video_core: Reimplement the buffer cache

Reimplement the buffer cache using cached bindings and page level
granularity for modification tracking. This also drops the usage of
shared pointers and virtual functions from the cache.

- Bindings are cached, allowing to skip work when the game changes few
  bits between draws.
- OpenGL Assembly shaders no longer copy when a region has been modified
  from the GPU to emulate constant buffers, instead GL_EXT_memory_object
  is used to alias sub-buffers within the same allocation.
- OpenGL Assembly shaders stream constant buffer data using
  glProgramBufferParametersIuivNV, from NV_parameter_buffer_object. In
  theory this should save one hash table resolve inside the driver
  compared to glBufferSubData.
- A new OpenGL stream buffer is implemented based on fences for drivers
  that are not Nvidia's proprietary, due to their low performance on
  partial glBufferSubData calls synchronized with 3D rendering (that
  some games use a lot).
- Most optimizations are shared between APIs now, allowing Vulkan to
  cache more bindings than before, skipping unnecesarry work.

This commit adds the necessary infrastructure to use Vulkan object from
OpenGL. Overall, it improves performance and fixes some bugs present on
the old cache. There are still some edge cases hit by some games that
harm performance on some vendors, this are planned to be fixed in later
commits.

1 files changed, 122 insertions, 452 deletions

diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index ea4ca9a82..52499ee4c 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -44,17 +44,10 @@ using VideoCore::Surface::PixelFormat;
 using VideoCore::Surface::SurfaceTarget;
 using VideoCore::Surface::SurfaceType;
 
-MICROPROFILE_DEFINE(OpenGL_VAO, "OpenGL", "Vertex Format Setup", MP_RGB(128, 128, 192));
-MICROPROFILE_DEFINE(OpenGL_VB, "OpenGL", "Vertex Buffer Setup", MP_RGB(128, 128, 192));
-MICROPROFILE_DEFINE(OpenGL_Shader, "OpenGL", "Shader Setup", MP_RGB(128, 128, 192));
-MICROPROFILE_DEFINE(OpenGL_UBO, "OpenGL", "Const Buffer Setup", MP_RGB(128, 128, 192));
-MICROPROFILE_DEFINE(OpenGL_Index, "OpenGL", "Index Buffer Setup", MP_RGB(128, 128, 192));
-MICROPROFILE_DEFINE(OpenGL_Texture, "OpenGL", "Texture Setup", MP_RGB(128, 128, 192));
-MICROPROFILE_DEFINE(OpenGL_Framebuffer, "OpenGL", "Framebuffer Setup", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_Drawing, "OpenGL", "Drawing", MP_RGB(128, 128, 192));
+MICROPROFILE_DEFINE(OpenGL_Clears, "OpenGL", "Clears", MP_RGB(128, 128, 192));
 MICROPROFILE_DEFINE(OpenGL_Blits, "OpenGL", "Blits", MP_RGB(128, 128, 192));
-MICROPROFILE_DEFINE(OpenGL_CacheManagement, "OpenGL", "Cache Mgmt", MP_RGB(100, 255, 100));
-MICROPROFILE_DEFINE(OpenGL_PrimitiveAssembly, "OpenGL", "Prim Asmbl", MP_RGB(255, 100, 100));
+MICROPROFILE_DEFINE(OpenGL_CacheManagement, "OpenGL", "Cache Management", MP_RGB(100, 255, 100));
 
 namespace {
 
@@ -101,20 +94,6 @@ TextureHandle GetTextureInfo(const Engine& engine, bool via_header_index, const
     return TextureHandle(engine.AccessConstBuffer32(shader_type, buffer, offset), via_header_index);
 }
 
-std::size_t GetConstBufferSize(const Tegra::Engines::ConstBufferInfo& buffer,
-                               const ConstBufferEntry& entry) {
-    if (!entry.IsIndirect()) {
-        return entry.GetSize();
-    }
-    if (buffer.size > Maxwell::MaxConstBufferSize) {
-        LOG_WARNING(Render_OpenGL, "Indirect constbuffer size {} exceeds maximum {}", buffer.size,
-                    Maxwell::MaxConstBufferSize);
-        return Maxwell::MaxConstBufferSize;
-    }
-
-    return buffer.size;
-}
-
 /// Translates hardware transform feedback indices
 /// @param location Hardware location
 /// @return Pair of ARB_transform_feedback3 token stream first and third arguments
@@ -147,14 +126,6 @@ void oglEnable(GLenum cap, bool state) {
     (state ? glEnable : glDisable)(cap);
 }
 
-void UpdateBindlessSSBOs(GLenum target, const BindlessSSBO* ssbos, size_t num_ssbos) {
-    if (num_ssbos == 0) {
-        return;
-    }
-    glProgramLocalParametersI4uivNV(target, 0, static_cast<GLsizei>(num_ssbos),
-                                    reinterpret_cast<const GLuint*>(ssbos));
-}
-
 ImageViewType ImageViewTypeFromEntry(const SamplerEntry& entry) {
     if (entry.is_buffer) {
         return ImageViewType::Buffer;
@@ -196,49 +167,35 @@ ImageViewType ImageViewTypeFromEntry(const ImageEntry& entry) {
 
 RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& emu_window_, Tegra::GPU& gpu_,
                                    Core::Memory::Memory& cpu_memory_, const Device& device_,
+                                   const Vulkan::Device* vulkan_device,
+                                   Vulkan::MemoryAllocator* vulkan_memory_allocator,
                                    ScreenInfo& screen_info_, ProgramManager& program_manager_,
                                    StateTracker& state_tracker_)
     : RasterizerAccelerated(cpu_memory_), gpu(gpu_), maxwell3d(gpu.Maxwell3D()),
       kepler_compute(gpu.KeplerCompute()), gpu_memory(gpu.MemoryManager()), device(device_),
       screen_info(screen_info_), program_manager(program_manager_), state_tracker(state_tracker_),
-      stream_buffer(device, state_tracker),
       texture_cache_runtime(device, program_manager, state_tracker),
       texture_cache(texture_cache_runtime, *this, maxwell3d, kepler_compute, gpu_memory),
+      buffer_cache_runtime(device, vulkan_device, vulkan_memory_allocator),
+      buffer_cache(*this, maxwell3d, kepler_compute, gpu_memory, cpu_memory_, buffer_cache_runtime),
       shader_cache(*this, emu_window_, gpu, maxwell3d, kepler_compute, gpu_memory, device),
       query_cache(*this, maxwell3d, gpu_memory),
-      buffer_cache(*this, gpu_memory, cpu_memory_, device, stream_buffer, state_tracker),
       fence_manager(*this, gpu, texture_cache, buffer_cache, query_cache),
       async_shaders(emu_window_) {
-    unified_uniform_buffer.Create();
-    glNamedBufferStorage(unified_uniform_buffer.handle, TOTAL_CONST_BUFFER_BYTES, nullptr, 0);
-
-    if (device.UseAssemblyShaders()) {
-        glCreateBuffers(static_cast<GLsizei>(staging_cbufs.size()), staging_cbufs.data());
-        for (const GLuint cbuf : staging_cbufs) {
-            glNamedBufferStorage(cbuf, static_cast<GLsizeiptr>(Maxwell::MaxConstBufferSize),
-                                 nullptr, 0);
-        }
-    }
     if (device.UseAsynchronousShaders()) {
         async_shaders.AllocateWorkers();
     }
 }
 
-RasterizerOpenGL::~RasterizerOpenGL() {
-    if (device.UseAssemblyShaders()) {
-        glDeleteBuffers(static_cast<GLsizei>(staging_cbufs.size()), staging_cbufs.data());
-    }
-}
+RasterizerOpenGL::~RasterizerOpenGL() = default;
 
-void RasterizerOpenGL::SetupVertexFormat() {
+void RasterizerOpenGL::SyncVertexFormats() {
     auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::VertexFormats]) {
         return;
     }
     flags[Dirty::VertexFormats] = false;
 
-    MICROPROFILE_SCOPE(OpenGL_VAO);
-
     // Use the vertex array as-is, assumes that the data is formatted correctly for OpenGL. Enables
     // the first 16 vertex attributes always, as we don't know which ones are actually used until
     // shader time. Note, Tegra technically supports 32, but we're capping this to 16 for now to
@@ -274,55 +231,7 @@ void RasterizerOpenGL::SetupVertexFormat() {
     }
 }
 
-void RasterizerOpenGL::SetupVertexBuffer() {
-    auto& flags = maxwell3d.dirty.flags;
-    if (!flags[Dirty::VertexBuffers]) {
-        return;
-    }
-    flags[Dirty::VertexBuffers] = false;
-
-    MICROPROFILE_SCOPE(OpenGL_VB);
-
-    const bool use_unified_memory = device.HasVertexBufferUnifiedMemory();
-
-    // Upload all guest vertex arrays sequentially to our buffer
-    const auto& regs = maxwell3d.regs;
-    for (std::size_t index = 0; index < NUM_SUPPORTED_VERTEX_BINDINGS; ++index) {
-        if (!flags[Dirty::VertexBuffer0 + index]) {
-            continue;
-        }
-        flags[Dirty::VertexBuffer0 + index] = false;
-
-        const auto& vertex_array = regs.vertex_array[index];
-        if (!vertex_array.IsEnabled()) {
-            continue;
-        }
-
-        const GPUVAddr start = vertex_array.StartAddress();
-        const GPUVAddr end = regs.vertex_array_limit[index].LimitAddress();
-        ASSERT(end >= start);
-
-        const GLuint gl_index = static_cast<GLuint>(index);
-        const u64 size = end - start;
-        if (size == 0) {
-            glBindVertexBuffer(gl_index, 0, 0, vertex_array.stride);
-            if (use_unified_memory) {
-                glBufferAddressRangeNV(GL_VERTEX_ATTRIB_ARRAY_ADDRESS_NV, gl_index, 0, 0);
-            }
-            continue;
-        }
-        const auto info = buffer_cache.UploadMemory(start, size);
-        if (use_unified_memory) {
-            glBindVertexBuffer(gl_index, 0, 0, vertex_array.stride);
-            glBufferAddressRangeNV(GL_VERTEX_ATTRIB_ARRAY_ADDRESS_NV, gl_index,
-                                   info.address + info.offset, size);
-        } else {
-            glBindVertexBuffer(gl_index, info.handle, info.offset, vertex_array.stride);
-        }
-    }
-}
-
-void RasterizerOpenGL::SetupVertexInstances() {
+void RasterizerOpenGL::SyncVertexInstances() {
     auto& flags = maxwell3d.dirty.flags;
     if (!flags[Dirty::VertexInstances]) {
         return;
@@ -343,17 +252,7 @@ void RasterizerOpenGL::SetupVertexInstances() {
     }
 }
 
-GLintptr RasterizerOpenGL::SetupIndexBuffer() {
-    MICROPROFILE_SCOPE(OpenGL_Index);
-    const auto& regs = maxwell3d.regs;
-    const std::size_t size = CalculateIndexBufferSize();
-    const auto info = buffer_cache.UploadMemory(regs.index_array.IndexStart(), size);
-    glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, info.handle);
-    return info.offset;
-}
-
-void RasterizerOpenGL::SetupShaders() {
-    MICROPROFILE_SCOPE(OpenGL_Shader);
+void RasterizerOpenGL::SetupShaders(bool is_indexed) {
     u32 clip_distances = 0;
 
     std::array<Shader*, Maxwell::MaxShaderStage> shaders{};
@@ -410,11 +309,19 @@ void RasterizerOpenGL::SetupShaders() {
         const size_t stage = index == 0 ? 0 : index - 1;
         shaders[stage] = shader;
 
-        SetupDrawConstBuffers(stage, shader);
-        SetupDrawGlobalMemory(stage, shader);
         SetupDrawTextures(shader, stage);
         SetupDrawImages(shader, stage);
 
+        buffer_cache.SetEnabledUniformBuffers(stage, shader->GetEntries().enabled_uniform_buffers);
+
+        buffer_cache.UnbindGraphicsStorageBuffers(stage);
+        u32 ssbo_index = 0;
+        for (const auto& buffer : shader->GetEntries().global_memory_entries) {
+            buffer_cache.BindGraphicsStorageBuffer(stage, ssbo_index, buffer.cbuf_index,
+                                                   buffer.cbuf_offset, buffer.is_written);
+            ++ssbo_index;
+        }
+
         // Workaround for Intel drivers.
         // When a clip distance is enabled but not set in the shader it crops parts of the screen
         // (sometimes it's half the screen, sometimes three quarters). To avoid this, enable the
@@ -430,43 +337,26 @@ void RasterizerOpenGL::SetupShaders() {
     SyncClipEnabled(clip_distances);
     maxwell3d.dirty.flags[Dirty::Shaders] = false;
 
+    buffer_cache.UpdateGraphicsBuffers(is_indexed);
+
     const std::span indices_span(image_view_indices.data(), image_view_indices.size());
     texture_cache.FillGraphicsImageViews(indices_span, image_view_ids);
 
+    buffer_cache.BindHostGeometryBuffers(is_indexed);
+
     size_t image_view_index = 0;
     size_t texture_index = 0;
     size_t image_index = 0;
     for (size_t stage = 0; stage < Maxwell::MaxShaderStage; ++stage) {
         const Shader* const shader = shaders[stage];
-        if (shader) {
-            const auto base = device.GetBaseBindings(stage);
-            BindTextures(shader->GetEntries(), base.sampler, base.image, image_view_index,
-                         texture_index, image_index);
-        }
-    }
-}
-
-std::size_t RasterizerOpenGL::CalculateVertexArraysSize() const {
-    const auto& regs = maxwell3d.regs;
-
-    std::size_t size = 0;
-    for (u32 index = 0; index < Maxwell::NumVertexArrays; ++index) {
-        if (!regs.vertex_array[index].IsEnabled())
+        if (!shader) {
             continue;
-
-        const GPUVAddr start = regs.vertex_array[index].StartAddress();
-        const GPUVAddr end = regs.vertex_array_limit[index].LimitAddress();
-
-        size += end - start;
-        ASSERT(end >= start);
+        }
+        buffer_cache.BindHostStageBuffers(stage);
+        const auto& base = device.GetBaseBindings(stage);
+        BindTextures(shader->GetEntries(), base.sampler, base.image, image_view_index,
+                     texture_index, image_index);
     }
-
-    return size;
-}
-
-std::size_t RasterizerOpenGL::CalculateIndexBufferSize() const {
-    return static_cast<std::size_t>(maxwell3d.regs.index_array.count) *
-           static_cast<std::size_t>(maxwell3d.regs.index_array.FormatSizeInBytes());
 }
 
 void RasterizerOpenGL::LoadDiskResources(u64 title_id, const std::atomic_bool& stop_loading,
@@ -475,6 +365,7 @@ void RasterizerOpenGL::LoadDiskResources(u64 title_id, const std::atomic_bool& s
 }
 
 void RasterizerOpenGL::Clear() {
+    MICROPROFILE_SCOPE(OpenGL_Clears);
     if (!maxwell3d.ShouldExecute()) {
         return;
     }
@@ -525,11 +416,9 @@ void RasterizerOpenGL::Clear() {
     }
     UNIMPLEMENTED_IF(regs.clear_flags.viewport);
 
-    {
-        auto lock = texture_cache.AcquireLock();
-        texture_cache.UpdateRenderTargets(true);
-        state_tracker.BindFramebuffer(texture_cache.GetFramebuffer()->Handle());
-    }
+    std::scoped_lock lock{texture_cache.mutex};
+    texture_cache.UpdateRenderTargets(true);
+    state_tracker.BindFramebuffer(texture_cache.GetFramebuffer()->Handle());
 
     if (use_color) {
         glClearBufferfv(GL_COLOR, regs.clear_buffers.RT, regs.clear_color);
@@ -541,7 +430,6 @@ void RasterizerOpenGL::Clear() {
     } else if (use_stencil) {
         glClearBufferiv(GL_STENCIL, 0, &regs.clear_stencil);
     }
-
     ++num_queued_commands;
 }
 
@@ -550,75 +438,12 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
 
     query_cache.UpdateCounters();
 
-    SyncViewport();
-    SyncRasterizeEnable();
-    SyncPolygonModes();
-    SyncColorMask();
-    SyncFragmentColorClampState();
-    SyncMultiSampleState();
-    SyncDepthTestState();
-    SyncDepthClamp();
-    SyncStencilTestState();
-    SyncBlendState();
-    SyncLogicOpState();
-    SyncCullMode();
-    SyncPrimitiveRestart();
-    SyncScissorTest();
-    SyncPointState();
-    SyncLineState();
-    SyncPolygonOffset();
-    SyncAlphaTest();
-    SyncFramebufferSRGB();
-
-    buffer_cache.Acquire();
-    current_cbuf = 0;
-
-    std::size_t buffer_size = CalculateVertexArraysSize();
-
-    // Add space for index buffer
-    if (is_indexed) {
-        buffer_size = Common::AlignUp(buffer_size, 4) + CalculateIndexBufferSize();
-    }
-
-    // Uniform space for the 5 shader stages
-    buffer_size =
-        Common::AlignUp<std::size_t>(buffer_size, 4) +
-        (sizeof(MaxwellUniformData) + device.GetUniformBufferAlignment()) * Maxwell::MaxShaderStage;
-
-    // Add space for at least 18 constant buffers
-    buffer_size += Maxwell::MaxConstBuffers *
-                   (Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment());
-
-    // Prepare the vertex array.
-    buffer_cache.Map(buffer_size);
-
-    // Prepare vertex array format.
-    SetupVertexFormat();
-
-    // Upload vertex and index data.
-    SetupVertexBuffer();
-    SetupVertexInstances();
-    GLintptr index_buffer_offset = 0;
-    if (is_indexed) {
-        index_buffer_offset = SetupIndexBuffer();
-    }
-
-    // Setup emulation uniform buffer.
-    if (!device.UseAssemblyShaders()) {
-        MaxwellUniformData ubo;
-        ubo.SetFromRegs(maxwell3d);
-        const auto info =
-            buffer_cache.UploadHostMemory(&ubo, sizeof(ubo), device.GetUniformBufferAlignment());
-        glBindBufferRange(GL_UNIFORM_BUFFER, EmulationUniformBlockBinding, info.handle, info.offset,
-                          static_cast<GLsizeiptr>(sizeof(ubo)));
-    }
+    SyncState();
 
     // Setup shaders and their used resources.
-    auto lock = texture_cache.AcquireLock();
-    SetupShaders();
+    std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex};
+    SetupShaders(is_indexed);
 
-    // Signal the buffer cache that we are not going to upload more things.
-    buffer_cache.Unmap();
     texture_cache.UpdateRenderTargets(false);
     state_tracker.BindFramebuffer(texture_cache.GetFramebuffer()->Handle());
     program_manager.BindGraphicsPipeline();
@@ -632,7 +457,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
     if (is_indexed) {
         const GLint base_vertex = static_cast<GLint>(maxwell3d.regs.vb_element_base);
         const GLsizei num_vertices = static_cast<GLsizei>(maxwell3d.regs.index_array.count);
-        const GLvoid* offset = reinterpret_cast<const GLvoid*>(index_buffer_offset);
+        const GLvoid* const offset = buffer_cache_runtime.IndexOffset();
         const GLenum format = MaxwellToGL::IndexFormat(maxwell3d.regs.index_array.format);
         if (num_instances == 1 && base_instance == 0 && base_vertex == 0) {
             glDrawElements(primitive_mode, num_vertices, format, offset);
@@ -672,22 +497,22 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
 }
 
 void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
-    buffer_cache.Acquire();
-    current_cbuf = 0;
-
     Shader* const kernel = shader_cache.GetComputeKernel(code_addr);
 
-    auto lock = texture_cache.AcquireLock();
+    std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex};
     BindComputeTextures(kernel);
 
-    const size_t buffer_size = Tegra::Engines::KeplerCompute::NumConstBuffers *
-                               (Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment());
-    buffer_cache.Map(buffer_size);
-
-    SetupComputeConstBuffers(kernel);
-    SetupComputeGlobalMemory(kernel);
-
-    buffer_cache.Unmap();
+    const auto& entries = kernel->GetEntries();
+    buffer_cache.SetEnabledComputeUniformBuffers(entries.enabled_uniform_buffers);
+    buffer_cache.UnbindComputeStorageBuffers();
+    u32 ssbo_index = 0;
+    for (const auto& buffer : entries.global_memory_entries) {
+        buffer_cache.BindComputeStorageBuffer(ssbo_index, buffer.cbuf_index, buffer.cbuf_offset,
+                                              buffer.is_written);
+        ++ssbo_index;
+    }
+    buffer_cache.UpdateComputeBuffers();
+    buffer_cache.BindHostComputeBuffers();
 
     const auto& launch_desc = kepler_compute.launch_description;
     glDispatchCompute(launch_desc.grid_dim_x, launch_desc.grid_dim_y, launch_desc.grid_dim_z);
@@ -703,6 +528,12 @@ void RasterizerOpenGL::Query(GPUVAddr gpu_addr, VideoCore::QueryType type,
     query_cache.Query(gpu_addr, type, timestamp);
 }
 
+void RasterizerOpenGL::BindGraphicsUniformBuffer(size_t stage, u32 index, GPUVAddr gpu_addr,
+                                                 u32 size) {
+    std::scoped_lock lock{buffer_cache.mutex};
+    buffer_cache.BindGraphicsUniformBuffer(stage, index, gpu_addr, size);
+}
+
 void RasterizerOpenGL::FlushAll() {}
 
 void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
@@ -711,19 +542,23 @@ void RasterizerOpenGL::FlushRegion(VAddr addr, u64 size) {
         return;
     }
     {
-        auto lock = texture_cache.AcquireLock();
+        std::scoped_lock lock{texture_cache.mutex};
         texture_cache.DownloadMemory(addr, size);
     }
-    buffer_cache.FlushRegion(addr, size);
+    {
+        std::scoped_lock lock{buffer_cache.mutex};
+        buffer_cache.DownloadMemory(addr, size);
+    }
     query_cache.FlushRegion(addr, size);
 }
 
 bool RasterizerOpenGL::MustFlushRegion(VAddr addr, u64 size) {
+    std::scoped_lock lock{buffer_cache.mutex, texture_cache.mutex};
     if (!Settings::IsGPULevelHigh()) {
-        return buffer_cache.MustFlushRegion(addr, size);
+        return buffer_cache.IsRegionGpuModified(addr, size);
     }
     return texture_cache.IsRegionGpuModified(addr, size) ||
-           buffer_cache.MustFlushRegion(addr, size);
+           buffer_cache.IsRegionGpuModified(addr, size);
 }
 
 void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
@@ -732,11 +567,14 @@ void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
         return;
     }
     {
-        auto lock = texture_cache.AcquireLock();
+        std::scoped_lock lock{texture_cache.mutex};
         texture_cache.WriteMemory(addr, size);
     }
+    {
+        std::scoped_lock lock{buffer_cache.mutex};
+        buffer_cache.WriteMemory(addr, size);
+    }
     shader_cache.InvalidateRegion(addr, size);
-    buffer_cache.InvalidateRegion(addr, size);
     query_cache.InvalidateRegion(addr, size);
 }
 
@@ -745,26 +583,35 @@ void RasterizerOpenGL::OnCPUWrite(VAddr addr, u64 size) {
     if (addr == 0 || size == 0) {
         return;
     }
+    shader_cache.OnCPUWrite(addr, size);
     {
-        auto lock = texture_cache.AcquireLock();
+        std::scoped_lock lock{texture_cache.mutex};
         texture_cache.WriteMemory(addr, size);
     }
-    shader_cache.OnCPUWrite(addr, size);
-    buffer_cache.OnCPUWrite(addr, size);
+    {
+        std::scoped_lock lock{buffer_cache.mutex};
+        buffer_cache.CachedWriteMemory(addr, size);
+    }
 }
 
 void RasterizerOpenGL::SyncGuestHost() {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
-    buffer_cache.SyncGuestHost();
     shader_cache.SyncGuestHost();
+    {
+        std::scoped_lock lock{buffer_cache.mutex};
+        buffer_cache.FlushCachedWrites();
+    }
 }
 
 void RasterizerOpenGL::UnmapMemory(VAddr addr, u64 size) {
     {
-        auto lock = texture_cache.AcquireLock();
+        std::scoped_lock lock{texture_cache.mutex};
         texture_cache.UnmapMemory(addr, size);
     }
-    buffer_cache.OnCPUWrite(addr, size);
+    {
+        std::scoped_lock lock{buffer_cache.mutex};
+        buffer_cache.WriteMemory(addr, size);
+    }
     shader_cache.OnCPUWrite(addr, size);
 }
 
@@ -799,14 +646,7 @@ void RasterizerOpenGL::FlushAndInvalidateRegion(VAddr addr, u64 size) {
 }
 
 void RasterizerOpenGL::WaitForIdle() {
-    // Place a barrier on everything that is not framebuffer related.
-    // This is related to another flag that is not currently implemented.
-    glMemoryBarrier(GL_VERTEX_ATTRIB_ARRAY_BARRIER_BIT | GL_ELEMENT_ARRAY_BARRIER_BIT |
-                    GL_UNIFORM_BARRIER_BIT | GL_TEXTURE_FETCH_BARRIER_BIT |
-                    GL_SHADER_IMAGE_ACCESS_BARRIER_BIT | GL_COMMAND_BARRIER_BIT |
-                    GL_PIXEL_BUFFER_BARRIER_BIT | GL_TEXTURE_UPDATE_BARRIER_BIT |
-                    GL_BUFFER_UPDATE_BARRIER_BIT | GL_TRANSFORM_FEEDBACK_BARRIER_BIT |
-                    GL_SHADER_STORAGE_BARRIER_BIT | GL_QUERY_BUFFER_BARRIER_BIT);
+    glMemoryBarrier(GL_ALL_BARRIER_BITS);
 }
 
 void RasterizerOpenGL::FragmentBarrier() {
@@ -831,18 +671,21 @@ void RasterizerOpenGL::TickFrame() {
     num_queued_commands = 0;
 
     fence_manager.TickFrame();
-    buffer_cache.TickFrame();
     {
-        auto lock = texture_cache.AcquireLock();
+        std::scoped_lock lock{texture_cache.mutex};
         texture_cache.TickFrame();
     }
+    {
+        std::scoped_lock lock{buffer_cache.mutex};
+        buffer_cache.TickFrame();
+    }
 }
 
 bool RasterizerOpenGL::AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Surface& src,
                                              const Tegra::Engines::Fermi2D::Surface& dst,
                                              const Tegra::Engines::Fermi2D::Config& copy_config) {
     MICROPROFILE_SCOPE(OpenGL_Blits);
-    auto lock = texture_cache.AcquireLock();
+    std::scoped_lock lock{texture_cache.mutex};
     texture_cache.BlitImage(dst, src, copy_config);
     return true;
 }
@@ -854,7 +697,7 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
     }
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
 
-    auto lock = texture_cache.AcquireLock();
+    std::scoped_lock lock{texture_cache.mutex};
     ImageView* const image_view{texture_cache.TryFindFramebufferImageView(framebuffer_addr)};
     if (!image_view) {
         return false;
@@ -921,166 +764,6 @@ void RasterizerOpenGL::BindTextures(const ShaderEntries& entries, GLuint base_te
     }
 }
 
-void RasterizerOpenGL::SetupDrawConstBuffers(std::size_t stage_index, Shader* shader) {
-    static constexpr std::array PARAMETER_LUT{
-        GL_VERTEX_PROGRAM_PARAMETER_BUFFER_NV,          GL_TESS_CONTROL_PROGRAM_PARAMETER_BUFFER_NV,
-        GL_TESS_EVALUATION_PROGRAM_PARAMETER_BUFFER_NV, GL_GEOMETRY_PROGRAM_PARAMETER_BUFFER_NV,
-        GL_FRAGMENT_PROGRAM_PARAMETER_BUFFER_NV,
-    };
-    MICROPROFILE_SCOPE(OpenGL_UBO);
-    const auto& stages = maxwell3d.state.shader_stages;
-    const auto& shader_stage = stages[stage_index];
-    const auto& entries = shader->GetEntries();
-    const bool use_unified = entries.use_unified_uniforms;
-    const std::size_t base_unified_offset = stage_index * NUM_CONST_BUFFERS_BYTES_PER_STAGE;
-
-    const auto base_bindings = device.GetBaseBindings(stage_index);
-    u32 binding = device.UseAssemblyShaders() ? 0 : base_bindings.uniform_buffer;
-    for (const auto& entry : entries.const_buffers) {
-        const u32 index = entry.GetIndex();
-        const auto& buffer = shader_stage.const_buffers[index];
-        SetupConstBuffer(PARAMETER_LUT[stage_index], binding, buffer, entry, use_unified,
-                         base_unified_offset + index * Maxwell::MaxConstBufferSize);
-        ++binding;
-    }
-    if (use_unified) {
-        const u32 index = static_cast<u32>(base_bindings.shader_storage_buffer +
-                                           entries.global_memory_entries.size());
-        glBindBufferRange(GL_SHADER_STORAGE_BUFFER, index, unified_uniform_buffer.handle,
-                          base_unified_offset, NUM_CONST_BUFFERS_BYTES_PER_STAGE);
-    }
-}
-
-void RasterizerOpenGL::SetupComputeConstBuffers(Shader* kernel) {
-    MICROPROFILE_SCOPE(OpenGL_UBO);
-    const auto& launch_desc = kepler_compute.launch_description;
-    const auto& entries = kernel->GetEntries();
-    const bool use_unified = entries.use_unified_uniforms;
-
-    u32 binding = 0;
-    for (const auto& entry : entries.const_buffers) {
-        const auto& config = launch_desc.const_buffer_config[entry.GetIndex()];
-        const std::bitset<8> mask = launch_desc.const_buffer_enable_mask.Value();
-        Tegra::Engines::ConstBufferInfo buffer;
-        buffer.address = config.Address();
-        buffer.size = config.size;
-        buffer.enabled = mask[entry.GetIndex()];
-        SetupConstBuffer(GL_COMPUTE_PROGRAM_PARAMETER_BUFFER_NV, binding, buffer, entry,
-                         use_unified, entry.GetIndex() * Maxwell::MaxConstBufferSize);
-        ++binding;
-    }
-    if (use_unified) {
-        const GLuint index = static_cast<GLuint>(entries.global_memory_entries.size());
-        glBindBufferRange(GL_SHADER_STORAGE_BUFFER, index, unified_uniform_buffer.handle, 0,
-                          NUM_CONST_BUFFERS_BYTES_PER_STAGE);
-    }
-}
-
-void RasterizerOpenGL::SetupConstBuffer(GLenum stage, u32 binding,
-                                        const Tegra::Engines::ConstBufferInfo& buffer,
-                                        const ConstBufferEntry& entry, bool use_unified,
-                                        std::size_t unified_offset) {
-    if (!buffer.enabled) {
-        // Set values to zero to unbind buffers
-        if (device.UseAssemblyShaders()) {
-            glBindBufferRangeNV(stage, entry.GetIndex(), 0, 0, 0);
-        } else {
-            glBindBufferRange(GL_UNIFORM_BUFFER, binding, 0, 0, sizeof(float));
-        }
-        return;
-    }
-
-    // Align the actual size so it ends up being a multiple of vec4 to meet the OpenGL std140
-    // UBO alignment requirements.
-    const std::size_t size = Common::AlignUp(GetConstBufferSize(buffer, entry), sizeof(GLvec4));
-
-    const bool fast_upload = !use_unified && device.HasFastBufferSubData();
-
-    const std::size_t alignment = use_unified ? 4 : device.GetUniformBufferAlignment();
-    const GPUVAddr gpu_addr = buffer.address;
-    auto info = buffer_cache.UploadMemory(gpu_addr, size, alignment, false, fast_upload);
-
-    if (device.UseAssemblyShaders()) {
-        UNIMPLEMENTED_IF(use_unified);
-        if (info.offset != 0) {
-            const GLuint staging_cbuf = staging_cbufs[current_cbuf++];
-            glCopyNamedBufferSubData(info.handle, staging_cbuf, info.offset, 0, size);
-            info.handle = staging_cbuf;
-            info.offset = 0;
-        }
-        glBindBufferRangeNV(stage, binding, info.handle, info.offset, size);
-        return;
-    }
-
-    if (use_unified) {
-        glCopyNamedBufferSubData(info.handle, unified_uniform_buffer.handle, info.offset,
-                                 unified_offset, size);
-    } else {
-        glBindBufferRange(GL_UNIFORM_BUFFER, binding, info.handle, info.offset, size);
-    }
-}
-
-void RasterizerOpenGL::SetupDrawGlobalMemory(std::size_t stage_index, Shader* shader) {
-    static constexpr std::array TARGET_LUT = {
-        GL_VERTEX_PROGRAM_NV,   GL_TESS_CONTROL_PROGRAM_NV, GL_TESS_EVALUATION_PROGRAM_NV,
-        GL_GEOMETRY_PROGRAM_NV, GL_FRAGMENT_PROGRAM_NV,
-    };
-    const auto& cbufs{maxwell3d.state.shader_stages[stage_index]};
-    const auto& entries{shader->GetEntries().global_memory_entries};
-
-    std::array<BindlessSSBO, 32> ssbos;
-    ASSERT(entries.size() < ssbos.size());
-
-    const bool assembly_shaders = device.UseAssemblyShaders();
-    u32 binding = assembly_shaders ? 0 : device.GetBaseBindings(stage_index).shader_storage_buffer;
-    for (const auto& entry : entries) {
-        const GPUVAddr addr{cbufs.const_buffers[entry.cbuf_index].address + entry.cbuf_offset};
-        const GPUVAddr gpu_addr{gpu_memory.Read<u64>(addr)};
-        const u32 size{gpu_memory.Read<u32>(addr + 8)};
-        SetupGlobalMemory(binding, entry, gpu_addr, size, &ssbos[binding]);
-        ++binding;
-    }
-    if (assembly_shaders) {
-        UpdateBindlessSSBOs(TARGET_LUT[stage_index], ssbos.data(), entries.size());
-    }
-}
-
-void RasterizerOpenGL::SetupComputeGlobalMemory(Shader* kernel) {
-    const auto& cbufs{kepler_compute.launch_description.const_buffer_config};
-    const auto& entries{kernel->GetEntries().global_memory_entries};
-
-    std::array<BindlessSSBO, 32> ssbos;
-    ASSERT(entries.size() < ssbos.size());
-
-    u32 binding = 0;
-    for (const auto& entry : entries) {
-        const GPUVAddr addr{cbufs[entry.cbuf_index].Address() + entry.cbuf_offset};
-        const GPUVAddr gpu_addr{gpu_memory.Read<u64>(addr)};
-        const u32 size{gpu_memory.Read<u32>(addr + 8)};
-        SetupGlobalMemory(binding, entry, gpu_addr, size, &ssbos[binding]);
-        ++binding;
-    }
-    if (device.UseAssemblyShaders()) {
-        UpdateBindlessSSBOs(GL_COMPUTE_PROGRAM_NV, ssbos.data(), ssbos.size());
-    }
-}
-
-void RasterizerOpenGL::SetupGlobalMemory(u32 binding, const GlobalMemoryEntry& entry,
-                                         GPUVAddr gpu_addr, size_t size, BindlessSSBO* ssbo) {
-    const size_t alignment{device.GetShaderStorageBufferAlignment()};
-    const auto info = buffer_cache.UploadMemory(gpu_addr, size, alignment, entry.is_written);
-    if (device.UseAssemblyShaders()) {
-        *ssbo = BindlessSSBO{
-            .address = static_cast<GLuint64EXT>(info.address + info.offset),
-            .length = static_cast<GLsizei>(size),
-            .padding = 0,
-        };
-    } else {
-        glBindBufferRange(GL_SHADER_STORAGE_BUFFER, binding, info.handle, info.offset,
-                          static_cast<GLsizeiptr>(size));
-    }
-}
-
 void RasterizerOpenGL::SetupDrawTextures(const Shader* shader, size_t stage_index) {
     const bool via_header_index =
         maxwell3d.regs.sampler_index == Maxwell::SamplerIndex::ViaHeaderIndex;
@@ -1128,6 +811,30 @@ void RasterizerOpenGL::SetupComputeImages(const Shader* shader) {
     }
 }
 
+void RasterizerOpenGL::SyncState() {
+    SyncViewport();
+    SyncRasterizeEnable();
+    SyncPolygonModes();
+    SyncColorMask();
+    SyncFragmentColorClampState();
+    SyncMultiSampleState();
+    SyncDepthTestState();
+    SyncDepthClamp();
+    SyncStencilTestState();
+    SyncBlendState();
+    SyncLogicOpState();
+    SyncCullMode();
+    SyncPrimitiveRestart();
+    SyncScissorTest();
+    SyncPointState();
+    SyncLineState();
+    SyncPolygonOffset();
+    SyncAlphaTest();
+    SyncFramebufferSRGB();
+    SyncVertexFormats();
+    SyncVertexInstances();
+}
+
 void RasterizerOpenGL::SyncViewport() {
     auto& flags = maxwell3d.dirty.flags;
     const auto& regs = maxwell3d.regs;
@@ -1163,9 +870,11 @@ void RasterizerOpenGL::SyncViewport() {
         if (regs.screen_y_control.y_negate != 0) {
             flip_y = !flip_y;
         }
-        glClipControl(flip_y ? GL_UPPER_LEFT : GL_LOWER_LEFT,
-                      regs.depth_mode == Maxwell::DepthMode::ZeroToOne ? GL_ZERO_TO_ONE
-                                                                       : GL_NEGATIVE_ONE_TO_ONE);
+        const bool is_zero_to_one = regs.depth_mode == Maxwell::DepthMode::ZeroToOne;
+        const GLenum origin = flip_y ? GL_UPPER_LEFT : GL_LOWER_LEFT;
+        const GLenum depth = is_zero_to_one ? GL_ZERO_TO_ONE : GL_NEGATIVE_ONE_TO_ONE;
+        state_tracker.ClipControl(origin, depth);
+        state_tracker.SetYNegate(regs.screen_y_control.y_negate != 0);
     }
 
     if (dirty_viewport) {
@@ -1649,36 +1358,13 @@ void RasterizerOpenGL::BeginTransformFeedback(GLenum primitive_mode) {
     if (regs.tfb_enabled == 0) {
         return;
     }
-
     if (device.UseAssemblyShaders()) {
         SyncTransformFeedback();
     }
-
     UNIMPLEMENTED_IF(regs.IsShaderConfigEnabled(Maxwell::ShaderProgram::TesselationControl) ||
                      regs.IsShaderConfigEnabled(Maxwell::ShaderProgram::TesselationEval) ||
                      regs.IsShaderConfigEnabled(Maxwell::ShaderProgram::Geometry));
-
-    for (std::size_t index = 0; index < Maxwell::NumTransformFeedbackBuffers; ++index) {
-        const auto& binding = regs.tfb_bindings[index];
-        if (!binding.buffer_enable) {
-            if (enabled_transform_feedback_buffers[index]) {
-                glBindBufferRange(GL_TRANSFORM_FEEDBACK_BUFFER, static_cast<GLuint>(index), 0, 0,
-                                  0);
-            }
-            enabled_transform_feedback_buffers[index] = false;
-            continue;
-        }
-        enabled_transform_feedback_buffers[index] = true;
-
-        auto& tfb_buffer = transform_feedback_buffers[index];
-        tfb_buffer.Create();
-
-        const GLuint handle = tfb_buffer.handle;
-        const std::size_t size = binding.buffer_size;
-        glNamedBufferData(handle, static_cast<GLsizeiptr>(size), nullptr, GL_STREAM_COPY);
-        glBindBufferRange(GL_TRANSFORM_FEEDBACK_BUFFER, static_cast<GLuint>(index), handle, 0,
-                          static_cast<GLsizeiptr>(size));
-    }
+    UNIMPLEMENTED_IF(primitive_mode != GL_POINTS);
 
     // We may have to call BeginTransformFeedbackNV here since they seem to call different
     // implementations on Nvidia's driver (the pointer is different) but we are using
@@ -1692,23 +1378,7 @@ void RasterizerOpenGL::EndTransformFeedback() {
     if (regs.tfb_enabled == 0) {
         return;
     }
-
     glEndTransformFeedback();
-
-    for (std::size_t index = 0; index < Maxwell::NumTransformFeedbackBuffers; ++index) {
-        const auto& binding = regs.tfb_bindings[index];
-        if (!binding.buffer_enable) {
-            continue;
-        }
-        UNIMPLEMENTED_IF(binding.buffer_offset != 0);
-
-        const GLuint handle = transform_feedback_buffers[index].handle;
-        const GPUVAddr gpu_addr = binding.Address();
-        const std::size_t size = binding.buffer_size;
-        const auto info = buffer_cache.UploadMemory(gpu_addr, size, 4, true);
-        glCopyNamedBufferSubData(handle, info.handle, 0, info.offset,
-                                 static_cast<GLsizeiptr>(size));
-    }
 }
 
 } // namespace OpenGL