Rasterizer: Implemented instanced rendering.

We keep track of the current instance and update an uniform in the shaders to let them know which instance they are.

Instanced vertex arrays are not yet implemented.

CMakeLists.txt

@@ -66,10 +66,12 @@ if (NOT ENABLE_GENERIC)
         detect_architecture("_M_AMD64" x86_64)
         detect_architecture("_M_IX86" x86)
         detect_architecture("_M_ARM" ARM)
+        detect_architecture("_M_ARM64" ARM64)
     else()
         detect_architecture("__x86_64__" x86_64)
         detect_architecture("__i386__" x86)
         detect_architecture("__arm__" ARM)
+        detect_architecture("__aarch64__" ARM64)
     endif()
 endif()
 

src/common/CMakeLists.txt

@@ -29,8 +29,6 @@ add_library(common STATIC
     assert.h
     bit_field.h
     bit_set.h
-    break_points.cpp
-    break_points.h
     cityhash.cpp
     cityhash.h
     color.h

src/common/break_points.cpp

@@ -1,90 +0,0 @@
-// Copyright 2013 Dolphin Emulator Project / 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <algorithm>
-#include <sstream>
-#include "common/break_points.h"
-
-bool BreakPoints::IsAddressBreakPoint(u32 iAddress) const {
-    auto cond = [&iAddress](const TBreakPoint& bp) { return bp.iAddress == iAddress; };
-    auto it = std::find_if(m_BreakPoints.begin(), m_BreakPoints.end(), cond);
-    return it != m_BreakPoints.end();
-}
-
-bool BreakPoints::IsTempBreakPoint(u32 iAddress) const {
-    auto cond = [&iAddress](const TBreakPoint& bp) {
-        return bp.iAddress == iAddress && bp.bTemporary;
-    };
-    auto it = std::find_if(m_BreakPoints.begin(), m_BreakPoints.end(), cond);
-    return it != m_BreakPoints.end();
-}
-
-BreakPoints::TBreakPointsStr BreakPoints::GetStrings() const {
-    TBreakPointsStr bps;
-    for (auto breakpoint : m_BreakPoints) {
-        if (!breakpoint.bTemporary) {
-            std::stringstream bp;
-            bp << std::hex << breakpoint.iAddress << " " << (breakpoint.bOn ? "n" : "");
-            bps.push_back(bp.str());
-        }
-    }
-
-    return bps;
-}
-
-void BreakPoints::AddFromStrings(const TBreakPointsStr& bps) {
-    for (auto bps_item : bps) {
-        TBreakPoint bp;
-        std::stringstream bpstr;
-        bpstr << std::hex << bps_item;
-        bpstr >> bp.iAddress;
-        bp.bOn = bps_item.find("n") != bps_item.npos;
-        bp.bTemporary = false;
-        Add(bp);
-    }
-}
-
-void BreakPoints::Add(const TBreakPoint& bp) {
-    if (!IsAddressBreakPoint(bp.iAddress)) {
-        m_BreakPoints.push_back(bp);
-        // if (jit)
-        //    jit->GetBlockCache()->InvalidateICache(bp.iAddress, 4);
-    }
-}
-
-void BreakPoints::Add(u32 em_address, bool temp) {
-    if (!IsAddressBreakPoint(em_address)) // only add new addresses
-    {
-        TBreakPoint pt; // breakpoint settings
-        pt.bOn = true;
-        pt.bTemporary = temp;
-        pt.iAddress = em_address;
-
-        m_BreakPoints.push_back(pt);
-
-        // if (jit)
-        //    jit->GetBlockCache()->InvalidateICache(em_address, 4);
-    }
-}
-
-void BreakPoints::Remove(u32 em_address) {
-    auto cond = [&em_address](const TBreakPoint& bp) { return bp.iAddress == em_address; };
-    auto it = std::find_if(m_BreakPoints.begin(), m_BreakPoints.end(), cond);
-    if (it != m_BreakPoints.end())
-        m_BreakPoints.erase(it);
-}
-
-void BreakPoints::Clear() {
-    // if (jit)
-    //{
-    //    std::for_each(m_BreakPoints.begin(), m_BreakPoints.end(),
-    //        [](const TBreakPoint& bp)
-    //        {
-    //            jit->GetBlockCache()->InvalidateICache(bp.iAddress, 4);
-    //        }
-    //    );
-    //}
-
-    m_BreakPoints.clear();
-}

src/common/break_points.h

@@ -1,49 +0,0 @@
-// Copyright 2013 Dolphin Emulator Project / 2014 Citra Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <string>
-#include <vector>
-#include "common/common_types.h"
-
-class DebugInterface;
-
-struct TBreakPoint {
-    u32 iAddress;
-    bool bOn;
-    bool bTemporary;
-};
-
-// Code breakpoints.
-class BreakPoints {
-public:
-    typedef std::vector<TBreakPoint> TBreakPoints;
-    typedef std::vector<std::string> TBreakPointsStr;
-
-    const TBreakPoints& GetBreakPoints() {
-        return m_BreakPoints;
-    }
-
-    TBreakPointsStr GetStrings() const;
-    void AddFromStrings(const TBreakPointsStr& bps);
-
-    // is address breakpoint
-    bool IsAddressBreakPoint(u32 iAddress) const;
-    bool IsTempBreakPoint(u32 iAddress) const;
-
-    // Add BreakPoint
-    void Add(u32 em_address, bool temp = false);
-    void Add(const TBreakPoint& bp);
-
-    // Remove Breakpoint
-    void Remove(u32 iAddress);
-    void Clear();
-
-    void DeleteByAddress(u32 Address);
-
-private:
-    TBreakPoints m_BreakPoints;
-    u32 m_iBreakOnCount;
-};

src/common/misc.cpp

@@ -4,7 +4,7 @@
 
 #include <cstddef>
 #ifdef _WIN32
-#include <Windows.h>
+#include <windows.h>
 #else
 #include <cerrno>
 #include <cstring>

src/core/frontend/emu_window.h

@@ -34,9 +34,9 @@ class EmuWindow {
 public:
     /// Data structure to store emuwindow configuration
     struct WindowConfig {
-        bool fullscreen;
-        int res_width;
-        int res_height;
+        bool fullscreen = false;
+        int res_width = 0;
+        int res_height = 0;
         std::pair<unsigned, unsigned> min_client_area_size;
     };
 

src/core/hle/kernel/svc.cpp

@@ -250,8 +250,11 @@ static ResultCode ArbitrateUnlock(VAddr mutex_addr) {
 }
 
 /// Break program execution
-static void Break(u64 unk_0, u64 unk_1, u64 unk_2) {
-    LOG_CRITICAL(Debug_Emulated, "Emulated program broke execution!");
+static void Break(u64 reason, u64 info1, u64 info2) {
+    LOG_CRITICAL(
+        Debug_Emulated,
+        "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
+        reason, info1, info2);
     ASSERT(false);
 }
 

src/core/hle/service/mm/mm_u.cpp

@@ -9,42 +9,63 @@
 
 namespace Service::MM {
 
+class MM_U final : public ServiceFramework<MM_U> {
+public:
+    explicit MM_U() : ServiceFramework{"mm:u"} {
+        // clang-format off
+        static const FunctionInfo functions[] = {
+            {0, &MM_U::Initialize, "InitializeOld"},
+            {1, &MM_U::Finalize, "FinalizeOld"},
+            {2, &MM_U::SetAndWait, "SetAndWaitOld"},
+            {3, &MM_U::Get, "GetOld"},
+            {4, &MM_U::Initialize, "Initialize"},
+            {5, &MM_U::Finalize, "Finalize"},
+            {6, &MM_U::SetAndWait, "SetAndWait"},
+            {7, &MM_U::Get, "Get"},
+        };
+        // clang-format on
+
+        RegisterHandlers(functions);
+    }
+
+private:
+    void Initialize(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_MM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    void Finalize(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_MM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    void SetAndWait(Kernel::HLERequestContext& ctx) {
+        IPC::RequestParser rp{ctx};
+        min = rp.Pop<u32>();
+        max = rp.Pop<u32>();
+        current = min;
+
+        LOG_WARNING(Service_MM, "(STUBBED) called, min=0x{:X}, max=0x{:X}", min, max);
+        IPC::ResponseBuilder rb{ctx, 2};
+        rb.Push(RESULT_SUCCESS);
+    }
+
+    void Get(Kernel::HLERequestContext& ctx) {
+        LOG_WARNING(Service_MM, "(STUBBED) called");
+        IPC::ResponseBuilder rb{ctx, 3};
+        rb.Push(RESULT_SUCCESS);
+        rb.Push(current);
+    }
+
+    u32 min{0};
+    u32 max{0};
+    u32 current{0};
+};
+
 void InstallInterfaces(SM::ServiceManager& service_manager) {
     std::make_shared<MM_U>()->InstallAsService(service_manager);
 }
 
-void MM_U::Initialize(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_MM, "(STUBBED) called");
-    IPC::ResponseBuilder rb{ctx, 2};
-    rb.Push(RESULT_SUCCESS);
-}
-
-void MM_U::SetAndWait(Kernel::HLERequestContext& ctx) {
-    IPC::RequestParser rp{ctx};
-    min = rp.Pop<u32>();
-    max = rp.Pop<u32>();
-    current = min;
-
-    LOG_WARNING(Service_MM, "(STUBBED) called, min=0x{:X}, max=0x{:X}", min, max);
-    IPC::ResponseBuilder rb{ctx, 2};
-    rb.Push(RESULT_SUCCESS);
-}
-
-void MM_U::Get(Kernel::HLERequestContext& ctx) {
-    LOG_WARNING(Service_MM, "(STUBBED) called");
-    IPC::ResponseBuilder rb{ctx, 3};
-    rb.Push(RESULT_SUCCESS);
-    rb.Push(current);
-}
-
-MM_U::MM_U() : ServiceFramework("mm:u") {
-    static const FunctionInfo functions[] = {
-        {0, nullptr, "InitializeOld"},        {1, nullptr, "FinalizeOld"},
-        {2, nullptr, "SetAndWaitOld"},        {3, nullptr, "GetOld"},
-        {4, &MM_U::Initialize, "Initialize"}, {5, nullptr, "Finalize"},
-        {6, &MM_U::SetAndWait, "SetAndWait"}, {7, &MM_U::Get, "Get"},
-    };
-    RegisterHandlers(functions);
-}
-
 } // namespace Service::MM

src/core/hle/service/mm/mm_u.h

@@ -8,21 +8,6 @@
 
 namespace Service::MM {
 
-class MM_U final : public ServiceFramework<MM_U> {
-public:
-    MM_U();
-    ~MM_U() = default;
-
-private:
-    void Initialize(Kernel::HLERequestContext& ctx);
-    void SetAndWait(Kernel::HLERequestContext& ctx);
-    void Get(Kernel::HLERequestContext& ctx);
-
-    u32 min{0};
-    u32 max{0};
-    u32 current{0};
-};
-
 /// Registers all MM services with the specified service manager.
 void InstallInterfaces(SM::ServiceManager& service_manager);
 

src/core/loader/deconstructed_rom_directory.cpp

@@ -118,7 +118,6 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(
 
     process->program_id = metadata.GetTitleID();
     process->svc_access_mask.set();
-    process->address_mappings = default_address_mappings;
     process->resource_limit =
         Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
     process->Run(Memory::PROCESS_IMAGE_VADDR, metadata.GetMainThreadPriority(),

src/core/loader/elf.cpp

@@ -398,7 +398,6 @@ ResultStatus AppLoader_ELF::Load(Kernel::SharedPtr<Kernel::Process>& process) {
 
     process->LoadModule(codeset, codeset->entrypoint);
     process->svc_access_mask.set();
-    process->address_mappings = default_address_mappings;
 
     // Attach the default resource limit (APPLICATION) to the process
     process->resource_limit =

src/core/loader/loader.cpp

@@ -17,12 +17,6 @@
 
 namespace Loader {
 
-const std::initializer_list<Kernel::AddressMapping> default_address_mappings = {
-    {0x1FF50000, 0x8000, true},    // part of DSP RAM
-    {0x1FF70000, 0x8000, true},    // part of DSP RAM
-    {0x1F000000, 0x600000, false}, // entire VRAM
-};
-
 FileType IdentifyFile(FileSys::VirtualFile file) {
     FileType type;
 

src/core/loader/loader.h

@@ -5,7 +5,6 @@
 #pragma once
 
 #include <algorithm>
-#include <initializer_list>
 #include <memory>
 #include <string>
 #include <utility>
@@ -207,12 +206,6 @@ protected:
     bool is_loaded = false;
 };
 
-/**
- * Common address mappings found in most games, used for binary formats that don't have this
- * information.
- */
-extern const std::initializer_list<Kernel::AddressMapping> default_address_mappings;
-
 /**
  * Identifies a bootable file and return a suitable loader
  * @param file The bootable file

src/core/loader/nro.cpp

@@ -186,7 +186,6 @@ ResultStatus AppLoader_NRO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     }
 
     process->svc_access_mask.set();
-    process->address_mappings = default_address_mappings;
     process->resource_limit =
         Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
     process->Run(base_addr, THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);

src/core/loader/nso.cpp

@@ -152,7 +152,6 @@ ResultStatus AppLoader_NSO::Load(Kernel::SharedPtr<Kernel::Process>& process) {
     LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", file->GetName(), Memory::PROCESS_IMAGE_VADDR);
 
     process->svc_access_mask.set();
-    process->address_mappings = default_address_mappings;
     process->resource_limit =
         Kernel::ResourceLimit::GetForCategory(Kernel::ResourceLimitCategory::APPLICATION);
     process->Run(Memory::PROCESS_IMAGE_VADDR, THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);

src/video_core/engines/maxwell_3d.cpp

@@ -222,6 +222,18 @@ void Maxwell3D::DrawArrays() {
         debug_context->OnEvent(Tegra::DebugContext::Event::FinishedPrimitiveBatch, nullptr);
     }
 
+    // Both instance configuration registers can not be set at the same time.
+    ASSERT_MSG(!regs.draw.instance_next || !regs.draw.instance_cont,
+               "Illegal combination of instancing parameters");
+
+    if (regs.draw.instance_next) {
+        // Increment the current instance *before* drawing.
+        state.current_instance += 1;
+    } else if (!regs.draw.instance_cont) {
+        // Reset the current instance to 0.
+        state.current_instance = 0;
+    }
+
     const bool is_indexed{regs.index_array.count && !regs.vertex_buffer.count};
     rasterizer.AccelerateDrawBatch(is_indexed);
 

src/video_core/engines/maxwell_3d.h

@@ -638,6 +638,8 @@ public:
                     union {
                         u32 vertex_begin_gl;
                         BitField<0, 16, PrimitiveTopology> topology;
+                        BitField<26, 1, u32> instance_next;
+                        BitField<27, 1, u32> instance_cont;
                     };
                 } draw;
 
@@ -830,6 +832,7 @@ public:
         };
 
         std::array<ShaderStageInfo, Regs::MaxShaderStage> shader_stages;
+        u32 current_instance = 0; ///< Current instance to be used to simulate instanced rendering.
     };
 
     State state{};

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -124,7 +124,7 @@ std::pair<u8*, GLintptr> RasterizerOpenGL::SetupVertexArrays(u8* array_ptr,
         glBindVertexBuffer(index, stream_buffer.GetHandle(), vertex_buffer_offset,
                            vertex_array.stride);
 
-        ASSERT_MSG(vertex_array.divisor == 0, "Vertex buffer divisor unimplemented");
+        ASSERT_MSG(vertex_array.divisor == 0, "Instanced vertex arrays are not supported");
     }
 
     // Use the vertex array as-is, assumes that the data is formatted correctly for OpenGL.
@@ -648,11 +648,11 @@ std::tuple<u8*, GLintptr, u32> RasterizerOpenGL::SetupConstBuffers(
 
         if (used_buffer.IsIndirect()) {
             // Buffer is accessed indirectly, so upload the entire thing
-            size = buffer.size * sizeof(float);
+            size = buffer.size;
 
             if (size > MaxConstbufferSize) {
-                LOG_ERROR(HW_GPU, "indirect constbuffer size {} exceeds maximum {}", size,
-                          MaxConstbufferSize);
+                LOG_CRITICAL(HW_GPU, "indirect constbuffer size {} exceeds maximum {}", size,
+                             MaxConstbufferSize);
                 size = MaxConstbufferSize;
             }
         } else {

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -119,7 +119,8 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_COMPRESSED_RGBA_BPTC_UNORM_ARB, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
      true},                                                                    // BC7U
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_4X4
-    {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},            // G8R8
+    {GL_RG8, GL_RG, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},            // G8R8U
+    {GL_RG8, GL_RG, GL_BYTE, ComponentType::SNorm, false},                     // G8R8S
     {GL_RGBA8, GL_BGRA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // BGRA8
     {GL_RGBA32F, GL_RGBA, GL_FLOAT, ComponentType::Float, false},              // RGBA32F
     {GL_RG32F, GL_RG, GL_FLOAT, ComponentType::Float, false},                  // RG32F
@@ -260,7 +261,8 @@ static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, Tegra::GPU
         MortonCopy<true, PixelFormat::DXN2SNORM>,
         MortonCopy<true, PixelFormat::BC7U>,
         MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
-        MortonCopy<true, PixelFormat::G8R8>,
+        MortonCopy<true, PixelFormat::G8R8U>,
+        MortonCopy<true, PixelFormat::G8R8S>,
         MortonCopy<true, PixelFormat::BGRA8>,
         MortonCopy<true, PixelFormat::RGBA32F>,
         MortonCopy<true, PixelFormat::RG32F>,
@@ -315,7 +317,8 @@ static constexpr std::array<void (*)(u32, u32, u32, std::vector<u8>&, Tegra::GPU
         nullptr,
         nullptr,
         nullptr,
-        MortonCopy<false, PixelFormat::G8R8>,
+        MortonCopy<false, PixelFormat::G8R8U>,
+        MortonCopy<false, PixelFormat::G8R8S>,
         MortonCopy<false, PixelFormat::BGRA8>,
         MortonCopy<false, PixelFormat::RGBA32F>,
         MortonCopy<false, PixelFormat::RG32F>,
@@ -461,7 +464,7 @@ static void ConvertS8Z24ToZ24S8(std::vector<u8>& data, u32 width, u32 height) {
 }
 
 static void ConvertG8R8ToR8G8(std::vector<u8>& data, u32 width, u32 height) {
-    const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::G8R8)};
+    const auto bpp{CachedSurface::GetGLBytesPerPixel(PixelFormat::G8R8U)};
     for (size_t y = 0; y < height; ++y) {
         for (size_t x = 0; x < width; ++x) {
             const size_t offset{bpp * (y * width + x)};
@@ -493,7 +496,8 @@ static void ConvertFormatAsNeeded_LoadGLBuffer(std::vector<u8>& data, PixelForma
         ConvertS8Z24ToZ24S8(data, width, height);
         break;
 
-    case PixelFormat::G8R8:
+    case PixelFormat::G8R8U:
+    case PixelFormat::G8R8S:
         // Convert the G8R8 color format to R8G8, as OpenGL does not support G8R8.
         ConvertG8R8ToR8G8(data, width, height);
         break;

src/video_core/renderer_opengl/gl_rasterizer_cache.h

@@ -43,36 +43,37 @@ struct SurfaceParams {
         DXN2SNORM = 17,
         BC7U = 18,
         ASTC_2D_4X4 = 19,
-        G8R8 = 20,
-        BGRA8 = 21,
-        RGBA32F = 22,
-        RG32F = 23,
-        R32F = 24,
-        R16F = 25,
-        R16UNORM = 26,
-        R16S = 27,
-        R16UI = 28,
-        R16I = 29,
-        RG16 = 30,
-        RG16F = 31,
-        RG16UI = 32,
-        RG16I = 33,
-        RG16S = 34,
-        RGB32F = 35,
-        SRGBA8 = 36,
-        RG8U = 37,
-        RG8S = 38,
-        RG32UI = 39,
-        R32UI = 40,
+        G8R8U = 20,
+        G8R8S = 21,
+        BGRA8 = 22,
+        RGBA32F = 23,
+        RG32F = 24,
+        R32F = 25,
+        R16F = 26,
+        R16UNORM = 27,
+        R16S = 28,
+        R16UI = 29,
+        R16I = 30,
+        RG16 = 31,
+        RG16F = 32,
+        RG16UI = 33,
+        RG16I = 34,
+        RG16S = 35,
+        RGB32F = 36,
+        SRGBA8 = 37,
+        RG8U = 38,
+        RG8S = 39,
+        RG32UI = 40,
+        R32UI = 41,
 
         MaxColorFormat,
 
         // DepthStencil formats
-        Z24S8 = 41,
-        S8Z24 = 42,
-        Z32F = 43,
-        Z16 = 44,
-        Z32FS8 = 45,
+        Z24S8 = 42,
+        S8Z24 = 43,
+        Z32F = 44,
+        Z16 = 45,
+        Z32FS8 = 46,
 
         MaxDepthStencilFormat,
 
@@ -130,7 +131,8 @@ struct SurfaceParams {
             4, // DXN2SNORM
             4, // BC7U
             4, // ASTC_2D_4X4
-            1, // G8R8
+            1, // G8R8U
+            1, // G8R8S
             1, // BGRA8
             1, // RGBA32F
             1, // RG32F
@@ -187,7 +189,8 @@ struct SurfaceParams {
             128, // DXN2SNORM
             128, // BC7U
             32,  // ASTC_2D_4X4
-            16,  // G8R8
+            16,  // G8R8U
+            16,  // G8R8S
             32,  // BGRA8
             128, // RGBA32F
             64,  // RG32F
@@ -341,7 +344,15 @@ struct SurfaceParams {
                          static_cast<u32>(component_type));
             UNREACHABLE();
         case Tegra::Texture::TextureFormat::G8R8:
-            return PixelFormat::G8R8;
+            switch (component_type) {
+            case Tegra::Texture::ComponentType::UNORM:
+                return PixelFormat::G8R8U;
+            case Tegra::Texture::ComponentType::SNORM:
+                return PixelFormat::G8R8S;
+            }
+            LOG_CRITICAL(HW_GPU, "Unimplemented component_type={}",
+                         static_cast<u32>(component_type));
+            UNREACHABLE();
         case Tegra::Texture::TextureFormat::R16_G16_B16_A16:
             return PixelFormat::RGBA16F;
         case Tegra::Texture::TextureFormat::BF10GF11RF11:
@@ -396,6 +407,8 @@ struct SurfaceParams {
             UNREACHABLE();
         case Tegra::Texture::TextureFormat::ZF32:
             return PixelFormat::Z32F;
+        case Tegra::Texture::TextureFormat::Z16:
+            return PixelFormat::Z16;
         case Tegra::Texture::TextureFormat::Z24S8:
             return PixelFormat::Z24S8;
         case Tegra::Texture::TextureFormat::DXT1:

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -383,15 +383,13 @@ public:
         }
     }
 
-    std::string GetUniformIndirect(u64 index, s64 offset, const Register& index_reg,
+    std::string GetUniformIndirect(u64 cbuf_index, s64 offset, const std::string& index_str,
                                    GLSLRegister::Type type) {
-        declr_const_buffers[index].MarkAsUsedIndirect(index, stage);
+        declr_const_buffers[cbuf_index].MarkAsUsedIndirect(cbuf_index, stage);
 
-        std::string final_offset = "((floatBitsToInt(" + GetRegister(index_reg, 0) + ") + " +
-                                   std::to_string(offset) + ") / 4)";
-
-        std::string value =
-            'c' + std::to_string(index) + '[' + final_offset + " / 4][" + final_offset + " % 4]";
+        std::string final_offset = fmt::format("({} + {})", index_str, offset / 4);
+        std::string value = 'c' + std::to_string(cbuf_index) + '[' + final_offset + " / 4][" +
+                            final_offset + " % 4]";
 
         if (type == GLSLRegister::Type::Float) {
             return value;
@@ -540,7 +538,7 @@ private:
             // vertex shader, and what's the value of the fourth element when inside a Tess Eval
             // shader.
             ASSERT(stage == Maxwell3D::Regs::ShaderStage::Vertex);
-            return "vec4(0, 0, uintBitsToFloat(gl_InstanceID), uintBitsToFloat(gl_VertexID))";
+            return "vec4(0, 0, uintBitsToFloat(instance_id.x), uintBitsToFloat(gl_VertexID))";
         default:
             const u32 index{static_cast<u32>(attribute) -
                             static_cast<u32>(Attribute::Index::Attribute_0)};
@@ -1355,11 +1353,16 @@ private:
             case OpCode::Id::LD_C: {
                 ASSERT_MSG(instr.ld_c.unknown == 0, "Unimplemented");
 
+                // Add an extra scope and declare the index register inside to prevent
+                // overwriting it in case it is used as an output of the LD instruction.
+                shader.AddLine("{");
+                ++shader.scope;
+
+                shader.AddLine("uint index = (" + regs.GetRegisterAsInteger(instr.gpr8, 0, false) +
+                               " / 4) & (MAX_CONSTBUFFER_ELEMENTS - 1);");
+
                 std::string op_a =
-                    regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 0, instr.gpr8,
-                                            GLSLRegister::Type::Float);
-                std::string op_b =
-                    regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 4, instr.gpr8,
+                    regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 0, "index",
                                             GLSLRegister::Type::Float);
 
                 switch (instr.ld_c.type.Value()) {
@@ -1367,16 +1370,22 @@ private:
                     regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
                     break;
 
-                case Tegra::Shader::UniformType::Double:
+                case Tegra::Shader::UniformType::Double: {
+                    std::string op_b =
+                        regs.GetUniformIndirect(instr.cbuf36.index, instr.cbuf36.offset + 4,
+                                                "index", GLSLRegister::Type::Float);
                     regs.SetRegisterToFloat(instr.gpr0, 0, op_a, 1, 1);
                     regs.SetRegisterToFloat(instr.gpr0.Value() + 1, 0, op_b, 1, 1);
                     break;
-
+                }
                 default:
                     LOG_CRITICAL(HW_GPU, "Unhandled type: {}",
                                  static_cast<unsigned>(instr.ld_c.type.Value()));
                     UNREACHABLE();
                 }
+
+                --shader.scope;
+                shader.AddLine("}");
                 break;
             }
             case OpCode::Id::ST_A: {

src/video_core/renderer_opengl/gl_shader_gen.cpp

@@ -38,6 +38,7 @@ out vec4 position;
 
 layout (std140) uniform vs_config {
     vec4 viewport_flip;
+    uvec4 instance_id;
 };
 
 void main() {
@@ -90,6 +91,7 @@ out vec4 color;
 
 layout (std140) uniform fs_config {
     vec4 viewport_flip;
+    uvec4 instance_id;
 };
 
 void main() {

src/video_core/renderer_opengl/gl_shader_manager.cpp

@@ -37,11 +37,16 @@ void SetShaderUniformBlockBindings(GLuint shader) {
 } // namespace Impl
 
 void MaxwellUniformData::SetFromRegs(const Maxwell3D::State::ShaderStageInfo& shader_stage) {
-    const auto& regs = Core::System::GetInstance().GPU().Maxwell3D().regs;
+    const auto& gpu = Core::System::GetInstance().GPU().Maxwell3D();
+    const auto& regs = gpu.regs;
+    const auto& state = gpu.state;
 
     // TODO(bunnei): Support more than one viewport
     viewport_flip[0] = regs.viewport_transform[0].scale_x < 0.0 ? -1.0f : 1.0f;
     viewport_flip[1] = regs.viewport_transform[0].scale_y < 0.0 ? -1.0f : 1.0f;
+
+    // We only assign the instance to the first component of the vector, the rest is just padding.
+    instance_id[0] = state.current_instance;
 }
 
 } // namespace GLShader

src/video_core/renderer_opengl/gl_shader_manager.h

@@ -24,14 +24,15 @@ void SetShaderUniformBlockBindings(GLuint shader);
 } // namespace Impl
 
 /// Uniform structure for the Uniform Buffer Object, all vectors must be 16-byte aligned
-// NOTE: Always keep a vec4 at the end. The GL spec is not clear wether the alignment at
+// NOTE: Always keep a vec4 at the end. The GL spec is not clear whether the alignment at
 //       the end of a uniform block is included in UNIFORM_BLOCK_DATA_SIZE or not.
 //       Not following that rule will cause problems on some AMD drivers.
 struct MaxwellUniformData {
     void SetFromRegs(const Maxwell3D::State::ShaderStageInfo& shader_stage);
     alignas(16) GLvec4 viewport_flip;
+    alignas(16) GLuvec4 instance_id;
 };
-static_assert(sizeof(MaxwellUniformData) == 16, "MaxwellUniformData structure size is incorrect");
+static_assert(sizeof(MaxwellUniformData) == 32, "MaxwellUniformData structure size is incorrect");
 static_assert(sizeof(MaxwellUniformData) < 16384,
               "MaxwellUniformData structure must be less than 16kb as per the OpenGL spec");
 

src/video_core/renderer_opengl/maxwell_to_gl.h

@@ -24,16 +24,25 @@ using Maxwell = Tegra::Engines::Maxwell3D::Regs;
 
 inline GLenum VertexType(Maxwell::VertexAttribute attrib) {
     switch (attrib.type) {
+    case Maxwell::VertexAttribute::Type::UnsignedInt:
     case Maxwell::VertexAttribute::Type::UnsignedNorm: {
 
         switch (attrib.size) {
         case Maxwell::VertexAttribute::Size::Size_8:
         case Maxwell::VertexAttribute::Size::Size_8_8:
+        case Maxwell::VertexAttribute::Size::Size_8_8_8:
         case Maxwell::VertexAttribute::Size::Size_8_8_8_8:
             return GL_UNSIGNED_BYTE;
+        case Maxwell::VertexAttribute::Size::Size_16:
         case Maxwell::VertexAttribute::Size::Size_16_16:
+        case Maxwell::VertexAttribute::Size::Size_16_16_16:
         case Maxwell::VertexAttribute::Size::Size_16_16_16_16:
             return GL_UNSIGNED_SHORT;
+        case Maxwell::VertexAttribute::Size::Size_32:
+        case Maxwell::VertexAttribute::Size::Size_32_32:
+        case Maxwell::VertexAttribute::Size::Size_32_32_32:
+        case Maxwell::VertexAttribute::Size::Size_32_32_32_32:
+            return GL_UNSIGNED_INT;
         case Maxwell::VertexAttribute::Size::Size_10_10_10_2:
             return GL_UNSIGNED_INT_2_10_10_10_REV;
         }
@@ -43,16 +52,25 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) {
         return {};
     }
 
+    case Maxwell::VertexAttribute::Type::SignedInt:
     case Maxwell::VertexAttribute::Type::SignedNorm: {
 
         switch (attrib.size) {
-        case Maxwell::VertexAttribute::Size::Size_32_32_32:
-            return GL_INT;
+        case Maxwell::VertexAttribute::Size::Size_8:
         case Maxwell::VertexAttribute::Size::Size_8_8:
+        case Maxwell::VertexAttribute::Size::Size_8_8_8:
         case Maxwell::VertexAttribute::Size::Size_8_8_8_8:
             return GL_BYTE;
+        case Maxwell::VertexAttribute::Size::Size_16:
         case Maxwell::VertexAttribute::Size::Size_16_16:
+        case Maxwell::VertexAttribute::Size::Size_16_16_16:
+        case Maxwell::VertexAttribute::Size::Size_16_16_16_16:
             return GL_SHORT;
+        case Maxwell::VertexAttribute::Size::Size_32:
+        case Maxwell::VertexAttribute::Size::Size_32_32:
+        case Maxwell::VertexAttribute::Size::Size_32_32_32:
+        case Maxwell::VertexAttribute::Size::Size_32_32_32_32:
+            return GL_INT;
         case Maxwell::VertexAttribute::Size::Size_10_10_10_2:
             return GL_INT_2_10_10_10_REV;
         }
@@ -62,9 +80,6 @@ inline GLenum VertexType(Maxwell::VertexAttribute attrib) {
         return {};
     }
 
-    case Maxwell::VertexAttribute::Type::UnsignedInt:
-        return GL_UNSIGNED_INT;
-
     case Maxwell::VertexAttribute::Type::Float:
         return GL_FLOAT;
     }