::CRectangle rect;
+      float znear;
+      float zfar;
+    } setViewport;
+    struct SetScissorCommand {
+      uint32_t x;
+      uint32_t y;
+      uint32_t w;
+      uint32_t h;
+    } setScissor;
+    ShaderDrawCommand draw;
+  } data;
+};
+
+zeus::CMatrix4f g_mv;
+zeus::CMatrix4f g_mvInv;
+zeus::CMatrix4f g_proj;
+metaforce::CFogState g_fogState;
+
+std::vector<Command> g_commands;
+
+bool get_dxt_compression_supported() noexcept { return g_Device.HasFeature(wgpu::FeatureName::TextureCompressionBC); }
 
 void update_model_view(const zeus::CMatrix4f& mv, const zeus::CMatrix4f& mv_inv) noexcept {
-  // TODO
+  g_mv = mv;
-}
+  g_mvInv = mv_inv;
-void update_projection(const zeus::CMatrix4f& proj) noexcept {
-  // TODO
-}
-void update_fog_state(const metaforce::CFogState& state) noexcept {
-  // TODO
 }
+void update_projection(const zeus::CMatrix4f& proj) noexcept { g_proj = proj; }
+void update_fog_state(const metaforce::CFogState& state) noexcept { g_fogState = state; }
 void set_viewport(const zeus::CRectangle& rect, float znear, float zfar) noexcept {
-  // TODO
+  g_commands.push_back({CommandType::SetViewport, {.setViewport = {rect, znear, zfar}}});
 }
 void set_scissor(uint32_t x, uint32_t y, uint32_t w, uint32_t h) noexcept {
-  // TODO
+  g_commands.push_back({CommandType::SetScissor, {.setScissor = {x, y, w, h}}});
 }
 
 void resolve_color(const ClipRect& rect, uint32_t bind, bool clear_depth) noexcept {
@@ -40,12 +81,12 @@ void queue_fog_volume_plane(const ArrayRef<zeus::CVector4f>& verts, uint8_t pass
 void queue_fog_volume_filter(const zeus::CColor& color, bool two_way) noexcept {
   // TODO
 }
-void queue_textured_quad_verts(CameraFilterType filter_type, TextureRef texture, ZTest z_comparison, bool z_test,
+void queue_textured_quad_verts(CameraFilterType filter_type, const TextureHandle& texture, ZTest z_comparison,
-                               const zeus::CColor& color, const ArrayRef<zeus::CVector3f>& pos,
+                               bool z_test, const zeus::CColor& color, const ArrayRef<zeus::CVector3f>& pos,
                                const ArrayRef<zeus::CVector2f>& uvs, float lod) noexcept {
   // TODO
 }
-void queue_textured_quad(CameraFilterType filter_type, TextureRef texture, ZTest z_comparison, bool z_test,
+void queue_textured_quad(CameraFilterType filter_type, const TextureHandle& texture, ZTest z_comparison, bool z_test,
                          const zeus::CColor& color, float uv_scale, const zeus::CRectangle& rect, float z) noexcept {
   // TODO
 }
@@ -57,25 +98,8 @@ void queue_colored_quad(CameraFilterType filter_type, ZTest z_comparison, bool z
                         const zeus::CRectangle& rect, float z) noexcept {
   // TODO
 }
-void queue_movie_player(TextureRef tex_y, TextureRef tex_u, TextureRef tex_v, const zeus::CColor& color, float h_pad,
+void queue_movie_player(const TextureHandle& tex_y, const TextureHandle& tex_u, const TextureHandle& tex_v,
-                        float v_pad) noexcept {
+                        const zeus::CColor& color, float h_pad, float v_pad) noexcept {
-  // TODO
-}
-
-std::shared_ptr<TextureHandle> new_static_texture_2d(uint32_t width, uint32_t height, uint32_t mips,
-                                                     TextureFormat format, ArrayRef<uint8_t> data,
-                                                     std::string_view label) noexcept {
-  return {}; // TODO
-}
-std::shared_ptr<TextureHandle> new_dynamic_texture_2d(uint32_t width, uint32_t height, uint32_t mips,
-                                                      TextureFormat format, std::string_view label) noexcept {
-  return {}; // TODO
-}
-std::shared_ptr<TextureHandle> new_render_texture(uint32_t width, uint32_t height, uint32_t color_bind_count,
-                                                  uint32_t depth_bind_count, std::string_view label) noexcept {
-  return {}; // TODO
-}
-void write_texture(TextureRef ref, ArrayRef<uint8_t> data) noexcept {
   // TODO
 }
 } // namespace aurora::gfx

aurora/lib/gfx/common.hpp

@@ -0,0 +1,48 @@
+#pragma once
+#include <aurora/gfx.hpp>
+
+#include <utility>
+
+#include <dawn/webgpu_cpp.h>
+#include <xxhash.h>
+#if defined(__x86_64__) || defined(__i386__) || defined(_M_IX86) || defined(_M_X64)
+#include <xxh_x86dispatch.h>
+#endif
+
+template <typename T>
+XXH64_hash_t xxh3_hash(const T& input, XXH64_hash_t seed = 0) {
+  return XXH3_64bits_withSeed(&input, sizeof(T), seed);
+}
+
+namespace aurora {
+extern wgpu::Device g_Device;
+extern wgpu::Queue g_Queue;
+} // namespace aurora
+
+namespace aurora::gfx {
+static logvisor::Module Log("aurora::gfx");
+
+extern zeus::CMatrix4f g_mv;
+extern zeus::CMatrix4f g_mvInv;
+extern zeus::CMatrix4f g_proj;
+extern metaforce::CFogState g_fogState;
+
+struct TextureRef {
+  wgpu::Texture texture;
+  wgpu::TextureView view;
+  wgpu::Extent3D size;
+  wgpu::TextureFormat format;
+
+  TextureRef(wgpu::Texture&& texture, wgpu::TextureView&& view, wgpu::Extent3D size, wgpu::TextureFormat format)
+  : texture(std::move(texture)), view(std::move(view)), size(size), format(format) {}
+};
+
+using PipelineRef = uint64_t;
+using Range = std::pair<uint64_t, uint64_t>;
+
+enum class ShaderType {
+  Aabb,
+  TexturedQuad,
+  MoviePlayer,
+};
+} // namespace aurora::gfx

aurora/lib/gfx/movie_player/shader.hpp

@@ -0,0 +1,29 @@
+#include "../common.hpp"
+
+namespace aurora::gfx::movie_player {
+struct DrawData {
+  PipelineRef pipeline;
+  Range vertRange;
+  Range uniformRange;
+  uint64_t bindGroupId;
+};
+
+struct PipelineConfig {
+  // nothing
+};
+const std::array INITIAL_PIPELINES{
+    PipelineConfig{},
+};
+
+struct State {
+  wgpu::ShaderModule shader;
+  wgpu::BindGroupLayout uniformLayout;
+  wgpu::BindGroup uniformBindGroup;
+  wgpu::BindGroupLayout textureLayout;
+  wgpu::Sampler sampler;
+  wgpu::PipelineLayout pipelineLayout;
+  // Transient state
+  std::unordered_map<uint64_t, wgpu::BindGroup> textureBindGroups;
+  std::vector<uint64_t> frameUsedTextures;
+};
+} // namespace aurora::gfx::movie_player

aurora/lib/gfx/movie_player/shader.wgsl

@@ -0,0 +1,42 @@
+struct Uniform {
+    xf: mat4x4<f32>;
+    color: vec4<f32>;
+};
+@group(0) @binding(0)
+var<uniform> ubuf: Uniform;
+@group(0) @binding(1)
+var tex_sampler: sampler;
+@group(1) @binding(0)
+var tex_y: texture_2d<f32>;
+@group(1) @binding(1)
+var tex_u: texture_2d<f32>;
+@group(1) @binding(2)
+var tex_v: texture_2d<f32>;
+
+struct VertexOutput {
+    @builtin(position) pos: vec4<f32>;
+    @location(0) uv: vec2<f32>;
+};
+
+@stage(vertex)
+fn vs_main(@location(0) in_pos: vec3<f32>, @location(1) in_uv: vec2<f32>) -> VertexOutput {
+    var out: VertexOutput;
+    out.pos = ubuf.xf * vec4<f32>(in_pos, 1.0);
+    out.uv = in_uv;
+    return out;
+}
+
+@stage(fragment)
+fn fs_main(in: VertexOutput) -> @location(0) vec4<f32> {
+    var yuv = vec3<f32>(
+        1.1643 * (textureSample(tex_y, tex_sampler, in.uv).x - 0.0625),
+        textureSample(tex_u, tex_sampler, in.uv).x - 0.5,
+        textureSample(tex_v, tex_sampler, in.uv).x - 0.5
+    );
+    return ubuf.color * vec4<f32>(
+        yuv.x + 1.5958 * yuv.z,
+        yuv.x - 0.39173 * yuv.y - 0.8129 * yuv.z,
+        yuv.x + 2.017 * yuv.y,
+        1.0
+    );
+}

aurora/lib/gfx/texture.cpp

@@ -0,0 +1,147 @@
+#include "common.hpp"
+
+#include <magic_enum.hpp>
+
+namespace aurora::gfx {
+static wgpu::TextureFormat to_wgpu(TextureFormat format) {
+  switch (format) {
+  case TextureFormat::RGBA8:
+    return wgpu::TextureFormat::RGBA8Unorm;
+  case TextureFormat::R8:
+    return wgpu::TextureFormat::R8Unorm;
+  case TextureFormat::R32Float:
+    return wgpu::TextureFormat::R32Float;
+  case TextureFormat::DXT1:
+    return wgpu::TextureFormat::BC1RGBAUnorm;
+  case TextureFormat::DXT3:
+    return wgpu::TextureFormat::BC3RGBAUnorm;
+  case TextureFormat::DXT5:
+    return wgpu::TextureFormat::BC5RGUnorm;
+  case TextureFormat::BPTC:
+    return wgpu::TextureFormat::BC7RGBAUnorm;
+  }
+}
+struct TextureFormatInfo {
+  uint8_t blockWidth;
+  uint8_t blockHeight;
+  uint8_t blockSize;
+  bool compressed;
+};
+static TextureFormatInfo format_info(wgpu::TextureFormat format) {
+  switch (format) {
+  case wgpu::TextureFormat::R8Unorm:
+    return {1, 1, 1, false};
+  case wgpu::TextureFormat::RGBA8Unorm:
+  case wgpu::TextureFormat::R32Float:
+    return {1, 1, 4, false};
+  case wgpu::TextureFormat::BC1RGBAUnorm:
+    return {4, 4, 8, true};
+  default:
+    Log.report(logvisor::Fatal, FMT_STRING("format_info: unimplemented format {}"), magic_enum::enum_name(format));
+    return {0, 0, 0, false};
+  }
+}
+static wgpu::Extent3D physical_size(wgpu::Extent3D size, TextureFormatInfo info) {
+  const uint32_t width = ((size.width + info.blockWidth - 1) / info.blockWidth) * info.blockWidth;
+  const uint32_t height = ((size.height + info.blockHeight - 1) / info.blockHeight) * info.blockHeight;
+  return {width, height, size.depthOrArrayLayers};
+}
+
+TextureHandle new_static_texture_2d(uint32_t width, uint32_t height, uint32_t mips, TextureFormat format,
+                                    ArrayRef<uint8_t> data, zstring_view label) noexcept {
+  auto handle = new_dynamic_texture_2d(width, height, mips, format, label);
+  const TextureRef& ref = *handle.ref;
+  uint32_t offset = 0;
+  for (uint32_t mip = 0; mip < mips; ++mip) {
+    const auto mipSize = wgpu::Extent3D{
+        .width = std::max(ref.size.width >> mip, 1u),
+        .height = std::max(ref.size.height >> mip, 1u),
+        .depthOrArrayLayers = ref.size.depthOrArrayLayers,
+    };
+    const auto info = format_info(ref.format);
+    const auto physicalSize = physical_size(mipSize, info);
+    const uint32_t widthBlocks = physicalSize.width / info.blockWidth;
+    const uint32_t heightBlocks = physicalSize.height / info.blockHeight;
+    const uint32_t bytesPerRow = widthBlocks * info.blockSize;
+    const uint32_t dataSize = bytesPerRow * heightBlocks * mipSize.depthOrArrayLayers;
+    if (offset + dataSize > data.size()) {
+      Log.report(logvisor::Fatal, FMT_STRING("new_static_texture_2d[{}]: expected at least {} bytes, got {}"), label,
+                 offset + dataSize, data.size());
+      return {};
+    }
+    const auto dstView = wgpu::ImageCopyTexture{
+        .texture = ref.texture,
+        .mipLevel = mip,
+    };
+    const auto dataLayout = wgpu::TextureDataLayout{
+        .bytesPerRow = bytesPerRow,
+        .rowsPerImage = heightBlocks,
+    };
+    g_Queue.WriteTexture(&dstView, data.data() + offset, dataSize, &dataLayout, &physicalSize);
+    offset += dataSize;
+  }
+  if (offset < data.size()) {
+    Log.report(logvisor::Warning, FMT_STRING("new_static_texture_2d[{}]: texture used {} bytes, but given {} bytes"),
+               label, offset, data.size());
+  }
+  return handle;
+}
+
+TextureHandle new_dynamic_texture_2d(uint32_t width, uint32_t height, uint32_t mips, TextureFormat format,
+                                     zstring_view label) noexcept {
+  const auto wgpuFormat = to_wgpu(format);
+  const auto size = wgpu::Extent3D{
+      .width = width,
+      .height = height,
+  };
+  const auto textureDescriptor = wgpu::TextureDescriptor{
+      .label = label.c_str(),
+      .usage = wgpu::TextureUsage::TextureBinding | wgpu::TextureUsage::CopyDst,
+      .size = size,
+      .format = wgpuFormat,
+      .mipLevelCount = mips,
+  };
+  const auto viewLabel = fmt::format(FMT_STRING("{} view"), label);
+  const auto textureViewDescriptor = wgpu::TextureViewDescriptor{
+      .label = viewLabel.c_str(),
+      .format = wgpuFormat,
+      .dimension = wgpu::TextureViewDimension::e2D,
+      .mipLevelCount = mips,
+  };
+  auto texture = g_Device.CreateTexture(&textureDescriptor);
+  auto textureView = texture.CreateView(&textureViewDescriptor);
+  return {std::make_shared<TextureRef>(std::move(texture), std::move(textureView), size, wgpuFormat)};
+}
+
+TextureHandle new_render_texture(uint32_t width, uint32_t height, uint32_t color_bind_count, uint32_t depth_bind_count,
+                                 zstring_view label) noexcept {
+  return {}; // TODO
+}
+
+// TODO accept mip/layer parameters
+void write_texture(const TextureHandle& handle, ArrayRef<uint8_t> data) noexcept {
+  const TextureRef& ref = *handle.ref;
+  const auto dstView = wgpu::ImageCopyTexture{
+      .texture = ref.texture,
+  };
+  if (ref.size.depthOrArrayLayers != 1) {
+    Log.report(logvisor::Fatal, FMT_STRING("write_texture: unsupported depth {}"), ref.size.depthOrArrayLayers);
+    return;
+  }
+  const auto info = format_info(ref.format);
+  const auto physicalSize = physical_size(ref.size, info);
+  const uint32_t widthBlocks = physicalSize.width / info.blockWidth;
+  const uint32_t heightBlocks = physicalSize.height / info.blockHeight;
+  const uint32_t bytesPerRow = widthBlocks * info.blockSize;
+  const uint32_t dataSize = bytesPerRow * heightBlocks * ref.size.depthOrArrayLayers;
+  if (dataSize > data.size()) {
+    Log.report(logvisor::Fatal, FMT_STRING("write_texture: expected at least {} bytes, got {}"), dataSize, data.size());
+    return;
+  }
+  const auto dataLayout = wgpu::TextureDataLayout{
+      .bytesPerRow = bytesPerRow,
+      .rowsPerImage = heightBlocks,
+  };
+  g_Queue.WriteTexture(&dstView, data.data(), dataSize, &dataLayout, &ref.size);
+}
+} // namespace aurora::gfx

extern/boo

@@ -1 +1 @@
-Subproject commit 693500c055a5c328dbf3de90c94033f8ca497bb3
+Subproject commit e458d1999cb613c5dc221c9805c81cf7e8a24fb6

extern/xxhash/CMakeLists.txt

@@ -0,0 +1,6 @@
+add_library(xxhash xxhash.c)
+if(CMAKE_SYSTEM_PROCESSOR MATCHES "x86_64|amd64|AMD64")
+  target_sources(xxhash PRIVATE xxh_x86dispatch.c)
+endif ()
+target_include_directories(xxhash PUBLIC ${CMAKE_CURRENT_SOURCE_DIR})
+target_compile_definitions(xxhash INTERFACE XXH_STATIC_LINKING_ONLY)

extern/xxhash/LICENSE

@@ -0,0 +1,26 @@
+xxHash Library
+Copyright (c) 2012-2021 Yann Collet
+All rights reserved.
+
+BSD 2-Clause License (https://www.opensource.org/licenses/bsd-license.php)
+
+Redistribution and use in source and binary forms, with or without modification,
+are permitted provided that the following conditions are met:
+
+* Redistributions of source code must retain the above copyright notice, this
+  list of conditions and the following disclaimer.
+
+* Redistributions in binary form must reproduce the above copyright notice, this
+  list of conditions and the following disclaimer in the documentation and/or
+  other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
+ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
+WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR
+ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
+(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
+LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
+ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

extern/xxhash/xxh_x86dispatch.c

@@ -0,0 +1,770 @@
+/*
+ * xxHash - Extremely Fast Hash algorithm
+ * Copyright (C) 2020-2021 Yann Collet
+ *
+ * BSD 2-Clause License (https://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *    * Redistributions of source code must retain the above copyright
+ *      notice, this list of conditions and the following disclaimer.
+ *    * Redistributions in binary form must reproduce the above
+ *      copyright notice, this list of conditions and the following disclaimer
+ *      in the documentation and/or other materials provided with the
+ *      distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * You can contact the author at:
+ *   - xxHash homepage: https://www.xxhash.com
+ *   - xxHash source repository: https://github.com/Cyan4973/xxHash
+ */
+
+
+/*!
+ * @file xxh_x86dispatch.c
+ *
+ * Automatic dispatcher code for the @ref XXH3_family on x86-based targets.
+ *
+ * Optional add-on.
+ *
+ * **Compile this file with the default flags for your target.** Do not compile
+ * with flags like `-mavx*`, `-march=native`, or `/arch:AVX*`, there will be
+ * an error. See @ref XXH_X86DISPATCH_ALLOW_AVX for details.
+ *
+ * @defgroup dispatch x86 Dispatcher
+ * @{
+ */
+
+#if defined (__cplusplus)
+extern "C" {
+#endif
+
+#if !(defined(__x86_64__) || defined(__i386__) || defined(_M_IX86) || defined(_M_X64))
+#  error "Dispatching is currently only supported on x86 and x86_64."
+#endif
+
+/*!
+ * @def XXH_X86DISPATCH_ALLOW_AVX
+ * @brief Disables the AVX sanity check.
+ *
+ * Don't compile xxh_x86dispatch.c with options like `-mavx*`, `-march=native`,
+ * or `/arch:AVX*`. It is intended to be compiled for the minimum target, and
+ * it selectively enables SSE2, AVX2, and AVX512 when it is needed.
+ *
+ * Using this option _globally_ allows this feature, and therefore makes it
+ * undefined behavior to execute on any CPU without said feature.
+ *
+ * Even if the source code isn't directly using AVX intrinsics in a function,
+ * the compiler can still generate AVX code from autovectorization and by
+ * "upgrading" SSE2 intrinsics to use the VEX prefixes (a.k.a. AVX128).
+ *
+ * Use the same flags that you use to compile the rest of the program; this
+ * file will safely generate SSE2, AVX2, and AVX512 without these flags.
+ *
+ * Define XXH_X86DISPATCH_ALLOW_AVX to ignore this check, and feel free to open
+ * an issue if there is a target in the future where AVX is a default feature.
+ */
+#ifdef XXH_DOXYGEN
+#  define XXH_X86DISPATCH_ALLOW_AVX
+#endif
+
+#if defined(__AVX__) && !defined(XXH_X86DISPATCH_ALLOW_AVX)
+#  error "Do not compile xxh_x86dispatch.c with AVX enabled! See the comment above."
+#endif
+
+#ifdef __has_include
+#  define XXH_HAS_INCLUDE(header) __has_include(header)
+#else
+#  define XXH_HAS_INCLUDE(header) 0
+#endif
+
+/*!
+ * @def XXH_DISPATCH_SCALAR
+ * @brief Enables/dispatching the scalar code path.
+ *
+ * If this is defined to 0, SSE2 support is assumed. This reduces code size
+ * when the scalar path is not needed.
+ *
+ * This is automatically defined to 0 when...
+ *   - SSE2 support is enabled in the compiler
+ *   - Targeting x86_64
+ *   - Targeting Android x86
+ *   - Targeting macOS
+ */
+#ifndef XXH_DISPATCH_SCALAR
+#  if defined(__SSE2__) || (defined(_M_IX86_FP) && _M_IX86_FP >= 2) /* SSE2 on by default */ \
+     || defined(__x86_64__) || defined(_M_X64) /* x86_64 */ \
+     || defined(__ANDROID__) || defined(__APPLEv__) /* Android or macOS */
+#     define XXH_DISPATCH_SCALAR 0 /* disable */
+#  else
+#     define XXH_DISPATCH_SCALAR 1
+#  endif
+#endif
+/*!
+ * @def XXH_DISPATCH_AVX2
+ * @brief Enables/disables dispatching for AVX2.
+ *
+ * This is automatically detected if it is not defined.
+ *  - GCC 4.7 and later are known to support AVX2, but >4.9 is required for
+ *    to get the AVX2 intrinsics and typedefs without -mavx -mavx2.
+ *  - Visual Studio 2013 Update 2 and later are known to support AVX2.
+ *  - The GCC/Clang internal header `<avx2intrin.h>` is detected. While this is
+ *    not allowed to be included directly, it still appears in the builtin
+ *    include path and is detectable with `__has_include`.
+ *
+ * @see XXH_AVX2
+ */
+#ifndef XXH_DISPATCH_AVX2
+#  if (defined(__GNUC__) && (__GNUC__ > 4)) /* GCC 5.0+ */ \
+   || (defined(_MSC_VER) && _MSC_VER >= 1900) /* VS 2015+ */ \
+   || (defined(_MSC_FULL_VER) && _MSC_FULL_VER >= 180030501) /* VS 2013 Update 2 */ \
+   || XXH_HAS_INCLUDE(<avx2intrin.h>) /* GCC/Clang internal header */
+#    define XXH_DISPATCH_AVX2 1   /* enable dispatch towards AVX2 */
+#  else
+#    define XXH_DISPATCH_AVX2 0
+#  endif
+#endif /* XXH_DISPATCH_AVX2 */
+
+/*!
+ * @def XXH_DISPATCH_AVX512
+ * @brief Enables/disables dispatching for AVX512.
+ *
+ * Automatically detected if one of the following conditions is met:
+ *  - GCC 4.9 and later are known to support AVX512.
+ *  - Visual Studio 2017  and later are known to support AVX2.
+ *  - The GCC/Clang internal header `<avx512fintrin.h>` is detected. While this
+ *    is not allowed to be included directly, it still appears in the builtin
+ *    include path and is detectable with `__has_include`.
+ *
+ * @see XXH_AVX512
+ */
+#ifndef XXH_DISPATCH_AVX512
+#  if (defined(__GNUC__) \
+       && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 9))) /* GCC 4.9+ */ \
+   || (defined(_MSC_VER) && _MSC_VER >= 1910) /* VS 2017+ */ \
+   || XXH_HAS_INCLUDE(<avx512fintrin.h>) /* GCC/Clang internal header */
+#    define XXH_DISPATCH_AVX512 1   /* enable dispatch towards AVX512 */
+#  else
+#    define XXH_DISPATCH_AVX512 0
+#  endif
+#endif /* XXH_DISPATCH_AVX512 */
+
+/*!
+ * @def XXH_TARGET_SSE2
+ * @brief Allows a function to be compiled with SSE2 intrinsics.
+ *
+ * Uses `__attribute__((__target__("sse2")))` on GCC to allow SSE2 to be used
+ * even with `-mno-sse2`.
+ *
+ * @def XXH_TARGET_AVX2
+ * @brief Like @ref XXH_TARGET_SSE2, but for AVX2.
+ *
+ * @def XXH_TARGET_AVX512
+ * @brief Like @ref XXH_TARGET_SSE2, but for AVX512.
+ */
+#if defined(__GNUC__)
+#  include <emmintrin.h> /* SSE2 */
+#  if XXH_DISPATCH_AVX2 || XXH_DISPATCH_AVX512
+#    include <immintrin.h> /* AVX2, AVX512F */
+#  endif
+#  define XXH_TARGET_SSE2 __attribute__((__target__("sse2")))
+#  define XXH_TARGET_AVX2 __attribute__((__target__("avx2")))
+#  define XXH_TARGET_AVX512 __attribute__((__target__("avx512f")))
+#elif defined(_MSC_VER)
+#  include <intrin.h>
+#  define XXH_TARGET_SSE2
+#  define XXH_TARGET_AVX2
+#  define XXH_TARGET_AVX512
+#else
+#  error "Dispatching is currently not supported for your compiler."
+#endif
+
+#ifdef XXH_DISPATCH_DEBUG
+/* debug logging */
+#  include <stdio.h>
+#  define XXH_debugPrint(str) { fprintf(stderr, "DEBUG: xxHash dispatch: %s \n", str); fflush(NULL); }
+#else
+#  define XXH_debugPrint(str) ((void)0)
+#  undef NDEBUG /* avoid redefinition */
+#  define NDEBUG
+#endif
+#include <assert.h>
+
+#define XXH_INLINE_ALL
+#define XXH_X86DISPATCH
+#include "xxhash.h"
+
+/*
+ * Support both AT&T and Intel dialects
+ *
+ * GCC doesn't convert AT&T syntax to Intel syntax, and will error out if
+ * compiled with -masm=intel. Instead, it supports dialect switching with
+ * curly braces: { AT&T syntax | Intel syntax }
+ *
+ * Clang's integrated assembler automatically converts AT&T syntax to Intel if
+ * needed, making the dialect switching useless (it isn't even supported).
+ *
+ * Note: Comments are written in the inline assembly itself.
+ */
+#ifdef __clang__
+#  define XXH_I_ATT(intel, att) att "\n\t"
+#else
+#  define XXH_I_ATT(intel, att) "{" att "|" intel "}\n\t"
+#endif
+
+/*!
+ * @internal
+ * @brief Runs CPUID.
+ *
+ * @param eax , ecx The parameters to pass to CPUID, %eax and %ecx respectively.
+ * @param abcd The array to store the result in, `{ eax, ebx, ecx, edx }`
+ */
+static void XXH_cpuid(xxh_u32 eax, xxh_u32 ecx, xxh_u32* abcd)
+{
+#if defined(_MSC_VER)
+    __cpuidex(abcd, eax, ecx);
+#else
+    xxh_u32 ebx, edx;
+# if defined(__i386__) && defined(__PIC__)
+    __asm__(
+        "# Call CPUID\n\t"
+        "#\n\t"
+        "# On 32-bit x86 with PIC enabled, we are not allowed to overwrite\n\t"
+        "# EBX, so we use EDI instead.\n\t"
+        XXH_I_ATT("mov     edi, ebx",   "movl    %%ebx, %%edi")
+        XXH_I_ATT("cpuid",              "cpuid"               )
+        XXH_I_ATT("xchg    edi, ebx",   "xchgl   %%ebx, %%edi")
+        : "=D" (ebx),
+# else
+    __asm__(
+        "# Call CPUID\n\t"
+        XXH_I_ATT("cpuid",              "cpuid")
+        : "=b" (ebx),
+# endif
+              "+a" (eax), "+c" (ecx), "=d" (edx));
+    abcd[0] = eax;
+    abcd[1] = ebx;
+    abcd[2] = ecx;
+    abcd[3] = edx;
+#endif
+}
+
+/*
+ * Modified version of Intel's guide
+ * https://software.intel.com/en-us/articles/how-to-detect-new-instruction-support-in-the-4th-generation-intel-core-processor-family
+ */
+
+#if XXH_DISPATCH_AVX2 || XXH_DISPATCH_AVX512
+/*!
+ * @internal
+ * @brief Runs `XGETBV`.
+ *
+ * While the CPU may support AVX2, the operating system might not properly save
+ * the full YMM/ZMM registers.
+ *
+ * xgetbv is used for detecting this: Any compliant operating system will define
+ * a set of flags in the xcr0 register indicating how it saves the AVX registers.
+ *
+ * You can manually disable this flag on Windows by running, as admin:
+ *
+ *   bcdedit.exe /set xsavedisable 1
+ *
+ * and rebooting. Run the same command with 0 to re-enable it.
+ */
+static xxh_u64 XXH_xgetbv(void)
+{
+#if defined(_MSC_VER)
+    return _xgetbv(0);  /* min VS2010 SP1 compiler is required */
+#else
+    xxh_u32 xcr0_lo, xcr0_hi;
+    __asm__(
+        "# Call XGETBV\n\t"
+        "#\n\t"
+        "# Older assemblers (e.g. macOS's ancient GAS version) don't support\n\t"
+        "# the XGETBV opcode, so we encode it by hand instead.\n\t"
+        "# See <https://github.com/asmjit/asmjit/issues/78> for details.\n\t"
+        ".byte   0x0f, 0x01, 0xd0\n\t"
+       : "=a" (xcr0_lo), "=d" (xcr0_hi) : "c" (0));
+    return xcr0_lo | ((xxh_u64)xcr0_hi << 32);
+#endif
+}
+#endif
+
+#define XXH_SSE2_CPUID_MASK (1 << 26)
+#define XXH_OSXSAVE_CPUID_MASK ((1 << 26) | (1 << 27))
+#define XXH_AVX2_CPUID_MASK (1 << 5)
+#define XXH_AVX2_XGETBV_MASK ((1 << 2) | (1 << 1))
+#define XXH_AVX512F_CPUID_MASK (1 << 16)
+#define XXH_AVX512F_XGETBV_MASK ((7 << 5) | (1 << 2) | (1 << 1))
+
+/*!
+ * @internal
+ * @brief Returns the best XXH3 implementation.
+ *
+ * Runs various CPUID/XGETBV tests to try and determine the best implementation.
+ *
+ * @return The best @ref XXH_VECTOR implementation.
+ * @see XXH_VECTOR_TYPES
+ */
+static int XXH_featureTest(void)
+{
+    xxh_u32 abcd[4];
+    xxh_u32 max_leaves;
+    int best = XXH_SCALAR;
+#if XXH_DISPATCH_AVX2 || XXH_DISPATCH_AVX512
+    xxh_u64 xgetbv_val;
+#endif
+#if defined(__GNUC__) && defined(__i386__)
+    xxh_u32 cpuid_supported;
+    __asm__(
+        "# For the sake of ruthless backwards compatibility, check if CPUID\n\t"
+        "# is supported in the EFLAGS on i386.\n\t"
+        "# This is not necessary on x86_64 - CPUID is mandatory.\n\t"
+        "#   The ID flag (bit 21) in the EFLAGS register indicates support\n\t"
+        "#   for the CPUID instruction. If a software procedure can set and\n\t"
+        "#   clear this flag, the processor executing the procedure supports\n\t"
+        "#   the CPUID instruction.\n\t"
+        "#   <https://c9x.me/x86/html/file_module_x86_id_45.html>\n\t"
+        "#\n\t"
+        "# Routine is from <https://wiki.osdev.org/CPUID>.\n\t"
+
+        "# Save EFLAGS\n\t"
+        XXH_I_ATT("pushfd",                           "pushfl"                    )
+        "# Store EFLAGS\n\t"
+        XXH_I_ATT("pushfd",                           "pushfl"                    )
+        "# Invert the ID bit in stored EFLAGS\n\t"
+        XXH_I_ATT("xor     dword ptr[esp], 0x200000", "xorl    $0x200000, (%%esp)")
+        "# Load stored EFLAGS (with ID bit inverted)\n\t"
+        XXH_I_ATT("popfd",                            "popfl"                     )
+        "# Store EFLAGS again (ID bit may or not be inverted)\n\t"
+        XXH_I_ATT("pushfd",                           "pushfl"                    )
+        "# eax = modified EFLAGS (ID bit may or may not be inverted)\n\t"
+        XXH_I_ATT("pop     eax",                      "popl    %%eax"             )
+        "# eax = whichever bits were changed\n\t"
+        XXH_I_ATT("xor     eax, dword ptr[esp]",      "xorl    (%%esp), %%eax"    )
+        "# Restore original EFLAGS\n\t"
+        XXH_I_ATT("popfd",                            "popfl"                     )
+        "# eax = zero if ID bit can't be changed, else non-zero\n\t"
+        XXH_I_ATT("and     eax, 0x200000",            "andl    $0x200000, %%eax"  )
+        : "=a" (cpuid_supported) :: "cc");
+
+    if (XXH_unlikely(!cpuid_supported)) {
+        XXH_debugPrint("CPUID support is not detected!");
+        return best;
+    }
+
+#endif
+    /* Check how many CPUID pages we have */
+    XXH_cpuid(0, 0, abcd);
+    max_leaves = abcd[0];
+
+    /* Shouldn't happen on hardware, but happens on some QEMU configs. */
+    if (XXH_unlikely(max_leaves == 0)) {
+        XXH_debugPrint("Max CPUID leaves == 0!");
+        return best;
+    }
+
+    /* Check for SSE2, OSXSAVE and xgetbv */
+    XXH_cpuid(1, 0, abcd);
+
+    /*
+     * Test for SSE2. The check is redundant on x86_64, but it doesn't hurt.
+     */
+    if (XXH_unlikely((abcd[3] & XXH_SSE2_CPUID_MASK) != XXH_SSE2_CPUID_MASK))
+        return best;
+
+    XXH_debugPrint("SSE2 support detected.");
+
+    best = XXH_SSE2;
+#if XXH_DISPATCH_AVX2 || XXH_DISPATCH_AVX512
+    /* Make sure we have enough leaves */
+    if (XXH_unlikely(max_leaves < 7))
+        return best;
+
+    /* Test for OSXSAVE and XGETBV */
+    if ((abcd[2] & XXH_OSXSAVE_CPUID_MASK) != XXH_OSXSAVE_CPUID_MASK)
+        return best;
+
+    /* CPUID check for AVX features */
+    XXH_cpuid(7, 0, abcd);
+
+    xgetbv_val = XXH_xgetbv();
+#if XXH_DISPATCH_AVX2
+    /* Validate that AVX2 is supported by the CPU */
+    if ((abcd[1] & XXH_AVX2_CPUID_MASK) != XXH_AVX2_CPUID_MASK)
+        return best;
+
+    /* Validate that the OS supports YMM registers */
+    if ((xgetbv_val & XXH_AVX2_XGETBV_MASK) != XXH_AVX2_XGETBV_MASK) {
+        XXH_debugPrint("AVX2 supported by the CPU, but not the OS.");
+        return best;
+    }
+
+    /* AVX2 supported */
+    XXH_debugPrint("AVX2 support detected.");
+    best = XXH_AVX2;
+#endif
+#if XXH_DISPATCH_AVX512
+    /* Check if AVX512F is supported by the CPU */
+    if ((abcd[1] & XXH_AVX512F_CPUID_MASK) != XXH_AVX512F_CPUID_MASK) {
+        XXH_debugPrint("AVX512F not supported by CPU");
+        return best;
+    }
+
+    /* Validate that the OS supports ZMM registers */
+    if ((xgetbv_val & XXH_AVX512F_XGETBV_MASK) != XXH_AVX512F_XGETBV_MASK) {
+        XXH_debugPrint("AVX512F supported by the CPU, but not the OS.");
+        return best;
+    }
+
+    /* AVX512F supported */
+    XXH_debugPrint("AVX512F support detected.");
+    best = XXH_AVX512;
+#endif
+#endif
+    return best;
+}
+
+
+/* ===   Vector implementations   === */
+
+/*!
+ * @internal
+ * @brief Defines the various dispatch functions.
+ *
+ * TODO: Consolidate?
+ *
+ * @param suffix The suffix for the functions, e.g. sse2 or scalar
+ * @param target XXH_TARGET_* or empty.
+ */
+#define XXH_DEFINE_DISPATCH_FUNCS(suffix, target)                             \
+                                                                              \
+/* ===   XXH3, default variants   === */                                      \
+                                                                              \
+XXH_NO_INLINE target XXH64_hash_t                                             \
+XXHL64_default_##suffix(const void* XXH_RESTRICT input, size_t len)           \
+{                                                                             \
+    return XXH3_hashLong_64b_internal(                                        \
+               input, len, XXH3_kSecret, sizeof(XXH3_kSecret),                \
+               XXH3_accumulate_512_##suffix, XXH3_scrambleAcc_##suffix        \
+    );                                                                        \
+}                                                                             \
+                                                                              \
+/* ===   XXH3, Seeded variants   === */                                       \
+                                                                              \
+XXH_NO_INLINE target XXH64_hash_t                                             \
+XXHL64_seed_##suffix(const void* XXH_RESTRICT input, size_t len,              \
+                     XXH64_hash_t seed)                                       \
+{                                                                             \
+    return XXH3_hashLong_64b_withSeed_internal(                               \
+                    input, len, seed, XXH3_accumulate_512_##suffix,           \
+                    XXH3_scrambleAcc_##suffix, XXH3_initCustomSecret_##suffix \
+    );                                                                        \
+}                                                                             \
+                                                                              \
+/* ===   XXH3, Secret variants   === */                                       \
+                                                                              \
+XXH_NO_INLINE target XXH64_hash_t                                             \
+XXHL64_secret_##suffix(const void* XXH_RESTRICT input, size_t len,            \
+                       const void* secret, size_t secretLen)                  \
+{                                                                             \
+    return XXH3_hashLong_64b_internal(                                        \
+                    input, len, secret, secretLen,                            \
+                    XXH3_accumulate_512_##suffix, XXH3_scrambleAcc_##suffix   \
+    );                                                                        \
+}                                                                             \
+                                                                              \
+/* ===   XXH3 update variants   === */                                        \
+                                                                              \
+XXH_NO_INLINE target XXH_errorcode                                            \
+XXH3_update_##suffix(XXH3_state_t* state, const void* input, size_t len)      \
+{                                                                             \
+    return XXH3_update(state, (const xxh_u8*)input, len,                      \
+                    XXH3_accumulate_512_##suffix, XXH3_scrambleAcc_##suffix); \
+}                                                                             \
+                                                                              \
+/* ===   XXH128 default variants   === */                                     \
+                                                                              \
+XXH_NO_INLINE target XXH128_hash_t                                            \
+XXHL128_default_##suffix(const void* XXH_RESTRICT input, size_t len)          \
+{                                                                             \
+    return XXH3_hashLong_128b_internal(                                       \
+                    input, len, XXH3_kSecret, sizeof(XXH3_kSecret),           \
+                    XXH3_accumulate_512_##suffix, XXH3_scrambleAcc_##suffix   \
+    );                                                                        \
+}                                                                             \
+                                                                              \
+/* ===   XXH128 Secret variants   === */                                      \
+                                                                              \
+XXH_NO_INLINE target XXH128_hash_t                                            \
+XXHL128_secret_##suffix(const void* XXH_RESTRICT input, size_t len,           \
+                        const void* XXH_RESTRICT secret, size_t secretLen)    \
+{                                                                             \
+    return XXH3_hashLong_128b_internal(                                       \
+                    input, len, (const xxh_u8*)secret, secretLen,             \
+                    XXH3_accumulate_512_##suffix, XXH3_scrambleAcc_##suffix); \
+}                                                                             \
+                                                                              \
+/* ===   XXH128 Seeded variants   === */                                      \
+                                                                              \
+XXH_NO_INLINE target XXH128_hash_t                                            \
+XXHL128_seed_##suffix(const void* XXH_RESTRICT input, size_t len,             \
+                      XXH64_hash_t seed)                                      \
+{                                                                             \
+    return XXH3_hashLong_128b_withSeed_internal(input, len, seed,             \
+                    XXH3_accumulate_512_##suffix, XXH3_scrambleAcc_##suffix,  \
+                    XXH3_initCustomSecret_##suffix);                          \
+}
+
+/* End XXH_DEFINE_DISPATCH_FUNCS */
+
+#if XXH_DISPATCH_SCALAR
+XXH_DEFINE_DISPATCH_FUNCS(scalar, /* nothing */)
+#endif
+XXH_DEFINE_DISPATCH_FUNCS(sse2, XXH_TARGET_SSE2)
+#if XXH_DISPATCH_AVX2
+XXH_DEFINE_DISPATCH_FUNCS(avx2, XXH_TARGET_AVX2)
+#endif
+#if XXH_DISPATCH_AVX512
+XXH_DEFINE_DISPATCH_FUNCS(avx512, XXH_TARGET_AVX512)
+#endif
+#undef XXH_DEFINE_DISPATCH_FUNCS
+
+/* ====    Dispatchers    ==== */
+
+typedef XXH64_hash_t (*XXH3_dispatchx86_hashLong64_default)(const void* XXH_RESTRICT, size_t);
+
+typedef XXH64_hash_t (*XXH3_dispatchx86_hashLong64_withSeed)(const void* XXH_RESTRICT, size_t, XXH64_hash_t);
+
+typedef XXH64_hash_t (*XXH3_dispatchx86_hashLong64_withSecret)(const void* XXH_RESTRICT, size_t, const void* XXH_RESTRICT, size_t);
+
+typedef XXH_errorcode (*XXH3_dispatchx86_update)(XXH3_state_t*, const void*, size_t);
+
+typedef struct {
+    XXH3_dispatchx86_hashLong64_default    hashLong64_default;
+    XXH3_dispatchx86_hashLong64_withSeed   hashLong64_seed;
+    XXH3_dispatchx86_hashLong64_withSecret hashLong64_secret;
+    XXH3_dispatchx86_update                update;
+} XXH_dispatchFunctions_s;
+
+#define XXH_NB_DISPATCHES 4
+
+/*!
+ * @internal
+ * @brief Table of dispatchers for @ref XXH3_64bits().
+ *
+ * @pre The indices must match @ref XXH_VECTOR_TYPE.
+ */
+static const XXH_dispatchFunctions_s XXH_kDispatch[XXH_NB_DISPATCHES] = {
+#if XXH_DISPATCH_SCALAR
+    /* Scalar */ { XXHL64_default_scalar, XXHL64_seed_scalar, XXHL64_secret_scalar, XXH3_update_scalar },
+#else
+    /* Scalar */ { NULL, NULL, NULL, NULL },
+#endif
+    /* SSE2   */ { XXHL64_default_sse2,   XXHL64_seed_sse2,   XXHL64_secret_sse2,   XXH3_update_sse2 },
+#if XXH_DISPATCH_AVX2
+    /* AVX2   */ { XXHL64_default_avx2,   XXHL64_seed_avx2,   XXHL64_secret_avx2,   XXH3_update_avx2 },
+#else
+    /* AVX2   */ { NULL, NULL, NULL, NULL },
+#endif
+#if XXH_DISPATCH_AVX512
+    /* AVX512 */ { XXHL64_default_avx512, XXHL64_seed_avx512, XXHL64_secret_avx512, XXH3_update_avx512 }
+#else
+    /* AVX512 */ { NULL, NULL, NULL, NULL }
+#endif
+};
+/*!
+ * @internal
+ * @brief The selected dispatch table for @ref XXH3_64bits().
+ */
+static XXH_dispatchFunctions_s XXH_g_dispatch = { NULL, NULL, NULL, NULL };
+
+
+typedef XXH128_hash_t (*XXH3_dispatchx86_hashLong128_default)(const void* XXH_RESTRICT, size_t);
+
+typedef XXH128_hash_t (*XXH3_dispatchx86_hashLong128_withSeed)(const void* XXH_RESTRICT, size_t, XXH64_hash_t);
+
+typedef XXH128_hash_t (*XXH3_dispatchx86_hashLong128_withSecret)(const void* XXH_RESTRICT, size_t, const void* XXH_RESTRICT, size_t);
+
+typedef struct {
+    XXH3_dispatchx86_hashLong128_default    hashLong128_default;
+    XXH3_dispatchx86_hashLong128_withSeed   hashLong128_seed;
+    XXH3_dispatchx86_hashLong128_withSecret hashLong128_secret;
+    XXH3_dispatchx86_update                 update;
+} XXH_dispatch128Functions_s;
+
+
+/*!
+ * @internal
+ * @brief Table of dispatchers for @ref XXH3_128bits().
+ *
+ * @pre The indices must match @ref XXH_VECTOR_TYPE.
+ */
+static const XXH_dispatch128Functions_s XXH_kDispatch128[XXH_NB_DISPATCHES] = {
+#if XXH_DISPATCH_SCALAR
+    /* Scalar */ { XXHL128_default_scalar, XXHL128_seed_scalar, XXHL128_secret_scalar, XXH3_update_scalar },
+#else
+    /* Scalar */ { NULL, NULL, NULL, NULL },
+#endif
+    /* SSE2   */ { XXHL128_default_sse2,   XXHL128_seed_sse2,   XXHL128_secret_sse2,   XXH3_update_sse2 },
+#if XXH_DISPATCH_AVX2
+    /* AVX2   */ { XXHL128_default_avx2,   XXHL128_seed_avx2,   XXHL128_secret_avx2,   XXH3_update_avx2 },
+#else
+    /* AVX2   */ { NULL, NULL, NULL, NULL },
+#endif
+#if XXH_DISPATCH_AVX512
+    /* AVX512 */ { XXHL128_default_avx512, XXHL128_seed_avx512, XXHL128_secret_avx512, XXH3_update_avx512 }
+#else
+    /* AVX512 */ { NULL, NULL, NULL, NULL }
+#endif
+};
+
+/*!
+ * @internal
+ * @brief The selected dispatch table for @ref XXH3_64bits().
+ */
+static XXH_dispatch128Functions_s XXH_g_dispatch128 = { NULL, NULL, NULL, NULL };
+
+/*!
+ * @internal
+ * @brief Runs a CPUID check and sets the correct dispatch tables.
+ */
+static void XXH_setDispatch(void)
+{
+    int vecID = XXH_featureTest();
+    XXH_STATIC_ASSERT(XXH_AVX512 == XXH_NB_DISPATCHES-1);
+    assert(XXH_SCALAR <= vecID && vecID <= XXH_AVX512);
+#if !XXH_DISPATCH_SCALAR
+    assert(vecID != XXH_SCALAR);
+#endif
+#if !XXH_DISPATCH_AVX512
+    assert(vecID != XXH_AVX512);
+#endif
+#if !XXH_DISPATCH_AVX2
+    assert(vecID != XXH_AVX2);
+#endif
+    XXH_g_dispatch = XXH_kDispatch[vecID];
+    XXH_g_dispatch128 = XXH_kDispatch128[vecID];
+}
+
+
+/* ====    XXH3 public functions    ==== */
+
+static XXH64_hash_t
+XXH3_hashLong_64b_defaultSecret_selection(const void* input, size_t len,
+                                          XXH64_hash_t seed64, const xxh_u8* secret, size_t secretLen)
+{
+    (void)seed64; (void)secret; (void)secretLen;
+    if (XXH_g_dispatch.hashLong64_default == NULL) XXH_setDispatch();
+    return XXH_g_dispatch.hashLong64_default(input, len);
+}
+
+XXH64_hash_t XXH3_64bits_dispatch(const void* input, size_t len)
+{
+    return XXH3_64bits_internal(input, len, 0, XXH3_kSecret, sizeof(XXH3_kSecret), XXH3_hashLong_64b_defaultSecret_selection);
+}
+
+static XXH64_hash_t
+XXH3_hashLong_64b_withSeed_selection(const void* input, size_t len,
+                                     XXH64_hash_t seed64, const xxh_u8* secret, size_t secretLen)
+{
+    (void)secret; (void)secretLen;
+    if (XXH_g_dispatch.hashLong64_seed == NULL) XXH_setDispatch();
+    return XXH_g_dispatch.hashLong64_seed(input, len, seed64);
+}
+
+XXH64_hash_t XXH3_64bits_withSeed_dispatch(const void* input, size_t len, XXH64_hash_t seed)
+{
+    return XXH3_64bits_internal(input, len, seed, XXH3_kSecret, sizeof(XXH3_kSecret), XXH3_hashLong_64b_withSeed_selection);
+}
+
+static XXH64_hash_t
+XXH3_hashLong_64b_withSecret_selection(const void* input, size_t len,
+                                       XXH64_hash_t seed64, const xxh_u8* secret, size_t secretLen)
+{
+    (void)seed64;
+    if (XXH_g_dispatch.hashLong64_secret == NULL) XXH_setDispatch();
+    return XXH_g_dispatch.hashLong64_secret(input, len, secret, secretLen);
+}
+
+XXH64_hash_t XXH3_64bits_withSecret_dispatch(const void* input, size_t len, const void* secret, size_t secretLen)
+{
+    return XXH3_64bits_internal(input, len, 0, secret, secretLen, XXH3_hashLong_64b_withSecret_selection);
+}
+
+XXH_errorcode
+XXH3_64bits_update_dispatch(XXH3_state_t* state, const void* input, size_t len)
+{
+    if (XXH_g_dispatch.update == NULL) XXH_setDispatch();
+    return XXH_g_dispatch.update(state, (const xxh_u8*)input, len);
+}
+
+
+/* ====    XXH128 public functions    ==== */
+
+static XXH128_hash_t
+XXH3_hashLong_128b_defaultSecret_selection(const void* input, size_t len,
+                                           XXH64_hash_t seed64, const void* secret, size_t secretLen)
+{
+    (void)seed64; (void)secret; (void)secretLen;
+    if (XXH_g_dispatch128.hashLong128_default == NULL) XXH_setDispatch();
+    return XXH_g_dispatch128.hashLong128_default(input, len);
+}
+
+XXH128_hash_t XXH3_128bits_dispatch(const void* input, size_t len)
+{
+    return XXH3_128bits_internal(input, len, 0, XXH3_kSecret, sizeof(XXH3_kSecret), XXH3_hashLong_128b_defaultSecret_selection);
+}
+
+static XXH128_hash_t
+XXH3_hashLong_128b_withSeed_selection(const void* input, size_t len,
+                                     XXH64_hash_t seed64, const void* secret, size_t secretLen)
+{
+    (void)secret; (void)secretLen;
+    if (XXH_g_dispatch128.hashLong128_seed == NULL) XXH_setDispatch();
+    return XXH_g_dispatch128.hashLong128_seed(input, len, seed64);
+}
+
+XXH128_hash_t XXH3_128bits_withSeed_dispatch(const void* input, size_t len, XXH64_hash_t seed)
+{
+    return XXH3_128bits_internal(input, len, seed, XXH3_kSecret, sizeof(XXH3_kSecret), XXH3_hashLong_128b_withSeed_selection);
+}
+
+static XXH128_hash_t
+XXH3_hashLong_128b_withSecret_selection(const void* input, size_t len,
+                                        XXH64_hash_t seed64, const void* secret, size_t secretLen)
+{
+    (void)seed64;
+    if (XXH_g_dispatch128.hashLong128_secret == NULL) XXH_setDispatch();
+    return XXH_g_dispatch128.hashLong128_secret(input, len, secret, secretLen);
+}
+
+XXH128_hash_t XXH3_128bits_withSecret_dispatch(const void* input, size_t len, const void* secret, size_t secretLen)
+{
+    return XXH3_128bits_internal(input, len, 0, secret, secretLen, XXH3_hashLong_128b_withSecret_selection);
+}
+
+XXH_errorcode
+XXH3_128bits_update_dispatch(XXH3_state_t* state, const void* input, size_t len)
+{
+    if (XXH_g_dispatch128.update == NULL) XXH_setDispatch();
+    return XXH_g_dispatch128.update(state, (const xxh_u8*)input, len);
+}
+
+#if defined (__cplusplus)
+}
+#endif
+/*! @} */

extern/xxhash/xxh_x86dispatch.h

@@ -0,0 +1,85 @@
+/*
+ * xxHash - XXH3 Dispatcher for x86-based targets
+ * Copyright (C) 2020-2021 Yann Collet
+ *
+ * BSD 2-Clause License (https://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *    * Redistributions of source code must retain the above copyright
+ *      notice, this list of conditions and the following disclaimer.
+ *    * Redistributions in binary form must reproduce the above
+ *      copyright notice, this list of conditions and the following disclaimer
+ *      in the documentation and/or other materials provided with the
+ *      distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * You can contact the author at:
+ *   - xxHash homepage: https://www.xxhash.com
+ *   - xxHash source repository: https://github.com/Cyan4973/xxHash
+ */
+
+#ifndef XXH_X86DISPATCH_H_13563687684
+#define XXH_X86DISPATCH_H_13563687684
+
+#include "xxhash.h"  /* XXH64_hash_t, XXH3_state_t */
+
+#if defined (__cplusplus)
+extern "C" {
+#endif
+
+XXH_PUBLIC_API XXH64_hash_t  XXH3_64bits_dispatch(const void* input, size_t len);
+XXH_PUBLIC_API XXH64_hash_t  XXH3_64bits_withSeed_dispatch(const void* input, size_t len, XXH64_hash_t seed);
+XXH_PUBLIC_API XXH64_hash_t  XXH3_64bits_withSecret_dispatch(const void* input, size_t len, const void* secret, size_t secretLen);
+XXH_PUBLIC_API XXH_errorcode XXH3_64bits_update_dispatch(XXH3_state_t* state, const void* input, size_t len);
+
+XXH_PUBLIC_API XXH128_hash_t XXH3_128bits_dispatch(const void* input, size_t len);
+XXH_PUBLIC_API XXH128_hash_t XXH3_128bits_withSeed_dispatch(const void* input, size_t len, XXH64_hash_t seed);
+XXH_PUBLIC_API XXH128_hash_t XXH3_128bits_withSecret_dispatch(const void* input, size_t len, const void* secret, size_t secretLen);
+XXH_PUBLIC_API XXH_errorcode XXH3_128bits_update_dispatch(XXH3_state_t* state, const void* input, size_t len);
+
+#if defined (__cplusplus)
+}
+#endif
+
+
+/* automatic replacement of XXH3 functions.
+ * can be disabled by setting XXH_DISPATCH_DISABLE_REPLACE */
+#ifndef XXH_DISPATCH_DISABLE_REPLACE
+
+# undef  XXH3_64bits
+# define XXH3_64bits XXH3_64bits_dispatch
+# undef  XXH3_64bits_withSeed
+# define XXH3_64bits_withSeed XXH3_64bits_withSeed_dispatch
+# undef  XXH3_64bits_withSecret
+# define XXH3_64bits_withSecret XXH3_64bits_withSecret_dispatch
+# undef  XXH3_64bits_update
+# define XXH3_64bits_update XXH3_64bits_update_dispatch
+
+# undef  XXH128
+# define XXH128 XXH3_128bits_withSeed_dispatch
+# undef  XXH3_128bits
+# define XXH3_128bits XXH3_128bits_dispatch
+# undef  XXH3_128bits_withSeed
+# define XXH3_128bits_withSeed XXH3_128bits_withSeed_dispatch
+# undef  XXH3_128bits_withSecret
+# define XXH3_128bits_withSecret XXH3_128bits_withSecret_dispatch
+# undef  XXH3_128bits_update
+# define XXH3_128bits_update XXH3_128bits_update_dispatch
+
+#endif /* XXH_DISPATCH_DISABLE_REPLACE */
+
+#endif /* XXH_X86DISPATCH_H_13563687684 */

extern/xxhash/xxhash.c

@@ -0,0 +1,43 @@
+/*
+ * xxHash - Extremely Fast Hash algorithm
+ * Copyright (C) 2012-2021 Yann Collet
+ *
+ * BSD 2-Clause License (https://www.opensource.org/licenses/bsd-license.php)
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are
+ * met:
+ *
+ *    * Redistributions of source code must retain the above copyright
+ *      notice, this list of conditions and the following disclaimer.
+ *    * Redistributions in binary form must reproduce the above
+ *      copyright notice, this list of conditions and the following disclaimer
+ *      in the documentation and/or other materials provided with the
+ *      distribution.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
+ * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
+ * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
+ * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
+ * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+ * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
+ * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
+ * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
+ * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+ * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ * You can contact the author at:
+ *   - xxHash homepage: https://www.xxhash.com
+ *   - xxHash source repository: https://github.com/Cyan4973/xxHash
+ */
+
+
+/*
+ * xxhash.c instantiates functions defined in xxhash.h
+ */
+
+#define XXH_STATIC_LINKING_ONLY   /* access advanced declarations */
+#define XXH_IMPLEMENTATION   /* access definitions */
+
+#include "xxhash.h"

hecl/include/hecl/Backend.hpp

@@ -5,9 +5,6 @@
 #include <string_view>
 #include <vector>
 
-//#include <boo/graphicsdev/IGraphicsDataFactory.hpp>
-#include <xxhash/xxhash.h>
-
 #include "hecl.hpp"
 
 namespace hecl::Backend {

hecl/include/hecl/hecl.hpp

@@ -44,7 +44,7 @@ extern "C" int rep_closefrom(int lower);
 #include "FourCC.hpp"
 #include "athena/Global.hpp"
 #include "logvisor/logvisor.hpp"
-#include "xxhash/xxhash.h"
+#include "xxhash.h"
 
 
 #if defined(__has_feature)

hecl/lib/CMakeLists.txt

@@ -68,7 +68,7 @@ add_library(hecl-full
         ${FIND_BLENDER_SOURCES})
 target_include_directories(hecl-full PUBLIC ../include)
 target_link_libraries(hecl-full PUBLIC ${HECL_APPLICATION_REPS_TARGETS_LIST}
-        hecl-blender-addon boo athena-core logvisor)
+        hecl-blender-addon boo athena-core logvisor xxhash)
 target_atdna(hecl-full atdna_HMDLMeta_full.cpp ../include/hecl/HMDLMeta.hpp)
 target_atdna(hecl-full atdna_CVar_full.cpp ../include/hecl/CVar.hpp)
 target_atdna(hecl-full atdna_SDNARead_full.cpp ../include/hecl/Blender/SDNARead.hpp)
@@ -80,7 +80,7 @@ add_library(hecl-light
         ${PLAT_SRCS}
         ${FIND_BLENDER_SOURCES})
 target_include_directories(hecl-light PUBLIC ../include)
-target_link_libraries(hecl-light PUBLIC ${HECL_APPLICATION_REPS_TARGETS_LIST} boo athena-core logvisor aurora)
+target_link_libraries(hecl-light PUBLIC ${HECL_APPLICATION_REPS_TARGETS_LIST} boo athena-core logvisor aurora xxhash)
 if (WIN32)
   # For FindBlender
   target_link_libraries(hecl-light PUBLIC Version)

visigen/VISIBuilder.hpp

@@ -2,7 +2,7 @@
 
 #include "VISIRenderer.hpp"
 #include "zeus/CAABox.hpp"
-#include "xxhash/xxhash.h"
+#include "xxhash.h"
 #include "athena/MemoryWriter.hpp"
 #include <unordered_map>