10 files changed, 78 insertions, 46 deletions

diff --git a/src/audio_core/cubeb_sink.cpp b/src/audio_core/cubeb_sink.cpp
index 7047ed9cf..c4e0e30fe 100644
--- a/src/audio_core/cubeb_sink.cpp
+++ b/src/audio_core/cubeb_sink.cpp
@@ -8,6 +8,7 @@
 #include "audio_core/cubeb_sink.h"
 #include "audio_core/stream.h"
 #include "audio_core/time_stretch.h"
+#include "common/assert.h"
 #include "common/logging/log.h"
 #include "common/ring_buffer.h"
 #include "core/settings.h"
@@ -65,12 +66,25 @@ public:
     void EnqueueSamples(u32 source_num_channels, const std::vector<s16>& samples) override {
         if (source_num_channels > num_channels) {
             // Downsample 6 channels to 2
+            ASSERT_MSG(source_num_channels == 6, "Channel count must be 6");
+
             std::vector<s16> buf;
             buf.reserve(samples.size() * num_channels / source_num_channels);
             for (std::size_t i = 0; i < samples.size(); i += source_num_channels) {
-                for (std::size_t ch = 0; ch < num_channels; ch++) {
-                    buf.push_back(samples[i + ch]);
-                }
+                // Downmixing implementation taken from the ATSC standard
+                const s16 left{samples[i + 0]};
+                const s16 right{samples[i + 1]};
+                const s16 center{samples[i + 2]};
+                const s16 surround_left{samples[i + 4]};
+                const s16 surround_right{samples[i + 5]};
+                // Not used in the ATSC reference implementation
+                [[maybe_unused]] const s16 low_frequency_effects { samples[i + 3] };
+
+                constexpr s32 clev{707}; // center mixing level coefficient
+                constexpr s32 slev{707}; // surround mixing level coefficient
+
+                buf.push_back(left + (clev * center / 1000) + (slev * surround_left / 1000));
+                buf.push_back(right + (clev * center / 1000) + (slev * surround_right / 1000));
             }
             queue.Push(buf);
             return;
diff --git a/src/core/frontend/framebuffer_layout.cpp b/src/core/frontend/framebuffer_layout.cpp
index 2dc795d56..68a0e0906 100644
--- a/src/core/frontend/framebuffer_layout.cpp
+++ b/src/core/frontend/framebuffer_layout.cpp
@@ -48,8 +48,8 @@ FramebufferLayout FrameLayoutFromResolutionScale(u32 res_scale) {
     u32 width, height;
 
     if (Settings::values.use_docked_mode) {
-        width = ScreenDocked::WidthDocked * res_scale;
-        height = ScreenDocked::HeightDocked * res_scale;
+        width = ScreenDocked::Width * res_scale;
+        height = ScreenDocked::Height * res_scale;
     } else {
         width = ScreenUndocked::Width * res_scale;
         height = ScreenUndocked::Height * res_scale;
diff --git a/src/core/frontend/framebuffer_layout.h b/src/core/frontend/framebuffer_layout.h
index e9d0a40d3..15ecfb13d 100644
--- a/src/core/frontend/framebuffer_layout.h
+++ b/src/core/frontend/framebuffer_layout.h
@@ -8,15 +8,15 @@
 
 namespace Layout {
 
-enum ScreenUndocked : u32 {
-    Width = 1280,
-    Height = 720,
-};
+namespace ScreenUndocked {
+constexpr u32 Width = 1280;
+constexpr u32 Height = 720;
+} // namespace ScreenUndocked
 
-enum ScreenDocked : u32 {
-    WidthDocked = 1920,
-    HeightDocked = 1080,
-};
+namespace ScreenDocked {
+constexpr u32 Width = 1920;
+constexpr u32 Height = 1080;
+} // namespace ScreenDocked
 
 enum class AspectRatio {
     Default,
diff --git a/src/core/hle/service/nvflinger/nvflinger.cpp b/src/core/hle/service/nvflinger/nvflinger.cpp
index 134152210..437bc5dee 100644
--- a/src/core/hle/service/nvflinger/nvflinger.cpp
+++ b/src/core/hle/service/nvflinger/nvflinger.cpp
@@ -191,8 +191,6 @@ void NVFlinger::Compose() {
         // Search for a queued buffer and acquire it
         auto buffer = buffer_queue.AcquireBuffer();
 
-        MicroProfileFlip();
-
         if (!buffer) {
             continue;
         }
@@ -206,6 +204,8 @@ void NVFlinger::Compose() {
             gpu.WaitFence(fence.id, fence.value);
         }
 
+        MicroProfileFlip();
+
         // Now send the buffer to the GPU for drawing.
         // TODO(Subv): Support more than just disp0. The display device selection is probably based
         // on which display we're drawing (Default, Internal, External, etc)
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 90124e114..8e48a6482 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -566,7 +566,7 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
     bind_ubo_pushbuffer.Bind();
     bind_ssbo_pushbuffer.Bind();
 
-    program_manager.Update();
+    program_manager.BindGraphicsPipeline();
 
     if (texture_cache.TextureBarrier()) {
         glTextureBarrier();
@@ -628,8 +628,7 @@ void RasterizerOpenGL::DispatchCompute(GPUVAddr code_addr) {
     const ProgramVariant variant(launch_desc.block_dim_x, launch_desc.block_dim_y,
                                  launch_desc.block_dim_z, launch_desc.shared_alloc,
                                  launch_desc.local_pos_alloc);
-    glUseProgramStages(program_manager.GetHandle(), GL_COMPUTE_SHADER_BIT,
-                       kernel->GetHandle(variant));
+    program_manager.BindComputeShader(kernel->GetHandle(variant));
 
     const std::size_t buffer_size =
         Tegra::Engines::KeplerCompute::NumConstBuffers *
diff --git a/src/video_core/renderer_opengl/gl_shader_manager.cpp b/src/video_core/renderer_opengl/gl_shader_manager.cpp
index 15f3cd066..9c7b0adbd 100644
--- a/src/video_core/renderer_opengl/gl_shader_manager.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_manager.cpp
@@ -2,21 +2,29 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <glad/glad.h>
+
 #include "common/common_types.h"
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/renderer_opengl/gl_shader_manager.h"
 
 namespace OpenGL::GLShader {
 
-using Tegra::Engines::Maxwell3D;
+ProgramManager::ProgramManager() = default;
 
 ProgramManager::~ProgramManager() = default;
 
 void ProgramManager::Create() {
-    pipeline.Create();
+    graphics_pipeline.Create();
+    glBindProgramPipeline(graphics_pipeline.handle);
 }
 
-void ProgramManager::Update() {
+void ProgramManager::BindGraphicsPipeline() {
+    if (!is_graphics_bound) {
+        is_graphics_bound = true;
+        glUseProgram(0);
+    }
+
     // Avoid updating the pipeline when values have no changed
     if (old_state == current_state) {
         return;
@@ -25,16 +33,21 @@ void ProgramManager::Update() {
     // Workaround for AMD bug
     static constexpr GLenum all_used_stages{GL_VERTEX_SHADER_BIT | GL_GEOMETRY_SHADER_BIT |
                                             GL_FRAGMENT_SHADER_BIT};
-    glUseProgramStages(pipeline.handle, all_used_stages, 0);
-
-    glUseProgramStages(pipeline.handle, GL_VERTEX_SHADER_BIT, current_state.vertex_shader);
-    glUseProgramStages(pipeline.handle, GL_GEOMETRY_SHADER_BIT, current_state.geometry_shader);
-    glUseProgramStages(pipeline.handle, GL_FRAGMENT_SHADER_BIT, current_state.fragment_shader);
+    const GLuint handle = graphics_pipeline.handle;
+    glUseProgramStages(handle, all_used_stages, 0);
+    glUseProgramStages(handle, GL_VERTEX_SHADER_BIT, current_state.vertex_shader);
+    glUseProgramStages(handle, GL_GEOMETRY_SHADER_BIT, current_state.geometry_shader);
+    glUseProgramStages(handle, GL_FRAGMENT_SHADER_BIT, current_state.fragment_shader);
 
     old_state = current_state;
 }
 
-void MaxwellUniformData::SetFromRegs(const Maxwell3D& maxwell) {
+void ProgramManager::BindComputeShader(GLuint program) {
+    is_graphics_bound = false;
+    glUseProgram(program);
+}
+
+void MaxwellUniformData::SetFromRegs(const Tegra::Engines::Maxwell3D& maxwell) {
     const auto& regs = maxwell.regs;
 
     // Y_NEGATE controls what value S2R returns for the Y_DIRECTION system value.
diff --git a/src/video_core/renderer_opengl/gl_shader_manager.h b/src/video_core/renderer_opengl/gl_shader_manager.h
index e94cd75aa..d2e47f2a9 100644
--- a/src/video_core/renderer_opengl/gl_shader_manager.h
+++ b/src/video_core/renderer_opengl/gl_shader_manager.h
@@ -28,11 +28,16 @@ static_assert(sizeof(MaxwellUniformData) < 16384,
 
 class ProgramManager {
 public:
+    explicit ProgramManager();
     ~ProgramManager();
 
     void Create();
 
-    void Update();
+    /// Updates the graphics pipeline and binds it.
+    void BindGraphicsPipeline();
+
+    /// Binds a compute shader.
+    void BindComputeShader(GLuint program);
 
     void UseVertexShader(GLuint program) {
         current_state.vertex_shader = program;
@@ -46,33 +51,27 @@ public:
         current_state.fragment_shader = program;
     }
 
-    GLuint GetHandle() const {
-        return pipeline.handle;
-    }
-
-    void UseTrivialFragmentShader() {
-        current_state.fragment_shader = 0;
-    }
-
 private:
     struct PipelineState {
-        bool operator==(const PipelineState& rhs) const {
+        bool operator==(const PipelineState& rhs) const noexcept {
             return vertex_shader == rhs.vertex_shader && fragment_shader == rhs.fragment_shader &&
                    geometry_shader == rhs.geometry_shader;
         }
 
-        bool operator!=(const PipelineState& rhs) const {
+        bool operator!=(const PipelineState& rhs) const noexcept {
             return !operator==(rhs);
         }
 
-        GLuint vertex_shader{};
-        GLuint fragment_shader{};
-        GLuint geometry_shader{};
+        GLuint vertex_shader = 0;
+        GLuint fragment_shader = 0;
+        GLuint geometry_shader = 0;
     };
 
-    OGLPipeline pipeline;
+    OGLPipeline graphics_pipeline;
+    OGLPipeline compute_pipeline;
     PipelineState current_state;
     PipelineState old_state;
+    bool is_graphics_bound = true;
 };
 
 } // namespace OpenGL::GLShader
diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp
index d03b29c22..12333e8c9 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.cpp
+++ b/src/video_core/renderer_opengl/renderer_opengl.cpp
@@ -443,7 +443,6 @@ void RendererOpenGL::InitOpenGLObjects() {
 
     // Create program pipeline
     program_manager.Create();
-    glBindProgramPipeline(program_manager.GetHandle());
 
     // Generate VBO handle for drawing
     vertex_buffer.Create();
@@ -597,7 +596,7 @@ void RendererOpenGL::DrawScreen(const Layout::FramebufferLayout& layout) {
     program_manager.UseVertexShader(vertex_program.handle);
     program_manager.UseGeometryShader(0);
     program_manager.UseFragmentShader(fragment_program.handle);
-    program_manager.Update();
+    program_manager.BindGraphicsPipeline();
 
     glEnable(GL_CULL_FACE);
     if (screen_info.display_srgb) {
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
index b402fb268..2bcb17b56 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
@@ -854,7 +854,7 @@ void RasterizerVulkan::SetupGraphicsTextures(const ShaderEntries& entries, std::
 
 void RasterizerVulkan::SetupGraphicsImages(const ShaderEntries& entries, std::size_t stage) {
     MICROPROFILE_SCOPE(Vulkan_Images);
-    const auto& gpu = system.GPU().KeplerCompute();
+    const auto& gpu = system.GPU().Maxwell3D();
     for (const auto& entry : entries.images) {
         const auto tic = GetTextureInfo(gpu, entry, stage).tic;
         SetupImage(tic, entry);
@@ -915,6 +915,13 @@ void RasterizerVulkan::SetupComputeImages(const ShaderEntries& entries) {
 
 void RasterizerVulkan::SetupConstBuffer(const ConstBufferEntry& entry,
                                         const Tegra::Engines::ConstBufferInfo& buffer) {
+    if (!buffer.enabled) {
+        // Set values to zero to unbind buffers
+        update_descriptor_queue.AddBuffer(buffer_cache.GetEmptyBuffer(sizeof(float)), 0,
+                                          sizeof(float));
+        return;
+    }
+
     // Align the size to avoid bad std140 interactions
     const std::size_t size =
         Common::AlignUp(CalculateConstBufferSize(entry, buffer), 4 * sizeof(float));
diff --git a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp
index 171d78afc..d9ea3cc21 100644
--- a/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp
+++ b/src/video_core/renderer_vulkan/vk_staging_buffer_pool.cpp
@@ -73,7 +73,8 @@ VKBuffer* VKStagingBufferPool::TryGetReservedBuffer(std::size_t size, bool host_
 VKBuffer& VKStagingBufferPool::CreateStagingBuffer(std::size_t size, bool host_visible) {
     const auto usage =
         vk::BufferUsageFlagBits::eTransferSrc | vk::BufferUsageFlagBits::eTransferDst |
-        vk::BufferUsageFlagBits::eStorageBuffer | vk::BufferUsageFlagBits::eIndexBuffer;
+        vk::BufferUsageFlagBits::eUniformBuffer | vk::BufferUsageFlagBits::eStorageBuffer |
+        vk::BufferUsageFlagBits::eIndexBuffer;
     const u32 log2 = Common::Log2Ceil64(size);
     const vk::BufferCreateInfo buffer_ci({}, 1ULL << log2, usage, vk::SharingMode::eExclusive, 0,
                                          nullptr);