Make Unimplemented Messages Log Error but not halt execution

src/common/assert.h

@@ -41,6 +41,13 @@ __declspec(noinline, noreturn)
         }                                                                                          \
     while (0)
 
+#define UNIMPLEMENTED_IF_MSG(_a_, ...)                                                             \
+    do                                                                                             \
+        if (!(_a_)) {                                                                              \
+            LOG_CRITICAL(Debug, "Unimplemented Code:\n" __VA_ARGS__);                                \
+        }                                                                                          \
+    while (0)
+
 #define UNREACHABLE() ASSERT_MSG(false, "Unreachable code!")
 #define UNREACHABLE_MSG(...) ASSERT_MSG(false, __VA_ARGS__)
 
@@ -52,8 +59,7 @@ __declspec(noinline, noreturn)
 #define DEBUG_ASSERT_MSG(_a_, _desc_, ...)
 #endif
 
-#define UNIMPLEMENTED() ASSERT_MSG(false, "Unimplemented code!")
-#define UNIMPLEMENTED_MSG(...) ASSERT_MSG(false, __VA_ARGS__)
+#define UNIMPLEMENTED() UNIMPLEMENTED_IF_MSG(false, "Unimplemented code!")
+#define UNIMPLEMENTED_MSG(...) UNIMPLEMENTED_IF_MSG(false, __VA_ARGS__)
 
-#define UNIMPLEMENTED_IF(cond) ASSERT_MSG(!(cond), "Unimplemented code!")
-#define UNIMPLEMENTED_IF_MSG(cond, ...) ASSERT_MSG(!(cond), __VA_ARGS__)
+#define UNIMPLEMENTED_IF(cond) UNIMPLEMENTED_IF_MSG(!(cond), "Unimplemented code!")

src/core/hle/kernel/svc.cpp

@@ -671,8 +671,7 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
         break;
     }
     default:
-        LOG_WARNING(Kernel_SVC, "(STUBBED) Unimplemented svcGetInfo id=0x{:016X}", info_id);
-        return ERR_INVALID_ENUM_VALUE;
+        UNIMPLEMENTED();
     }
 
     return RESULT_SUCCESS;

src/core/hle/service/sm/sm.cpp

@@ -63,7 +63,7 @@ ResultVal<Kernel::SharedPtr<Kernel::ServerPort>> ServiceManager::RegisterService
     return MakeResult<Kernel::SharedPtr<Kernel::ServerPort>>(std::move(server_port));
 }
 
-ResultCode ServiceManager::UnregisterService(const std::string& name) {
+ResultCode ServiceManager::UnregisterService(std::string name) {
     CASCADE_CODE(ValidateServiceName(name));
 
     const auto iter = registered_services.find(name);

src/core/hle/service/sm/sm.h

@@ -50,7 +50,7 @@ public:
 
     ResultVal<Kernel::SharedPtr<Kernel::ServerPort>> RegisterService(std::string name,
                                                                      unsigned int max_sessions);
-    ResultCode UnregisterService(const std::string& name);
+    ResultCode UnregisterService(std::string name);
     ResultVal<Kernel::SharedPtr<Kernel::ClientPort>> GetServicePort(const std::string& name);
     ResultVal<Kernel::SharedPtr<Kernel::ClientSession>> ConnectToService(const std::string& name);
 

src/video_core/CMakeLists.txt

@@ -21,8 +21,6 @@ add_library(video_core STATIC
     macro_interpreter.h
     memory_manager.cpp
     memory_manager.h
-    morton.cpp
-    morton.h
     rasterizer_cache.cpp
     rasterizer_cache.h
     rasterizer_interface.h
@@ -30,8 +28,6 @@ add_library(video_core STATIC
     renderer_base.h
     renderer_opengl/gl_buffer_cache.cpp
     renderer_opengl/gl_buffer_cache.h
-    renderer_opengl/gl_global_cache.cpp
-    renderer_opengl/gl_global_cache.h
     renderer_opengl/gl_primitive_assembler.cpp
     renderer_opengl/gl_primitive_assembler.h
     renderer_opengl/gl_rasterizer.cpp
@@ -66,6 +62,7 @@ add_library(video_core STATIC
     textures/decoders.cpp
     textures/decoders.h
     textures/texture.h
+    utils.h
     video_core.cpp
     video_core.h
 )

src/video_core/engines/maxwell_3d.cpp

@@ -319,11 +319,6 @@ void Maxwell3D::DrawArrays() {
     }
 }
 
-bool operator<(const Maxwell3D::GlobalMemoryDescriptor& lhs,
-               const Maxwell3D::GlobalMemoryDescriptor& rhs) {
-    return std::tie(lhs.cbuf_index, lhs.cbuf_offset) < std::tie(rhs.cbuf_index, rhs.cbuf_offset);
-}
-
 void Maxwell3D::ProcessCBBind(Regs::ShaderStage stage) {
     // Bind the buffer currently in CB_ADDRESS to the specified index in the desired shader stage.
     auto& shader = state.shader_stages[static_cast<std::size_t>(stage)];

src/video_core/engines/maxwell_3d.h

@@ -5,7 +5,6 @@
 #pragma once
 
 #include <array>
-#include <set>
 #include <unordered_map>
 #include <vector>
 #include "common/assert.h"
@@ -32,12 +31,6 @@ public:
     explicit Maxwell3D(VideoCore::RasterizerInterface& rasterizer, MemoryManager& memory_manager);
     ~Maxwell3D() = default;
 
-    /// Structure representing a global memory region
-    struct GlobalMemoryDescriptor {
-        u64 cbuf_index;
-        u64 cbuf_offset;
-    };
-
     /// Register structure of the Maxwell3D engine.
     /// TODO(Subv): This structure will need to be made bigger as more registers are discovered.
     struct Regs {
@@ -1044,8 +1037,6 @@ public:
 
         std::array<ShaderStageInfo, Regs::MaxShaderStage> shader_stages;
         u32 current_instance = 0; ///< Current instance to be used to simulate instanced rendering.
-
-        std::set<GlobalMemoryDescriptor> global_memory_uniforms;
     };
 
     State state{};
@@ -1078,9 +1069,6 @@ public:
         return macro_memory;
     }
 
-    std::string CreateGlobalMemoryRegion(std::tuple<u64, u64, u64> iadd_data);
-    std::set<std::pair<u64, u64>> ListGlobalMemoryRegions() const;
-
 private:
     void InitializeRegisterDefaults();
 
@@ -1135,9 +1123,6 @@ private:
     void DrawArrays();
 };
 
-bool operator<(const Maxwell3D::GlobalMemoryDescriptor& lhs,
-               const Maxwell3D::GlobalMemoryDescriptor& rhs);
-
 #define ASSERT_REG_POSITION(field_name, position)                                                  \
     static_assert(offsetof(Maxwell3D::Regs, field_name) == position * 4,                           \
                   "Field " #field_name " has invalid position")

src/video_core/engines/shader_bytecode.h

@@ -1,4 +1,4 @@
-// Copyright 2018 yuzu Emulator Project
+// Copyright 2018 yuzu Emulator Project
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
@@ -82,8 +82,6 @@ union Attribute {
         Position = 7,
         Attribute_0 = 8,
         Attribute_31 = 39,
-        ClipDistances0123 = 44,
-        ClipDistances4567 = 45,
         PointCoord = 46,
         // This attribute contains a tuple of (~, ~, InstanceId, VertexId) when inside a vertex
         // shader, and a tuple of (TessCoord.x, TessCoord.y, TessCoord.z, ~) when inside a Tess Eval
@@ -208,8 +206,6 @@ enum class UniformType : u64 {
     SignedShort = 3,
     Single = 4,
     Double = 5,
-    Quad = 6,
-    UnsignedQuad = 7,
 };
 
 enum class StoreType : u64 {
@@ -370,11 +366,6 @@ enum class HalfPrecision : u64 {
     FMZ = 2,
 };
 
-enum class R2pMode : u64 {
-    Pr = 0,
-    Cc = 1,
-};
-
 enum class IpaInterpMode : u64 {
     Linear = 0,
     Perspective = 1,
@@ -781,12 +772,6 @@ union Instruction {
         BitField<44, 2, u64> unknown;
     } st_l;
 
-    union {
-        BitField<48, 3, UniformType> type;
-        BitField<46, 2, u64> cache_mode;
-        BitField<20, 24, s64> offset_immediate;
-    } ld_g;
-
     union {
         BitField<0, 3, u64> pred0;
         BitField<3, 3, u64> pred3;
@@ -869,12 +854,6 @@ union Instruction {
         BitField<39, 3, u64> pred39;
     } hsetp2;
 
-    union {
-        BitField<40, 1, R2pMode> mode;
-        BitField<41, 2, u64> byte;
-        BitField<20, 7, u64> immediate_mask;
-    } r2p;
-
     union {
         BitField<39, 3, u64> pred39;
         BitField<42, 1, u64> neg_pred;
@@ -1277,7 +1256,6 @@ public:
         BFE_C,
         BFE_R,
         BFE_IMM,
-        BFI_IMM_R,
         BRA,
         PBK,
         LD_A,
@@ -1403,7 +1381,6 @@ public:
         PSETP,
         PSET,
         CSETP,
-        R2P_IMM,
         XMAD_IMM,
         XMAD_CR,
         XMAD_RC,
@@ -1419,7 +1396,6 @@ public:
         ArithmeticHalf,
         ArithmeticHalfImmediate,
         Bfe,
-        Bfi,
         Shift,
         Ffma,
         Hfma2,
@@ -1434,7 +1410,6 @@ public:
         HalfSetPredicate,
         PredicateSetPredicate,
         PredicateSetRegister,
-        RegisterSetPredicate,
         Conversion,
         Xmad,
         Unknown,
@@ -1638,7 +1613,6 @@ private:
             INST("0100110000000---", Id::BFE_C, Type::Bfe, "BFE_C"),
             INST("0101110000000---", Id::BFE_R, Type::Bfe, "BFE_R"),
             INST("0011100-00000---", Id::BFE_IMM, Type::Bfe, "BFE_IMM"),
-            INST("0011011-11110---", Id::BFI_IMM_R, Type::Bfi, "BFI_IMM_R"),
             INST("0100110001000---", Id::LOP_C, Type::ArithmeticInteger, "LOP_C"),
             INST("0101110001000---", Id::LOP_R, Type::ArithmeticInteger, "LOP_R"),
             INST("0011100001000---", Id::LOP_IMM, Type::ArithmeticInteger, "LOP_IMM"),
@@ -1673,7 +1647,6 @@ private:
             INST("0101000010001---", Id::PSET, Type::PredicateSetRegister, "PSET"),
             INST("0101000010010---", Id::PSETP, Type::PredicateSetPredicate, "PSETP"),
             INST("010100001010----", Id::CSETP, Type::PredicateSetPredicate, "CSETP"),
-            INST("0011100-11110---", Id::R2P_IMM, Type::RegisterSetPredicate, "R2P_IMM"),
             INST("0011011-00------", Id::XMAD_IMM, Type::Xmad, "XMAD_IMM"),
             INST("0100111---------", Id::XMAD_CR, Type::Xmad, "XMAD_CR"),
             INST("010100010-------", Id::XMAD_RC, Type::Xmad, "XMAD_RC"),

src/video_core/engines/shader_header.h

@@ -62,16 +62,7 @@ struct Header {
             INSERT_PADDING_BYTES(1);  // ImapSystemValuesB
             INSERT_PADDING_BYTES(16); // ImapGenericVector[32]
             INSERT_PADDING_BYTES(2);  // ImapColor
-            union {
-                BitField<0, 8, u16> clip_distances;
-                BitField<8, 1, u16> point_sprite_s;
-                BitField<9, 1, u16> point_sprite_t;
-                BitField<10, 1, u16> fog_coordinate;
-                BitField<12, 1, u16> tessellation_eval_point_u;
-                BitField<13, 1, u16> tessellation_eval_point_v;
-                BitField<14, 1, u16> instance_id;
-                BitField<15, 1, u16> vertex_id;
-            };
+            INSERT_PADDING_BYTES(2);  // ImapSystemValuesC
             INSERT_PADDING_BYTES(5);  // ImapFixedFncTexture[10]
             INSERT_PADDING_BYTES(1);  // ImapReserved
             INSERT_PADDING_BYTES(3);  // OmapSystemValuesA

src/video_core/morton.cpp

@@ -1,353 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <array>
-#include <cstring>
-#include "common/assert.h"
-#include "common/common_types.h"
-#include "core/memory.h"
-#include "video_core/morton.h"
-#include "video_core/surface.h"
-#include "video_core/textures/decoders.h"
-
-namespace VideoCore {
-
-using Surface::GetBytesPerPixel;
-using Surface::PixelFormat;
-
-using MortonCopyFn = void (*)(u32, u32, u32, u32, u32, u8*, std::size_t, VAddr);
-using ConversionArray = std::array<MortonCopyFn, Surface::MaxPixelFormat>;
-
-template <bool morton_to_linear, PixelFormat format>
-static void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth, u32 depth,
-                       u8* buffer, std::size_t buffer_size, VAddr addr) {
-    constexpr u32 bytes_per_pixel = GetBytesPerPixel(format);
-
-    // With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
-    // pixel values.
-    const u32 tile_size_x{GetDefaultBlockWidth(format)};
-    const u32 tile_size_y{GetDefaultBlockHeight(format)};
-
-    if constexpr (morton_to_linear) {
-        Tegra::Texture::UnswizzleTexture(buffer, addr, tile_size_x, tile_size_y, bytes_per_pixel,
-                                         stride, height, depth, block_height, block_depth);
-    } else {
-        Tegra::Texture::CopySwizzledData((stride + tile_size_x - 1) / tile_size_x,
-                                         (height + tile_size_y - 1) / tile_size_y, depth,
-                                         bytes_per_pixel, bytes_per_pixel, Memory::GetPointer(addr),
-                                         buffer, false, block_height, block_depth);
-    }
-}
-
-static constexpr ConversionArray morton_to_linear_fns = {
-    // clang-format off
-        MortonCopy<true, PixelFormat::ABGR8U>,
-        MortonCopy<true, PixelFormat::ABGR8S>,
-        MortonCopy<true, PixelFormat::ABGR8UI>,
-        MortonCopy<true, PixelFormat::B5G6R5U>,
-        MortonCopy<true, PixelFormat::A2B10G10R10U>,
-        MortonCopy<true, PixelFormat::A1B5G5R5U>,
-        MortonCopy<true, PixelFormat::R8U>,
-        MortonCopy<true, PixelFormat::R8UI>,
-        MortonCopy<true, PixelFormat::RGBA16F>,
-        MortonCopy<true, PixelFormat::RGBA16U>,
-        MortonCopy<true, PixelFormat::RGBA16UI>,
-        MortonCopy<true, PixelFormat::R11FG11FB10F>,
-        MortonCopy<true, PixelFormat::RGBA32UI>,
-        MortonCopy<true, PixelFormat::DXT1>,
-        MortonCopy<true, PixelFormat::DXT23>,
-        MortonCopy<true, PixelFormat::DXT45>,
-        MortonCopy<true, PixelFormat::DXN1>,
-        MortonCopy<true, PixelFormat::DXN2UNORM>,
-        MortonCopy<true, PixelFormat::DXN2SNORM>,
-        MortonCopy<true, PixelFormat::BC7U>,
-        MortonCopy<true, PixelFormat::BC6H_UF16>,
-        MortonCopy<true, PixelFormat::BC6H_SF16>,
-        MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
-        MortonCopy<true, PixelFormat::G8R8U>,
-        MortonCopy<true, PixelFormat::G8R8S>,
-        MortonCopy<true, PixelFormat::BGRA8>,
-        MortonCopy<true, PixelFormat::RGBA32F>,
-        MortonCopy<true, PixelFormat::RG32F>,
-        MortonCopy<true, PixelFormat::R32F>,
-        MortonCopy<true, PixelFormat::R16F>,
-        MortonCopy<true, PixelFormat::R16U>,
-        MortonCopy<true, PixelFormat::R16S>,
-        MortonCopy<true, PixelFormat::R16UI>,
-        MortonCopy<true, PixelFormat::R16I>,
-        MortonCopy<true, PixelFormat::RG16>,
-        MortonCopy<true, PixelFormat::RG16F>,
-        MortonCopy<true, PixelFormat::RG16UI>,
-        MortonCopy<true, PixelFormat::RG16I>,
-        MortonCopy<true, PixelFormat::RG16S>,
-        MortonCopy<true, PixelFormat::RGB32F>,
-        MortonCopy<true, PixelFormat::RGBA8_SRGB>,
-        MortonCopy<true, PixelFormat::RG8U>,
-        MortonCopy<true, PixelFormat::RG8S>,
-        MortonCopy<true, PixelFormat::RG32UI>,
-        MortonCopy<true, PixelFormat::R32UI>,
-        MortonCopy<true, PixelFormat::ASTC_2D_8X8>,
-        MortonCopy<true, PixelFormat::ASTC_2D_8X5>,
-        MortonCopy<true, PixelFormat::ASTC_2D_5X4>,
-        MortonCopy<true, PixelFormat::BGRA8_SRGB>,
-        MortonCopy<true, PixelFormat::DXT1_SRGB>,
-        MortonCopy<true, PixelFormat::DXT23_SRGB>,
-        MortonCopy<true, PixelFormat::DXT45_SRGB>,
-        MortonCopy<true, PixelFormat::BC7U_SRGB>,
-        MortonCopy<true, PixelFormat::ASTC_2D_4X4_SRGB>,
-        MortonCopy<true, PixelFormat::ASTC_2D_8X8_SRGB>,
-        MortonCopy<true, PixelFormat::ASTC_2D_8X5_SRGB>,
-        MortonCopy<true, PixelFormat::ASTC_2D_5X4_SRGB>,
-        MortonCopy<true, PixelFormat::ASTC_2D_5X5>,
-        MortonCopy<true, PixelFormat::ASTC_2D_5X5_SRGB>,
-        MortonCopy<true, PixelFormat::ASTC_2D_10X8>,
-        MortonCopy<true, PixelFormat::ASTC_2D_10X8_SRGB>,
-        MortonCopy<true, PixelFormat::Z32F>,
-        MortonCopy<true, PixelFormat::Z16>,
-        MortonCopy<true, PixelFormat::Z24S8>,
-        MortonCopy<true, PixelFormat::S8Z24>,
-        MortonCopy<true, PixelFormat::Z32FS8>,
-    // clang-format on
-};
-
-static constexpr ConversionArray linear_to_morton_fns = {
-    // clang-format off
-        MortonCopy<false, PixelFormat::ABGR8U>,
-        MortonCopy<false, PixelFormat::ABGR8S>,
-        MortonCopy<false, PixelFormat::ABGR8UI>,
-        MortonCopy<false, PixelFormat::B5G6R5U>,
-        MortonCopy<false, PixelFormat::A2B10G10R10U>,
-        MortonCopy<false, PixelFormat::A1B5G5R5U>,
-        MortonCopy<false, PixelFormat::R8U>,
-        MortonCopy<false, PixelFormat::R8UI>,
-        MortonCopy<false, PixelFormat::RGBA16F>,
-        MortonCopy<false, PixelFormat::RGBA16U>,
-        MortonCopy<false, PixelFormat::RGBA16UI>,
-        MortonCopy<false, PixelFormat::R11FG11FB10F>,
-        MortonCopy<false, PixelFormat::RGBA32UI>,
-        MortonCopy<false, PixelFormat::DXT1>,
-        MortonCopy<false, PixelFormat::DXT23>,
-        MortonCopy<false, PixelFormat::DXT45>,
-        MortonCopy<false, PixelFormat::DXN1>,
-        MortonCopy<false, PixelFormat::DXN2UNORM>,
-        MortonCopy<false, PixelFormat::DXN2SNORM>,
-        MortonCopy<false, PixelFormat::BC7U>,
-        MortonCopy<false, PixelFormat::BC6H_UF16>,
-        MortonCopy<false, PixelFormat::BC6H_SF16>,
-        // TODO(Subv): Swizzling ASTC formats are not supported
-        nullptr,
-        MortonCopy<false, PixelFormat::G8R8U>,
-        MortonCopy<false, PixelFormat::G8R8S>,
-        MortonCopy<false, PixelFormat::BGRA8>,
-        MortonCopy<false, PixelFormat::RGBA32F>,
-        MortonCopy<false, PixelFormat::RG32F>,
-        MortonCopy<false, PixelFormat::R32F>,
-        MortonCopy<false, PixelFormat::R16F>,
-        MortonCopy<false, PixelFormat::R16U>,
-        MortonCopy<false, PixelFormat::R16S>,
-        MortonCopy<false, PixelFormat::R16UI>,
-        MortonCopy<false, PixelFormat::R16I>,
-        MortonCopy<false, PixelFormat::RG16>,
-        MortonCopy<false, PixelFormat::RG16F>,
-        MortonCopy<false, PixelFormat::RG16UI>,
-        MortonCopy<false, PixelFormat::RG16I>,
-        MortonCopy<false, PixelFormat::RG16S>,
-        MortonCopy<false, PixelFormat::RGB32F>,
-        MortonCopy<false, PixelFormat::RGBA8_SRGB>,
-        MortonCopy<false, PixelFormat::RG8U>,
-        MortonCopy<false, PixelFormat::RG8S>,
-        MortonCopy<false, PixelFormat::RG32UI>,
-        MortonCopy<false, PixelFormat::R32UI>,
-        nullptr,
-        nullptr,
-        nullptr,
-        MortonCopy<false, PixelFormat::BGRA8_SRGB>,
-        MortonCopy<false, PixelFormat::DXT1_SRGB>,
-        MortonCopy<false, PixelFormat::DXT23_SRGB>,
-        MortonCopy<false, PixelFormat::DXT45_SRGB>,
-        MortonCopy<false, PixelFormat::BC7U_SRGB>,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        nullptr,
-        MortonCopy<false, PixelFormat::Z32F>,
-        MortonCopy<false, PixelFormat::Z16>,
-        MortonCopy<false, PixelFormat::Z24S8>,
-        MortonCopy<false, PixelFormat::S8Z24>,
-        MortonCopy<false, PixelFormat::Z32FS8>,
-    // clang-format on
-};
-
-constexpr MortonCopyFn GetSwizzleFunction(MortonSwizzleMode mode, Surface::PixelFormat format) {
-    switch (mode) {
-    case MortonSwizzleMode::MortonToLinear:
-        return morton_to_linear_fns[static_cast<std::size_t>(format)];
-    case MortonSwizzleMode::LinearToMorton:
-        return linear_to_morton_fns[static_cast<std::size_t>(format)];
-    }
-}
-
-/// 8x8 Z-Order coordinate from 2D coordinates
-static u32 MortonInterleave(u32 x, u32 y) {
-    static const u32 xlut[] = {0x00, 0x01, 0x04, 0x05, 0x10, 0x11, 0x14, 0x15};
-    static const u32 ylut[] = {0x00, 0x02, 0x08, 0x0a, 0x20, 0x22, 0x28, 0x2a};
-    return xlut[x % 8] + ylut[y % 8];
-}
-
-/// Calculates the offset of the position of the pixel in Morton order
-static u32 GetMortonOffset(u32 x, u32 y, u32 bytes_per_pixel) {
-    // Images are split into 8x8 tiles. Each tile is composed of four 4x4 subtiles each
-    // of which is composed of four 2x2 subtiles each of which is composed of four texels.
-    // Each structure is embedded into the next-bigger one in a diagonal pattern, e.g.
-    // texels are laid out in a 2x2 subtile like this:
-    // 2 3
-    // 0 1
-    //
-    // The full 8x8 tile has the texels arranged like this:
-    //
-    // 42 43 46 47 58 59 62 63
-    // 40 41 44 45 56 57 60 61
-    // 34 35 38 39 50 51 54 55
-    // 32 33 36 37 48 49 52 53
-    // 10 11 14 15 26 27 30 31
-    // 08 09 12 13 24 25 28 29
-    // 02 03 06 07 18 19 22 23
-    // 00 01 04 05 16 17 20 21
-    //
-    // This pattern is what's called Z-order curve, or Morton order.
-
-    const unsigned int block_height = 8;
-    const unsigned int coarse_x = x & ~7;
-
-    u32 i = MortonInterleave(x, y);
-
-    const unsigned int offset = coarse_x * block_height;
-
-    return (i + offset) * bytes_per_pixel;
-}
-
-static u32 MortonInterleave128(u32 x, u32 y) {
-    // 128x128 Z-Order coordinate from 2D coordinates
-    static constexpr u32 xlut[] = {
-        0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042,
-        0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809,
-        0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000,
-        0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043,
-        0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a,
-        0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001,
-        0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048,
-        0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b,
-        0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002,
-        0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049,
-        0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840,
-        0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003,
-        0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a,
-        0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841,
-        0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008,
-        0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b,
-        0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842,
-        0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009,
-        0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800,
-        0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843,
-        0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a,
-        0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801,
-        0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848,
-        0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b,
-        0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802,
-        0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849,
-        0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040,
-        0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803,
-        0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a,
-        0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041,
-        0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808,
-        0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b,
-        0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042,
-        0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809,
-        0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b,
-    };
-    static constexpr u32 ylut[] = {
-        0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090,
-        0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124,
-        0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200,
-        0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294,
-        0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330,
-        0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404,
-        0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0,
-        0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534,
-        0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610,
-        0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4,
-        0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780,
-        0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014,
-        0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0,
-        0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184,
-        0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220,
-        0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4,
-        0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390,
-        0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424,
-        0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500,
-        0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594,
-        0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630,
-        0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704,
-        0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0,
-        0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034,
-        0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110,
-        0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4,
-        0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280,
-        0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314,
-        0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0,
-        0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484,
-        0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520,
-        0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4,
-        0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690,
-        0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724,
-        0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4,
-    };
-    return xlut[x % 128] + ylut[y % 128];
-}
-
-static u32 GetMortonOffset128(u32 x, u32 y, u32 bytes_per_pixel) {
-    // Calculates the offset of the position of the pixel in Morton order
-    // Framebuffer images are split into 128x128 tiles.
-
-    constexpr u32 block_height = 128;
-    const u32 coarse_x = x & ~127;
-
-    const u32 i = MortonInterleave128(x, y);
-
-    const u32 offset = coarse_x * block_height;
-
-    return (i + offset) * bytes_per_pixel;
-}
-
-void MortonSwizzle(MortonSwizzleMode mode, Surface::PixelFormat format, u32 stride,
-                   u32 block_height, u32 height, u32 block_depth, u32 depth, u8* buffer,
-                   std::size_t buffer_size, VAddr addr) {
-
-    GetSwizzleFunction(mode, format)(stride, block_height, height, block_depth, depth, buffer,
-                                     buffer_size, addr);
-}
-
-void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixel, u32 linear_bytes_per_pixel,
-                         u8* morton_data, u8* linear_data, bool morton_to_linear) {
-    u8* data_ptrs[2];
-    for (u32 y = 0; y < height; ++y) {
-        for (u32 x = 0; x < width; ++x) {
-            const u32 coarse_y = y & ~127;
-            const u32 morton_offset =
-                GetMortonOffset128(x, y, bytes_per_pixel) + coarse_y * width * bytes_per_pixel;
-            const u32 linear_pixel_index = (x + y * width) * linear_bytes_per_pixel;
-
-            data_ptrs[morton_to_linear ? 1 : 0] = morton_data + morton_offset;
-            data_ptrs[morton_to_linear ? 0 : 1] = &linear_data[linear_pixel_index];
-
-            std::memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel);
-        }
-    }
-}
-
-} // namespace VideoCore

src/video_core/morton.h

@@ -1,21 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include "common/common_types.h"
-#include "video_core/surface.h"
-
-namespace VideoCore {
-
-enum class MortonSwizzleMode { MortonToLinear, LinearToMorton };
-
-void MortonSwizzle(MortonSwizzleMode mode, VideoCore::Surface::PixelFormat format, u32 stride,
-                   u32 block_height, u32 height, u32 block_depth, u32 depth, u8* buffer,
-                   std::size_t buffer_size, VAddr addr);
-
-void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixel, u32 linear_bytes_per_pixel,
-                         u8* morton_data, u8* linear_data, bool morton_to_linear);
-
-} // namespace VideoCore

src/video_core/renderer_opengl/gl_global_cache.cpp

@@ -1,96 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "common/assert.h"
-#include "core/core.h"
-#include "core/memory.h"
-#include "video_core/engines/maxwell_3d.h"
-#include "video_core/renderer_opengl/gl_global_cache.h"
-#include "video_core/renderer_opengl/gl_rasterizer.h"
-#include "video_core/renderer_opengl/gl_shader_cache.h"
-#include "video_core/renderer_opengl/gl_shader_manager.h"
-#include "video_core/renderer_opengl/utils.h"
-
-namespace OpenGL {
-
-CachedGlobalRegion::CachedGlobalRegion(VAddr addr, u32 size) : addr{addr}, size{size} {
-    buffer.Create();
-    LabelGLObject(GL_BUFFER, buffer.handle, addr);
-}
-
-/// Helper function to get the maximum size we can use for an OpenGL uniform block
-static u32 GetMaxUniformBlockSize() {
-    GLint max_size{};
-    glGetIntegerv(GL_MAX_UNIFORM_BLOCK_SIZE, &max_size);
-    return static_cast<u32>(max_size);
-}
-
-void CachedGlobalRegion::Reload(u32 size_) {
-    static const u32 max_size{GetMaxUniformBlockSize()};
-
-    size = size_;
-    if (size > max_size) {
-        size = max_size;
-        LOG_CRITICAL(HW_GPU, "Global region size {} exceeded max UBO size of {}!", size_, max_size);
-    }
-
-    glBindBuffer(GL_UNIFORM_BUFFER, buffer.handle);
-    glBufferData(GL_UNIFORM_BUFFER, size, Memory::GetPointer(addr), GL_DYNAMIC_DRAW);
-}
-
-GlobalRegion GlobalRegionCacheOpenGL::TryGetReservedGlobalRegion(VAddr addr, u32 size) const {
-    auto search{reserve.find(addr)};
-    if (search == reserve.end()) {
-        return {};
-    }
-    return search->second;
-}
-
-GlobalRegion GlobalRegionCacheOpenGL::GetUncachedGlobalRegion(VAddr addr, u32 size) {
-    GlobalRegion region{TryGetReservedGlobalRegion(addr, size)};
-    if (!region) {
-        // No reserved surface available, create a new one and reserve it
-        region = std::make_shared<CachedGlobalRegion>(addr, size);
-        ReserveGlobalRegion(region);
-    }
-    region->Reload(size);
-    return region;
-}
-
-void GlobalRegionCacheOpenGL::ReserveGlobalRegion(const GlobalRegion& region) {
-    reserve[region->GetAddr()] = region;
-}
-
-GlobalRegionCacheOpenGL::GlobalRegionCacheOpenGL(RasterizerOpenGL& rasterizer)
-    : RasterizerCache{rasterizer} {}
-
-GlobalRegion GlobalRegionCacheOpenGL::GetGlobalRegion(
-    const Tegra::Engines::Maxwell3D::GlobalMemoryDescriptor& global_region,
-    Tegra::Engines::Maxwell3D::Regs::ShaderStage stage) {
-    auto& gpu{Core::System::GetInstance().GPU()};
-    const auto cbufs = gpu.Maxwell3D().state.shader_stages[static_cast<u64>(stage)];
-    const auto cbuf_addr{gpu.MemoryManager().GpuToCpuAddress(
-        cbufs.const_buffers[global_region.cbuf_index].address + global_region.cbuf_offset)};
-
-    ASSERT(cbuf_addr);
-
-    const auto actual_addr_gpu = Memory::Read64(*cbuf_addr);
-    const auto size = Memory::Read32(*cbuf_addr + 8);
-    const auto actual_addr{gpu.MemoryManager().GpuToCpuAddress(actual_addr_gpu)};
-
-    ASSERT(actual_addr);
-
-    // Look up global region in the cache based on address
-    GlobalRegion region{TryGet(*actual_addr)};
-
-    if (!region) {
-        // No global region found - create a new one
-        region = GetUncachedGlobalRegion(*actual_addr, size);
-        Register(region);
-    }
-
-    return region;
-}
-
-} // namespace OpenGL

src/video_core/renderer_opengl/gl_global_cache.h

@@ -1,89 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <memory>
-#include <string>
-#include <unordered_map>
-#include <fmt/format.h>
-#include "common/common_types.h"
-#include "video_core/engines/maxwell_3d.h"
-#include "video_core/rasterizer_cache.h"
-#include "video_core/renderer_opengl/gl_resource_manager.h"
-
-namespace OpenGL {
-
-class RasterizerOpenGL;
-class CachedGlobalRegion;
-using GlobalRegion = std::shared_ptr<CachedGlobalRegion>;
-
-/// Helper class for caching global region uniform locations
-class CachedGlobalRegionUniform {
-public:
-    explicit CachedGlobalRegionUniform(std::size_t index) : index{index} {}
-
-    std::string GetName() const {
-        return fmt::format("global_memory_region_declblock_{}", index);
-    }
-
-    u32 GetHash() const {
-        // This needs to be unique from ConstBufferEntry::GetHash and SamplerEntry::GetHash
-        return (static_cast<u32>(index) << 16) | 0xFFFF;
-    }
-
-private:
-    std::size_t index{};
-};
-
-class CachedGlobalRegion final : public RasterizerCacheObject {
-public:
-    CachedGlobalRegion(VAddr addr, u32 size);
-
-    /// Gets the address of the shader in guest memory, required for cache management
-    VAddr GetAddr() const {
-        return addr;
-    }
-
-    /// Gets the size of the shader in guest memory, required for cache management
-    std::size_t GetSizeInBytes() const {
-        return size;
-    }
-
-    /// Gets the GL program handle for the buffer
-    GLuint GetBufferHandle() const {
-        return buffer.handle;
-    }
-
-    /// Reloads the global region from guest memory
-    void Reload(u32 size_);
-
-    // We do not have to flush this cache as things in it are never modified by us.
-    void Flush() override {}
-
-private:
-    VAddr addr;
-    u32 size;
-
-    OGLBuffer buffer;
-};
-
-class GlobalRegionCacheOpenGL final : public RasterizerCache<GlobalRegion> {
-public:
-    explicit GlobalRegionCacheOpenGL(RasterizerOpenGL& rasterizer);
-
-    /// Gets the current specified shader stage program
-    GlobalRegion GetGlobalRegion(
-        const Tegra::Engines::Maxwell3D::GlobalMemoryDescriptor& descriptor,
-        Tegra::Engines::Maxwell3D::Regs::ShaderStage stage);
-
-private:
-    GlobalRegion TryGetReservedGlobalRegion(VAddr addr, u32 size) const;
-    GlobalRegion GetUncachedGlobalRegion(VAddr addr, u32 size);
-    void ReserveGlobalRegion(const GlobalRegion& region);
-
-    std::unordered_map<VAddr, GlobalRegion> reserve;
-};
-
-} // namespace OpenGL

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -81,7 +81,7 @@ struct DrawParameters {
 
 RasterizerOpenGL::RasterizerOpenGL(Core::Frontend::EmuWindow& window, ScreenInfo& info)
     : res_cache{*this}, shader_cache{*this}, emu_window{window}, screen_info{info},
-      buffer_cache(*this, STREAM_BUFFER_SIZE), global_cache{*this} {
+      buffer_cache(*this, STREAM_BUFFER_SIZE) {
     // Create sampler objects
     for (std::size_t i = 0; i < texture_samplers.size(); ++i) {
         texture_samplers[i].Create();
@@ -267,7 +267,7 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
 
     // Next available bindpoints to use when uploading the const buffers and textures to the GLSL
     // shaders. The constbuffer bindpoint starts after the shader stage configuration bind points.
-    u32 current_buffer_bindpoint = Tegra::Engines::Maxwell3D::Regs::MaxShaderStage;
+    u32 current_constbuffer_bindpoint = Tegra::Engines::Maxwell3D::Regs::MaxShaderStage;
     u32 current_texture_bindpoint = 0;
 
     for (std::size_t index = 0; index < Maxwell::MaxShaderProgram; ++index) {
@@ -321,14 +321,9 @@ void RasterizerOpenGL::SetupShaders(GLenum primitive_mode) {
         }
 
         // Configure the const buffers for this shader stage.
-        current_buffer_bindpoint =
+        current_constbuffer_bindpoint =
             SetupConstBuffers(static_cast<Maxwell::ShaderStage>(stage), shader, primitive_mode,
-                              current_buffer_bindpoint);
-
-        // Configure global memory regions for this shader stage.
-        current_buffer_bindpoint =
-            SetupGlobalRegions(static_cast<Maxwell::ShaderStage>(stage), shader, primitive_mode,
-                               current_buffer_bindpoint);
+                              current_constbuffer_bindpoint);
 
         // Configure the textures for this shader stage.
         current_texture_bindpoint = SetupTextures(static_cast<Maxwell::ShaderStage>(stage), shader,
@@ -700,7 +695,6 @@ void RasterizerOpenGL::InvalidateRegion(VAddr addr, u64 size) {
     MICROPROFILE_SCOPE(OpenGL_CacheManagement);
     res_cache.InvalidateRegion(addr, size);
     shader_cache.InvalidateRegion(addr, size);
-    global_cache.InvalidateRegion(addr, size);
     buffer_cache.InvalidateRegion(addr, size);
 }
 
@@ -925,29 +919,6 @@ u32 RasterizerOpenGL::SetupConstBuffers(Maxwell::ShaderStage stage, Shader& shad
     return current_bindpoint + static_cast<u32>(entries.size());
 }
 
-u32 RasterizerOpenGL::SetupGlobalRegions(Maxwell::ShaderStage stage, Shader& shader,
-                                         GLenum primitive_mode, u32 current_bindpoint) {
-    std::size_t global_region_index{};
-    const auto& maxwell3d{Core::System::GetInstance().GPU().Maxwell3D()};
-    for (const auto& global_region : maxwell3d.state.global_memory_uniforms) {
-        const auto& region{
-            global_cache.GetGlobalRegion(global_region, static_cast<Maxwell::ShaderStage>(stage))};
-        const GLenum b_index{
-            shader->GetProgramResourceIndex(CachedGlobalRegionUniform{global_region_index})};
-
-        if (b_index != GL_INVALID_INDEX) {
-            glBindBufferBase(GL_UNIFORM_BUFFER, current_bindpoint, region->GetBufferHandle());
-            glUniformBlockBinding(shader->GetProgramHandle(primitive_mode), b_index,
-                                  current_bindpoint);
-            ++current_bindpoint;
-        }
-
-        ++global_region_index;
-    }
-
-    return current_bindpoint;
-}
-
 u32 RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, Shader& shader,
                                     GLenum primitive_mode, u32 current_unit) {
     MICROPROFILE_SCOPE(OpenGL_Texture);

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -23,7 +23,6 @@
 #include "video_core/rasterizer_cache.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/renderer_opengl/gl_buffer_cache.h"
-#include "video_core/renderer_opengl/gl_global_cache.h"
 #include "video_core/renderer_opengl/gl_primitive_assembler.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
@@ -119,7 +118,7 @@ private:
                                bool using_depth_fb = true, bool preserve_contents = true,
                                std::optional<std::size_t> single_color_target = {});
 
-    /**
+    /*
      * Configures the current constbuffers to use for the draw command.
      * @param stage The shader stage to configure buffers for.
      * @param shader The shader object that contains the specified stage.
@@ -129,17 +128,7 @@ private:
     u32 SetupConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, Shader& shader,
                           GLenum primitive_mode, u32 current_bindpoint);
 
-    /**
-     * Configures the current global memory regions to use for the draw command.
-     * @param stage The shader stage to configure buffers for.
-     * @param shader The shader object that contains the specified stage.
-     * @param current_bindpoint The offset at which to start counting new buffer bindpoints.
-     * @returns The next available bindpoint for use in the next shader stage.
-     */
-    u32 SetupGlobalRegions(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage, Shader& shader,
-                           GLenum primitive_mode, u32 current_bindpoint);
-
-    /**
+    /*
      * Configures the current textures to use for the draw command.
      * @param stage The shader stage to configure textures for.
      * @param shader The shader object that contains the specified stage.
@@ -204,7 +193,6 @@ private:
 
     RasterizerCacheOpenGL res_cache;
     ShaderCacheOpenGL shader_cache;
-    GlobalRegionCacheOpenGL global_cache;
 
     Core::Frontend::EmuWindow& emu_window;
 

src/video_core/renderer_opengl/gl_rasterizer_cache.cpp

@@ -15,7 +15,6 @@
 #include "core/memory.h"
 #include "core/settings.h"
 #include "video_core/engines/maxwell_3d.h"
-#include "video_core/morton.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_state.h"
@@ -23,11 +22,10 @@
 #include "video_core/surface.h"
 #include "video_core/textures/astc.h"
 #include "video_core/textures/decoders.h"
+#include "video_core/utils.h"
 
 namespace OpenGL {
 
-using VideoCore::MortonSwizzle;
-using VideoCore::MortonSwizzleMode;
 using VideoCore::Surface::ComponentTypeFromDepthFormat;
 using VideoCore::Surface::ComponentTypeFromRenderTarget;
 using VideoCore::Surface::ComponentTypeFromTexture;
@@ -372,7 +370,174 @@ MathUtil::Rectangle<u32> SurfaceParams::GetRect(u32 mip_level) const {
     return {0, actual_height, MipWidth(mip_level), 0};
 }
 
-void SwizzleFunc(const MortonSwizzleMode& mode, const SurfaceParams& params,
+template <bool morton_to_gl, PixelFormat format>
+void MortonCopy(u32 stride, u32 block_height, u32 height, u32 block_depth, u32 depth, u8* gl_buffer,
+                std::size_t gl_buffer_size, VAddr addr) {
+    constexpr u32 bytes_per_pixel = GetBytesPerPixel(format);
+
+    // With the BCn formats (DXT and DXN), each 4x4 tile is swizzled instead of just individual
+    // pixel values.
+    const u32 tile_size_x{GetDefaultBlockWidth(format)};
+    const u32 tile_size_y{GetDefaultBlockHeight(format)};
+
+    if (morton_to_gl) {
+        Tegra::Texture::UnswizzleTexture(gl_buffer, addr, tile_size_x, tile_size_y, bytes_per_pixel,
+                                         stride, height, depth, block_height, block_depth);
+    } else {
+        Tegra::Texture::CopySwizzledData((stride + tile_size_x - 1) / tile_size_x,
+                                         (height + tile_size_y - 1) / tile_size_y, depth,
+                                         bytes_per_pixel, bytes_per_pixel, Memory::GetPointer(addr),
+                                         gl_buffer, false, block_height, block_depth);
+    }
+}
+
+using GLConversionArray = std::array<void (*)(u32, u32, u32, u32, u32, u8*, std::size_t, VAddr),
+                                     VideoCore::Surface::MaxPixelFormat>;
+
+static constexpr GLConversionArray morton_to_gl_fns = {
+    // clang-format off
+        MortonCopy<true, PixelFormat::ABGR8U>,
+        MortonCopy<true, PixelFormat::ABGR8S>,
+        MortonCopy<true, PixelFormat::ABGR8UI>,
+        MortonCopy<true, PixelFormat::B5G6R5U>,
+        MortonCopy<true, PixelFormat::A2B10G10R10U>,
+        MortonCopy<true, PixelFormat::A1B5G5R5U>,
+        MortonCopy<true, PixelFormat::R8U>,
+        MortonCopy<true, PixelFormat::R8UI>,
+        MortonCopy<true, PixelFormat::RGBA16F>,
+        MortonCopy<true, PixelFormat::RGBA16U>,
+        MortonCopy<true, PixelFormat::RGBA16UI>,
+        MortonCopy<true, PixelFormat::R11FG11FB10F>,
+        MortonCopy<true, PixelFormat::RGBA32UI>,
+        MortonCopy<true, PixelFormat::DXT1>,
+        MortonCopy<true, PixelFormat::DXT23>,
+        MortonCopy<true, PixelFormat::DXT45>,
+        MortonCopy<true, PixelFormat::DXN1>,
+        MortonCopy<true, PixelFormat::DXN2UNORM>,
+        MortonCopy<true, PixelFormat::DXN2SNORM>,
+        MortonCopy<true, PixelFormat::BC7U>,
+        MortonCopy<true, PixelFormat::BC6H_UF16>,
+        MortonCopy<true, PixelFormat::BC6H_SF16>,
+        MortonCopy<true, PixelFormat::ASTC_2D_4X4>,
+        MortonCopy<true, PixelFormat::G8R8U>,
+        MortonCopy<true, PixelFormat::G8R8S>,
+        MortonCopy<true, PixelFormat::BGRA8>,
+        MortonCopy<true, PixelFormat::RGBA32F>,
+        MortonCopy<true, PixelFormat::RG32F>,
+        MortonCopy<true, PixelFormat::R32F>,
+        MortonCopy<true, PixelFormat::R16F>,
+        MortonCopy<true, PixelFormat::R16U>,
+        MortonCopy<true, PixelFormat::R16S>,
+        MortonCopy<true, PixelFormat::R16UI>,
+        MortonCopy<true, PixelFormat::R16I>,
+        MortonCopy<true, PixelFormat::RG16>,
+        MortonCopy<true, PixelFormat::RG16F>,
+        MortonCopy<true, PixelFormat::RG16UI>,
+        MortonCopy<true, PixelFormat::RG16I>,
+        MortonCopy<true, PixelFormat::RG16S>,
+        MortonCopy<true, PixelFormat::RGB32F>,
+        MortonCopy<true, PixelFormat::RGBA8_SRGB>,
+        MortonCopy<true, PixelFormat::RG8U>,
+        MortonCopy<true, PixelFormat::RG8S>,
+        MortonCopy<true, PixelFormat::RG32UI>,
+        MortonCopy<true, PixelFormat::R32UI>,
+        MortonCopy<true, PixelFormat::ASTC_2D_8X8>,
+        MortonCopy<true, PixelFormat::ASTC_2D_8X5>,
+        MortonCopy<true, PixelFormat::ASTC_2D_5X4>,
+        MortonCopy<true, PixelFormat::BGRA8_SRGB>,
+        MortonCopy<true, PixelFormat::DXT1_SRGB>,
+        MortonCopy<true, PixelFormat::DXT23_SRGB>,
+        MortonCopy<true, PixelFormat::DXT45_SRGB>,
+        MortonCopy<true, PixelFormat::BC7U_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_4X4_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_8X8_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_8X5_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_5X4_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_5X5>,
+        MortonCopy<true, PixelFormat::ASTC_2D_5X5_SRGB>,
+        MortonCopy<true, PixelFormat::ASTC_2D_10X8>,
+        MortonCopy<true, PixelFormat::ASTC_2D_10X8_SRGB>,
+        MortonCopy<true, PixelFormat::Z32F>,
+        MortonCopy<true, PixelFormat::Z16>,
+        MortonCopy<true, PixelFormat::Z24S8>,
+        MortonCopy<true, PixelFormat::S8Z24>,
+        MortonCopy<true, PixelFormat::Z32FS8>,
+    // clang-format on
+};
+
+static constexpr GLConversionArray gl_to_morton_fns = {
+    // clang-format off
+        MortonCopy<false, PixelFormat::ABGR8U>,
+        MortonCopy<false, PixelFormat::ABGR8S>,
+        MortonCopy<false, PixelFormat::ABGR8UI>,
+        MortonCopy<false, PixelFormat::B5G6R5U>,
+        MortonCopy<false, PixelFormat::A2B10G10R10U>,
+        MortonCopy<false, PixelFormat::A1B5G5R5U>,
+        MortonCopy<false, PixelFormat::R8U>,
+        MortonCopy<false, PixelFormat::R8UI>,
+        MortonCopy<false, PixelFormat::RGBA16F>,
+        MortonCopy<false, PixelFormat::RGBA16U>,
+        MortonCopy<false, PixelFormat::RGBA16UI>,
+        MortonCopy<false, PixelFormat::R11FG11FB10F>,
+        MortonCopy<false, PixelFormat::RGBA32UI>,
+        MortonCopy<false, PixelFormat::DXT1>,
+        MortonCopy<false, PixelFormat::DXT23>,
+        MortonCopy<false, PixelFormat::DXT45>,
+        MortonCopy<false, PixelFormat::DXN1>,
+        MortonCopy<false, PixelFormat::DXN2UNORM>,
+        MortonCopy<false, PixelFormat::DXN2SNORM>,
+        MortonCopy<false, PixelFormat::BC7U>,
+        MortonCopy<false, PixelFormat::BC6H_UF16>,
+        MortonCopy<false, PixelFormat::BC6H_SF16>,
+        // TODO(Subv): Swizzling ASTC formats are not supported
+        nullptr,
+        MortonCopy<false, PixelFormat::G8R8U>,
+        MortonCopy<false, PixelFormat::G8R8S>,
+        MortonCopy<false, PixelFormat::BGRA8>,
+        MortonCopy<false, PixelFormat::RGBA32F>,
+        MortonCopy<false, PixelFormat::RG32F>,
+        MortonCopy<false, PixelFormat::R32F>,
+        MortonCopy<false, PixelFormat::R16F>,
+        MortonCopy<false, PixelFormat::R16U>,
+        MortonCopy<false, PixelFormat::R16S>,
+        MortonCopy<false, PixelFormat::R16UI>,
+        MortonCopy<false, PixelFormat::R16I>,
+        MortonCopy<false, PixelFormat::RG16>,
+        MortonCopy<false, PixelFormat::RG16F>,
+        MortonCopy<false, PixelFormat::RG16UI>,
+        MortonCopy<false, PixelFormat::RG16I>,
+        MortonCopy<false, PixelFormat::RG16S>,
+        MortonCopy<false, PixelFormat::RGB32F>,
+        MortonCopy<false, PixelFormat::RGBA8_SRGB>,
+        MortonCopy<false, PixelFormat::RG8U>,
+        MortonCopy<false, PixelFormat::RG8S>,
+        MortonCopy<false, PixelFormat::RG32UI>,
+        MortonCopy<false, PixelFormat::R32UI>,
+        nullptr,
+        nullptr,
+        nullptr,
+        MortonCopy<false, PixelFormat::BGRA8_SRGB>,
+        MortonCopy<false, PixelFormat::DXT1_SRGB>,
+        MortonCopy<false, PixelFormat::DXT23_SRGB>,
+        MortonCopy<false, PixelFormat::DXT45_SRGB>,
+        MortonCopy<false, PixelFormat::BC7U_SRGB>,
+        nullptr,
+        nullptr,
+        nullptr,
+        nullptr,
+        nullptr,
+        nullptr,
+        nullptr,
+        nullptr,
+        MortonCopy<false, PixelFormat::Z32F>,
+        MortonCopy<false, PixelFormat::Z16>,
+        MortonCopy<false, PixelFormat::Z24S8>,
+        MortonCopy<false, PixelFormat::S8Z24>,
+        MortonCopy<false, PixelFormat::Z32FS8>,
+    // clang-format on
+};
+
+void SwizzleFunc(const GLConversionArray& functions, const SurfaceParams& params,
                  std::vector<u8>& gl_buffer, u32 mip_level) {
     u32 depth = params.MipDepth(mip_level);
     if (params.target == SurfaceTarget::Texture2D) {
@@ -385,19 +550,19 @@ void SwizzleFunc(const MortonSwizzleMode& mode, const SurfaceParams& params,
         const u64 layer_size = params.LayerMemorySize();
         const u64 gl_size = params.LayerSizeGL(mip_level);
         for (u32 i = 0; i < params.depth; i++) {
-            MortonSwizzle(mode, params.pixel_format, params.MipWidth(mip_level),
-                          params.MipBlockHeight(mip_level), params.MipHeight(mip_level),
-                          params.MipBlockDepth(mip_level), 1, gl_buffer.data() + offset_gl, gl_size,
-                          params.addr + offset);
+            functions[static_cast<std::size_t>(params.pixel_format)](
+                params.MipWidth(mip_level), params.MipBlockHeight(mip_level),
+                params.MipHeight(mip_level), params.MipBlockDepth(mip_level), 1,
+                gl_buffer.data() + offset_gl, gl_size, params.addr + offset);
             offset += layer_size;
             offset_gl += gl_size;
         }
     } else {
         const u64 offset = params.GetMipmapLevelOffset(mip_level);
-        MortonSwizzle(mode, params.pixel_format, params.MipWidth(mip_level),
-                      params.MipBlockHeight(mip_level), params.MipHeight(mip_level),
-                      params.MipBlockDepth(mip_level), depth, gl_buffer.data(), gl_buffer.size(),
-                      params.addr + offset);
+        functions[static_cast<std::size_t>(params.pixel_format)](
+            params.MipWidth(mip_level), params.MipBlockHeight(mip_level),
+            params.MipHeight(mip_level), params.MipBlockDepth(mip_level), depth, gl_buffer.data(),
+            gl_buffer.size(), params.addr + offset);
     }
 }
 
@@ -831,7 +996,7 @@ void CachedSurface::LoadGLBuffer() {
         ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}",
                    params.block_width, static_cast<u32>(params.target));
         for (u32 i = 0; i < params.max_mip_level; i++)
-            SwizzleFunc(MortonSwizzleMode::MortonToLinear, params, gl_buffer[i], i);
+            SwizzleFunc(morton_to_gl_fns, params, gl_buffer[i], i);
     } else {
         const auto texture_src_data{Memory::GetPointer(params.addr)};
         const auto texture_src_data_end{texture_src_data + params.size_in_bytes_gl};
@@ -870,7 +1035,7 @@ void CachedSurface::FlushGLBuffer() {
         ASSERT_MSG(params.block_width == 1, "Block width is defined as {} on texture type {}",
                    params.block_width, static_cast<u32>(params.target));
 
-        SwizzleFunc(MortonSwizzleMode::LinearToMorton, params, gl_buffer[0], 0);
+        SwizzleFunc(gl_to_morton_fns, params, gl_buffer[0], 0);
     } else {
         std::memcpy(Memory::GetPointer(GetAddr()), gl_buffer[0].data(), GetSizeInBytes());
     }

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -98,6 +98,18 @@ CachedShader::CachedShader(VAddr addr, Maxwell::ShaderProgram program_type)
     }
 }
 
+GLuint CachedShader::GetProgramResourceIndex(const GLShader::ConstBufferEntry& buffer) {
+    const auto search{resource_cache.find(buffer.GetHash())};
+    if (search == resource_cache.end()) {
+        const GLuint index{
+            glGetProgramResourceIndex(program.handle, GL_UNIFORM_BLOCK, buffer.GetName().c_str())};
+        resource_cache[buffer.GetHash()] = index;
+        return index;
+    }
+
+    return search->second;
+}
+
 GLint CachedShader::GetUniformLocation(const GLShader::SamplerEntry& sampler) {
     const auto search{uniform_cache.find(sampler.GetHash())};
     if (search == uniform_cache.end()) {

src/video_core/renderer_opengl/gl_shader_cache.h

@@ -71,18 +71,7 @@ public:
     }
 
     /// Gets the GL program resource location for the specified resource, caching as needed
-    template <typename T>
-    GLuint GetProgramResourceIndex(const T& buffer) {
-        const auto& search{resource_cache.find(buffer.GetHash())};
-        if (search == resource_cache.end()) {
-            const GLuint index{glGetProgramResourceIndex(program.handle, GL_UNIFORM_BLOCK,
-                                                         buffer.GetName().c_str())};
-            resource_cache[buffer.GetHash()] = index;
-            return index;
-        }
-
-        return search->second;
-    }
+    GLuint GetProgramResourceIndex(const GLShader::ConstBufferEntry& buffer);
 
     /// Gets the GL uniform location for the specified resource, caching as needed
     GLint GetUniformLocation(const GLShader::SamplerEntry& sampler);

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -13,7 +13,6 @@
 
 #include "common/assert.h"
 #include "common/common_types.h"
-#include "core/core.h"
 #include "video_core/engines/shader_bytecode.h"
 #include "video_core/engines/shader_header.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
@@ -501,42 +500,27 @@ public:
                                       const Register& buf_reg) {
         const std::string dest = GetOutputAttribute(attribute);
         const std::string src = GetRegisterAsFloat(val_reg);
-        if (dest.empty())
-            return;
 
-        // Can happen with unknown/unimplemented output attributes, in which case we ignore the
-        // instruction for now.
-        if (stage == Maxwell3D::Regs::ShaderStage::Geometry) {
-            // TODO(Rodrigo): nouveau sets some attributes after setting emitting a geometry
-            // shader. These instructions use a dirty register as buffer index, to avoid some
-            // drivers from complaining about out of boundary writes, guard them.
-            const std::string buf_index{"((" + GetRegisterAsInteger(buf_reg) + ") % " +
-                                        std::to_string(MAX_GEOMETRY_BUFFERS) + ')'};
-            shader.AddLine("amem[" + buf_index + "][" +
-                           std::to_string(static_cast<u32>(attribute)) + ']' + GetSwizzle(elem) +
-                           " = " + src + ';');
-            return;
-        }
-
-        switch (attribute) {
-        case Attribute::Index::ClipDistances0123:
-        case Attribute::Index::ClipDistances4567: {
-            const u64 index = attribute == Attribute::Index::ClipDistances4567 ? 4 : 0 + elem;
-            UNIMPLEMENTED_IF_MSG(
-                ((header.vtg.clip_distances >> index) & 1) == 0,
-                "Shader is setting gl_ClipDistance{} without enabling it in the header", index);
-
-            fixed_pipeline_output_attributes_used.insert(attribute);
-            shader.AddLine(dest + '[' + std::to_string(index) + "] = " + src + ';');
-            break;
-        }
-        case Attribute::Index::PointSize:
-            fixed_pipeline_output_attributes_used.insert(attribute);
-            shader.AddLine(dest + " = " + src + ';');
-            break;
-        default:
-            shader.AddLine(dest + GetSwizzle(elem) + " = " + src + ';');
-            break;
+        if (!dest.empty()) {
+            // Can happen with unknown/unimplemented output attributes, in which case we ignore the
+            // instruction for now.
+            if (stage == Maxwell3D::Regs::ShaderStage::Geometry) {
+                // TODO(Rodrigo): nouveau sets some attributes after setting emitting a geometry
+                // shader. These instructions use a dirty register as buffer index, to avoid some
+                // drivers from complaining about out of boundary writes, guard them.
+                const std::string buf_index{"((" + GetRegisterAsInteger(buf_reg) + ") % " +
+                                            std::to_string(MAX_GEOMETRY_BUFFERS) + ')'};
+                shader.AddLine("amem[" + buf_index + "][" +
+                               std::to_string(static_cast<u32>(attribute)) + ']' +
+                               GetSwizzle(elem) + " = " + src + ';');
+            } else {
+                if (attribute == Attribute::Index::PointSize) {
+                    fixed_pipeline_output_attributes_used.insert(attribute);
+                    shader.AddLine(dest + " = " + src + ';');
+                } else {
+                    shader.AddLine(dest + GetSwizzle(elem) + " = " + src + ';');
+                }
+            }
         }
     }
 
@@ -586,7 +570,6 @@ public:
         GenerateInputAttrs();
         GenerateOutputAttrs();
         GenerateConstBuffers();
-        GenerateGlobalRegions();
         GenerateSamplers();
         GenerateGeometry();
     }
@@ -708,21 +691,6 @@ private:
         declarations.AddNewLine();
     }
 
-    /// Generates declarations for global memory regions.
-    void GenerateGlobalRegions() {
-        const auto& regions{
-            Core::System::GetInstance().GPU().Maxwell3D().state.global_memory_uniforms};
-        for (std::size_t i = 0; i < regions.size(); ++i) {
-            declarations.AddLine("layout(std140) uniform " +
-                                 fmt::format("global_memory_region_declblock_{}", i));
-            declarations.AddLine('{');
-            declarations.AddLine("    vec4 global_memory_region_" + std::to_string(i) + "[0x400];");
-            declarations.AddLine("};");
-            declarations.AddNewLine();
-        }
-        declarations.AddNewLine();
-    }
-
     /// Generates declarations for samplers.
     void GenerateSamplers() {
         const auto& samplers = GetSamplers();
@@ -772,19 +740,12 @@ private:
     void GenerateVertex() {
         if (stage != Maxwell3D::Regs::ShaderStage::Vertex)
             return;
-        bool clip_distances_declared = false;
-
         declarations.AddLine("out gl_PerVertex {");
         ++declarations.scope;
         declarations.AddLine("vec4 gl_Position;");
         for (auto& o : fixed_pipeline_output_attributes_used) {
             if (o == Attribute::Index::PointSize)
                 declarations.AddLine("float gl_PointSize;");
-            if (!clip_distances_declared && (o == Attribute::Index::ClipDistances0123 ||
-                                             o == Attribute::Index::ClipDistances4567)) {
-                declarations.AddLine("float gl_ClipDistance[];");
-                clip_distances_declared = true;
-            }
         }
         --declarations.scope;
         declarations.AddLine("};");
@@ -955,10 +916,6 @@ private:
             return "gl_PointSize";
         case Attribute::Index::Position:
             return "position";
-        case Attribute::Index::ClipDistances0123:
-        case Attribute::Index::ClipDistances4567: {
-            return "gl_ClipDistance";
-        }
         default:
             const u32 index{static_cast<u32>(attribute) -
                             static_cast<u32>(Attribute::Index::Attribute_0)};
@@ -1309,15 +1266,7 @@ private:
         regs.SetRegisterToInteger(dest, true, 0, result, 1, 1);
     }
 
-    void WriteTexsInstruction(const Instruction& instr, const std::string& coord,
-                              const std::string& texture) {
-        // Add an extra scope and declare the texture coords inside to prevent
-        // overwriting them in case they are used as outputs of the texs instruction.
-        shader.AddLine('{');
-        ++shader.scope;
-        shader.AddLine(coord);
-        shader.AddLine("vec4 texture_tmp = " + texture + ';');
-
+    void WriteTexsInstruction(const Instruction& instr, const std::string& texture) {
         // TEXS has two destination registers and a swizzle. The first two elements in the swizzle
         // go into gpr0+0 and gpr0+1, and the rest goes into gpr28+0 and gpr28+1
 
@@ -1329,19 +1278,17 @@ private:
 
             if (written_components < 2) {
                 // Write the first two swizzle components to gpr0 and gpr0+1
-                regs.SetRegisterToFloat(instr.gpr0, component, "texture_tmp", 1, 4, false,
+                regs.SetRegisterToFloat(instr.gpr0, component, texture, 1, 4, false,
                                         written_components % 2);
             } else {
                 ASSERT(instr.texs.HasTwoDestinations());
                 // Write the rest of the swizzle components to gpr28 and gpr28+1
-                regs.SetRegisterToFloat(instr.gpr28, component, "texture_tmp", 1, 4, false,
+                regs.SetRegisterToFloat(instr.gpr28, component, texture, 1, 4, false,
                                         written_components % 2);
             }
 
             ++written_components;
         }
-        --shader.scope;
-        shader.AddLine('}');
     }
 
     static u32 TextureCoordinates(Tegra::Shader::TextureType texture_type) {
@@ -1738,26 +1685,6 @@ private:
 
             break;
         }
-        case OpCode::Type::Bfi: {
-            UNIMPLEMENTED_IF(instr.generates_cc);
-
-            const auto [base, packed_shift] = [&]() -> std::tuple<std::string, std::string> {
-                switch (opcode->get().GetId()) {
-                case OpCode::Id::BFI_IMM_R:
-                    return {regs.GetRegisterAsInteger(instr.gpr39, 0, false),
-                            std::to_string(instr.alu.GetSignedImm20_20())};
-                default:
-                    UNREACHABLE();
-                }
-            }();
-            const std::string offset = '(' + packed_shift + " & 0xff)";
-            const std::string bits = "((" + packed_shift + " >> 8) & 0xff)";
-            const std::string insert = regs.GetRegisterAsInteger(instr.gpr8, 0, false);
-            regs.SetRegisterToInteger(
-                instr.gpr0, false, 0,
-                "bitfieldInsert(" + base + ", " + insert + ", " + offset + ", " + bits + ')', 1, 1);
-            break;
-        }
         case OpCode::Type::Shift: {
             std::string op_a = regs.GetRegisterAsInteger(instr.gpr8, 0, true);
             std::string op_b;
@@ -1851,11 +1778,6 @@ private:
                 } else {
                     op_b += regs.GetUniform(instr.cbuf34.index, instr.cbuf34.offset,
                                             GLSLRegister::Type::Integer);
-                    if (opcode->get().GetId() == OpCode::Id::IADD_C) {
-                        s_last_iadd = last_iadd;
-                        last_iadd = IADDReference{instr.gpr8.Value(), instr.cbuf34.index,
-                                                  instr.cbuf34.offset};
-                    }
                 }
             }
 
@@ -2588,83 +2510,61 @@ private:
                 const bool depth_compare =
                     instr.tex.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC);
                 u32 num_coordinates = TextureCoordinates(texture_type);
-                u32 start_index = 0;
-                std::string array_elem;
-                if (is_array) {
-                    array_elem = regs.GetRegisterAsInteger(instr.gpr8);
-                    start_index = 1;
-                }
-                const auto process_mode = instr.tex.GetTextureProcessMode();
-                u32 start_index_b = 0;
-                std::string lod_value;
-                if (process_mode != Tegra::Shader::TextureProcessMode::LZ &&
-                    process_mode != Tegra::Shader::TextureProcessMode::None) {
-                    start_index_b = 1;
-                    lod_value = regs.GetRegisterAsFloat(instr.gpr20);
-                }
-
-                std::string depth_value;
-                if (depth_compare) {
-                    depth_value = regs.GetRegisterAsFloat(instr.gpr20.Value() + start_index_b);
-                }
-
-                bool depth_compare_extra = false;
+                if (depth_compare)
+                    num_coordinates += 1;
 
                 switch (num_coordinates) {
                 case 1: {
-                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index);
                     if (is_array) {
-                        if (depth_compare) {
-                            coord = "vec3 coords = vec3(" + x + ", " + depth_value + ", " +
-                                    array_elem + ");";
-                        } else {
-                            coord = "vec2 coords = vec2(" + x + ", " + array_elem + ");";
-                        }
+                        const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        coord = "vec2 coords = vec2(" + x + ", " + index + ");";
                     } else {
-                        if (depth_compare) {
-                            coord = "vec2 coords = vec2(" + x + ", " + depth_value + ");";
-                        } else {
-                            coord = "float coords = " + x + ';';
-                        }
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                        coord = "float coords = " + x + ';';
                     }
                     break;
                 }
                 case 2: {
-                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index);
-                    const std::string y =
-                        regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index + 1);
                     if (is_array) {
-                        if (depth_compare) {
-                            coord = "vec4 coords = vec4(" + x + ", " + y + ", " + depth_value +
-                                    ", " + array_elem + ");";
-                        } else {
-                            coord = "vec3 coords = vec3(" + x + ", " + y + ", " + array_elem + ");";
-                        }
+                        const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
+                        coord = "vec3 coords = vec3(" + x + ", " + y + ", " + index + ");";
                     } else {
-                        if (depth_compare) {
-                            coord =
-                                "vec3 coords = vec3(" + x + ", " + y + ", " + depth_value + ");";
-                        } else {
-                            coord = "vec2 coords = vec2(" + x + ", " + y + ");";
-                        }
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        coord = "vec2 coords = vec2(" + x + ", " + y + ");";
                     }
                     break;
                 }
                 case 3: {
-                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index);
-                    const std::string y =
-                        regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index + 1);
-                    const std::string z =
-                        regs.GetRegisterAsFloat(instr.gpr8.Value() + start_index + 2);
-                    if (is_array) {
-                        depth_compare_extra = depth_compare;
-                        coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " +
-                                array_elem + ");";
-                    } else {
-                        if (depth_compare) {
-                            coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " +
-                                    depth_value + ");";
+                    if (depth_compare) {
+                        if (is_array) {
+                            const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                            const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                            const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
+                            const std::string z = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
+                            coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + index +
+                                    ");";
                         } else {
+                            const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                            const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                            const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
+                            coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
+                        }
+                    } else {
+                        if (is_array) {
+                            const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                            const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                            const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
+                            const std::string z = regs.GetRegisterAsFloat(instr.gpr8.Value() + 3);
+                            coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + index +
+                                    ");";
+                        } else {
+                            const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                            const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                            const std::string z = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
                             coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
                         }
                     }
@@ -2680,85 +2580,82 @@ private:
                     coord = "vec2 coords = vec2(" + x + ", " + y + ");";
                     texture_type = Tegra::Shader::TextureType::Texture2D;
                 }
+                // TODO: make sure coordinates are always indexed to gpr8 and gpr20 is always bias
+                // or lod.
 
                 const std::string sampler =
                     GetSampler(instr.sampler, texture_type, is_array, depth_compare);
                 // Add an extra scope and declare the texture coords inside to prevent
                 // overwriting them in case they are used as outputs of the texs instruction.
 
-                shader.AddLine('{');
+                shader.AddLine("{");
                 ++shader.scope;
                 shader.AddLine(coord);
                 std::string texture;
 
                 switch (instr.tex.GetTextureProcessMode()) {
                 case Tegra::Shader::TextureProcessMode::None: {
-                    if (!depth_compare_extra) {
-                        texture = "texture(" + sampler + ", coords)";
-                    } else {
-                        texture = "texture(" + sampler + ", coords, " + depth_value + ')';
-                    }
+                    texture = "texture(" + sampler + ", coords)";
                     break;
                 }
                 case Tegra::Shader::TextureProcessMode::LZ: {
-                    if (!depth_compare_extra) {
-                        texture = "textureLod(" + sampler + ", coords, 0.0)";
-                    } else {
-                        texture = "texture(" + sampler + ", coords, " + depth_value + ')';
-                    }
+                    texture = "textureLod(" + sampler + ", coords, 0.0)";
                     break;
                 }
                 case Tegra::Shader::TextureProcessMode::LB:
                 case Tegra::Shader::TextureProcessMode::LBA: {
+                    const std::string bias = [&]() {
+                        if (depth_compare) {
+                            if (is_array)
+                                return regs.GetRegisterAsFloat(instr.gpr20.Value() + 2);
+                            else
+                                return regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
+                        } else {
+                            return regs.GetRegisterAsFloat(instr.gpr20);
+                        }
+                    }();
+                    shader.AddLine("float bias = " + bias + ';');
+
                     // TODO: Figure if A suffix changes the equation at all.
-                    if (!depth_compare_extra) {
-                        texture = "texture(" + sampler + ", coords, " + lod_value + ')';
-                    } else {
-                        texture = "texture(" + sampler + ", coords, " + depth_value + ')';
-                        LOG_WARNING(HW_GPU,
-                                    "OpenGL Limitation: can't set bias value along depth compare");
-                    }
+                    texture = "texture(" + sampler + ", coords, bias)";
                     break;
                 }
                 case Tegra::Shader::TextureProcessMode::LL:
                 case Tegra::Shader::TextureProcessMode::LLA: {
+                    const std::string lod = [&]() {
+                        if (num_coordinates <= 2) {
+                            return regs.GetRegisterAsFloat(instr.gpr20);
+                        } else {
+                            return regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
+                        }
+                    }();
+                    shader.AddLine("float lod = " + lod + ';');
+
                     // TODO: Figure if A suffix changes the equation at all.
-                    if (!depth_compare_extra) {
-                        texture = "textureLod(" + sampler + ", coords, " + lod_value + ')';
-                    } else {
-                        texture = "texture(" + sampler + ", coords, " + depth_value + ')';
-                        LOG_WARNING(HW_GPU,
-                                    "OpenGL Limitation: can't set lod value along depth compare");
-                    }
+                    texture = "textureLod(" + sampler + ", coords, lod)";
                     break;
                 }
                 default: {
-                    if (!depth_compare_extra) {
-                        texture = "texture(" + sampler + ", coords)";
-                    } else {
-                        texture = "texture(" + sampler + ", coords, " + depth_value + ')';
-                    }
+                    texture = "texture(" + sampler + ", coords)";
                     UNIMPLEMENTED_MSG("Unhandled texture process mode {}",
                                       static_cast<u32>(instr.tex.GetTextureProcessMode()));
                 }
                 }
                 if (!depth_compare) {
-                    shader.AddLine("vec4 texture_tmp = " + texture + ';');
                     std::size_t dest_elem{};
                     for (std::size_t elem = 0; elem < 4; ++elem) {
                         if (!instr.tex.IsComponentEnabled(elem)) {
                             // Skip disabled components
                             continue;
                         }
-                        regs.SetRegisterToFloat(instr.gpr0, elem, "texture_tmp", 1, 4, false,
-                                                dest_elem);
+                        regs.SetRegisterToFloat(instr.gpr0, elem, texture, 1, 4, false, dest_elem);
                         ++dest_elem;
                     }
                 } else {
                     regs.SetRegisterToFloat(instr.gpr0, 0, texture, 1, 1, false);
                 }
                 --shader.scope;
-                shader.AddLine('}');
+                shader.AddLine("}");
                 break;
             }
             case OpCode::Id::TEXS: {
@@ -2771,76 +2668,41 @@ private:
                 const bool depth_compare =
                     instr.texs.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC);
                 u32 num_coordinates = TextureCoordinates(texture_type);
-                const auto process_mode = instr.texs.GetTextureProcessMode();
-                std::string lod_value;
-                std::string coord;
-                u32 lod_offset = 0;
-                if (process_mode == Tegra::Shader::TextureProcessMode::LL) {
-                    if (num_coordinates > 2) {
-                        lod_value = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
-                        lod_offset = 2;
-                    } else {
-                        lod_value = regs.GetRegisterAsFloat(instr.gpr20);
-                        lod_offset = 1;
-                    }
-                }
+                if (depth_compare)
+                    num_coordinates += 1;
+
+                // Scope to avoid variable name overlaps.
+                shader.AddLine('{');
+                ++shader.scope;
 
                 switch (num_coordinates) {
-                case 1: {
-                    coord = "float coords = " + regs.GetRegisterAsFloat(instr.gpr8) + ';';
-                    break;
-                }
                 case 2: {
                     if (is_array) {
-                        if (depth_compare) {
-                            const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
-                            const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                            const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
-                            const std::string z = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
-                            coord = "vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " + index +
-                                    ");";
-                        } else {
-                            const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
-                            const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                            const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
-                            coord = "vec3 coords = vec3(" + x + ", " + y + ", " + index + ");";
-                        }
+                        const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
+                        shader.AddLine("vec3 coords = vec3(" + x + ", " + y + ", " + index + ");");
                     } else {
-                        if (lod_offset != 0) {
-                            if (depth_compare) {
-                                const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                                const std::string y =
-                                    regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                                const std::string z =
-                                    regs.GetRegisterAsFloat(instr.gpr20.Value() + lod_offset);
-                                coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
-                            } else {
-                                const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                                const std::string y =
-                                    regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                                coord = "vec2 coords = vec2(" + x + ", " + y + ");";
-                            }
-                        } else {
-                            if (depth_compare) {
-                                const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                                const std::string y =
-                                    regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                                const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
-                                coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
-                            } else {
-                                const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                                const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
-                                coord = "vec2 coords = vec2(" + x + ", " + y + ");";
-                            }
-                        }
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
+                        shader.AddLine("vec2 coords = vec2(" + x + ", " + y + ");");
                     }
                     break;
                 }
                 case 3: {
-                    const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
-                    const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                    const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
-                    coord = "vec3 coords = vec3(" + x + ", " + y + ", " + z + ");";
+                    if (is_array) {
+                        const std::string index = regs.GetRegisterAsInteger(instr.gpr8);
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 2);
+                        const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
+                        shader.AddLine("vec4 coords = vec4(" + x + ", " + y + ", " + z + ", " +
+                                       index + ");");
+                    } else {
+                        const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
+                        const std::string y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
+                        const std::string z = regs.GetRegisterAsFloat(instr.gpr20);
+                        shader.AddLine("vec3 coords = vec3(" + x + ", " + y + ", " + z + ");");
+                    }
                     break;
                 }
                 default:
@@ -2850,14 +2712,14 @@ private:
                     // Fallback to interpreting as a 2D texture for now
                     const std::string x = regs.GetRegisterAsFloat(instr.gpr8);
                     const std::string y = regs.GetRegisterAsFloat(instr.gpr20);
-                    coord = "vec2 coords = vec2(" + x + ", " + y + ");";
+                    shader.AddLine("vec2 coords = vec2(" + x + ", " + y + ");");
                     texture_type = Tegra::Shader::TextureType::Texture2D;
                     is_array = false;
                 }
                 const std::string sampler =
                     GetSampler(instr.sampler, texture_type, is_array, depth_compare);
                 std::string texture;
-                switch (process_mode) {
+                switch (instr.texs.GetTextureProcessMode()) {
                 case Tegra::Shader::TextureProcessMode::None: {
                     texture = "texture(" + sampler + ", coords)";
                     break;
@@ -2871,7 +2733,8 @@ private:
                     break;
                 }
                 case Tegra::Shader::TextureProcessMode::LL: {
-                    texture = "textureLod(" + sampler + ", coords, " + lod_value + ')';
+                    const std::string op_c = regs.GetRegisterAsFloat(instr.gpr20.Value() + 1);
+                    texture = "textureLod(" + sampler + ", coords, " + op_c + ')';
                     break;
                 }
                 default: {
@@ -2881,11 +2744,13 @@ private:
                 }
                 }
                 if (!depth_compare) {
-                    WriteTexsInstruction(instr, coord, texture);
+                    WriteTexsInstruction(instr, texture);
                 } else {
-                    WriteTexsInstruction(instr, coord, "vec4(" + texture + ')');
+                    WriteTexsInstruction(instr, "vec4(" + texture + ')');
                 }
 
+                shader.AddLine('}');
+                --shader.scope;
                 break;
             }
             case OpCode::Id::TLDS: {
@@ -2907,12 +2772,11 @@ private:
                 // Scope to avoid variable name overlaps.
                 shader.AddLine('{');
                 ++shader.scope;
-                std::string coords;
 
                 switch (texture_type) {
                 case Tegra::Shader::TextureType::Texture1D: {
                     const std::string x = regs.GetRegisterAsInteger(instr.gpr8);
-                    coords = "float coords = " + x + ';';
+                    shader.AddLine("int coords = " + x + ';');
                     break;
                 }
                 case Tegra::Shader::TextureType::Texture2D: {
@@ -2920,8 +2784,7 @@ private:
 
                     const std::string x = regs.GetRegisterAsInteger(instr.gpr8);
                     const std::string y = regs.GetRegisterAsInteger(instr.gpr20);
-                    // shader.AddLine("ivec2 coords = ivec2(" + x + ", " + y + ");");
-                    coords = "ivec2 coords = ivec2(" + x + ", " + y + ");";
+                    shader.AddLine("ivec2 coords = ivec2(" + x + ", " + y + ");");
                     extra_op_offset = 1;
                     break;
                 }
@@ -2949,7 +2812,7 @@ private:
                                       static_cast<u32>(instr.tlds.GetTextureProcessMode()));
                 }
                 }
-                WriteTexsInstruction(instr, coords, texture);
+                WriteTexsInstruction(instr, texture);
 
                 --shader.scope;
                 shader.AddLine('}');
@@ -3008,17 +2871,14 @@ private:
 
                 const std::string texture = "textureGather(" + sampler + ", coords, " +
                                             std::to_string(instr.tld4.component) + ')';
-
                 if (!depth_compare) {
-                    shader.AddLine("vec4 texture_tmp = " + texture + ';');
                     std::size_t dest_elem{};
                     for (std::size_t elem = 0; elem < 4; ++elem) {
                         if (!instr.tex.IsComponentEnabled(elem)) {
                             // Skip disabled components
                             continue;
                         }
-                        regs.SetRegisterToFloat(instr.gpr0, elem, "texture_tmp", 1, 4, false,
-                                                dest_elem);
+                        regs.SetRegisterToFloat(instr.gpr0, elem, texture, 1, 4, false, dest_elem);
                         ++dest_elem;
                     }
                 } else {
@@ -3039,7 +2899,6 @@ private:
                 // Scope to avoid variable name overlaps.
                 shader.AddLine('{');
                 ++shader.scope;
-                std::string coords;
 
                 const bool depth_compare =
                     instr.tld4s.UsesMiscMode(Tegra::Shader::TextureMiscMode::DC);
@@ -3049,19 +2908,20 @@ private:
                 const std::string sampler = GetSampler(
                     instr.sampler, Tegra::Shader::TextureType::Texture2D, false, depth_compare);
                 if (!depth_compare) {
-                    coords = "vec2 coords = vec2(" + op_a + ", " + op_b + ");";
+                    shader.AddLine("vec2 coords = vec2(" + op_a + ", " + op_b + ");");
                 } else {
                     // Note: TLD4S coordinate encoding works just like TEXS's
-                    const std::string op_y = regs.GetRegisterAsFloat(instr.gpr8.Value() + 1);
-                    coords = "vec3 coords = vec3(" + op_a + ", " + op_y + ", " + op_b + ");";
+                    shader.AddLine(
+                        "float op_y = " + regs.GetRegisterAsFloat(instr.gpr8.Value() + 1) + ';');
+                    shader.AddLine("vec3 coords = vec3(" + op_a + ", op_y, " + op_b + ");");
                 }
                 const std::string texture = "textureGather(" + sampler + ", coords, " +
                                             std::to_string(instr.tld4s.component) + ')';
 
                 if (!depth_compare) {
-                    WriteTexsInstruction(instr, coords, texture);
+                    WriteTexsInstruction(instr, texture);
                 } else {
-                    WriteTexsInstruction(instr, coords, "vec4(" + texture + ')');
+                    WriteTexsInstruction(instr, "vec4(" + texture + ')');
                 }
 
                 --shader.scope;
@@ -3148,72 +3008,6 @@ private:
                 shader.AddLine('}');
                 break;
             }
-            case OpCode::Id::LDG: {
-                // Determine number of GPRs to fill with data
-                u64 count = 1;
-
-                switch (instr.ld_g.type) {
-                case Tegra::Shader::UniformType::Single:
-                    count = 1;
-                    break;
-                case Tegra::Shader::UniformType::Double:
-                    count = 2;
-                    break;
-                case Tegra::Shader::UniformType::Quad:
-                case Tegra::Shader::UniformType::UnsignedQuad:
-                    count = 4;
-                    break;
-                default:
-                    UNREACHABLE_MSG("Unimplemented LDG size!");
-                }
-
-                auto [gpr_index, index, offset] = last_iadd;
-
-                // The last IADD might be the upper u32 of address, so instead take the one before
-                // that.
-                if (gpr_index == Register::ZeroIndex) {
-                    gpr_index = s_last_iadd.out;
-                    index = s_last_iadd.cbuf_index;
-                    offset = s_last_iadd.cbuf_offset;
-                }
-
-                const auto gpr = regs.GetRegisterAsInteger(gpr_index);
-                const auto constbuffer =
-                    regs.GetUniform(index, offset, GLSLRegister::Type::UnsignedInteger);
-
-                Core::System::GetInstance().GPU().Maxwell3D().state.global_memory_uniforms.insert(
-                    {index, offset * 4});
-                const auto memory = fmt::format("global_memory_region_{}",
-                                                Core::System::GetInstance()
-                                                        .GPU()
-                                                        .Maxwell3D()
-                                                        .state.global_memory_uniforms.size() -
-                                                    1);
-
-                const auto immediate = std::to_string(instr.ld_g.offset_immediate.Value());
-                const auto o_register = regs.GetRegisterAsInteger(instr.gpr8, 0, false);
-                const auto address = "( " + immediate + " + " + o_register + " )";
-                const auto base_sub = address + " - " + constbuffer;
-
-                // New scope to prevent potential conflicts
-                shader.AddLine('{');
-                ++shader.scope;
-
-                shader.AddLine("uint final_offset = " + base_sub + ";");
-                for (std::size_t out = 0; out < count; ++out) {
-                    const u64 reg_id = instr.gpr0.Value() + out;
-                    const auto this_memory =
-                        fmt::format("{}[(final_offset + {}) / 16][((final_offset + {}) / 4) % 4]",
-                                    memory, out * 4, out * 4);
-
-                    regs.SetRegisterToFloat(reg_id, 0, this_memory, 1, 1);
-                }
-
-                --shader.scope;
-                shader.AddLine('}');
-
-                break;
-            }
             default: {
                 UNIMPLEMENTED_MSG("Unhandled memory instruction: {}", opcode->get().GetName());
             }
@@ -3423,34 +3217,6 @@ private:
             }
             break;
         }
-        case OpCode::Type::RegisterSetPredicate: {
-            UNIMPLEMENTED_IF(instr.r2p.mode != Tegra::Shader::R2pMode::Pr);
-
-            const std::string apply_mask = [&]() {
-                switch (opcode->get().GetId()) {
-                case OpCode::Id::R2P_IMM:
-                    return std::to_string(instr.r2p.immediate_mask);
-                default:
-                    UNREACHABLE();
-                }
-            }();
-            const std::string mask = '(' + regs.GetRegisterAsInteger(instr.gpr8, 0, false) +
-                                     " >> " + std::to_string(instr.r2p.byte) + ')';
-
-            constexpr u64 programmable_preds = 7;
-            for (u64 pred = 0; pred < programmable_preds; ++pred) {
-                const auto shift = std::to_string(1 << pred);
-
-                shader.AddLine("if ((" + apply_mask + " & " + shift + ") != 0) {");
-                ++shader.scope;
-
-                SetPredicate(pred, '(' + mask + " & " + shift + ") != 0");
-
-                --shader.scope;
-                shader.AddLine('}');
-            }
-            break;
-        }
         case OpCode::Type::FloatSet: {
             const std::string op_a = GetOperandAbsNeg(regs.GetRegisterAsFloat(instr.gpr8),
                                                       instr.fset.abs_a != 0, instr.fset.neg_a != 0);
@@ -4011,18 +3777,9 @@ private:
     ShaderWriter declarations;
     GLSLRegisterManager regs{shader, declarations, stage, suffix, header};
 
-    struct IADDReference {
-        Register out;
-        u64 cbuf_index;
-        u64 cbuf_offset;
-    };
-
-    IADDReference last_iadd{};
-    IADDReference s_last_iadd{};
-
     // Declarations
     std::set<std::string> declr_predicates;
-};
+}; // namespace OpenGL::GLShader::Decompiler
 
 std::string GetCommonDeclarations() {
     return fmt::format("#define MAX_CONSTBUFFER_ELEMENTS {}\n",

src/video_core/renderer_opengl/gl_shader_gen.h

@@ -57,8 +57,7 @@ public:
     }
 
     u32 GetHash() const {
-        // This needs to be unique from CachedGlobalRegionUniform::GetHash
-        return (static_cast<u32>(stage) << 12) | index;
+        return (static_cast<u32>(stage) << 16) | index;
     }
 
 private:
@@ -139,8 +138,7 @@ public:
     }
 
     u32 GetHash() const {
-        // This needs to be unique from CachedGlobalRegionUniform::GetHash
-        return (static_cast<u32>(stage) << 12) | static_cast<u32>(sampler_index);
+        return (static_cast<u32>(stage) << 16) | static_cast<u32>(sampler_index);
     }
 
     static std::string GetArrayName(Maxwell::ShaderStage stage) {

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -19,9 +19,9 @@
 #include "core/settings.h"
 #include "core/telemetry_session.h"
 #include "core/tracer/recorder.h"
-#include "video_core/morton.h"
 #include "video_core/renderer_opengl/gl_rasterizer.h"
 #include "video_core/renderer_opengl/renderer_opengl.h"
+#include "video_core/utils.h"
 
 namespace OpenGL {
 

src/video_core/utils.h

@@ -0,0 +1,164 @@
+// Copyright 2014 Citra Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace VideoCore {
+
+// 8x8 Z-Order coordinate from 2D coordinates
+static inline u32 MortonInterleave(u32 x, u32 y) {
+    static const u32 xlut[] = {0x00, 0x01, 0x04, 0x05, 0x10, 0x11, 0x14, 0x15};
+    static const u32 ylut[] = {0x00, 0x02, 0x08, 0x0a, 0x20, 0x22, 0x28, 0x2a};
+    return xlut[x % 8] + ylut[y % 8];
+}
+
+/**
+ * Calculates the offset of the position of the pixel in Morton order
+ */
+static inline u32 GetMortonOffset(u32 x, u32 y, u32 bytes_per_pixel) {
+    // Images are split into 8x8 tiles. Each tile is composed of four 4x4 subtiles each
+    // of which is composed of four 2x2 subtiles each of which is composed of four texels.
+    // Each structure is embedded into the next-bigger one in a diagonal pattern, e.g.
+    // texels are laid out in a 2x2 subtile like this:
+    // 2 3
+    // 0 1
+    //
+    // The full 8x8 tile has the texels arranged like this:
+    //
+    // 42 43 46 47 58 59 62 63
+    // 40 41 44 45 56 57 60 61
+    // 34 35 38 39 50 51 54 55
+    // 32 33 36 37 48 49 52 53
+    // 10 11 14 15 26 27 30 31
+    // 08 09 12 13 24 25 28 29
+    // 02 03 06 07 18 19 22 23
+    // 00 01 04 05 16 17 20 21
+    //
+    // This pattern is what's called Z-order curve, or Morton order.
+
+    const unsigned int block_height = 8;
+    const unsigned int coarse_x = x & ~7;
+
+    u32 i = VideoCore::MortonInterleave(x, y);
+
+    const unsigned int offset = coarse_x * block_height;
+
+    return (i + offset) * bytes_per_pixel;
+}
+
+static inline u32 MortonInterleave128(u32 x, u32 y) {
+    // 128x128 Z-Order coordinate from 2D coordinates
+    static constexpr u32 xlut[] = {
+        0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042,
+        0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809,
+        0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000,
+        0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043,
+        0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a,
+        0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001,
+        0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048,
+        0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b,
+        0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002,
+        0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049,
+        0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840,
+        0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003,
+        0x0008, 0x0009, 0x000a, 0x000b, 0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a,
+        0x004b, 0x0800, 0x0801, 0x0802, 0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841,
+        0x0842, 0x0843, 0x0848, 0x0849, 0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008,
+        0x1009, 0x100a, 0x100b, 0x1040, 0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b,
+        0x1800, 0x1801, 0x1802, 0x1803, 0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842,
+        0x1843, 0x1848, 0x1849, 0x184a, 0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009,
+        0x200a, 0x200b, 0x2040, 0x2041, 0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800,
+        0x2801, 0x2802, 0x2803, 0x2808, 0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843,
+        0x2848, 0x2849, 0x284a, 0x284b, 0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a,
+        0x300b, 0x3040, 0x3041, 0x3042, 0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801,
+        0x3802, 0x3803, 0x3808, 0x3809, 0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848,
+        0x3849, 0x384a, 0x384b, 0x0000, 0x0001, 0x0002, 0x0003, 0x0008, 0x0009, 0x000a, 0x000b,
+        0x0040, 0x0041, 0x0042, 0x0043, 0x0048, 0x0049, 0x004a, 0x004b, 0x0800, 0x0801, 0x0802,
+        0x0803, 0x0808, 0x0809, 0x080a, 0x080b, 0x0840, 0x0841, 0x0842, 0x0843, 0x0848, 0x0849,
+        0x084a, 0x084b, 0x1000, 0x1001, 0x1002, 0x1003, 0x1008, 0x1009, 0x100a, 0x100b, 0x1040,
+        0x1041, 0x1042, 0x1043, 0x1048, 0x1049, 0x104a, 0x104b, 0x1800, 0x1801, 0x1802, 0x1803,
+        0x1808, 0x1809, 0x180a, 0x180b, 0x1840, 0x1841, 0x1842, 0x1843, 0x1848, 0x1849, 0x184a,
+        0x184b, 0x2000, 0x2001, 0x2002, 0x2003, 0x2008, 0x2009, 0x200a, 0x200b, 0x2040, 0x2041,
+        0x2042, 0x2043, 0x2048, 0x2049, 0x204a, 0x204b, 0x2800, 0x2801, 0x2802, 0x2803, 0x2808,
+        0x2809, 0x280a, 0x280b, 0x2840, 0x2841, 0x2842, 0x2843, 0x2848, 0x2849, 0x284a, 0x284b,
+        0x3000, 0x3001, 0x3002, 0x3003, 0x3008, 0x3009, 0x300a, 0x300b, 0x3040, 0x3041, 0x3042,
+        0x3043, 0x3048, 0x3049, 0x304a, 0x304b, 0x3800, 0x3801, 0x3802, 0x3803, 0x3808, 0x3809,
+        0x380a, 0x380b, 0x3840, 0x3841, 0x3842, 0x3843, 0x3848, 0x3849, 0x384a, 0x384b,
+    };
+    static constexpr u32 ylut[] = {
+        0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090,
+        0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124,
+        0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200,
+        0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294,
+        0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330,
+        0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404,
+        0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0,
+        0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534,
+        0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610,
+        0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4,
+        0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780,
+        0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014,
+        0x0020, 0x0024, 0x0030, 0x0034, 0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0,
+        0x00b4, 0x0100, 0x0104, 0x0110, 0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184,
+        0x0190, 0x0194, 0x01a0, 0x01a4, 0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220,
+        0x0224, 0x0230, 0x0234, 0x0280, 0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4,
+        0x0300, 0x0304, 0x0310, 0x0314, 0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390,
+        0x0394, 0x03a0, 0x03a4, 0x03b0, 0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424,
+        0x0430, 0x0434, 0x0480, 0x0484, 0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500,
+        0x0504, 0x0510, 0x0514, 0x0520, 0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594,
+        0x05a0, 0x05a4, 0x05b0, 0x05b4, 0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630,
+        0x0634, 0x0680, 0x0684, 0x0690, 0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704,
+        0x0710, 0x0714, 0x0720, 0x0724, 0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0,
+        0x07a4, 0x07b0, 0x07b4, 0x0000, 0x0004, 0x0010, 0x0014, 0x0020, 0x0024, 0x0030, 0x0034,
+        0x0080, 0x0084, 0x0090, 0x0094, 0x00a0, 0x00a4, 0x00b0, 0x00b4, 0x0100, 0x0104, 0x0110,
+        0x0114, 0x0120, 0x0124, 0x0130, 0x0134, 0x0180, 0x0184, 0x0190, 0x0194, 0x01a0, 0x01a4,
+        0x01b0, 0x01b4, 0x0200, 0x0204, 0x0210, 0x0214, 0x0220, 0x0224, 0x0230, 0x0234, 0x0280,
+        0x0284, 0x0290, 0x0294, 0x02a0, 0x02a4, 0x02b0, 0x02b4, 0x0300, 0x0304, 0x0310, 0x0314,
+        0x0320, 0x0324, 0x0330, 0x0334, 0x0380, 0x0384, 0x0390, 0x0394, 0x03a0, 0x03a4, 0x03b0,
+        0x03b4, 0x0400, 0x0404, 0x0410, 0x0414, 0x0420, 0x0424, 0x0430, 0x0434, 0x0480, 0x0484,
+        0x0490, 0x0494, 0x04a0, 0x04a4, 0x04b0, 0x04b4, 0x0500, 0x0504, 0x0510, 0x0514, 0x0520,
+        0x0524, 0x0530, 0x0534, 0x0580, 0x0584, 0x0590, 0x0594, 0x05a0, 0x05a4, 0x05b0, 0x05b4,
+        0x0600, 0x0604, 0x0610, 0x0614, 0x0620, 0x0624, 0x0630, 0x0634, 0x0680, 0x0684, 0x0690,
+        0x0694, 0x06a0, 0x06a4, 0x06b0, 0x06b4, 0x0700, 0x0704, 0x0710, 0x0714, 0x0720, 0x0724,
+        0x0730, 0x0734, 0x0780, 0x0784, 0x0790, 0x0794, 0x07a0, 0x07a4, 0x07b0, 0x07b4,
+    };
+    return xlut[x % 128] + ylut[y % 128];
+}
+
+static inline u32 GetMortonOffset128(u32 x, u32 y, u32 bytes_per_pixel) {
+    // Calculates the offset of the position of the pixel in Morton order
+    // Framebuffer images are split into 128x128 tiles.
+
+    const unsigned int block_height = 128;
+    const unsigned int coarse_x = x & ~127;
+
+    u32 i = MortonInterleave128(x, y);
+
+    const unsigned int offset = coarse_x * block_height;
+
+    return (i + offset) * bytes_per_pixel;
+}
+
+static inline void MortonCopyPixels128(u32 width, u32 height, u32 bytes_per_pixel,
+                                       u32 gl_bytes_per_pixel, u8* morton_data, u8* gl_data,
+                                       bool morton_to_gl) {
+    u8* data_ptrs[2];
+    for (unsigned y = 0; y < height; ++y) {
+        for (unsigned x = 0; x < width; ++x) {
+            const u32 coarse_y = y & ~127;
+            u32 morton_offset =
+                GetMortonOffset128(x, y, bytes_per_pixel) + coarse_y * width * bytes_per_pixel;
+            u32 gl_pixel_index = (x + y * width) * gl_bytes_per_pixel;
+
+            data_ptrs[morton_to_gl] = morton_data + morton_offset;
+            data_ptrs[!morton_to_gl] = &gl_data[gl_pixel_index];
+
+            memcpy(data_ptrs[0], data_ptrs[1], bytes_per_pixel);
+        }
+    }
+}
+
+} // namespace VideoCore