arm_dynarmic_cp15: Implement CNTPCT

arm_dynarmic_cp15: Update CP15
arm_dynarmic_32: InterpreterFallback should never happen
2020-06-17 17:10:24 +01:00 · 2020-06-17 17:10:24 +01:00 · 2020-06-17 17:10:24 +01:00 · 2020-06-17 11:20:52 -04:00 · 2020-06-16 23:44:48 -04:00 · 2020-06-16 22:27:27 -04:00
58 changed files with 2756 additions and 729 deletions
@@ -13,6 +13,9 @@
 [submodule "soundtouch"]
    path = externals/soundtouch
    url = https://github.com/citra-emu/ext-soundtouch.git
+[submodule "libressl"]
+    path = externals/libressl
+    url = https://github.com/citra-emu/ext-libressl-portable.git
 [submodule "discord-rpc"]
    path = externals/discord-rpc
    url = https://github.com/discordapp/discord-rpc.git
@@ -152,7 +152,6 @@ macro(yuzu_find_packages)
        "Boost             1.71        boost/1.72.0"
        "Catch2            2.11        catch2/2.11.0"
        "fmt               6.2         fmt/6.2.0"
-        "OpenSSL           1.1         openssl/1.1.1f"
    # can't use until https://github.com/bincrafters/community/issues/1173
        #"libzip            1.5         libzip/1.5.2@bincrafters/stable"
        "lz4               1.8         lz4/1.9.2"
@@ -312,15 +311,6 @@ elseif (TARGET Boost::boost)
    add_library(boost ALIAS Boost::boost)
 endif()

-if (NOT TARGET OpenSSL::SSL)
-    set_target_properties(OpenSSL::OpenSSL PROPERTIES IMPORTED_GLOBAL TRUE)
-    add_library(OpenSSL::SSL ALIAS OpenSSL::OpenSSL)
-endif()
-if (NOT TARGET OpenSSL::Crypto)
-    set_target_properties(OpenSSL::OpenSSL PROPERTIES IMPORTED_GLOBAL TRUE)
-    add_library(OpenSSL::Crypto ALIAS OpenSSL::OpenSSL)
-endif()
-
 if (TARGET sdl2::sdl2)
    # imported from the conan generated sdl2Config.cmake
    set_target_properties(sdl2::sdl2 PROPERTIES IMPORTED_GLOBAL TRUE)
@@ -51,6 +51,8 @@ endif()
 # The variable SRC_DIR must be passed into the script (since it uses the current build directory for all values of CMAKE_*_DIR)
 set(VIDEO_CORE "${SRC_DIR}/src/video_core")
 set(HASH_FILES
+    "${VIDEO_CORE}/renderer_opengl/gl_arb_decompiler.cpp"
+    "${VIDEO_CORE}/renderer_opengl/gl_arb_decompiler.h"
    "${VIDEO_CORE}/renderer_opengl/gl_shader_cache.cpp"
    "${VIDEO_CORE}/renderer_opengl/gl_shader_cache.h"
    "${VIDEO_CORE}/renderer_opengl/gl_shader_decompiler.cpp"
@@ -673,10 +673,6 @@ QTabWidget::pane {
    border-bottom-left-radius: 2px;
 }

-QTabWidget::tab-bar {
-    overflow: visible;
-}
-
 QTabBar {
    qproperty-drawBase: 0;
    border-radius: 3px;
@@ -4,6 +4,13 @@ list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/CMakeModules")
 list(APPEND CMAKE_MODULE_PATH "${PROJECT_SOURCE_DIR}/externals/find-modules")
 include(DownloadExternals)

+# xbyak
+if (ARCHITECTURE_x86 OR ARCHITECTURE_x86_64)
+    add_library(xbyak INTERFACE)
+    target_include_directories(xbyak SYSTEM INTERFACE ./xbyak/xbyak)
+    target_compile_definitions(xbyak INTERFACE XBYAK_NO_OP_NAMES)
+endif()
+
 # Catch
 add_library(catch-single-include INTERFACE)
 target_include_directories(catch-single-include INTERFACE catch/single_include)
@@ -66,6 +73,15 @@ if (NOT LIBZIP_FOUND)
 endif()

 if (ENABLE_WEB_SERVICE)
+    # LibreSSL
+    set(LIBRESSL_SKIP_INSTALL ON CACHE BOOL "")
+    add_subdirectory(libressl EXCLUDE_FROM_ALL)
+    target_include_directories(ssl INTERFACE ./libressl/include)
+    target_compile_definitions(ssl PRIVATE -DHAVE_INET_NTOP)
+    get_directory_property(OPENSSL_LIBRARIES
+        DIRECTORY libressl
+        DEFINITION OPENSSL_LIBS)
+
    # lurlparser
    add_subdirectory(lurlparser EXCLUDE_FROM_ALL)

@@ -73,13 +89,5 @@ if (ENABLE_WEB_SERVICE)
    add_library(httplib INTERFACE)
    target_include_directories(httplib INTERFACE ./httplib)
    target_compile_definitions(httplib INTERFACE -DCPPHTTPLIB_OPENSSL_SUPPORT)
-    target_link_libraries(httplib INTERFACE OpenSSL::SSL OpenSSL::Crypto)
-endif()
-
-if (NOT TARGET xbyak)
-    if (ARCHITECTURE_x86 OR ARCHITECTURE_x86_64)
-        add_library(xbyak INTERFACE)
-        target_include_directories(xbyak SYSTEM INTERFACE ./xbyak/xbyak)
-        target_compile_definitions(xbyak INTERFACE XBYAK_NO_OP_NAMES)
-    endif()
+    target_link_libraries(httplib INTERFACE ${OPENSSL_LIBRARIES})
 endif()
@@ -32,6 +32,8 @@ add_custom_command(OUTPUT scm_rev.cpp
    DEPENDS
      # WARNING! It was too much work to try and make a common location for this list,
      # so if you need to change it, please update CMakeModules/GenerateSCMRev.cmake as well
+      "${VIDEO_CORE}/renderer_opengl/gl_arb_decompiler.cpp"
+      "${VIDEO_CORE}/renderer_opengl/gl_arb_decompiler.h"
      "${VIDEO_CORE}/renderer_opengl/gl_shader_cache.cpp"
      "${VIDEO_CORE}/renderer_opengl/gl_shader_cache.h"
      "${VIDEO_CORE}/renderer_opengl/gl_shader_decompiler.cpp"
@@ -11,7 +11,7 @@

 namespace Common::X64 {

-inline int RegToIndex(const Xbyak::Reg& reg) {
+inline std::size_t RegToIndex(const Xbyak::Reg& reg) {
    using Kind = Xbyak::Reg::Kind;
    ASSERT_MSG((reg.getKind() & (Kind::REG | Kind::XMM)) != 0,
               "RegSet only support GPRs and XMM registers.");
@@ -19,17 +19,17 @@ inline int RegToIndex(const Xbyak::Reg& reg) {
    return reg.getIdx() + (reg.getKind() == Kind::REG ? 0 : 16);
 }

-inline Xbyak::Reg64 IndexToReg64(int reg_index) {
+inline Xbyak::Reg64 IndexToReg64(std::size_t reg_index) {
    ASSERT(reg_index < 16);
-    return Xbyak::Reg64(reg_index);
+    return Xbyak::Reg64(static_cast<int>(reg_index));
 }

-inline Xbyak::Xmm IndexToXmm(int reg_index) {
+inline Xbyak::Xmm IndexToXmm(std::size_t reg_index) {
    ASSERT(reg_index >= 16 && reg_index < 32);
-    return Xbyak::Xmm(reg_index - 16);
+    return Xbyak::Xmm(static_cast<int>(reg_index - 16));
 }

-inline Xbyak::Reg IndexToReg(int reg_index) {
+inline Xbyak::Reg IndexToReg(std::size_t reg_index) {
    if (reg_index < 16) {
        return IndexToReg64(reg_index);
    } else {
@@ -151,9 +151,13 @@ constexpr size_t ABI_SHADOW_SPACE = 0;

 #endif

-inline void ABI_CalculateFrameSize(std::bitset<32> regs, size_t rsp_alignment,
-                                   size_t needed_frame_size, s32* out_subtraction,
-                                   s32* out_xmm_offset) {
+struct ABIFrameInfo {
+    s32 subtraction;
+    s32 xmm_offset;
+};
+
+inline ABIFrameInfo ABI_CalculateFrameSize(std::bitset<32> regs, size_t rsp_alignment,
+                                           size_t needed_frame_size) {
    const auto count = (regs & ABI_ALL_GPRS).count();
    rsp_alignment -= count * 8;
    size_t subtraction = 0;
@@ -170,33 +174,28 @@ inline void ABI_CalculateFrameSize(std::bitset<32> regs, size_t rsp_alignment,
    rsp_alignment -= subtraction;
    subtraction += rsp_alignment & 0xF;

-    *out_subtraction = (s32)subtraction;
-    *out_xmm_offset = (s32)(subtraction - xmm_base_subtraction);
+    return ABIFrameInfo{static_cast<s32>(subtraction),
+                        static_cast<s32>(subtraction - xmm_base_subtraction)};
 }

 inline size_t ABI_PushRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::bitset<32> regs,
                                              size_t rsp_alignment, size_t needed_frame_size = 0) {
-    s32 subtraction, xmm_offset;
-    ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
+    auto frame_info = ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size);
+
    for (std::size_t i = 0; i < regs.size(); ++i) {
        if (regs[i] && ABI_ALL_GPRS[i]) {
-            code.push(IndexToReg64(static_cast<int>(i)));
-        }
-    }
-    if (subtraction != 0) {
-        code.sub(code.rsp, subtraction);
-    }
-
-    for (int i = 0; i < regs.count(); i++) {
-        if (regs.test(i) & ABI_ALL_GPRS.test(i)) {
            code.push(IndexToReg64(i));
        }
    }

+    if (frame_info.subtraction != 0) {
+        code.sub(code.rsp, frame_info.subtraction);
+    }
+
    for (std::size_t i = 0; i < regs.size(); ++i) {
        if (regs[i] && ABI_ALL_XMMS[i]) {
-            code.movaps(code.xword[code.rsp + xmm_offset], IndexToXmm(static_cast<int>(i)));
-            xmm_offset += 0x10;
+            code.movaps(code.xword[code.rsp + frame_info.xmm_offset], IndexToXmm(i));
+            frame_info.xmm_offset += 0x10;
        }
    }

@@ -205,59 +204,23 @@ inline size_t ABI_PushRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::b

 inline void ABI_PopRegistersAndAdjustStack(Xbyak::CodeGenerator& code, std::bitset<32> regs,
                                           size_t rsp_alignment, size_t needed_frame_size = 0) {
-    s32 subtraction, xmm_offset;
-    ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
+    auto frame_info = ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size);

    for (std::size_t i = 0; i < regs.size(); ++i) {
        if (regs[i] && ABI_ALL_XMMS[i]) {
-            code.movaps(IndexToXmm(static_cast<int>(i)), code.xword[code.rsp + xmm_offset]);
-            xmm_offset += 0x10;
+            code.movaps(IndexToXmm(i), code.xword[code.rsp + frame_info.xmm_offset]);
+            frame_info.xmm_offset += 0x10;
        }
    }

-    if (subtraction != 0) {
-        code.add(code.rsp, subtraction);
+    if (frame_info.subtraction != 0) {
+        code.add(code.rsp, frame_info.subtraction);
    }

    // GPRs need to be popped in reverse order
-    for (int i = 15; i >= 0; i--) {
-        if (regs[i]) {
-            code.pop(IndexToReg64(i));
-        }
-    }
-}
-
-inline size_t ABI_PushRegistersAndAdjustStackGPS(Xbyak::CodeGenerator& code, std::bitset<32> regs,
-                                                 size_t rsp_alignment,
-                                                 size_t needed_frame_size = 0) {
-    s32 subtraction, xmm_offset;
-    ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
-
-    for (std::size_t i = 0; i < regs.size(); ++i) {
+    for (std::size_t j = 0; j < regs.size(); ++j) {
+        const std::size_t i = regs.size() - j - 1;
        if (regs[i] && ABI_ALL_GPRS[i]) {
-            code.push(IndexToReg64(static_cast<int>(i)));
-        }
-    }
-
-    if (subtraction != 0) {
-        code.sub(code.rsp, subtraction);
-    }
-
-    return ABI_SHADOW_SPACE;
-}
-
-inline void ABI_PopRegistersAndAdjustStackGPS(Xbyak::CodeGenerator& code, std::bitset<32> regs,
-                                              size_t rsp_alignment, size_t needed_frame_size = 0) {
-    s32 subtraction, xmm_offset;
-    ABI_CalculateFrameSize(regs, rsp_alignment, needed_frame_size, &subtraction, &xmm_offset);
-
-    if (subtraction != 0) {
-        code.add(code.rsp, subtraction);
-    }
-
-    // GPRs need to be popped in reverse order
-    for (int i = 15; i >= 0; i--) {
-        if (regs[i]) {
            code.pop(IndexToReg64(i));
        }
    }
@@ -50,7 +50,8 @@ public:
    }

    void InterpreterFallback(u32 pc, std::size_t num_instructions) override {
-        UNIMPLEMENTED();
+        UNIMPLEMENTED_MSG("This should never happen, pc = {:08X}, code = {:08X}", pc,
+                          MemoryReadCode(pc));
    }

    void ExceptionRaised(u32 pc, Dynarmic::A32::Exception exception) override {
@@ -89,8 +90,6 @@ public:

    ARM_Dynarmic_32& parent;
    std::size_t num_interpreted_instructions{};
-    u64 tpidrro_el0{};
-    u64 tpidr_el0{};
 };

 std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable& page_table,
@@ -99,7 +98,7 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable&
    config.callbacks = cb.get();
    // TODO(bunnei): Implement page table for 32-bit
    // config.page_table = &page_table.pointers;
-    config.coprocessors[15] = std::make_shared<DynarmicCP15>((u32*)&CP15_regs[0]);
+    config.coprocessors[15] = cp15;
    config.define_unpredictable_behaviour = true;
    return std::make_unique<Dynarmic::A32::Jit>(config);
 }
@@ -112,13 +111,13 @@ void ARM_Dynarmic_32::Run() {
 }

 void ARM_Dynarmic_32::Step() {
-    cb->InterpreterFallback(jit->Regs()[15], 1);
+    jit->Step();
 }

 ARM_Dynarmic_32::ARM_Dynarmic_32(System& system, ExclusiveMonitor& exclusive_monitor,
                                 std::size_t core_index)
-    : ARM_Interface{system},
-      cb(std::make_unique<DynarmicCallbacks32>(*this)), core_index{core_index},
+    : ARM_Interface{system}, cb(std::make_unique<DynarmicCallbacks32>(*this)),
+      cp15(std::make_shared<DynarmicCP15>(*this)), core_index{core_index},
      exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}

 ARM_Dynarmic_32::~ARM_Dynarmic_32() = default;
@@ -154,19 +153,19 @@ void ARM_Dynarmic_32::SetPSTATE(u32 cpsr) {
 }

 u64 ARM_Dynarmic_32::GetTlsAddress() const {
-    return CP15_regs[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)];
+    return cp15->uro;
 }

 void ARM_Dynarmic_32::SetTlsAddress(VAddr address) {
-    CP15_regs[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)] = static_cast<u32>(address);
+    cp15->uro = static_cast<u32>(address);
 }

 u64 ARM_Dynarmic_32::GetTPIDR_EL0() const {
-    return cb->tpidr_el0;
+    return cp15->uprw;
 }

 void ARM_Dynarmic_32::SetTPIDR_EL0(u64 value) {
-    cb->tpidr_el0 = value;
+    cp15->uprw = value;
 }

 void ARM_Dynarmic_32::SaveContext(ThreadContext32& ctx) {
@@ -22,6 +22,7 @@ class Memory;
 namespace Core {

 class DynarmicCallbacks32;
+class DynarmicCP15;
 class DynarmicExclusiveMonitor;
 class System;

@@ -66,12 +67,14 @@ private:
        std::unordered_map<JitCacheKey, std::shared_ptr<Dynarmic::A32::Jit>, Common::PairHash>;

    friend class DynarmicCallbacks32;
+    friend class DynarmicCP15;
+
    std::unique_ptr<DynarmicCallbacks32> cb;
    JitCacheType jit_cache;
    std::shared_ptr<Dynarmic::A32::Jit> jit;
+    std::shared_ptr<DynarmicCP15> cp15;
    std::size_t core_index;
    DynarmicExclusiveMonitor& exclusive_monitor;
-    std::array<u32, 84> CP15_regs{};
 };

 } // namespace Core
@@ -2,79 +2,132 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.

+#include <fmt/format.h>
+#include "common/logging/log.h"
+#include "core/arm/dynarmic/arm_dynarmic_32.h"
 #include "core/arm/dynarmic/arm_dynarmic_cp15.h"
+#include "core/core.h"
+#include "core/core_timing.h"
+#include "core/core_timing_util.h"

 using Callback = Dynarmic::A32::Coprocessor::Callback;
 using CallbackOrAccessOneWord = Dynarmic::A32::Coprocessor::CallbackOrAccessOneWord;
 using CallbackOrAccessTwoWords = Dynarmic::A32::Coprocessor::CallbackOrAccessTwoWords;

+template <>
+struct fmt::formatter<Dynarmic::A32::CoprocReg> {
+    constexpr auto parse(format_parse_context& ctx) {
+        return ctx.begin();
+    }
+    template <typename FormatContext>
+    auto format(const Dynarmic::A32::CoprocReg& reg, FormatContext& ctx) {
+        return format_to(ctx.out(), "cp{}", static_cast<size_t>(reg));
+    }
+};
+
+namespace Core {
+
+static u32 dummy_value;
+
 std::optional<Callback> DynarmicCP15::CompileInternalOperation(bool two, unsigned opc1,
                                                               CoprocReg CRd, CoprocReg CRn,
                                                               CoprocReg CRm, unsigned opc2) {
+    LOG_CRITICAL(Core_ARM, "CP15: cdp{} p15, {}, {}, {}, {}, {}", two ? "2" : "", opc1, CRd, CRn,
+                 CRm, opc2);
    return {};
 }

 CallbackOrAccessOneWord DynarmicCP15::CompileSendOneWord(bool two, unsigned opc1, CoprocReg CRn,
                                                         CoprocReg CRm, unsigned opc2) {
-    // TODO(merry): Privileged CP15 registers
-
    if (!two && CRn == CoprocReg::C7 && opc1 == 0 && CRm == CoprocReg::C5 && opc2 == 4) {
+        // CP15_FLUSH_PREFETCH_BUFFER
        // This is a dummy write, we ignore the value written here.
-        return &CP15[static_cast<std::size_t>(CP15Register::CP15_FLUSH_PREFETCH_BUFFER)];
+        return &dummy_value;
    }

    if (!two && CRn == CoprocReg::C7 && opc1 == 0 && CRm == CoprocReg::C10) {
        switch (opc2) {
        case 4:
+            // CP15_DATA_SYNC_BARRIER
            // This is a dummy write, we ignore the value written here.
-            return &CP15[static_cast<std::size_t>(CP15Register::CP15_DATA_SYNC_BARRIER)];
+            return &dummy_value;
        case 5:
+            // CP15_DATA_MEMORY_BARRIER
            // This is a dummy write, we ignore the value written here.
-            return &CP15[static_cast<std::size_t>(CP15Register::CP15_DATA_MEMORY_BARRIER)];
-        default:
-            return {};
+            return &dummy_value;
        }
    }

    if (!two && CRn == CoprocReg::C13 && opc1 == 0 && CRm == CoprocReg::C0 && opc2 == 2) {
-        return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_UPRW)];
+        // CP15_THREAD_UPRW
+        return &uprw;
    }

+    LOG_CRITICAL(Core_ARM, "CP15: mcr{} p15, {}, <Rt>, {}, {}, {}", two ? "2" : "", opc1, CRn, CRm,
+                 opc2);
    return {};
 }

 CallbackOrAccessTwoWords DynarmicCP15::CompileSendTwoWords(bool two, unsigned opc, CoprocReg CRm) {
+    LOG_CRITICAL(Core_ARM, "CP15: mcrr{} p15, {}, <Rt>, <Rt2>, {}", two ? "2" : "", opc, CRm);
    return {};
 }

 CallbackOrAccessOneWord DynarmicCP15::CompileGetOneWord(bool two, unsigned opc1, CoprocReg CRn,
                                                        CoprocReg CRm, unsigned opc2) {
-    // TODO(merry): Privileged CP15 registers
-
    if (!two && CRn == CoprocReg::C13 && opc1 == 0 && CRm == CoprocReg::C0) {
        switch (opc2) {
        case 2:
-            return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_UPRW)];
+            // CP15_THREAD_UPRW
+            return &uprw;
        case 3:
-            return &CP15[static_cast<std::size_t>(CP15Register::CP15_THREAD_URO)];
-        default:
-            return {};
+            // CP15_THREAD_URO
+            return &uro;
        }
    }

+    LOG_CRITICAL(Core_ARM, "CP15: mrc{} p15, {}, <Rt>, {}, {}, {}", two ? "2" : "", opc1, CRn, CRm,
+                 opc2);
    return {};
 }

 CallbackOrAccessTwoWords DynarmicCP15::CompileGetTwoWords(bool two, unsigned opc, CoprocReg CRm) {
+    if (!two && opc == 0 && CRm == CoprocReg::C14) {
+        // CNTPCT
+        const auto callback = static_cast<u64 (*)(Dynarmic::A32::Jit*, void*, u32, u32)>(
+            [](Dynarmic::A32::Jit*, void* arg, u32, u32) -> u64 {
+                ARM_Dynarmic_32& parent = *(ARM_Dynarmic_32*)arg;
+                return Timing::CpuCyclesToClockCycles(parent.system.CoreTiming().GetTicks());
+            });
+        return Dynarmic::A32::Coprocessor::Callback{callback, (void*)&parent};
+    }
+
+    LOG_CRITICAL(Core_ARM, "CP15: mrrc{} p15, {}, <Rt>, <Rt2>, {}", two ? "2" : "", opc, CRm);
    return {};
 }

 std::optional<Callback> DynarmicCP15::CompileLoadWords(bool two, bool long_transfer, CoprocReg CRd,
                                                       std::optional<u8> option) {
+    if (option) {
+        LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...], {}", two ? "2" : "",
+                     long_transfer ? "l" : "", CRd, *option);
+    } else {
+        LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...]", two ? "2" : "",
+                     long_transfer ? "l" : "", CRd);
+    }
    return {};
 }

 std::optional<Callback> DynarmicCP15::CompileStoreWords(bool two, bool long_transfer, CoprocReg CRd,
                                                        std::optional<u8> option) {
+    if (option) {
+        LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...], {}", two ? "2" : "",
+                     long_transfer ? "l" : "", CRd, *option);
+    } else {
+        LOG_CRITICAL(Core_ARM, "CP15: mrrc{}{} p15, {}, [...]", two ? "2" : "",
+                     long_transfer ? "l" : "", CRd);
+    }
    return {};
 }
+
+} // namespace Core
@@ -10,128 +10,15 @@
 #include <dynarmic/A32/coprocessor.h>
 #include "common/common_types.h"

-enum class CP15Register {
-    // c0 - Information registers
-    CP15_MAIN_ID,
-    CP15_CACHE_TYPE,
-    CP15_TCM_STATUS,
-    CP15_TLB_TYPE,
-    CP15_CPU_ID,
-    CP15_PROCESSOR_FEATURE_0,
-    CP15_PROCESSOR_FEATURE_1,
-    CP15_DEBUG_FEATURE_0,
-    CP15_AUXILIARY_FEATURE_0,
-    CP15_MEMORY_MODEL_FEATURE_0,
-    CP15_MEMORY_MODEL_FEATURE_1,
-    CP15_MEMORY_MODEL_FEATURE_2,
-    CP15_MEMORY_MODEL_FEATURE_3,
-    CP15_ISA_FEATURE_0,
-    CP15_ISA_FEATURE_1,
-    CP15_ISA_FEATURE_2,
-    CP15_ISA_FEATURE_3,
-    CP15_ISA_FEATURE_4,
+namespace Core {

-    // c1 - Control registers
-    CP15_CONTROL,
-    CP15_AUXILIARY_CONTROL,
-    CP15_COPROCESSOR_ACCESS_CONTROL,
-
-    // c2 - Translation table registers
-    CP15_TRANSLATION_BASE_TABLE_0,
-    CP15_TRANSLATION_BASE_TABLE_1,
-    CP15_TRANSLATION_BASE_CONTROL,
-    CP15_DOMAIN_ACCESS_CONTROL,
-    CP15_RESERVED,
-
-    // c5 - Fault status registers
-    CP15_FAULT_STATUS,
-    CP15_INSTR_FAULT_STATUS,
-    CP15_COMBINED_DATA_FSR = CP15_FAULT_STATUS,
-    CP15_INST_FSR,
-
-    // c6 - Fault Address registers
-    CP15_FAULT_ADDRESS,
-    CP15_COMBINED_DATA_FAR = CP15_FAULT_ADDRESS,
-    CP15_WFAR,
-    CP15_IFAR,
-
-    // c7 - Cache operation registers
-    CP15_WAIT_FOR_INTERRUPT,
-    CP15_PHYS_ADDRESS,
-    CP15_INVALIDATE_INSTR_CACHE,
-    CP15_INVALIDATE_INSTR_CACHE_USING_MVA,
-    CP15_INVALIDATE_INSTR_CACHE_USING_INDEX,
-    CP15_FLUSH_PREFETCH_BUFFER,
-    CP15_FLUSH_BRANCH_TARGET_CACHE,
-    CP15_FLUSH_BRANCH_TARGET_CACHE_ENTRY,
-    CP15_INVALIDATE_DATA_CACHE,
-    CP15_INVALIDATE_DATA_CACHE_LINE_USING_MVA,
-    CP15_INVALIDATE_DATA_CACHE_LINE_USING_INDEX,
-    CP15_INVALIDATE_DATA_AND_INSTR_CACHE,
-    CP15_CLEAN_DATA_CACHE,
-    CP15_CLEAN_DATA_CACHE_LINE_USING_MVA,
-    CP15_CLEAN_DATA_CACHE_LINE_USING_INDEX,
-    CP15_DATA_SYNC_BARRIER,
-    CP15_DATA_MEMORY_BARRIER,
-    CP15_CLEAN_AND_INVALIDATE_DATA_CACHE,
-    CP15_CLEAN_AND_INVALIDATE_DATA_CACHE_LINE_USING_MVA,
-    CP15_CLEAN_AND_INVALIDATE_DATA_CACHE_LINE_USING_INDEX,
-
-    // c8 - TLB operations
-    CP15_INVALIDATE_ITLB,
-    CP15_INVALIDATE_ITLB_SINGLE_ENTRY,
-    CP15_INVALIDATE_ITLB_ENTRY_ON_ASID_MATCH,
-    CP15_INVALIDATE_ITLB_ENTRY_ON_MVA,
-    CP15_INVALIDATE_DTLB,
-    CP15_INVALIDATE_DTLB_SINGLE_ENTRY,
-    CP15_INVALIDATE_DTLB_ENTRY_ON_ASID_MATCH,
-    CP15_INVALIDATE_DTLB_ENTRY_ON_MVA,
-    CP15_INVALIDATE_UTLB,
-    CP15_INVALIDATE_UTLB_SINGLE_ENTRY,
-    CP15_INVALIDATE_UTLB_ENTRY_ON_ASID_MATCH,
-    CP15_INVALIDATE_UTLB_ENTRY_ON_MVA,
-
-    // c9 - Data cache lockdown register
-    CP15_DATA_CACHE_LOCKDOWN,
-
-    // c10 - TLB/Memory map registers
-    CP15_TLB_LOCKDOWN,
-    CP15_PRIMARY_REGION_REMAP,
-    CP15_NORMAL_REGION_REMAP,
-
-    // c13 - Thread related registers
-    CP15_PID,
-    CP15_CONTEXT_ID,
-    CP15_THREAD_UPRW, // Thread ID register - User/Privileged Read/Write
-    CP15_THREAD_URO,  // Thread ID register - User Read Only (Privileged R/W)
-    CP15_THREAD_PRW,  // Thread ID register - Privileged R/W only.
-
-    // c15 - Performance and TLB lockdown registers
-    CP15_PERFORMANCE_MONITOR_CONTROL,
-    CP15_CYCLE_COUNTER,
-    CP15_COUNT_0,
-    CP15_COUNT_1,
-    CP15_READ_MAIN_TLB_LOCKDOWN_ENTRY,
-    CP15_WRITE_MAIN_TLB_LOCKDOWN_ENTRY,
-    CP15_MAIN_TLB_LOCKDOWN_VIRT_ADDRESS,
-    CP15_MAIN_TLB_LOCKDOWN_PHYS_ADDRESS,
-    CP15_MAIN_TLB_LOCKDOWN_ATTRIBUTE,
-    CP15_TLB_DEBUG_CONTROL,
-
-    // Skyeye defined
-    CP15_TLB_FAULT_ADDR,
-    CP15_TLB_FAULT_STATUS,
-
-    // Not an actual register.
-    // All registers should be defined above this.
-    CP15_REGISTER_COUNT,
-};
+class ARM_Dynarmic_32;

 class DynarmicCP15 final : public Dynarmic::A32::Coprocessor {
 public:
    using CoprocReg = Dynarmic::A32::CoprocReg;

-    explicit DynarmicCP15(u32* cp15) : CP15(cp15){};
+    explicit DynarmicCP15(ARM_Dynarmic_32& parent) : parent(parent) {}

    std::optional<Callback> CompileInternalOperation(bool two, unsigned opc1, CoprocReg CRd,
                                                     CoprocReg CRn, CoprocReg CRm,
@@ -147,6 +34,9 @@ public:
    std::optional<Callback> CompileStoreWords(bool two, bool long_transfer, CoprocReg CRd,
                                              std::optional<u8> option) override;

-private:
-    u32* CP15{};
+    ARM_Dynarmic_32& parent;
+    u32 uprw;
+    u32 uro;
 };
+
+} // namespace Core
@@ -132,7 +132,8 @@ std::shared_ptr<ResourceLimit> Process::GetResourceLimit() const {

 u64 Process::GetTotalPhysicalMemoryAvailable() const {
    const u64 capacity{resource_limit->GetCurrentResourceValue(ResourceType::PhysicalMemory) +
-                       page_table->GetTotalHeapSize() + image_size + main_thread_stack_size};
+                       page_table->GetTotalHeapSize() + GetSystemResourceSize() + image_size +
+                       main_thread_stack_size};

    if (capacity < memory_usage_capacity) {
        return capacity;
@@ -146,7 +147,8 @@ u64 Process::GetTotalPhysicalMemoryAvailableWithoutSystemResource() const {
 }

 u64 Process::GetTotalPhysicalMemoryUsed() const {
-    return image_size + main_thread_stack_size + page_table->GetTotalHeapSize();
+    return image_size + main_thread_stack_size + page_table->GetTotalHeapSize() +
+           GetSystemResourceSize();
 }

 u64 Process::GetTotalPhysicalMemoryUsedWithoutSystemResource() const {
@@ -25,7 +25,7 @@ u32 nvhost_ctrl_gpu::ioctl(Ioctl command, const std::vector<u8>& input,
    case IoctlCommand::IocGetCharacteristicsCommand:
        return GetCharacteristics(input, output, output2, version);
    case IoctlCommand::IocGetTPCMasksCommand:
-        return GetTPCMasks(input, output);
+        return GetTPCMasks(input, output, output2, version);
    case IoctlCommand::IocGetActiveSlotMaskCommand:
        return GetActiveSlotMask(input, output);
    case IoctlCommand::IocZcullGetCtxSizeCommand:
@@ -98,17 +98,22 @@ u32 nvhost_ctrl_gpu::GetCharacteristics(const std::vector<u8>& input, std::vecto
    return 0;
 }

-u32 nvhost_ctrl_gpu::GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output) {
+u32 nvhost_ctrl_gpu::GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output,
+                                 std::vector<u8>& output2, IoctlVersion version) {
    IoctlGpuGetTpcMasksArgs params{};
    std::memcpy(&params, input.data(), input.size());
-    LOG_INFO(Service_NVDRV, "called, mask=0x{:X}, mask_buf_addr=0x{:X}", params.mask_buf_size,
-             params.mask_buf_addr);
-    // TODO(ogniK): Confirm value on hardware
-    if (params.mask_buf_size)
-        params.tpc_mask_size = 4 * 1; // 4 * num_gpc
-    else
-        params.tpc_mask_size = 0;
-    std::memcpy(output.data(), &params, sizeof(params));
+    LOG_DEBUG(Service_NVDRV, "called, mask_buffer_size=0x{:X}", params.mask_buffer_size);
+    if (params.mask_buffer_size != 0) {
+        params.tcp_mask = 3;
+    }
+
+    if (version == IoctlVersion::Version3) {
+        std::memcpy(output.data(), input.data(), output.size());
+        std::memcpy(output2.data(), &params.tcp_mask, output2.size());
+    } else {
+        std::memcpy(output.data(), &params, output.size());
+    }
+
    return 0;
 }

@@ -92,16 +92,11 @@ private:
                  "IoctlCharacteristics is incorrect size");

    struct IoctlGpuGetTpcMasksArgs {
-        /// [in]  TPC mask buffer size reserved by userspace. Should be at least
-        /// sizeof(__u32) * fls(gpc_mask) to receive TPC mask for each GPC.
-        /// [out] full kernel buffer size
-        u32_le mask_buf_size;
-        u32_le reserved;
-
-        /// [in]  pointer to TPC mask buffer. It will receive one 32-bit TPC mask per GPC or 0 if
-        /// GPC is not enabled or not present. This parameter is ignored if mask_buf_size is 0.
-        u64_le mask_buf_addr;
-        u64_le tpc_mask_size; // Nintendo add this?
+        u32_le mask_buffer_size{};
+        INSERT_PADDING_WORDS(1);
+        u64_le mask_buffer_address{};
+        u32_le tcp_mask{};
+        INSERT_PADDING_WORDS(1);
    };
    static_assert(sizeof(IoctlGpuGetTpcMasksArgs) == 24,
                  "IoctlGpuGetTpcMasksArgs is incorrect size");
@@ -166,7 +161,8 @@ private:

    u32 GetCharacteristics(const std::vector<u8>& input, std::vector<u8>& output,
                           std::vector<u8>& output2, IoctlVersion version);
-    u32 GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output);
+    u32 GetTPCMasks(const std::vector<u8>& input, std::vector<u8>& output, std::vector<u8>& output2,
+                    IoctlVersion version);
    u32 GetActiveSlotMask(const std::vector<u8>& input, std::vector<u8>& output);
    u32 ZCullGetCtxSize(const std::vector<u8>& input, std::vector<u8>& output);
    u32 ZCullGetInfo(const std::vector<u8>& input, std::vector<u8>& output);
@@ -437,7 +437,7 @@ struct Values {
    bool renderer_debug;
    int vulkan_device;

-    float resolution_factor;
+    u16 resolution_factor{1};
    int aspect_ratio;
    int max_anisotropy;
    bool use_frame_limit;
@@ -52,6 +52,8 @@ add_library(video_core STATIC
    rasterizer_interface.h
    renderer_base.cpp
    renderer_base.h
+    renderer_opengl/gl_arb_decompiler.cpp
+    renderer_opengl/gl_arb_decompiler.h
    renderer_opengl/gl_buffer_cache.cpp
    renderer_opengl/gl_buffer_cache.h
    renderer_opengl/gl_device.cpp
@@ -15,48 +15,47 @@ namespace VideoCommon {

 class BufferBlock {
 public:
-    bool Overlaps(const VAddr start, const VAddr end) const {
+    bool Overlaps(VAddr start, VAddr end) const {
        return (cpu_addr < end) && (cpu_addr_end > start);
    }

-    bool IsInside(const VAddr other_start, const VAddr other_end) const {
+    bool IsInside(VAddr other_start, VAddr other_end) const {
        return cpu_addr <= other_start && other_end <= cpu_addr_end;
    }

-    std::size_t GetOffset(const VAddr in_addr) {
+    std::size_t Offset(VAddr in_addr) const {
        return static_cast<std::size_t>(in_addr - cpu_addr);
    }

-    VAddr GetCpuAddr() const {
+    VAddr CpuAddr() const {
        return cpu_addr;
    }

-    VAddr GetCpuAddrEnd() const {
+    VAddr CpuAddrEnd() const {
        return cpu_addr_end;
    }

-    void SetCpuAddr(const VAddr new_addr) {
+    void SetCpuAddr(VAddr new_addr) {
        cpu_addr = new_addr;
        cpu_addr_end = new_addr + size;
    }

-    std::size_t GetSize() const {
+    std::size_t Size() const {
        return size;
    }

+    u64 Epoch() const {
+        return epoch;
+    }
+
    void SetEpoch(u64 new_epoch) {
        epoch = new_epoch;
    }

-    u64 GetEpoch() {
-        return epoch;
-    }
-
 protected:
-    explicit BufferBlock(VAddr cpu_addr, const std::size_t size) : size{size} {
-        SetCpuAddr(cpu_addr);
+    explicit BufferBlock(VAddr cpu_addr_, std::size_t size_) : size{size_} {
+        SetCpuAddr(cpu_addr_);
    }
-    ~BufferBlock() = default;

 private:
    VAddr cpu_addr{};
@@ -30,12 +30,16 @@

 namespace VideoCommon {

-template <typename OwnerBuffer, typename BufferType, typename StreamBuffer>
+template <typename Buffer, typename BufferType, typename StreamBuffer>
 class BufferCache {
    using IntervalSet = boost::icl::interval_set<VAddr>;
    using IntervalType = typename IntervalSet::interval_type;
    using VectorMapInterval = boost::container::small_vector<MapInterval*, 1>;

+    static constexpr u64 WRITE_PAGE_BIT = 11;
+    static constexpr u64 BLOCK_PAGE_BITS = 21;
+    static constexpr u64 BLOCK_PAGE_SIZE = 1ULL << BLOCK_PAGE_BITS;
+
 public:
    using BufferInfo = std::pair<BufferType, u64>;

@@ -82,7 +86,7 @@ public:
            }
        }

-        OwnerBuffer block = GetBlock(cpu_addr, size);
+        Buffer* const block = GetBlock(cpu_addr, size);
        MapInterval* const map = MapAddress(block, gpu_addr, cpu_addr, size);
        if (!map) {
            return {GetEmptyBuffer(size), 0};
@@ -98,7 +102,7 @@ public:
            }
        }

-        return {ToHandle(block), static_cast<u64>(block->GetOffset(cpu_addr))};
+        return {block->Handle(), static_cast<u64>(block->Offset(cpu_addr))};
    }

    /// Uploads from a host memory. Returns the OpenGL buffer where it's located and its offset.
@@ -110,31 +114,37 @@ public:
        });
    }

-    void Map(std::size_t max_size) {
+    /// Prepares the buffer cache for data uploading
+    /// @param max_size Maximum number of bytes that will be uploaded
+    /// @return True when a stream buffer invalidation was required, false otherwise
+    bool Map(std::size_t max_size) {
        std::lock_guard lock{mutex};

+        bool invalidated;
        std::tie(buffer_ptr, buffer_offset_base, invalidated) = stream_buffer->Map(max_size, 4);
        buffer_offset = buffer_offset_base;
+
+        return invalidated;
    }

-    /// Finishes the upload stream, returns true on bindings invalidation.
-    bool Unmap() {
+    /// Finishes the upload stream
+    void Unmap() {
        std::lock_guard lock{mutex};
-
        stream_buffer->Unmap(buffer_offset - buffer_offset_base);
-        return std::exchange(invalidated, false);
    }

+    /// Function called at the end of each frame, inteded for deferred operations
    void TickFrame() {
        ++epoch;
+
        while (!pending_destruction.empty()) {
            // Delay at least 4 frames before destruction.
            // This is due to triple buffering happening on some drivers.
            static constexpr u64 epochs_to_destroy = 5;
-            if (pending_destruction.front()->GetEpoch() + epochs_to_destroy > epoch) {
+            if (pending_destruction.front()->Epoch() + epochs_to_destroy > epoch) {
                break;
            }
-            pending_destruction.pop_front();
+            pending_destruction.pop();
        }
    }

@@ -249,23 +259,21 @@ public:

 protected:
    explicit BufferCache(VideoCore::RasterizerInterface& rasterizer, Core::System& system,
-                         std::unique_ptr<StreamBuffer> stream_buffer)
-        : rasterizer{rasterizer}, system{system}, stream_buffer{std::move(stream_buffer)},
-          stream_buffer_handle{this->stream_buffer->GetHandle()} {}
+                         std::unique_ptr<StreamBuffer> stream_buffer_)
+        : rasterizer{rasterizer}, system{system}, stream_buffer{std::move(stream_buffer_)},
+          stream_buffer_handle{stream_buffer->Handle()} {}

    ~BufferCache() = default;

-    virtual BufferType ToHandle(const OwnerBuffer& storage) = 0;
+    virtual std::shared_ptr<Buffer> CreateBlock(VAddr cpu_addr, std::size_t size) = 0;

-    virtual OwnerBuffer CreateBlock(VAddr cpu_addr, std::size_t size) = 0;
-
-    virtual void UploadBlockData(const OwnerBuffer& buffer, std::size_t offset, std::size_t size,
+    virtual void UploadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size,
                                 const u8* data) = 0;

-    virtual void DownloadBlockData(const OwnerBuffer& buffer, std::size_t offset, std::size_t size,
+    virtual void DownloadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size,
                                   u8* data) = 0;

-    virtual void CopyBlock(const OwnerBuffer& src, const OwnerBuffer& dst, std::size_t src_offset,
+    virtual void CopyBlock(const Buffer& src, const Buffer& dst, std::size_t src_offset,
                           std::size_t dst_offset, std::size_t size) = 0;

    virtual BufferInfo ConstBufferUpload(const void* raw_pointer, std::size_t size) {
@@ -321,7 +329,7 @@ protected:
    }

 private:
-    MapInterval* MapAddress(const OwnerBuffer& block, GPUVAddr gpu_addr, VAddr cpu_addr,
+    MapInterval* MapAddress(const Buffer* block, GPUVAddr gpu_addr, VAddr cpu_addr,
                            std::size_t size) {
        const VectorMapInterval overlaps = GetMapsInRange(cpu_addr, size);
        if (overlaps.empty()) {
@@ -329,11 +337,11 @@ private:
            const VAddr cpu_addr_end = cpu_addr + size;
            if (memory_manager.IsGranularRange(gpu_addr, size)) {
                u8* host_ptr = memory_manager.GetPointer(gpu_addr);
-                UploadBlockData(block, block->GetOffset(cpu_addr), size, host_ptr);
+                UploadBlockData(*block, block->Offset(cpu_addr), size, host_ptr);
            } else {
                staging_buffer.resize(size);
                memory_manager.ReadBlockUnsafe(gpu_addr, staging_buffer.data(), size);
-                UploadBlockData(block, block->GetOffset(cpu_addr), size, staging_buffer.data());
+                UploadBlockData(*block, block->Offset(cpu_addr), size, staging_buffer.data());
            }
            return Register(MapInterval(cpu_addr, cpu_addr_end, gpu_addr));
        }
@@ -376,7 +384,7 @@ private:
        return map;
    }

-    void UpdateBlock(const OwnerBuffer& block, VAddr start, VAddr end,
+    void UpdateBlock(const Buffer* block, VAddr start, VAddr end,
                     const VectorMapInterval& overlaps) {
        const IntervalType base_interval{start, end};
        IntervalSet interval_set{};
@@ -386,13 +394,13 @@ private:
            interval_set.subtract(subtract);
        }
        for (auto& interval : interval_set) {
-            std::size_t size = interval.upper() - interval.lower();
-            if (size > 0) {
-                staging_buffer.resize(size);
-                system.Memory().ReadBlockUnsafe(interval.lower(), staging_buffer.data(), size);
-                UploadBlockData(block, block->GetOffset(interval.lower()), size,
-                                staging_buffer.data());
+            const std::size_t size = interval.upper() - interval.lower();
+            if (size == 0) {
+                continue;
            }
+            staging_buffer.resize(size);
+            system.Memory().ReadBlockUnsafe(interval.lower(), staging_buffer.data(), size);
+            UploadBlockData(*block, block->Offset(interval.lower()), size, staging_buffer.data());
        }
    }

@@ -422,10 +430,14 @@ private:
    }

    void FlushMap(MapInterval* map) {
+        const auto it = blocks.find(map->start >> BLOCK_PAGE_BITS);
+        ASSERT_OR_EXECUTE(it != blocks.end(), return;);
+
+        std::shared_ptr<Buffer> block = it->second;
+
        const std::size_t size = map->end - map->start;
-        OwnerBuffer block = blocks[map->start >> block_page_bits];
        staging_buffer.resize(size);
-        DownloadBlockData(block, block->GetOffset(map->start), size, staging_buffer.data());
+        DownloadBlockData(*block, block->Offset(map->start), size, staging_buffer.data());
        system.Memory().WriteBlockUnsafe(map->start, staging_buffer.data(), size);
        map->MarkAsModified(false, 0);
    }
@@ -448,97 +460,89 @@ private:
        buffer_offset = offset_aligned;
    }

-    OwnerBuffer EnlargeBlock(OwnerBuffer buffer) {
-        const std::size_t old_size = buffer->GetSize();
-        const std::size_t new_size = old_size + block_page_size;
-        const VAddr cpu_addr = buffer->GetCpuAddr();
-        OwnerBuffer new_buffer = CreateBlock(cpu_addr, new_size);
-        CopyBlock(buffer, new_buffer, 0, 0, old_size);
-        buffer->SetEpoch(epoch);
-        pending_destruction.push_back(buffer);
+    std::shared_ptr<Buffer> EnlargeBlock(std::shared_ptr<Buffer> buffer) {
+        const std::size_t old_size = buffer->Size();
+        const std::size_t new_size = old_size + BLOCK_PAGE_SIZE;
+        const VAddr cpu_addr = buffer->CpuAddr();
+        std::shared_ptr<Buffer> new_buffer = CreateBlock(cpu_addr, new_size);
+        CopyBlock(*buffer, *new_buffer, 0, 0, old_size);
+        QueueDestruction(std::move(buffer));
+
        const VAddr cpu_addr_end = cpu_addr + new_size - 1;
-        u64 page_start = cpu_addr >> block_page_bits;
-        const u64 page_end = cpu_addr_end >> block_page_bits;
-        while (page_start <= page_end) {
-            blocks[page_start] = new_buffer;
-            ++page_start;
+        const u64 page_end = cpu_addr_end >> BLOCK_PAGE_BITS;
+        for (u64 page_start = cpu_addr >> BLOCK_PAGE_BITS; page_start <= page_end; ++page_start) {
+            blocks.insert_or_assign(page_start, new_buffer);
        }
+
        return new_buffer;
    }

-    OwnerBuffer MergeBlocks(OwnerBuffer first, OwnerBuffer second) {
-        const std::size_t size_1 = first->GetSize();
-        const std::size_t size_2 = second->GetSize();
-        const VAddr first_addr = first->GetCpuAddr();
-        const VAddr second_addr = second->GetCpuAddr();
+    std::shared_ptr<Buffer> MergeBlocks(std::shared_ptr<Buffer> first,
+                                        std::shared_ptr<Buffer> second) {
+        const std::size_t size_1 = first->Size();
+        const std::size_t size_2 = second->Size();
+        const VAddr first_addr = first->CpuAddr();
+        const VAddr second_addr = second->CpuAddr();
        const VAddr new_addr = std::min(first_addr, second_addr);
        const std::size_t new_size = size_1 + size_2;
-        OwnerBuffer new_buffer = CreateBlock(new_addr, new_size);
-        CopyBlock(first, new_buffer, 0, new_buffer->GetOffset(first_addr), size_1);
-        CopyBlock(second, new_buffer, 0, new_buffer->GetOffset(second_addr), size_2);
-        first->SetEpoch(epoch);
-        second->SetEpoch(epoch);
-        pending_destruction.push_back(first);
-        pending_destruction.push_back(second);
+
+        std::shared_ptr<Buffer> new_buffer = CreateBlock(new_addr, new_size);
+        CopyBlock(*first, *new_buffer, 0, new_buffer->Offset(first_addr), size_1);
+        CopyBlock(*second, *new_buffer, 0, new_buffer->Offset(second_addr), size_2);
+        QueueDestruction(std::move(first));
+        QueueDestruction(std::move(second));
+
        const VAddr cpu_addr_end = new_addr + new_size - 1;
-        u64 page_start = new_addr >> block_page_bits;
-        const u64 page_end = cpu_addr_end >> block_page_bits;
-        while (page_start <= page_end) {
-            blocks[page_start] = new_buffer;
-            ++page_start;
+        const u64 page_end = cpu_addr_end >> BLOCK_PAGE_BITS;
+        for (u64 page_start = new_addr >> BLOCK_PAGE_BITS; page_start <= page_end; ++page_start) {
+            blocks.insert_or_assign(page_start, new_buffer);
        }
        return new_buffer;
    }

-    OwnerBuffer GetBlock(const VAddr cpu_addr, const std::size_t size) {
-        OwnerBuffer found;
+    Buffer* GetBlock(VAddr cpu_addr, std::size_t size) {
+        std::shared_ptr<Buffer> found;
+
        const VAddr cpu_addr_end = cpu_addr + size - 1;
-        u64 page_start = cpu_addr >> block_page_bits;
-        const u64 page_end = cpu_addr_end >> block_page_bits;
-        while (page_start <= page_end) {
+        const u64 page_end = cpu_addr_end >> BLOCK_PAGE_BITS;
+        for (u64 page_start = cpu_addr >> BLOCK_PAGE_BITS; page_start <= page_end; ++page_start) {
            auto it = blocks.find(page_start);
            if (it == blocks.end()) {
                if (found) {
                    found = EnlargeBlock(found);
-                } else {
-                    const VAddr start_addr = (page_start << block_page_bits);
-                    found = CreateBlock(start_addr, block_page_size);
-                    blocks[page_start] = found;
-                }
-            } else {
-                if (found) {
-                    if (found == it->second) {
-                        ++page_start;
-                        continue;
-                    }
-                    found = MergeBlocks(found, it->second);
-                } else {
-                    found = it->second;
+                    continue;
                }
+                const VAddr start_addr = page_start << BLOCK_PAGE_BITS;
+                found = CreateBlock(start_addr, BLOCK_PAGE_SIZE);
+                blocks.insert_or_assign(page_start, found);
+                continue;
+            }
+            if (!found) {
+                found = it->second;
+                continue;
+            }
+            if (found != it->second) {
+                found = MergeBlocks(std::move(found), it->second);
            }
-            ++page_start;
        }
-        return found;
+        return found.get();
    }

-    void MarkRegionAsWritten(const VAddr start, const VAddr end) {
-        u64 page_start = start >> write_page_bit;
-        const u64 page_end = end >> write_page_bit;
-        while (page_start <= page_end) {
+    void MarkRegionAsWritten(VAddr start, VAddr end) {
+        const u64 page_end = end >> WRITE_PAGE_BIT;
+        for (u64 page_start = start >> WRITE_PAGE_BIT; page_start <= page_end; ++page_start) {
            auto it = written_pages.find(page_start);
            if (it != written_pages.end()) {
                it->second = it->second + 1;
            } else {
-                written_pages[page_start] = 1;
+                written_pages.insert_or_assign(page_start, 1);
            }
-            ++page_start;
        }
    }

-    void UnmarkRegionAsWritten(const VAddr start, const VAddr end) {
-        u64 page_start = start >> write_page_bit;
-        const u64 page_end = end >> write_page_bit;
-        while (page_start <= page_end) {
+    void UnmarkRegionAsWritten(VAddr start, VAddr end) {
+        const u64 page_end = end >> WRITE_PAGE_BIT;
+        for (u64 page_start = start >> WRITE_PAGE_BIT; page_start <= page_end; ++page_start) {
            auto it = written_pages.find(page_start);
            if (it != written_pages.end()) {
                if (it->second > 1) {
@@ -547,22 +551,24 @@ private:
                    written_pages.erase(it);
                }
            }
-            ++page_start;
        }
    }

-    bool IsRegionWritten(const VAddr start, const VAddr end) const {
-        u64 page_start = start >> write_page_bit;
-        const u64 page_end = end >> write_page_bit;
-        while (page_start <= page_end) {
+    bool IsRegionWritten(VAddr start, VAddr end) const {
+        const u64 page_end = end >> WRITE_PAGE_BIT;
+        for (u64 page_start = start >> WRITE_PAGE_BIT; page_start <= page_end; ++page_start) {
            if (written_pages.count(page_start) > 0) {
                return true;
            }
-            ++page_start;
        }
        return false;
    }

+    void QueueDestruction(std::shared_ptr<Buffer> buffer) {
+        buffer->SetEpoch(epoch);
+        pending_destruction.push(std::move(buffer));
+    }
+
    void MarkForAsyncFlush(MapInterval* map) {
        if (!uncommitted_flushes) {
            uncommitted_flushes = std::make_shared<std::unordered_set<MapInterval*>>();
@@ -574,9 +580,7 @@ private:
    Core::System& system;

    std::unique_ptr<StreamBuffer> stream_buffer;
-    BufferType stream_buffer_handle{};
-
-    bool invalidated = false;
+    BufferType stream_buffer_handle;

    u8* buffer_ptr = nullptr;
    u64 buffer_offset = 0;
@@ -586,18 +590,15 @@ private:
    boost::intrusive::set<MapInterval, boost::intrusive::compare<MapIntervalCompare>>
        mapped_addresses;

-    static constexpr u64 write_page_bit = 11;
    std::unordered_map<u64, u32> written_pages;
+    std::unordered_map<u64, std::shared_ptr<Buffer>> blocks;

-    static constexpr u64 block_page_bits = 21;
-    static constexpr u64 block_page_size = 1ULL << block_page_bits;
-    std::unordered_map<u64, OwnerBuffer> blocks;
-
-    std::list<OwnerBuffer> pending_destruction;
+    std::queue<std::shared_ptr<Buffer>> pending_destruction;
    u64 epoch = 0;
    u64 modified_ticks = 0;

    std::vector<u8> staging_buffer;
+
    std::list<MapInterval*> marked_for_unregister;

    std::shared_ptr<std::unordered_set<MapInterval*>> uncommitted_flushes;
@@ -93,6 +93,7 @@ public:
    virtual SamplerDescriptor AccessBoundSampler(ShaderType stage, u64 offset) const = 0;
    virtual SamplerDescriptor AccessBindlessSampler(ShaderType stage, u64 const_buffer,
                                                    u64 offset) const = 0;
+    virtual SamplerDescriptor AccessSampler(u32 handle) const = 0;
    virtual u32 GetBoundBuffer() const = 0;

    virtual VideoCore::GuestDriverProfile& AccessGuestDriverProfile() = 0;
@@ -92,8 +92,11 @@ SamplerDescriptor KeplerCompute::AccessBindlessSampler(ShaderType stage, u64 con
    ASSERT(stage == ShaderType::Compute);
    const auto& tex_info_buffer = launch_description.const_buffer_config[const_buffer];
    const GPUVAddr tex_info_address = tex_info_buffer.Address() + offset;
+    return AccessSampler(memory_manager.Read<u32>(tex_info_address));
+}

-    const Texture::TextureHandle tex_handle{memory_manager.Read<u32>(tex_info_address)};
+SamplerDescriptor KeplerCompute::AccessSampler(u32 handle) const {
+    const Texture::TextureHandle tex_handle{handle};
    const Texture::FullTextureInfo tex_info = GetTextureInfo(tex_handle);
    SamplerDescriptor result = SamplerDescriptor::FromTIC(tex_info.tic);
    result.is_shadow.Assign(tex_info.tsc.depth_compare_enabled.Value());
@@ -219,6 +219,8 @@ public:
    SamplerDescriptor AccessBindlessSampler(ShaderType stage, u64 const_buffer,
                                            u64 offset) const override;

+    SamplerDescriptor AccessSampler(u32 handle) const override;
+
    u32 GetBoundBuffer() const override {
        return regs.tex_cb_index;
    }
@@ -740,8 +740,11 @@ SamplerDescriptor Maxwell3D::AccessBindlessSampler(ShaderType stage, u64 const_b
    const auto& shader = state.shader_stages[static_cast<std::size_t>(stage)];
    const auto& tex_info_buffer = shader.const_buffers[const_buffer];
    const GPUVAddr tex_info_address = tex_info_buffer.address + offset;
+    return AccessSampler(memory_manager.Read<u32>(tex_info_address));
+}

-    const Texture::TextureHandle tex_handle{memory_manager.Read<u32>(tex_info_address)};
+SamplerDescriptor Maxwell3D::AccessSampler(u32 handle) const {
+    const Texture::TextureHandle tex_handle{handle};
    const Texture::FullTextureInfo tex_info = GetTextureInfo(tex_handle);
    SamplerDescriptor result = SamplerDescriptor::FromTIC(tex_info.tic);
    result.is_shadow.Assign(tex_info.tsc.depth_compare_enabled.Value());
@@ -1404,6 +1404,8 @@ public:
    SamplerDescriptor AccessBindlessSampler(ShaderType stage, u64 const_buffer,
                                            u64 offset) const override;

+    SamplerDescriptor AccessSampler(u32 handle) const override;
+
    u32 GetBoundBuffer() const override {
        return regs.tex_cb_index;
    }
@@ -14,22 +14,16 @@ MICROPROFILE_DEFINE(MacroJitCompile, "GPU", "Compile macro JIT", MP_RGB(173, 255
 MICROPROFILE_DEFINE(MacroJitExecute, "GPU", "Execute macro JIT", MP_RGB(255, 255, 0));

 namespace Tegra {
-static const Xbyak::Reg64 PARAMETERS = Xbyak::util::r9;
-static const Xbyak::Reg64 REGISTERS = Xbyak::util::r10;
-static const Xbyak::Reg64 STATE = Xbyak::util::r11;
-static const Xbyak::Reg64 NEXT_PARAMETER = Xbyak::util::r12;
-static const Xbyak::Reg32 RESULT = Xbyak::util::r13d;
-static const Xbyak::Reg64 RESULT_64 = Xbyak::util::r13;
+static const Xbyak::Reg64 STATE = Xbyak::util::rbx;
+static const Xbyak::Reg32 RESULT = Xbyak::util::ebp;
+static const Xbyak::Reg64 PARAMETERS = Xbyak::util::r12;
 static const Xbyak::Reg32 METHOD_ADDRESS = Xbyak::util::r14d;
-static const Xbyak::Reg64 METHOD_ADDRESS_64 = Xbyak::util::r14;
 static const Xbyak::Reg64 BRANCH_HOLDER = Xbyak::util::r15;

 static const std::bitset<32> PERSISTENT_REGISTERS = Common::X64::BuildRegSet({
-    PARAMETERS,
-    REGISTERS,
    STATE,
-    NEXT_PARAMETER,
    RESULT,
+    PARAMETERS,
    METHOD_ADDRESS,
    BRANCH_HOLDER,
 });
@@ -53,8 +47,7 @@ void MacroJITx64Impl::Execute(const std::vector<u32>& parameters, u32 method) {
    JITState state{};
    state.maxwell3d = &maxwell3d;
    state.registers = {};
-    state.parameters = parameters.data();
-    program(&state);
+    program(&state, parameters.data());
 }

 void MacroJITx64Impl::Compile_ALU(Macro::Opcode opcode) {
@@ -64,18 +57,18 @@ void MacroJITx64Impl::Compile_ALU(Macro::Opcode opcode) {
    const bool is_move_operation = !is_a_zero && is_b_zero;
    const bool has_zero_register = is_a_zero || is_b_zero;

-    Xbyak::Reg64 src_a;
+    Xbyak::Reg32 src_a;
    Xbyak::Reg32 src_b;

    if (!optimizer.zero_reg_skip) {
-        src_a = Compile_GetRegister(opcode.src_a, RESULT_64);
-        src_b = Compile_GetRegister(opcode.src_b, ebx);
+        src_a = Compile_GetRegister(opcode.src_a, RESULT);
+        src_b = Compile_GetRegister(opcode.src_b, eax);
    } else {
        if (!is_a_zero) {
-            src_a = Compile_GetRegister(opcode.src_a, RESULT_64);
+            src_a = Compile_GetRegister(opcode.src_a, RESULT);
        }
        if (!is_b_zero) {
-            src_b = Compile_GetRegister(opcode.src_b, ebx);
+            src_b = Compile_GetRegister(opcode.src_b, eax);
        }
    }
    Xbyak::Label skip_carry{};
@@ -302,22 +295,22 @@ void MacroJITx64Impl::Compile_Read(Macro::Opcode opcode) {
            sub(result, opcode.immediate * -1);
        }
    }
-    Common::X64::ABI_PushRegistersAndAdjustStackGPS(*this, PersistentCallerSavedRegs(), 0);
+    Common::X64::ABI_PushRegistersAndAdjustStack(*this, PersistentCallerSavedRegs(), 0);
    mov(Common::X64::ABI_PARAM1, qword[STATE]);
    mov(Common::X64::ABI_PARAM2, RESULT);
    Common::X64::CallFarFunction(*this, &Read);
-    Common::X64::ABI_PopRegistersAndAdjustStackGPS(*this, PersistentCallerSavedRegs(), 0);
+    Common::X64::ABI_PopRegistersAndAdjustStack(*this, PersistentCallerSavedRegs(), 0);
    mov(RESULT, Common::X64::ABI_RETURN.cvt32());
    Compile_ProcessResult(opcode.result_operation, opcode.dst);
 }

 void Tegra::MacroJITx64Impl::Compile_Send(Xbyak::Reg32 value) {
-    Common::X64::ABI_PushRegistersAndAdjustStackGPS(*this, PersistentCallerSavedRegs(), 0);
+    Common::X64::ABI_PushRegistersAndAdjustStack(*this, PersistentCallerSavedRegs(), 0);
    mov(Common::X64::ABI_PARAM1, qword[STATE]);
    mov(Common::X64::ABI_PARAM2, METHOD_ADDRESS);
    mov(Common::X64::ABI_PARAM3, value);
    Common::X64::CallFarFunction(*this, &Send);
-    Common::X64::ABI_PopRegistersAndAdjustStackGPS(*this, PersistentCallerSavedRegs(), 0);
+    Common::X64::ABI_PopRegistersAndAdjustStack(*this, PersistentCallerSavedRegs(), 0);

    Xbyak::Label dont_process{};
    // Get increment
@@ -329,7 +322,7 @@ void Tegra::MacroJITx64Impl::Compile_Send(Xbyak::Reg32 value) {
    and_(METHOD_ADDRESS, 0xfff);
    shr(ecx, 12);
    and_(ecx, 0x3f);
-    lea(eax, ptr[rcx + METHOD_ADDRESS_64]);
+    lea(eax, ptr[rcx + METHOD_ADDRESS.cvt64()]);
    sal(ecx, 12);
    or_(eax, ecx);

@@ -421,19 +414,15 @@ void MacroJITx64Impl::Compile() {
    bool keep_executing = true;
    labels.fill(Xbyak::Label());

-    Common::X64::ABI_PushRegistersAndAdjustStackGPS(*this, Common::X64::ABI_ALL_CALLEE_SAVED, 8);
+    Common::X64::ABI_PushRegistersAndAdjustStack(*this, Common::X64::ABI_ALL_CALLEE_SAVED, 8);
    // JIT state
    mov(STATE, Common::X64::ABI_PARAM1);
-    mov(PARAMETERS, qword[Common::X64::ABI_PARAM1 +
-                          static_cast<Xbyak::uint32>(offsetof(JITState, parameters))]);
-    mov(REGISTERS, Common::X64::ABI_PARAM1);
-    add(REGISTERS, static_cast<Xbyak::uint32>(offsetof(JITState, registers)));
+    mov(PARAMETERS, Common::X64::ABI_PARAM2);
    xor_(RESULT, RESULT);
    xor_(METHOD_ADDRESS, METHOD_ADDRESS);
-    xor_(NEXT_PARAMETER, NEXT_PARAMETER);
    xor_(BRANCH_HOLDER, BRANCH_HOLDER);

-    mov(dword[REGISTERS + 4], Compile_FetchParameter());
+    mov(dword[STATE + offsetof(JITState, registers) + 4], Compile_FetchParameter());

    // Track get register for zero registers and mark it as no-op
    optimizer.zero_reg_skip = true;
@@ -463,7 +452,7 @@ void MacroJITx64Impl::Compile() {

    L(end_of_code);

-    Common::X64::ABI_PopRegistersAndAdjustStackGPS(*this, Common::X64::ABI_ALL_CALLEE_SAVED, 8);
+    Common::X64::ABI_PopRegistersAndAdjustStack(*this, Common::X64::ABI_ALL_CALLEE_SAVED, 8);
    ret();
    ready();
    program = getCode<ProgramType>();
@@ -537,8 +526,8 @@ bool MacroJITx64Impl::Compile_NextInstruction() {
 }

 Xbyak::Reg32 Tegra::MacroJITx64Impl::Compile_FetchParameter() {
-    mov(eax, dword[PARAMETERS + NEXT_PARAMETER * sizeof(u32)]);
-    inc(NEXT_PARAMETER);
+    mov(eax, dword[PARAMETERS]);
+    add(PARAMETERS, sizeof(u32));
    return eax;
 }

@@ -547,31 +536,12 @@ Xbyak::Reg32 MacroJITx64Impl::Compile_GetRegister(u32 index, Xbyak::Reg32 dst) {
        // Register 0 is always zero
        xor_(dst, dst);
    } else {
-        mov(dst, dword[REGISTERS + index * sizeof(u32)]);
+        mov(dst, dword[STATE + offsetof(JITState, registers) + index * sizeof(u32)]);
    }

    return dst;
 }

-Xbyak::Reg64 Tegra::MacroJITx64Impl::Compile_GetRegister(u32 index, Xbyak::Reg64 dst) {
-    if (index == 0) {
-        // Register 0 is always zero
-        xor_(dst, dst);
-    } else {
-        mov(dst, dword[REGISTERS + index * sizeof(u32)]);
-    }
-
-    return dst;
-}
-
-void Tegra::MacroJITx64Impl::Compile_WriteCarry(Xbyak::Reg64 dst) {
-    Xbyak::Label zero{}, end{};
-    xor_(ecx, ecx);
-    shr(dst, 32);
-    setne(cl);
-    mov(dword[STATE + offsetof(JITState, carry_flag)], ecx);
-}
-
 void MacroJITx64Impl::Compile_ProcessResult(Macro::ResultOperation operation, u32 reg) {
    auto SetRegister = [=](u32 reg, Xbyak::Reg32 result) {
        // Register 0 is supposed to always return 0. NOP is implemented as a store to the zero
@@ -579,7 +549,7 @@ void MacroJITx64Impl::Compile_ProcessResult(Macro::ResultOperation operation, u3
        if (reg == 0) {
            return;
        }
-        mov(dword[REGISTERS + reg * sizeof(u32)], result);
+        mov(dword[STATE + offsetof(JITState, registers) + reg * sizeof(u32)], result);
    };
    auto SetMethodAddress = [=](Xbyak::Reg32 reg) { mov(METHOD_ADDRESS, reg); };

@@ -55,8 +55,6 @@ private:

    Xbyak::Reg32 Compile_FetchParameter();
    Xbyak::Reg32 Compile_GetRegister(u32 index, Xbyak::Reg32 dst);
-    Xbyak::Reg64 Compile_GetRegister(u32 index, Xbyak::Reg64 dst);
-    void Compile_WriteCarry(Xbyak::Reg64 dst);

    void Compile_ProcessResult(Macro::ResultOperation operation, u32 reg);
    void Compile_Send(Xbyak::Reg32 value);
@@ -67,11 +65,10 @@ private:
    struct JITState {
        Engines::Maxwell3D* maxwell3d{};
        std::array<u32, Macro::NUM_MACRO_REGISTERS> registers{};
-        const u32* parameters{};
        u32 carry_flag{};
    };
    static_assert(offsetof(JITState, maxwell3d) == 0, "Maxwell3D is not at 0x0");
-    using ProgramType = void (*)(JITState*);
+    using ProgramType = void (*)(JITState*, const u32*);

    struct OptimizerState {
        bool can_skip_carry{};
@@ -85,8 +82,8 @@ private:
    std::optional<Macro::Opcode> next_opcode{};
    ProgramType program{nullptr};

-    std::array<Xbyak::Label, MAX_CODE_SIZE> labels{};
-    std::array<Xbyak::Label, MAX_CODE_SIZE> delay_skip{};
+    std::array<Xbyak::Label, MAX_CODE_SIZE> labels;
+    std::array<Xbyak::Label, MAX_CODE_SIZE> delay_skip;
    Xbyak::Label end_of_code{};

    bool is_delay_slot{};
@@ -0,0 +1,29 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <string>
+#include <string_view>
+
+#include "common/common_types.h"
+
+namespace Tegra::Engines {
+enum class ShaderType : u32;
+}
+
+namespace VideoCommon::Shader {
+class ShaderIR;
+class Registry;
+} // namespace VideoCommon::Shader
+
+namespace OpenGL {
+
+class Device;
+
+std::string DecompileAssemblyShader(const Device& device, const VideoCommon::Shader::ShaderIR& ir,
+                                    const VideoCommon::Shader::Registry& registry,
+                                    Tegra::Engines::ShaderType stage, std::string_view identifier);
+
+} // namespace OpenGL
@@ -22,13 +22,12 @@ using Maxwell = Tegra::Engines::Maxwell3D::Regs;

 MICROPROFILE_DEFINE(OpenGL_Buffer_Download, "OpenGL", "Buffer Download", MP_RGB(192, 192, 128));

-CachedBufferBlock::CachedBufferBlock(VAddr cpu_addr, const std::size_t size)
-    : VideoCommon::BufferBlock{cpu_addr, size} {
+Buffer::Buffer(VAddr cpu_addr, const std::size_t size) : VideoCommon::BufferBlock{cpu_addr, size} {
    gl_buffer.Create();
    glNamedBufferData(gl_buffer.handle, static_cast<GLsizeiptr>(size), nullptr, GL_DYNAMIC_DRAW);
 }

-CachedBufferBlock::~CachedBufferBlock() = default;
+Buffer::~Buffer() = default;

 OGLBufferCache::OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system,
                               const Device& device, std::size_t stream_size)
@@ -48,12 +47,8 @@ OGLBufferCache::~OGLBufferCache() {
    glDeleteBuffers(static_cast<GLsizei>(std::size(cbufs)), std::data(cbufs));
 }

-Buffer OGLBufferCache::CreateBlock(VAddr cpu_addr, std::size_t size) {
-    return std::make_shared<CachedBufferBlock>(cpu_addr, size);
-}
-
-GLuint OGLBufferCache::ToHandle(const Buffer& buffer) {
-    return buffer->GetHandle();
+std::shared_ptr<Buffer> OGLBufferCache::CreateBlock(VAddr cpu_addr, std::size_t size) {
+    return std::make_shared<Buffer>(cpu_addr, size);
 }

 GLuint OGLBufferCache::GetEmptyBuffer(std::size_t) {
@@ -62,7 +57,7 @@ GLuint OGLBufferCache::GetEmptyBuffer(std::size_t) {

 void OGLBufferCache::UploadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size,
                                     const u8* data) {
-    glNamedBufferSubData(buffer->GetHandle(), static_cast<GLintptr>(offset),
+    glNamedBufferSubData(buffer.Handle(), static_cast<GLintptr>(offset),
                         static_cast<GLsizeiptr>(size), data);
 }

@@ -70,20 +65,20 @@ void OGLBufferCache::DownloadBlockData(const Buffer& buffer, std::size_t offset,
                                       u8* data) {
    MICROPROFILE_SCOPE(OpenGL_Buffer_Download);
    glMemoryBarrier(GL_BUFFER_UPDATE_BARRIER_BIT);
-    glGetNamedBufferSubData(buffer->GetHandle(), static_cast<GLintptr>(offset),
+    glGetNamedBufferSubData(buffer.Handle(), static_cast<GLintptr>(offset),
                            static_cast<GLsizeiptr>(size), data);
 }

 void OGLBufferCache::CopyBlock(const Buffer& src, const Buffer& dst, std::size_t src_offset,
                               std::size_t dst_offset, std::size_t size) {
-    glCopyNamedBufferSubData(src->GetHandle(), dst->GetHandle(), static_cast<GLintptr>(src_offset),
+    glCopyNamedBufferSubData(src.Handle(), dst.Handle(), static_cast<GLintptr>(src_offset),
                             static_cast<GLintptr>(dst_offset), static_cast<GLsizeiptr>(size));
 }

 OGLBufferCache::BufferInfo OGLBufferCache::ConstBufferUpload(const void* raw_pointer,
                                                             std::size_t size) {
    DEBUG_ASSERT(cbuf_cursor < std::size(cbufs));
-    const GLuint& cbuf = cbufs[cbuf_cursor++];
+    const GLuint cbuf = cbufs[cbuf_cursor++];
    glNamedBufferSubData(cbuf, 0, static_cast<GLsizeiptr>(size), raw_pointer);
    return {cbuf, 0};
 }
@@ -23,17 +23,12 @@ class Device;
 class OGLStreamBuffer;
 class RasterizerOpenGL;

-class CachedBufferBlock;
-
-using Buffer = std::shared_ptr<CachedBufferBlock>;
-using GenericBufferCache = VideoCommon::BufferCache<Buffer, GLuint, OGLStreamBuffer>;
-
-class CachedBufferBlock : public VideoCommon::BufferBlock {
+class Buffer : public VideoCommon::BufferBlock {
 public:
-    explicit CachedBufferBlock(VAddr cpu_addr, const std::size_t size);
-    ~CachedBufferBlock();
+    explicit Buffer(VAddr cpu_addr, const std::size_t size);
+    ~Buffer();

-    GLuint GetHandle() const {
+    GLuint Handle() const {
        return gl_buffer.handle;
    }

@@ -41,6 +36,7 @@ private:
    OGLBuffer gl_buffer;
 };

+using GenericBufferCache = VideoCommon::BufferCache<Buffer, GLuint, OGLStreamBuffer>;
 class OGLBufferCache final : public GenericBufferCache {
 public:
    explicit OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system,
@@ -54,9 +50,7 @@ public:
    }

 protected:
-    Buffer CreateBlock(VAddr cpu_addr, std::size_t size) override;
-
-    GLuint ToHandle(const Buffer& buffer) override;
+    std::shared_ptr<Buffer> CreateBlock(VAddr cpu_addr, std::size_t size) override;

    void UploadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size,
                         const u8* data) override;
@@ -213,6 +213,7 @@ Device::Device()
    has_component_indexing_bug = is_amd;
    has_precise_bug = TestPreciseBug();
    has_fast_buffer_sub_data = is_nvidia && !disable_fast_buffer_sub_data;
+    has_nv_viewport_array2 = GLAD_GL_NV_viewport_array2;
    use_assembly_shaders = Settings::values.use_assembly_shaders && GLAD_GL_NV_gpu_program5 &&
                           GLAD_GL_NV_compute_program5 && GLAD_GL_NV_transform_feedback &&
                           GLAD_GL_NV_transform_feedback2;
@@ -88,6 +88,10 @@ public:
        return has_fast_buffer_sub_data;
    }

+    bool HasNvViewportArray2() const {
+        return has_nv_viewport_array2;
+    }
+
    bool UseAssemblyShaders() const {
        return use_assembly_shaders;
    }
@@ -111,6 +115,7 @@ private:
    bool has_component_indexing_bug{};
    bool has_precise_bug{};
    bool has_fast_buffer_sub_data{};
+    bool has_nv_viewport_array2{};
    bool use_assembly_shaders{};
 };

@@ -66,10 +66,22 @@ constexpr std::size_t NumSupportedVertexAttributes = 16;
 template <typename Engine, typename Entry>
 Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry,
                                               ShaderType shader_type, std::size_t index = 0) {
-    if (entry.is_bindless) {
-        const auto tex_handle = engine.AccessConstBuffer32(shader_type, entry.buffer, entry.offset);
-        return engine.GetTextureInfo(tex_handle);
+    if constexpr (std::is_same_v<Entry, SamplerEntry>) {
+        if (entry.is_separated) {
+            const u32 buffer_1 = entry.buffer;
+            const u32 buffer_2 = entry.secondary_buffer;
+            const u32 offset_1 = entry.offset;
+            const u32 offset_2 = entry.secondary_offset;
+            const u32 handle_1 = engine.AccessConstBuffer32(shader_type, buffer_1, offset_1);
+            const u32 handle_2 = engine.AccessConstBuffer32(shader_type, buffer_2, offset_2);
+            return engine.GetTextureInfo(handle_1 | handle_2);
+        }
    }
+    if (entry.is_bindless) {
+        const u32 handle = engine.AccessConstBuffer32(shader_type, entry.buffer, entry.offset);
+        return engine.GetTextureInfo(handle);
+    }
+
    const auto& gpu_profile = engine.AccessGuestDriverProfile();
    const u32 offset = entry.offset + static_cast<u32>(index * gpu_profile.GetTextureHandlerSize());
    if constexpr (std::is_same_v<Engine, Tegra::Engines::Maxwell3D>) {
@@ -605,7 +617,16 @@ void RasterizerOpenGL::Draw(bool is_indexed, bool is_instanced) {
                   (Maxwell::MaxConstBufferSize + device.GetUniformBufferAlignment());

    // Prepare the vertex array.
-    buffer_cache.Map(buffer_size);
+    const bool invalidated = buffer_cache.Map(buffer_size);
+
+    if (invalidated) {
+        // When the stream buffer has been invalidated, we have to consider vertex buffers as dirty
+        auto& dirty = gpu.dirty.flags;
+        dirty[Dirty::VertexBuffers] = true;
+        for (int index = Dirty::VertexBuffer0; index <= Dirty::VertexBuffer31; ++index) {
+            dirty[index] = true;
+        }
+    }

    // Prepare vertex array format.
    SetupVertexFormat();
@@ -20,6 +20,7 @@
 #include "video_core/engines/maxwell_3d.h"
 #include "video_core/engines/shader_type.h"
 #include "video_core/memory_manager.h"
+#include "video_core/renderer_opengl/gl_arb_decompiler.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_decompiler.h"
@@ -148,7 +149,8 @@ ProgramSharedPtr BuildShader(const Device& device, ShaderType shader_type, u64 u
    auto program = std::make_shared<ProgramHandle>();

    if (device.UseAssemblyShaders()) {
-        const std::string arb = "Not implemented";
+        const std::string arb =
+            DecompileAssemblyShader(device, ir, registry, shader_type, shader_id);

        GLuint& arb_prog = program->assembly_program.handle;

@@ -29,6 +29,8 @@ using VideoCommon::Shader::KeyMap;

 namespace {

+using VideoCommon::Shader::SeparateSamplerKey;
+
 using ShaderCacheVersionHash = std::array<u8, 64>;

 struct ConstBufferKey {
@@ -37,18 +39,26 @@ struct ConstBufferKey {
    u32 value = 0;
 };

-struct BoundSamplerKey {
+struct BoundSamplerEntry {
    u32 offset = 0;
    Tegra::Engines::SamplerDescriptor sampler;
 };

-struct BindlessSamplerKey {
+struct SeparateSamplerEntry {
+    u32 cbuf1 = 0;
+    u32 cbuf2 = 0;
+    u32 offset1 = 0;
+    u32 offset2 = 0;
+    Tegra::Engines::SamplerDescriptor sampler;
+};
+
+struct BindlessSamplerEntry {
    u32 cbuf = 0;
    u32 offset = 0;
    Tegra::Engines::SamplerDescriptor sampler;
 };

-constexpr u32 NativeVersion = 20;
+constexpr u32 NativeVersion = 21;

 ShaderCacheVersionHash GetShaderCacheVersionHash() {
    ShaderCacheVersionHash hash{};
@@ -87,12 +97,14 @@ bool ShaderDiskCacheEntry::Load(FileUtil::IOFile& file) {
    u32 texture_handler_size_value;
    u32 num_keys;
    u32 num_bound_samplers;
+    u32 num_separate_samplers;
    u32 num_bindless_samplers;
    if (file.ReadArray(&unique_identifier, 1) != 1 || file.ReadArray(&bound_buffer, 1) != 1 ||
        file.ReadArray(&is_texture_handler_size_known, 1) != 1 ||
        file.ReadArray(&texture_handler_size_value, 1) != 1 ||
        file.ReadArray(&graphics_info, 1) != 1 || file.ReadArray(&compute_info, 1) != 1 ||
        file.ReadArray(&num_keys, 1) != 1 || file.ReadArray(&num_bound_samplers, 1) != 1 ||
+        file.ReadArray(&num_separate_samplers, 1) != 1 ||
        file.ReadArray(&num_bindless_samplers, 1) != 1) {
        return false;
    }
@@ -101,23 +113,32 @@ bool ShaderDiskCacheEntry::Load(FileUtil::IOFile& file) {
    }

    std::vector<ConstBufferKey> flat_keys(num_keys);
-    std::vector<BoundSamplerKey> flat_bound_samplers(num_bound_samplers);
-    std::vector<BindlessSamplerKey> flat_bindless_samplers(num_bindless_samplers);
+    std::vector<BoundSamplerEntry> flat_bound_samplers(num_bound_samplers);
+    std::vector<SeparateSamplerEntry> flat_separate_samplers(num_separate_samplers);
+    std::vector<BindlessSamplerEntry> flat_bindless_samplers(num_bindless_samplers);
    if (file.ReadArray(flat_keys.data(), flat_keys.size()) != flat_keys.size() ||
        file.ReadArray(flat_bound_samplers.data(), flat_bound_samplers.size()) !=
            flat_bound_samplers.size() ||
+        file.ReadArray(flat_separate_samplers.data(), flat_separate_samplers.size()) !=
+            flat_separate_samplers.size() ||
        file.ReadArray(flat_bindless_samplers.data(), flat_bindless_samplers.size()) !=
            flat_bindless_samplers.size()) {
        return false;
    }
-    for (const auto& key : flat_keys) {
-        keys.insert({{key.cbuf, key.offset}, key.value});
+    for (const auto& entry : flat_keys) {
+        keys.insert({{entry.cbuf, entry.offset}, entry.value});
    }
-    for (const auto& key : flat_bound_samplers) {
-        bound_samplers.emplace(key.offset, key.sampler);
+    for (const auto& entry : flat_bound_samplers) {
+        bound_samplers.emplace(entry.offset, entry.sampler);
    }
-    for (const auto& key : flat_bindless_samplers) {
-        bindless_samplers.insert({{key.cbuf, key.offset}, key.sampler});
+    for (const auto& entry : flat_separate_samplers) {
+        SeparateSamplerKey key;
+        key.buffers = {entry.cbuf1, entry.cbuf2};
+        key.offsets = {entry.offset1, entry.offset2};
+        separate_samplers.emplace(key, entry.sampler);
+    }
+    for (const auto& entry : flat_bindless_samplers) {
+        bindless_samplers.insert({{entry.cbuf, entry.offset}, entry.sampler});
    }

    return true;
@@ -142,6 +163,7 @@ bool ShaderDiskCacheEntry::Save(FileUtil::IOFile& file) const {
        file.WriteObject(graphics_info) != 1 || file.WriteObject(compute_info) != 1 ||
        file.WriteObject(static_cast<u32>(keys.size())) != 1 ||
        file.WriteObject(static_cast<u32>(bound_samplers.size())) != 1 ||
+        file.WriteObject(static_cast<u32>(separate_samplers.size())) != 1 ||
        file.WriteObject(static_cast<u32>(bindless_samplers.size())) != 1) {
        return false;
    }
@@ -152,22 +174,34 @@ bool ShaderDiskCacheEntry::Save(FileUtil::IOFile& file) const {
        flat_keys.push_back(ConstBufferKey{address.first, address.second, value});
    }

-    std::vector<BoundSamplerKey> flat_bound_samplers;
+    std::vector<BoundSamplerEntry> flat_bound_samplers;
    flat_bound_samplers.reserve(bound_samplers.size());
    for (const auto& [address, sampler] : bound_samplers) {
-        flat_bound_samplers.push_back(BoundSamplerKey{address, sampler});
+        flat_bound_samplers.push_back(BoundSamplerEntry{address, sampler});
    }

-    std::vector<BindlessSamplerKey> flat_bindless_samplers;
+    std::vector<SeparateSamplerEntry> flat_separate_samplers;
+    flat_separate_samplers.reserve(separate_samplers.size());
+    for (const auto& [key, sampler] : separate_samplers) {
+        SeparateSamplerEntry entry;
+        std::tie(entry.cbuf1, entry.cbuf2) = key.buffers;
+        std::tie(entry.offset1, entry.offset2) = key.offsets;
+        entry.sampler = sampler;
+        flat_separate_samplers.push_back(entry);
+    }
+
+    std::vector<BindlessSamplerEntry> flat_bindless_samplers;
    flat_bindless_samplers.reserve(bindless_samplers.size());
    for (const auto& [address, sampler] : bindless_samplers) {
        flat_bindless_samplers.push_back(
-            BindlessSamplerKey{address.first, address.second, sampler});
+            BindlessSamplerEntry{address.first, address.second, sampler});
    }

    return file.WriteArray(flat_keys.data(), flat_keys.size()) == flat_keys.size() &&
           file.WriteArray(flat_bound_samplers.data(), flat_bound_samplers.size()) ==
               flat_bound_samplers.size() &&
+           file.WriteArray(flat_separate_samplers.data(), flat_separate_samplers.size()) ==
+               flat_separate_samplers.size() &&
           file.WriteArray(flat_bindless_samplers.data(), flat_bindless_samplers.size()) ==
               flat_bindless_samplers.size();
 }
@@ -57,6 +57,7 @@ struct ShaderDiskCacheEntry {
    VideoCommon::Shader::ComputeInfo compute_info;
    VideoCommon::Shader::KeyMap keys;
    VideoCommon::Shader::BoundSamplerMap bound_samplers;
+    VideoCommon::Shader::SeparateSamplerMap separate_samplers;
    VideoCommon::Shader::BindlessSamplerMap bindless_samplers;
 };

@@ -49,14 +49,6 @@ OGLStreamBuffer::~OGLStreamBuffer() {
    gl_buffer.Release();
 }

-GLuint OGLStreamBuffer::GetHandle() const {
-    return gl_buffer.handle;
-}
-
-GLsizeiptr OGLStreamBuffer::GetSize() const {
-    return buffer_size;
-}
-
 std::tuple<u8*, GLintptr, bool> OGLStreamBuffer::Map(GLsizeiptr size, GLintptr alignment) {
    ASSERT(size <= buffer_size);
    ASSERT(alignment <= buffer_size);
@@ -17,9 +17,6 @@ public:
                             bool use_persistent = true);
    ~OGLStreamBuffer();

-    GLuint GetHandle() const;
-    GLsizeiptr GetSize() const;
-
    /*
     * Allocates a linear chunk of memory in the GPU buffer with at least "size" bytes
     * and the optional alignment requirement.
@@ -32,6 +29,14 @@ public:

    void Unmap(GLsizeiptr size);

+    GLuint Handle() const {
+        return gl_buffer.handle;
+    }
+
+    GLsizeiptr Size() const {
+        return buffer_size;
+    }
+
 private:
    OGLBuffer gl_buffer;

@@ -37,8 +37,8 @@ std::unique_ptr<VKStreamBuffer> CreateStreamBuffer(const VKDevice& device, VKSch

 } // Anonymous namespace

-CachedBufferBlock::CachedBufferBlock(const VKDevice& device, VKMemoryManager& memory_manager,
-                                     VAddr cpu_addr, std::size_t size)
+Buffer::Buffer(const VKDevice& device, VKMemoryManager& memory_manager, VAddr cpu_addr,
+               std::size_t size)
    : VideoCommon::BufferBlock{cpu_addr, size} {
    VkBufferCreateInfo ci;
    ci.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO;
@@ -54,7 +54,7 @@ CachedBufferBlock::CachedBufferBlock(const VKDevice& device, VKMemoryManager& me
    buffer.commit = memory_manager.Commit(buffer.handle, false);
 }

-CachedBufferBlock::~CachedBufferBlock() = default;
+Buffer::~Buffer() = default;

 VKBufferCache::VKBufferCache(VideoCore::RasterizerInterface& rasterizer, Core::System& system,
                             const VKDevice& device, VKMemoryManager& memory_manager,
@@ -67,12 +67,8 @@ VKBufferCache::VKBufferCache(VideoCore::RasterizerInterface& rasterizer, Core::S

 VKBufferCache::~VKBufferCache() = default;

-Buffer VKBufferCache::CreateBlock(VAddr cpu_addr, std::size_t size) {
-    return std::make_shared<CachedBufferBlock>(device, memory_manager, cpu_addr, size);
-}
-
-VkBuffer VKBufferCache::ToHandle(const Buffer& buffer) {
-    return buffer->GetHandle();
+std::shared_ptr<Buffer> VKBufferCache::CreateBlock(VAddr cpu_addr, std::size_t size) {
+    return std::make_shared<Buffer>(device, memory_manager, cpu_addr, size);
 }

 VkBuffer VKBufferCache::GetEmptyBuffer(std::size_t size) {
@@ -91,7 +87,7 @@ void VKBufferCache::UploadBlockData(const Buffer& buffer, std::size_t offset, st
    std::memcpy(staging.commit->Map(size), data, size);

    scheduler.RequestOutsideRenderPassOperationContext();
-    scheduler.Record([staging = *staging.handle, buffer = buffer->GetHandle(), offset,
+    scheduler.Record([staging = *staging.handle, buffer = buffer.Handle(), offset,
                      size](vk::CommandBuffer cmdbuf) {
        cmdbuf.CopyBuffer(staging, buffer, VkBufferCopy{0, offset, size});

@@ -114,7 +110,7 @@ void VKBufferCache::DownloadBlockData(const Buffer& buffer, std::size_t offset,
                                      u8* data) {
    const auto& staging = staging_pool.GetUnusedBuffer(size, true);
    scheduler.RequestOutsideRenderPassOperationContext();
-    scheduler.Record([staging = *staging.handle, buffer = buffer->GetHandle(), offset,
+    scheduler.Record([staging = *staging.handle, buffer = buffer.Handle(), offset,
                      size](vk::CommandBuffer cmdbuf) {
        VkBufferMemoryBarrier barrier;
        barrier.sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER;
@@ -141,8 +137,8 @@ void VKBufferCache::DownloadBlockData(const Buffer& buffer, std::size_t offset,
 void VKBufferCache::CopyBlock(const Buffer& src, const Buffer& dst, std::size_t src_offset,
                              std::size_t dst_offset, std::size_t size) {
    scheduler.RequestOutsideRenderPassOperationContext();
-    scheduler.Record([src_buffer = src->GetHandle(), dst_buffer = dst->GetHandle(), src_offset,
-                      dst_offset, size](vk::CommandBuffer cmdbuf) {
+    scheduler.Record([src_buffer = src.Handle(), dst_buffer = dst.Handle(), src_offset, dst_offset,
+                      size](vk::CommandBuffer cmdbuf) {
        cmdbuf.CopyBuffer(src_buffer, dst_buffer, VkBufferCopy{src_offset, dst_offset, size});

        std::array<VkBufferMemoryBarrier, 2> barriers;
@@ -23,13 +23,13 @@ class VKDevice;
 class VKMemoryManager;
 class VKScheduler;

-class CachedBufferBlock final : public VideoCommon::BufferBlock {
+class Buffer final : public VideoCommon::BufferBlock {
 public:
-    explicit CachedBufferBlock(const VKDevice& device, VKMemoryManager& memory_manager,
-                               VAddr cpu_addr, std::size_t size);
-    ~CachedBufferBlock();
+    explicit Buffer(const VKDevice& device, VKMemoryManager& memory_manager, VAddr cpu_addr,
+                    std::size_t size);
+    ~Buffer();

-    VkBuffer GetHandle() const {
+    VkBuffer Handle() const {
        return *buffer.handle;
    }

@@ -37,8 +37,6 @@ private:
    VKBuffer buffer;
 };

-using Buffer = std::shared_ptr<CachedBufferBlock>;
-
 class VKBufferCache final : public VideoCommon::BufferCache<Buffer, VkBuffer, VKStreamBuffer> {
 public:
    explicit VKBufferCache(VideoCore::RasterizerInterface& rasterizer, Core::System& system,
@@ -49,9 +47,7 @@ public:
    VkBuffer GetEmptyBuffer(std::size_t size) override;

 protected:
-    VkBuffer ToHandle(const Buffer& buffer) override;
-
-    Buffer CreateBlock(VAddr cpu_addr, std::size_t size) override;
+    std::shared_ptr<Buffer> CreateBlock(VAddr cpu_addr, std::size_t size) override;

    void UploadBlockData(const Buffer& buffer, std::size_t offset, std::size_t size,
                         const u8* data) override;
@@ -118,6 +118,17 @@ template <typename Engine, typename Entry>
 Tegra::Texture::FullTextureInfo GetTextureInfo(const Engine& engine, const Entry& entry,
                                               std::size_t stage, std::size_t index = 0) {
    const auto stage_type = static_cast<Tegra::Engines::ShaderType>(stage);
+    if constexpr (std::is_same_v<Entry, SamplerEntry>) {
+        if (entry.is_separated) {
+            const u32 buffer_1 = entry.buffer;
+            const u32 buffer_2 = entry.secondary_buffer;
+            const u32 offset_1 = entry.offset;
+            const u32 offset_2 = entry.secondary_offset;
+            const u32 handle_1 = engine.AccessConstBuffer32(stage_type, buffer_1, offset_1);
+            const u32 handle_2 = engine.AccessConstBuffer32(stage_type, buffer_2, offset_2);
+            return engine.GetTextureInfo(handle_1 | handle_2);
+        }
+    }
    if (entry.is_bindless) {
        const auto tex_handle = engine.AccessConstBuffer32(stage_type, entry.buffer, entry.offset);
        return engine.GetTextureInfo(tex_handle);
@@ -35,7 +35,7 @@ public:
    /// Ensures that "size" bytes of memory are available to the GPU, potentially recording a copy.
    void Unmap(u64 size);

-    VkBuffer GetHandle() const {
+    VkBuffer Handle() const {
        return *buffer;
    }

@@ -357,13 +357,11 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
    return pc;
 }

-ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo(SamplerInfo info, u32 offset,
-                                               std::optional<u32> buffer) {
+ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo(
+    SamplerInfo info, std::optional<Tegra::Engines::SamplerDescriptor> sampler) {
    if (info.IsComplete()) {
        return info;
    }
-    const auto sampler = buffer ? registry.ObtainBindlessSampler(*buffer, offset)
-                                : registry.ObtainBoundSampler(offset);
    if (!sampler) {
        LOG_WARNING(HW_GPU, "Unknown sampler info");
        info.type = info.type.value_or(Tegra::Shader::TextureType::Texture2D);
@@ -381,8 +379,8 @@ ShaderIR::SamplerInfo ShaderIR::GetSamplerInfo(SamplerInfo info, u32 offset,

 std::optional<Sampler> ShaderIR::GetSampler(Tegra::Shader::Sampler sampler,
                                            SamplerInfo sampler_info) {
-    const auto offset = static_cast<u32>(sampler.index.Value());
-    const auto info = GetSamplerInfo(sampler_info, offset);
+    const u32 offset = static_cast<u32>(sampler.index.Value());
+    const auto info = GetSamplerInfo(sampler_info, registry.ObtainBoundSampler(offset));

    // If this sampler has already been used, return the existing mapping.
    const auto it = std::find_if(used_samplers.begin(), used_samplers.end(),
@@ -404,20 +402,19 @@ std::optional<Sampler> ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg,
    const Node sampler_register = GetRegister(reg);
    const auto [base_node, tracked_sampler_info] =
        TrackBindlessSampler(sampler_register, global_code, static_cast<s64>(global_code.size()));
-    ASSERT(base_node != nullptr);
-    if (base_node == nullptr) {
+    if (!base_node) {
+        UNREACHABLE();
        return std::nullopt;
    }

-    if (const auto bindless_sampler_info =
-            std::get_if<BindlessSamplerNode>(&*tracked_sampler_info)) {
-        const u32 buffer = bindless_sampler_info->GetIndex();
-        const u32 offset = bindless_sampler_info->GetOffset();
-        info = GetSamplerInfo(info, offset, buffer);
+    if (const auto sampler_info = std::get_if<BindlessSamplerNode>(&*tracked_sampler_info)) {
+        const u32 buffer = sampler_info->index;
+        const u32 offset = sampler_info->offset;
+        info = GetSamplerInfo(info, registry.ObtainBindlessSampler(buffer, offset));

        // If this sampler has already been used, return the existing mapping.
        const auto it = std::find_if(used_samplers.begin(), used_samplers.end(),
-                                     [buffer = buffer, offset = offset](const Sampler& entry) {
+                                     [buffer, offset](const Sampler& entry) {
                                         return entry.buffer == buffer && entry.offset == offset;
                                     });
        if (it != used_samplers.end()) {
@@ -431,10 +428,32 @@ std::optional<Sampler> ShaderIR::GetBindlessSampler(Tegra::Shader::Register reg,
        return used_samplers.emplace_back(next_index, offset, buffer, *info.type, *info.is_array,
                                          *info.is_shadow, *info.is_buffer, false);
    }
-    if (const auto array_sampler_info = std::get_if<ArraySamplerNode>(&*tracked_sampler_info)) {
-        const u32 base_offset = array_sampler_info->GetBaseOffset() / 4;
-        index_var = GetCustomVariable(array_sampler_info->GetIndexVar());
-        info = GetSamplerInfo(info, base_offset);
+    if (const auto sampler_info = std::get_if<SeparateSamplerNode>(&*tracked_sampler_info)) {
+        const std::pair indices = sampler_info->indices;
+        const std::pair offsets = sampler_info->offsets;
+        info = GetSamplerInfo(info, registry.ObtainSeparateSampler(indices, offsets));
+
+        // Try to use an already created sampler if it exists
+        const auto it = std::find_if(
+            used_samplers.begin(), used_samplers.end(), [indices, offsets](const Sampler& entry) {
+                return offsets == std::pair{entry.offset, entry.secondary_offset} &&
+                       indices == std::pair{entry.buffer, entry.secondary_buffer};
+            });
+        if (it != used_samplers.end()) {
+            ASSERT(it->is_separated && it->type == info.type && it->is_array == info.is_array &&
+                   it->is_shadow == info.is_shadow && it->is_buffer == info.is_buffer);
+            return *it;
+        }
+
+        // Otherwise create a new mapping for this sampler
+        const u32 next_index = static_cast<u32>(used_samplers.size());
+        return used_samplers.emplace_back(next_index, offsets, indices, *info.type, *info.is_array,
+                                          *info.is_shadow, *info.is_buffer);
+    }
+    if (const auto sampler_info = std::get_if<ArraySamplerNode>(&*tracked_sampler_info)) {
+        const u32 base_offset = sampler_info->base_offset / 4;
+        index_var = GetCustomVariable(sampler_info->bindless_var);
+        info = GetSamplerInfo(info, registry.ObtainBoundSampler(base_offset));

        // If this sampler has already been used, return the existing mapping.
        const auto it = std::find_if(
@@ -275,10 +275,11 @@ using Node = std::shared_ptr<NodeData>;
 using Node4 = std::array<Node, 4>;
 using NodeBlock = std::vector<Node>;

-class BindlessSamplerNode;
-class ArraySamplerNode;
+struct ArraySamplerNode;
+struct BindlessSamplerNode;
+struct SeparateSamplerNode;

-using TrackSamplerData = std::variant<BindlessSamplerNode, ArraySamplerNode>;
+using TrackSamplerData = std::variant<BindlessSamplerNode, SeparateSamplerNode, ArraySamplerNode>;
 using TrackSampler = std::shared_ptr<TrackSamplerData>;

 struct Sampler {
@@ -288,63 +289,51 @@ struct Sampler {
        : index{index}, offset{offset}, type{type}, is_array{is_array}, is_shadow{is_shadow},
          is_buffer{is_buffer}, is_indexed{is_indexed} {}

+    /// Separate sampler constructor
+    constexpr explicit Sampler(u32 index, std::pair<u32, u32> offsets, std::pair<u32, u32> buffers,
+                               Tegra::Shader::TextureType type, bool is_array, bool is_shadow,
+                               bool is_buffer)
+        : index{index}, offset{offsets.first}, secondary_offset{offsets.second},
+          buffer{buffers.first}, secondary_buffer{buffers.second}, type{type}, is_array{is_array},
+          is_shadow{is_shadow}, is_buffer{is_buffer}, is_separated{true} {}
+
    /// Bindless samplers constructor
    constexpr explicit Sampler(u32 index, u32 offset, u32 buffer, Tegra::Shader::TextureType type,
                               bool is_array, bool is_shadow, bool is_buffer, bool is_indexed)
        : index{index}, offset{offset}, buffer{buffer}, type{type}, is_array{is_array},
          is_shadow{is_shadow}, is_buffer{is_buffer}, is_bindless{true}, is_indexed{is_indexed} {}

-    u32 index = 0;  ///< Emulated index given for the this sampler.
-    u32 offset = 0; ///< Offset in the const buffer from where the sampler is being read.
-    u32 buffer = 0; ///< Buffer where the bindless sampler is being read (unused on bound samplers).
-    u32 size = 1;   ///< Size of the sampler.
+    u32 index = 0;            ///< Emulated index given for the this sampler.
+    u32 offset = 0;           ///< Offset in the const buffer from where the sampler is being read.
+    u32 secondary_offset = 0; ///< Secondary offset in the const buffer.
+    u32 buffer = 0;           ///< Buffer where the bindless sampler is read.
+    u32 secondary_buffer = 0; ///< Secondary buffer where the bindless sampler is read.
+    u32 size = 1;             ///< Size of the sampler.

    Tegra::Shader::TextureType type{}; ///< The type used to sample this texture (Texture2D, etc)
-    bool is_array = false;    ///< Whether the texture is being sampled as an array texture or not.
-    bool is_shadow = false;   ///< Whether the texture is being sampled as a depth texture or not.
-    bool is_buffer = false;   ///< Whether the texture is a texture buffer without sampler.
-    bool is_bindless = false; ///< Whether this sampler belongs to a bindless texture or not.
-    bool is_indexed = false;  ///< Whether this sampler is an indexed array of textures.
+    bool is_array = false;     ///< Whether the texture is being sampled as an array texture or not.
+    bool is_shadow = false;    ///< Whether the texture is being sampled as a depth texture or not.
+    bool is_buffer = false;    ///< Whether the texture is a texture buffer without sampler.
+    bool is_bindless = false;  ///< Whether this sampler belongs to a bindless texture or not.
+    bool is_indexed = false;   ///< Whether this sampler is an indexed array of textures.
+    bool is_separated = false; ///< Whether the image and sampler is separated or not.
 };

 /// Represents a tracked bindless sampler into a direct const buffer
-class ArraySamplerNode final {
-public:
-    explicit ArraySamplerNode(u32 index, u32 base_offset, u32 bindless_var)
-        : index{index}, base_offset{base_offset}, bindless_var{bindless_var} {}
-
-    constexpr u32 GetIndex() const {
-        return index;
-    }
-
-    constexpr u32 GetBaseOffset() const {
-        return base_offset;
-    }
-
-    constexpr u32 GetIndexVar() const {
-        return bindless_var;
-    }
-
-private:
+struct ArraySamplerNode {
    u32 index;
    u32 base_offset;
    u32 bindless_var;
 };

+/// Represents a tracked separate sampler image pair that was folded statically
+struct SeparateSamplerNode {
+    std::pair<u32, u32> indices;
+    std::pair<u32, u32> offsets;
+};
+
 /// Represents a tracked bindless sampler into a direct const buffer
-class BindlessSamplerNode final {
-public:
-    explicit BindlessSamplerNode(u32 index, u32 offset) : index{index}, offset{offset} {}
-
-    constexpr u32 GetIndex() const {
-        return index;
-    }
-
-    constexpr u32 GetOffset() const {
-        return offset;
-    }
-
-private:
+struct BindlessSamplerNode {
    u32 index;
    u32 offset;
 };
@@ -48,7 +48,7 @@ Node MakeNode(Args&&... args) {
 template <typename T, typename... Args>
 TrackSampler MakeTrackSampler(Args&&... args) {
    static_assert(std::is_convertible_v<T, TrackSamplerData>);
-    return std::make_shared<TrackSamplerData>(T(std::forward<Args>(args)...));
+    return std::make_shared<TrackSamplerData>(T{std::forward<Args>(args)...});
 }

 template <typename... Args>
@@ -93,6 +93,26 @@ std::optional<SamplerDescriptor> Registry::ObtainBoundSampler(u32 offset) {
    return value;
 }

+std::optional<Tegra::Engines::SamplerDescriptor> Registry::ObtainSeparateSampler(
+    std::pair<u32, u32> buffers, std::pair<u32, u32> offsets) {
+    SeparateSamplerKey key;
+    key.buffers = buffers;
+    key.offsets = offsets;
+    const auto iter = separate_samplers.find(key);
+    if (iter != separate_samplers.end()) {
+        return iter->second;
+    }
+    if (!engine) {
+        return std::nullopt;
+    }
+
+    const u32 handle_1 = engine->AccessConstBuffer32(stage, key.buffers.first, key.offsets.first);
+    const u32 handle_2 = engine->AccessConstBuffer32(stage, key.buffers.second, key.offsets.second);
+    const SamplerDescriptor value = engine->AccessSampler(handle_1 | handle_2);
+    separate_samplers.emplace(key, value);
+    return value;
+}
+
 std::optional<Tegra::Engines::SamplerDescriptor> Registry::ObtainBindlessSampler(u32 buffer,
                                                                                 u32 offset) {
    const std::pair key = {buffer, offset};
@@ -19,8 +19,39 @@

 namespace VideoCommon::Shader {

+struct SeparateSamplerKey {
+    std::pair<u32, u32> buffers;
+    std::pair<u32, u32> offsets;
+};
+
+} // namespace VideoCommon::Shader
+
+namespace std {
+
+template <>
+struct hash<VideoCommon::Shader::SeparateSamplerKey> {
+    std::size_t operator()(const VideoCommon::Shader::SeparateSamplerKey& key) const noexcept {
+        return std::hash<u32>{}(key.buffers.first ^ key.buffers.second ^ key.offsets.first ^
+                                key.offsets.second);
+    }
+};
+
+template <>
+struct equal_to<VideoCommon::Shader::SeparateSamplerKey> {
+    bool operator()(const VideoCommon::Shader::SeparateSamplerKey& lhs,
+                    const VideoCommon::Shader::SeparateSamplerKey& rhs) const noexcept {
+        return lhs.buffers == rhs.buffers && lhs.offsets == rhs.offsets;
+    }
+};
+
+} // namespace std
+
+namespace VideoCommon::Shader {
+
 using KeyMap = std::unordered_map<std::pair<u32, u32>, u32, Common::PairHash>;
 using BoundSamplerMap = std::unordered_map<u32, Tegra::Engines::SamplerDescriptor>;
+using SeparateSamplerMap =
+    std::unordered_map<SeparateSamplerKey, Tegra::Engines::SamplerDescriptor>;
 using BindlessSamplerMap =
    std::unordered_map<std::pair<u32, u32>, Tegra::Engines::SamplerDescriptor, Common::PairHash>;

@@ -73,6 +104,9 @@ public:

    std::optional<Tegra::Engines::SamplerDescriptor> ObtainBoundSampler(u32 offset);

+    std::optional<Tegra::Engines::SamplerDescriptor> ObtainSeparateSampler(
+        std::pair<u32, u32> buffers, std::pair<u32, u32> offsets);
+
    std::optional<Tegra::Engines::SamplerDescriptor> ObtainBindlessSampler(u32 buffer, u32 offset);

    /// Inserts a key.
@@ -128,6 +162,7 @@ private:
    Tegra::Engines::ConstBufferEngineInterface* engine = nullptr;
    KeyMap keys;
    BoundSamplerMap bound_samplers;
+    SeparateSamplerMap separate_samplers;
    BindlessSamplerMap bindless_samplers;
    u32 bound_buffer;
    GraphicsInfo graphics_info;
@@ -330,8 +330,8 @@ private:
    OperationCode GetPredicateCombiner(Tegra::Shader::PredOperation operation);

    /// Queries the missing sampler info from the execution context.
-    SamplerInfo GetSamplerInfo(SamplerInfo info, u32 offset,
-                               std::optional<u32> buffer = std::nullopt);
+    SamplerInfo GetSamplerInfo(SamplerInfo info,
+                               std::optional<Tegra::Engines::SamplerDescriptor> sampler);

    /// Accesses a texture sampler.
    std::optional<Sampler> GetSampler(Tegra::Shader::Sampler sampler, SamplerInfo info);
@@ -409,8 +409,14 @@ private:

    std::tuple<Node, u32, u32> TrackCbuf(Node tracked, const NodeBlock& code, s64 cursor) const;

-    std::tuple<Node, TrackSampler> TrackBindlessSampler(Node tracked, const NodeBlock& code,
-                                                        s64 cursor);
+    std::pair<Node, TrackSampler> TrackBindlessSampler(Node tracked, const NodeBlock& code,
+                                                       s64 cursor);
+
+    std::pair<Node, TrackSampler> HandleBindlessIndirectRead(const CbufNode& cbuf,
+                                                             const OperationNode& operation,
+                                                             Node gpr, Node base_offset,
+                                                             Node tracked, const NodeBlock& code,
+                                                             s64 cursor);

    std::optional<u32> TrackImmediate(Node tracked, const NodeBlock& code, s64 cursor) const;

@@ -14,6 +14,7 @@
 namespace VideoCommon::Shader {

 namespace {
+
 std::pair<Node, s64> FindOperation(const NodeBlock& code, s64 cursor,
                                   OperationCode operation_code) {
    for (; cursor >= 0; --cursor) {
@@ -63,7 +64,8 @@ bool AmendNodeCv(std::size_t amend_index, Node node) {
    if (const auto operation = std::get_if<OperationNode>(&*node)) {
        operation->SetAmendIndex(amend_index);
        return true;
-    } else if (const auto conditional = std::get_if<ConditionalNode>(&*node)) {
+    }
+    if (const auto conditional = std::get_if<ConditionalNode>(&*node)) {
        conditional->SetAmendIndex(amend_index);
        return true;
    }
@@ -72,40 +74,27 @@ bool AmendNodeCv(std::size_t amend_index, Node node) {

 } // Anonymous namespace

-std::tuple<Node, TrackSampler> ShaderIR::TrackBindlessSampler(Node tracked, const NodeBlock& code,
-                                                              s64 cursor) {
+std::pair<Node, TrackSampler> ShaderIR::TrackBindlessSampler(Node tracked, const NodeBlock& code,
+                                                             s64 cursor) {
    if (const auto cbuf = std::get_if<CbufNode>(&*tracked)) {
+        const u32 cbuf_index = cbuf->GetIndex();
+
        // Constant buffer found, test if it's an immediate
        const auto& offset = cbuf->GetOffset();
        if (const auto immediate = std::get_if<ImmediateNode>(&*offset)) {
-            auto track =
-                MakeTrackSampler<BindlessSamplerNode>(cbuf->GetIndex(), immediate->GetValue());
+            auto track = MakeTrackSampler<BindlessSamplerNode>(cbuf_index, immediate->GetValue());
            return {tracked, track};
        }
        if (const auto operation = std::get_if<OperationNode>(&*offset)) {
            const u32 bound_buffer = registry.GetBoundBuffer();
-            if (bound_buffer != cbuf->GetIndex()) {
+            if (bound_buffer != cbuf_index) {
                return {};
            }
-            const auto pair = DecoupleIndirectRead(*operation);
-            if (!pair) {
-                return {};
+            if (const std::optional pair = DecoupleIndirectRead(*operation)) {
+                auto [gpr, base_offset] = *pair;
+                return HandleBindlessIndirectRead(*cbuf, *operation, gpr, base_offset, tracked,
+                                                  code, cursor);
            }
-            auto [gpr, base_offset] = *pair;
-            const auto offset_inm = std::get_if<ImmediateNode>(&*base_offset);
-            const auto& gpu_driver = registry.AccessGuestDriverProfile();
-            const u32 bindless_cv = NewCustomVariable();
-            Node op =
-                Operation(OperationCode::UDiv, gpr, Immediate(gpu_driver.GetTextureHandlerSize()));
-
-            const Node cv_node = GetCustomVariable(bindless_cv);
-            Node amend_op = Operation(OperationCode::Assign, cv_node, std::move(op));
-            const std::size_t amend_index = DeclareAmend(std::move(amend_op));
-            AmendNodeCv(amend_index, code[cursor]);
-            // TODO Implement Bindless Index custom variable
-            auto track = MakeTrackSampler<ArraySamplerNode>(cbuf->GetIndex(),
-                                                            offset_inm->GetValue(), bindless_cv);
-            return {tracked, track};
        }
        return {};
    }
@@ -122,10 +111,23 @@ std::tuple<Node, TrackSampler> ShaderIR::TrackBindlessSampler(Node tracked, cons
        return TrackBindlessSampler(source, code, new_cursor);
    }
    if (const auto operation = std::get_if<OperationNode>(&*tracked)) {
-        for (std::size_t i = operation->GetOperandsCount(); i > 0; --i) {
-            if (auto found = TrackBindlessSampler((*operation)[i - 1], code, cursor);
-                std::get<0>(found)) {
-                // Cbuf found in operand.
+        const OperationNode& op = *operation;
+
+        const OperationCode opcode = operation->GetCode();
+        if (opcode == OperationCode::IBitwiseOr || opcode == OperationCode::UBitwiseOr) {
+            ASSERT(op.GetOperandsCount() == 2);
+            auto [node_a, index_a, offset_a] = TrackCbuf(op[0], code, cursor);
+            auto [node_b, index_b, offset_b] = TrackCbuf(op[1], code, cursor);
+            if (node_a && node_b) {
+                auto track = MakeTrackSampler<SeparateSamplerNode>(std::pair{index_a, index_b},
+                                                                   std::pair{offset_a, offset_b});
+                return {tracked, std::move(track)};
+            }
+        }
+        std::size_t i = op.GetOperandsCount();
+        while (i--) {
+            if (auto found = TrackBindlessSampler(op[i - 1], code, cursor); std::get<0>(found)) {
+                // Constant buffer found in operand.
                return found;
            }
        }
@@ -139,6 +141,26 @@ std::tuple<Node, TrackSampler> ShaderIR::TrackBindlessSampler(Node tracked, cons
    return {};
 }

+std::pair<Node, TrackSampler> ShaderIR::HandleBindlessIndirectRead(
+    const CbufNode& cbuf, const OperationNode& operation, Node gpr, Node base_offset, Node tracked,
+    const NodeBlock& code, s64 cursor) {
+    const auto offset_imm = std::get<ImmediateNode>(*base_offset);
+    const auto& gpu_driver = registry.AccessGuestDriverProfile();
+    const u32 bindless_cv = NewCustomVariable();
+    const u32 texture_handler_size = gpu_driver.GetTextureHandlerSize();
+    Node op = Operation(OperationCode::UDiv, gpr, Immediate(texture_handler_size));
+
+    Node cv_node = GetCustomVariable(bindless_cv);
+    Node amend_op = Operation(OperationCode::Assign, std::move(cv_node), std::move(op));
+    const std::size_t amend_index = DeclareAmend(std::move(amend_op));
+    AmendNodeCv(amend_index, code[cursor]);
+
+    // TODO: Implement bindless index custom variable
+    auto track =
+        MakeTrackSampler<ArraySamplerNode>(cbuf.GetIndex(), offset_imm.GetValue(), bindless_cv);
+    return {tracked, track};
+}
+
 std::tuple<Node, u32, u32> ShaderIR::TrackCbuf(Node tracked, const NodeBlock& code,
                                               s64 cursor) const {
    if (const auto cbuf = std::get_if<CbufNode>(&*tracked)) {
@@ -631,13 +631,11 @@ void Config::ReadRendererValues() {
        static_cast<Settings::RendererBackend>(ReadSetting(QStringLiteral("backend"), 0).toInt());
    Settings::values.renderer_debug = ReadSetting(QStringLiteral("debug"), false).toBool();
    Settings::values.vulkan_device = ReadSetting(QStringLiteral("vulkan_device"), 0).toInt();
-    Settings::values.resolution_factor =
-        ReadSetting(QStringLiteral("resolution_factor"), 1.0).toFloat();
    Settings::values.aspect_ratio = ReadSetting(QStringLiteral("aspect_ratio"), 0).toInt();
    Settings::values.max_anisotropy = ReadSetting(QStringLiteral("max_anisotropy"), 0).toInt();
    Settings::values.use_frame_limit =
        ReadSetting(QStringLiteral("use_frame_limit"), true).toBool();
-    Settings::values.frame_limit = ReadSetting(QStringLiteral("frame_limit"), 100).toInt();
+    Settings::values.frame_limit = ReadSetting(QStringLiteral("frame_limit"), 100).toUInt();
    Settings::values.use_disk_shader_cache =
        ReadSetting(QStringLiteral("use_disk_shader_cache"), true).toBool();
    const int gpu_accuracy_level = ReadSetting(QStringLiteral("gpu_accuracy"), 0).toInt();
@@ -722,8 +720,6 @@ void Config::ReadUIValues() {
            .toString();
    UISettings::values.enable_discord_presence =
        ReadSetting(QStringLiteral("enable_discord_presence"), true).toBool();
-    UISettings::values.screenshot_resolution_factor =
-        static_cast<u16>(ReadSetting(QStringLiteral("screenshot_resolution_factor"), 0).toUInt());
    UISettings::values.select_user_on_boot =
        ReadSetting(QStringLiteral("select_user_on_boot"), false).toBool();

@@ -1082,8 +1078,6 @@ void Config::SaveRendererValues() {
    WriteSetting(QStringLiteral("backend"), static_cast<int>(Settings::values.renderer_backend), 0);
    WriteSetting(QStringLiteral("debug"), Settings::values.renderer_debug, false);
    WriteSetting(QStringLiteral("vulkan_device"), Settings::values.vulkan_device, 0);
-    WriteSetting(QStringLiteral("resolution_factor"),
-                 static_cast<double>(Settings::values.resolution_factor), 1.0);
    WriteSetting(QStringLiteral("aspect_ratio"), Settings::values.aspect_ratio, 0);
    WriteSetting(QStringLiteral("max_anisotropy"), Settings::values.max_anisotropy, 0);
    WriteSetting(QStringLiteral("use_frame_limit"), Settings::values.use_frame_limit, true);
@@ -1159,8 +1153,6 @@ void Config::SaveUIValues() {
                 QString::fromUtf8(UISettings::themes[0].second));
    WriteSetting(QStringLiteral("enable_discord_presence"),
                 UISettings::values.enable_discord_presence, true);
-    WriteSetting(QStringLiteral("screenshot_resolution_factor"),
-                 UISettings::values.screenshot_resolution_factor, 0);
    WriteSetting(QStringLiteral("select_user_on_boot"), UISettings::values.select_user_on_boot,
                 false);

@@ -19,47 +19,6 @@
 #include "video_core/renderer_vulkan/renderer_vulkan.h"
 #endif

-namespace {
-enum class Resolution : int {
-    Auto,
-    Scale1x,
-    Scale2x,
-    Scale3x,
-    Scale4x,
-};
-
-float ToResolutionFactor(Resolution option) {
-    switch (option) {
-    case Resolution::Auto:
-        return 0.f;
-    case Resolution::Scale1x:
-        return 1.f;
-    case Resolution::Scale2x:
-        return 2.f;
-    case Resolution::Scale3x:
-        return 3.f;
-    case Resolution::Scale4x:
-        return 4.f;
-    }
-    return 0.f;
-}
-
-Resolution FromResolutionFactor(float factor) {
-    if (factor == 0.f) {
-        return Resolution::Auto;
-    } else if (factor == 1.f) {
-        return Resolution::Scale1x;
-    } else if (factor == 2.f) {
-        return Resolution::Scale2x;
-    } else if (factor == 3.f) {
-        return Resolution::Scale3x;
-    } else if (factor == 4.f) {
-        return Resolution::Scale4x;
-    }
-    return Resolution::Auto;
-}
-} // Anonymous namespace
-
 ConfigureGraphics::ConfigureGraphics(QWidget* parent)
    : QWidget(parent), ui(new Ui::ConfigureGraphics) {
    vulkan_device = Settings::values.vulkan_device;
@@ -99,8 +58,6 @@ void ConfigureGraphics::SetConfiguration() {

    ui->api->setEnabled(runtime_lock);
    ui->api->setCurrentIndex(static_cast<int>(Settings::values.renderer_backend));
-    ui->resolution_factor_combobox->setCurrentIndex(
-        static_cast<int>(FromResolutionFactor(Settings::values.resolution_factor)));
    ui->aspect_ratio_combobox->setCurrentIndex(Settings::values.aspect_ratio);
    ui->use_disk_shader_cache->setEnabled(runtime_lock);
    ui->use_disk_shader_cache->setChecked(Settings::values.use_disk_shader_cache);
@@ -114,8 +71,6 @@ void ConfigureGraphics::SetConfiguration() {
 void ConfigureGraphics::ApplyConfiguration() {
    Settings::values.renderer_backend = GetCurrentGraphicsBackend();
    Settings::values.vulkan_device = vulkan_device;
-    Settings::values.resolution_factor =
-        ToResolutionFactor(static_cast<Resolution>(ui->resolution_factor_combobox->currentIndex()));
    Settings::values.aspect_ratio = ui->aspect_ratio_combobox->currentIndex();
    Settings::values.use_disk_shader_cache = ui->use_disk_shader_cache->isChecked();
    Settings::values.use_asynchronous_gpu_emulation =
@@ -84,46 +84,6 @@
          </property>
         </widget>
        </item>
-        <item>
-         <layout class="QHBoxLayout" name="horizontalLayout_2">
-          <item>
-           <widget class="QLabel" name="label">
-            <property name="text">
-             <string>Internal Resolution:</string>
-            </property>
-           </widget>
-          </item>
-          <item>
-           <widget class="QComboBox" name="resolution_factor_combobox">
-            <item>
-             <property name="text">
-              <string>Auto (Window Size)</string>
-             </property>
-            </item>
-            <item>
-             <property name="text">
-              <string>Native (1280x720)</string>
-             </property>
-            </item>
-            <item>
-             <property name="text">
-              <string>2x Native (2560x1440)</string>
-             </property>
-            </item>
-            <item>
-             <property name="text">
-              <string>3x Native (3840x2160)</string>
-             </property>
-            </item>
-            <item>
-             <property name="text">
-              <string>4x Native (5120x2880)</string>
-             </property>
-            </item>
-           </widget>
-          </item>
-         </layout>
-        </item>
        <item>
         <layout class="QHBoxLayout" name="horizontalLayout_6">
          <item>
@@ -12,9 +12,6 @@ ConfigureGraphicsAdvanced::ConfigureGraphicsAdvanced(QWidget* parent)

    ui->setupUi(this);

-    // TODO: Remove this after assembly shaders are fully integrated
-    ui->use_assembly_shaders->setVisible(false);
-
    SetConfiguration();
 }

@@ -689,10 +689,7 @@ void GMainWindow::InitializeHotkeys() {
                Settings::values.use_frame_limit = !Settings::values.use_frame_limit;
                UpdateStatusBar();
            });
-    // TODO: Remove this comment/static whenever the next major release of
-    // MSVC occurs and we make it a requirement (see:
-    // https://developercommunity.visualstudio.com/content/problem/93922/constexprs-are-trying-to-be-captured-in-lambda-fun.html)
-    static constexpr u16 SPEED_LIMIT_STEP = 5;
+    constexpr u16 SPEED_LIMIT_STEP = 5;
    connect(hotkey_registry.GetHotkey(main_window, QStringLiteral("Increase Speed Limit"), this),
            &QShortcut::activated, this, [&] {
                if (Settings::values.frame_limit < 9999 - SPEED_LIMIT_STEP) {
@@ -380,8 +380,6 @@ void Config::ReadValues() {
    Settings::values.renderer_debug = sdl2_config->GetBoolean("Renderer", "debug", false);
    Settings::values.vulkan_device = sdl2_config->GetInteger("Renderer", "vulkan_device", 0);

-    Settings::values.resolution_factor =
-        static_cast<float>(sdl2_config->GetReal("Renderer", "resolution_factor", 1.0));
    Settings::values.aspect_ratio =
        static_cast<int>(sdl2_config->GetInteger("Renderer", "aspect_ratio", 0));
    Settings::values.max_anisotropy =
@@ -117,11 +117,6 @@ use_hw_renderer =
 # 0: Interpreter (slow), 1 (default): JIT (fast)
 use_shader_jit =

-# Resolution scale factor
-# 0: Auto (scales resolution to window size), 1: Native Switch screen resolution, Otherwise a scale
-# factor for the Switch resolution
-resolution_factor =
-
 # Aspect ratio
 # 0: Default (16:9), 1: Force 4:3, 2: Force 21:9, 3: Stretch to Window
 aspect_ratio =
@@ -116,8 +116,6 @@ void Config::ReadValues() {
    Settings::values.use_multi_core = sdl2_config->GetBoolean("Core", "use_multi_core", false);

    // Renderer
-    Settings::values.resolution_factor =
-        static_cast<float>(sdl2_config->GetReal("Renderer", "resolution_factor", 1.0));
    Settings::values.aspect_ratio =
        static_cast<int>(sdl2_config->GetInteger("Renderer", "aspect_ratio", 0));
    Settings::values.max_anisotropy =
@@ -21,11 +21,6 @@ use_hw_renderer =
 # 0: Interpreter (slow), 1 (default): JIT (fast)
 use_shader_jit =

-# Resolution scale factor
-# 0: Auto (scales resolution to window size), 1: Native Switch screen resolution, Otherwise a scale
-# factor for the Switch resolution
-resolution_factor =
-
 # Aspect ratio
 # 0: Default (16:9), 1: Force 4:3, 2: Force 21:9, 3: Stretch to Window
 aspect_ratio =
Author	SHA1	Message	Date
MerryMage	52bcfac116	arm_dynarmic_cp15: Implement CNTPCT	2020-06-17 17:10:24 +01:00
MerryMage	109df7705f	arm_dynarmic_cp15: Update CP15	2020-06-17 17:10:24 +01:00
MerryMage	32a127faaa	arm_dynarmic_32: InterpreterFallback should never happen	2020-06-17 17:10:24 +01:00
bunnei	a8ac99b619	Merge pull request #4086 from MerryMage/abi xbyak_abi: Cleanup	2020-06-17 11:20:52 -04:00
bunnei	b660ef6c8a	Merge pull request #4089 from MerryMage/macrojit-cleanup-1 macro_jit_x64: Cleanup	2020-06-16 23:44:48 -04:00
bunnei	0f57bbfa3f	Merge pull request #3976 from Neodyblue/qdarkstyle_fix_prop qt_themes: remove unknown qss property from dark theme	2020-06-16 22:27:27 -04:00
bunnei	2a3d4cad63	externals: Revert to libressl, as build is broken with find_package(OpenSSL). (#4093 ) * externals: Revert to libressl, as build is broken with find_package(OpenSLL). * fixup! externals: Revert to libressl, as build is broken with find_package(OpenSLL). * fixup! externals: Revert to libressl, as build is broken with find_package(OpenSLL).	2020-06-16 21:46:19 -04:00
bunnei	798ec003ce	Merge pull request #4041 from ReinUsesLisp/arb-decomp gl_arb_decompiler: Implement an assembly shader decompiler	2020-06-16 14:56:23 -04:00
bunnei	f22d02083c	Merge pull request #3966 from Morph1984/hide-internal-resolution-ui yuzu/frontend: Remove internal resolution option	2020-06-16 14:12:17 -04:00
bunnei	ed2cd9d8f3	Merge pull request #4091 from MerryMage/cmakelists-xbyak-order CMakeLists: xbyak comes before dynarmic	2020-06-15 22:39:48 -04:00
Rodrigo Locatti	0bd9bc7201	Merge pull request #4066 from ReinUsesLisp/shared-ptr-buf buffer_cache: Avoid passing references of shared pointers and misc style changes	2020-06-15 22:29:32 -03:00
MerryMage	256cb2979b	CMakeLists: xbyak comes before dynarmic	2020-06-15 22:37:27 +01:00
MerryMage	cf0aad7d6a	macro_jit_x64: Remove NEXT_PARAMETER Not required, as PARAMETERS can just be incremented directly.	2020-06-15 21:19:38 +01:00
MerryMage	1799f4e774	macro_jit_x64: Remove unused function Compile_WriteCarry	2020-06-15 21:19:38 +01:00
MerryMage	c09a9e5cc7	macro_jit_x64: Select better registers All registers are now callee-save registers. RBX and RBP selected for STATE and RESULT because these are most commonly accessed; this is to avoid the REX prefix. RBP not used for STATE because there are some SIB restrictions, RBX emits smaller code.	2020-06-15 21:19:38 +01:00
MerryMage	79aa7b3ace	macro_jit_x64: Remove REGISTERS Unnecessary since this is just an offset from STATE.	2020-06-15 21:00:59 +01:00
MerryMage	35db6e1c68	macro_jit_x64: Remove JITState::parameters This can be passed in as an argument instead.	2020-06-15 20:55:02 +01:00
MerryMage	389549b80d	macro_jit_x64: Remove METHOD_ADDRESS_64 Unnecessary variable.	2020-06-15 20:51:33 +01:00
MerryMage	a6a43a5ae0	macro_jit_x64: Remove RESULT_64 This Reg64 codepath has the exact same behaviour as the Reg32 one.	2020-06-15 20:35:08 +01:00
MerryMage	7c6203dc5e	xbyak_abi: Prefer returning a struct to using out parameters in ABI_CalculateFrameSize	2020-06-15 19:07:11 +01:00
MerryMage	36362e9695	xbyak_abi: Register indexes should be unsigned	2020-06-15 19:07:11 +01:00
MerryMage	d563017dfe	xbyak_abi: Remove *GPS variants of stack manipulation functions	2020-06-15 18:59:54 +01:00
MerryMage	4417770ba9	xbyak_abi: Fix ABI_PushRegistersAndAdjustStack Pushing GPRs twice.	2020-06-15 18:59:01 +01:00
David	5c9dee2c94	Merge pull request #4085 from ReinUsesLisp/gcc-times video_core/macro_jit_x64: Remove initializer in member variable	2020-06-15 23:05:21 +10:00
ReinUsesLisp	6e5d8aac4d	video_core/macro_jit_x64: Remove initializer in member variable Fix build time issues on gcc. Confirmed through asan that avoiding this initialization is safe.	2020-06-15 05:17:55 -03:00
bunnei	55ebf68636	Merge pull request #4070 from ogniK5377/GetTPCMasks-fix nvdrv: Fix GetTPCMasks for ioctl3	2020-06-14 20:12:45 -04:00
bunnei	89d11f2268	Merge pull request #4069 from ogniK5377/total-phys-mem kernel: Account for system resource size for memory usage	2020-06-14 00:44:34 -04:00
bunnei	92021a344c	Merge pull request #4064 from ReinUsesLisp/invalidate-buffers gl_rasterizer: Mark vertex buffers as dirty after buffer cache invalidation	2020-06-14 00:29:16 -04:00
bunnei	c2ea1e1bcb	Merge pull request #4049 from ReinUsesLisp/separate-samplers shader/texture: Join separate image and sampler pairs offline	2020-06-13 13:48:27 -04:00
ReinUsesLisp	87011a97f9	gl_arb_decompiler: Implement FSwizzleAdd	2020-06-11 22:12:07 -03:00
ReinUsesLisp	a63a0daa5e	gl_arb_decompiler: Implement an assembly shader decompiler Emit code compatible with NV_gpu_program5. This should emit code compatible with Fermi, but it wasn't tested on that architecture. Pascal has some issues not present on Turing GPUs.	2020-06-11 22:12:07 -03:00
ReinUsesLisp	d89888389d	yuzu/configuration: Show assembly shaders check box	2020-06-10 19:04:53 -03:00
David Marcec	b15cbf9bcf	nvdrv: Fix GetTPCMasks for ioctl3 Fixes animal crossing svcBreak on launch	2020-06-10 18:36:42 +10:00
David Marcec	74ff1db758	kernel: Account for system resource size for memory usage GetTotalPhysicalMemoryAvailableWithoutSystemResource & GetTotalPhysicalMemoryUsedWithoutSystemResource seem to subtract the resource size from the usage.	2020-06-10 14:49:00 +10:00
ReinUsesLisp	6508cdd003	buffer_cache: Avoid passing references of shared pointers and misc style changes Instead of using as template argument a shared pointer, use the underlying type and manage shared pointers explicitly. This can make removing shared pointers from the cache more easy. While we are at it, make some misc style changes and general improvements (like insert_or_assign instead of operator[] + operator=).	2020-06-09 18:30:49 -03:00
ReinUsesLisp	7646f2c21d	gl_rasterizer: Mark vertex buffers as dirty after buffer cache invalidation Vertex buffers bindings become invalid after the stream buffer is invalidated. We were originally doing this, but it got lost at some point. - Fixes Animal Crossing: New Horizons, but it affects everything.	2020-06-08 20:24:16 -03:00
ReinUsesLisp	6e122f0b2c	buffer_cache: Return stream buffer invalidation in Map instead of Unmap We have to invalidate whatever cache is being used before uploading the data, hence it makes more sense to return this on Map instead of Unmap.	2020-06-08 20:22:31 -03:00
Morph	03fad5ebe8	yuzu/frontend: Remove internal resolution option	2020-06-06 15:56:14 -04:00
ReinUsesLisp	5b2b6d594c	shader/texture: Join separate image and sampler pairs offline Games using D3D idioms can join images and samplers when a shader executes, instead of baking them into a combined sampler image. This is also possible on Vulkan. One approach to this solution would be to use separate samplers on Vulkan and leave this unimplemented on OpenGL, but we can't do this because there's no consistent way of determining which constant buffer holds a sampler and which one an image. We could in theory find the first bit and if it's in the TIC area, it's an image; but this falls apart when an image or sampler handle use an index of zero. The used approach is to track for a LOP.OR operation (this is done at an IR level, not at an ISA level), track again the constant buffers used as source and store this pair. Then, outside of shader execution, join the sample and image pair with a bitwise or operation. This approach won't work on games that truly use separate samplers in a meaningful way. For example, pooling textures in a 2D array and determining at runtime what sampler to use. This invalidates OpenGL's disk shader cache :) - Used mostly by D3D ports to Switch	2020-06-05 00:24:51 -03:00
ReinUsesLisp	e1438f8e91	shader/track: Move bindless tracking to a separate function	2020-06-04 23:02:55 -03:00
Neodyblue	ea14af2164	qt_themes: remove unknown qss property from dark theme Qdarkstyle's qss file uses an overflow property. According to `https://doc.qt.io/qt-5/stylesheet-reference.html`, the property `overflow` doesn't exist, which leads to a warning message in the console.	2020-05-21 17:51:53 -07:00