Allow picking a Compatibility Profile for OpenGL.

This option allows picking the compatibility profile since a lot of bugs
are fixed in it. We devs will use this option to easierly debug current
problems in our Core implementation.:wq

src/core/arm/arm_interface.h

@@ -7,6 +7,10 @@
 #include <array>
 #include "common/common_types.h"
 
+namespace Common {
+struct PageTable;
+}
+
 namespace Kernel {
 enum class VMAPermission : u8;
 }
@@ -49,8 +53,14 @@ public:
     /// Clear all instruction cache
     virtual void ClearInstructionCache() = 0;
 
-    /// Notify CPU emulation that page tables have changed
-    virtual void PageTableChanged() = 0;
+    /// Notifies CPU emulation that the current page table has changed.
+    ///
+    /// @param new_page_table                 The new page table.
+    /// @param new_address_space_size_in_bits The new usable size of the address space in bits.
+    ///                                       This can be either 32, 36, or 39 on official software.
+    ///
+    virtual void PageTableChanged(Common::PageTable& new_page_table,
+                                  std::size_t new_address_space_size_in_bits) = 0;
 
     /**
      * Set the Program Counter to an address

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -14,7 +14,6 @@
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
 #include "core/gdbstub/gdbstub.h"
-#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/svc.h"
 #include "core/hle/kernel/vm_manager.h"
@@ -129,18 +128,16 @@ public:
     u64 tpidr_el0 = 0;
 };
 
-std::unique_ptr<Dynarmic::A64::Jit> ARM_Dynarmic::MakeJit() const {
-    auto* current_process = system.Kernel().CurrentProcess();
-    auto** const page_table = current_process->VMManager().page_table.pointers.data();
-
+std::unique_ptr<Dynarmic::A64::Jit> ARM_Dynarmic::MakeJit(Common::PageTable& page_table,
+                                                          std::size_t address_space_bits) const {
     Dynarmic::A64::UserConfig config;
 
     // Callbacks
     config.callbacks = cb.get();
 
     // Memory
-    config.page_table = reinterpret_cast<void**>(page_table);
-    config.page_table_address_space_bits = current_process->VMManager().GetAddressSpaceWidth();
+    config.page_table = reinterpret_cast<void**>(page_table.pointers.data());
+    config.page_table_address_space_bits = address_space_bits;
     config.silently_mirror_page_table = false;
 
     // Multi-process state
@@ -176,12 +173,7 @@ ARM_Dynarmic::ARM_Dynarmic(System& system, ExclusiveMonitor& exclusive_monitor,
                            std::size_t core_index)
     : cb(std::make_unique<ARM_Dynarmic_Callbacks>(*this)), inner_unicorn{system},
       core_index{core_index}, system{system},
-      exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {
-    ThreadContext ctx{};
-    inner_unicorn.SaveContext(ctx);
-    PageTableChanged();
-    LoadContext(ctx);
-}
+      exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}
 
 ARM_Dynarmic::~ARM_Dynarmic() = default;
 
@@ -276,8 +268,9 @@ void ARM_Dynarmic::ClearExclusiveState() {
     jit->ClearExclusiveState();
 }
 
-void ARM_Dynarmic::PageTableChanged() {
-    jit = MakeJit();
+void ARM_Dynarmic::PageTableChanged(Common::PageTable& page_table,
+                                    std::size_t new_address_space_size_in_bits) {
+    jit = MakeJit(page_table, new_address_space_size_in_bits);
 }
 
 DynarmicExclusiveMonitor::DynarmicExclusiveMonitor(std::size_t core_count) : monitor(core_count) {}

src/core/arm/dynarmic/arm_dynarmic.h

@@ -48,10 +48,12 @@ public:
     void ClearExclusiveState() override;
 
     void ClearInstructionCache() override;
-    void PageTableChanged() override;
+    void PageTableChanged(Common::PageTable& new_page_table,
+                          std::size_t new_address_space_size_in_bits) override;
 
 private:
-    std::unique_ptr<Dynarmic::A64::Jit> MakeJit() const;
+    std::unique_ptr<Dynarmic::A64::Jit> MakeJit(Common::PageTable& page_table,
+                                                std::size_t address_space_bits) const;
 
     friend class ARM_Dynarmic_Callbacks;
     std::unique_ptr<ARM_Dynarmic_Callbacks> cb;

src/core/arm/unicorn/arm_unicorn.h

@@ -41,7 +41,7 @@ public:
     void Run() override;
     void Step() override;
     void ClearInstructionCache() override;
-    void PageTableChanged() override{};
+    void PageTableChanged(Common::PageTable&, std::size_t) override {}
     void RecordBreak(GDBStub::BreakpointAddress bkpt);
 
 private:

src/core/core.cpp

@@ -3,9 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <array>
-#include <map>
 #include <memory>
-#include <thread>
 #include <utility>
 
 #include "common/file_util.h"
@@ -38,8 +36,6 @@
 #include "frontend/applets/software_keyboard.h"
 #include "frontend/applets/web_browser.h"
 #include "video_core/debug_utils/debug_utils.h"
-#include "video_core/gpu_asynch.h"
-#include "video_core/gpu_synch.h"
 #include "video_core/renderer_base.h"
 #include "video_core/video_core.h"
 
@@ -81,7 +77,7 @@ FileSys::VirtualFile GetGameFileFromPath(const FileSys::VirtualFilesystem& vfs,
     return vfs->OpenFile(path, FileSys::Mode::Read);
 }
 struct System::Impl {
-    explicit Impl(System& system) : kernel{system} {}
+    explicit Impl(System& system) : kernel{system}, cpu_core_manager{system} {}
 
     Cpu& CurrentCpuCore() {
         return cpu_core_manager.GetCurrentCore();
@@ -99,6 +95,7 @@ struct System::Impl {
         LOG_DEBUG(HW_Memory, "initialized OK");
 
         core_timing.Initialize();
+        cpu_core_manager.Initialize();
         kernel.Initialize();
 
         const auto current_time = std::chrono::duration_cast<std::chrono::seconds>(
@@ -120,9 +117,6 @@ struct System::Impl {
         if (web_browser == nullptr)
             web_browser = std::make_unique<Core::Frontend::DefaultWebBrowserApplet>();
 
-        auto main_process = Kernel::Process::Create(system, "main");
-        kernel.MakeCurrentProcess(main_process.get());
-
         telemetry_session = std::make_unique<Core::TelemetrySession>();
         service_manager = std::make_shared<Service::SM::ServiceManager>();
 
@@ -134,16 +128,10 @@ struct System::Impl {
             return ResultStatus::ErrorVideoCore;
         }
 
+        gpu_core = VideoCore::CreateGPU(system);
+
         is_powered_on = true;
 
-        if (Settings::values.use_asynchronous_gpu_emulation) {
-            gpu_core = std::make_unique<VideoCommon::GPUAsynch>(system, *renderer);
-        } else {
-            gpu_core = std::make_unique<VideoCommon::GPUSynch>(system, *renderer);
-        }
-
-        cpu_core_manager.Initialize(system);
-
         LOG_DEBUG(Core, "Initialized OK");
 
         // Reset counters and set time origin to current frame
@@ -179,7 +167,8 @@ struct System::Impl {
             return init_result;
         }
 
-        const Loader::ResultStatus load_result{app_loader->Load(*kernel.CurrentProcess())};
+        auto main_process = Kernel::Process::Create(system, "main");
+        const auto [load_result, load_parameters] = app_loader->Load(*main_process);
         if (load_result != Loader::ResultStatus::Success) {
             LOG_CRITICAL(Core, "Failed to load ROM (Error {})!", static_cast<int>(load_result));
             Shutdown();
@@ -187,6 +176,16 @@ struct System::Impl {
             return static_cast<ResultStatus>(static_cast<u32>(ResultStatus::ErrorLoader) +
                                              static_cast<u32>(load_result));
         }
+        kernel.MakeCurrentProcess(main_process.get());
+
+        // Main process has been loaded and been made current.
+        // Begin GPU and CPU execution.
+        gpu_core->Start();
+        cpu_core_manager.StartThreads();
+
+        // All threads are started, begin main process execution, now that we're in the clear.
+        main_process->Run(load_parameters->main_thread_priority,
+                          load_parameters->main_thread_stack_size);
 
         status = ResultStatus::Success;
         return status;

src/core/cpu_core_manager.cpp

@@ -19,17 +19,19 @@ void RunCpuCore(const System& system, Cpu& cpu_state) {
 }
 } // Anonymous namespace
 
-CpuCoreManager::CpuCoreManager() = default;
+CpuCoreManager::CpuCoreManager(System& system) : system{system} {}
 CpuCoreManager::~CpuCoreManager() = default;
 
-void CpuCoreManager::Initialize(System& system) {
+void CpuCoreManager::Initialize() {
     barrier = std::make_unique<CpuBarrier>();
     exclusive_monitor = Cpu::MakeExclusiveMonitor(cores.size());
 
     for (std::size_t index = 0; index < cores.size(); ++index) {
         cores[index] = std::make_unique<Cpu>(system, *exclusive_monitor, *barrier, index);
     }
+}
 
+void CpuCoreManager::StartThreads() {
     // Create threads for CPU cores 1-3, and build thread_to_cpu map
     // CPU core 0 is run on the main thread
     thread_to_cpu[std::this_thread::get_id()] = cores[0].get();

src/core/cpu_core_manager.h

@@ -18,7 +18,7 @@ class System;
 
 class CpuCoreManager {
 public:
-    CpuCoreManager();
+    explicit CpuCoreManager(System& system);
     CpuCoreManager(const CpuCoreManager&) = delete;
     CpuCoreManager(CpuCoreManager&&) = delete;
 
@@ -27,7 +27,8 @@ public:
     CpuCoreManager& operator=(const CpuCoreManager&) = delete;
     CpuCoreManager& operator=(CpuCoreManager&&) = delete;
 
-    void Initialize(System& system);
+    void Initialize();
+    void StartThreads();
     void Shutdown();
 
     Cpu& GetCore(std::size_t index);
@@ -54,6 +55,8 @@ private:
 
     /// Map of guest threads to CPU cores
     std::map<std::thread::id, Cpu*> thread_to_cpu;
+
+    System& system;
 };
 
 } // namespace Core

src/core/hle/kernel/kernel.cpp

@@ -182,7 +182,12 @@ void KernelCore::AppendNewProcess(SharedPtr<Process> process) {
 
 void KernelCore::MakeCurrentProcess(Process* process) {
     impl->current_process = process;
-    Memory::SetCurrentPageTable(&process->VMManager().page_table);
+
+    if (process == nullptr) {
+        return;
+    }
+
+    Memory::SetCurrentPageTable(*process);
 }
 
 Process* KernelCore::CurrentProcess() {

src/core/hle/kernel/process.cpp

@@ -28,21 +28,20 @@ namespace {
  *
  * @param owner_process The parent process for the main thread
  * @param kernel The kernel instance to create the main thread under.
- * @param entry_point The address at which the thread should start execution
  * @param priority The priority to give the main thread
  */
-void SetupMainThread(Process& owner_process, KernelCore& kernel, VAddr entry_point, u32 priority) {
-    // Initialize new "main" thread
-    const VAddr stack_top = owner_process.VMManager().GetTLSIORegionEndAddress();
+void SetupMainThread(Process& owner_process, KernelCore& kernel, u32 priority) {
+    const auto& vm_manager = owner_process.VMManager();
+    const VAddr entry_point = vm_manager.GetCodeRegionBaseAddress();
+    const VAddr stack_top = vm_manager.GetTLSIORegionEndAddress();
     auto thread_res = Thread::Create(kernel, "main", entry_point, priority, 0,
                                      owner_process.GetIdealCore(), stack_top, owner_process);
 
     SharedPtr<Thread> thread = std::move(thread_res).Unwrap();
 
     // Register 1 must be a handle to the main thread
-    const Handle guest_handle = owner_process.GetHandleTable().Create(thread).Unwrap();
-    thread->SetGuestHandle(guest_handle);
-    thread->GetContext().cpu_registers[1] = guest_handle;
+    const Handle thread_handle = owner_process.GetHandleTable().Create(thread).Unwrap();
+    thread->GetContext().cpu_registers[1] = thread_handle;
 
     // Threads by default are dormant, wake up the main thread so it runs when the scheduler fires
     thread->ResumeFromWait();
@@ -106,8 +105,6 @@ ResultCode Process::LoadFromMetadata(const FileSys::ProgramMetadata& metadata) {
     is_64bit_process = metadata.Is64BitProgram();
 
     vm_manager.Reset(metadata.GetAddressSpaceType());
-    // Ensure that the potentially resized page table is seen by CPU backends.
-    Memory::SetCurrentPageTable(&vm_manager.page_table);
 
     const auto& caps = metadata.GetKernelCapabilities();
     const auto capability_init_result =
@@ -119,7 +116,7 @@ ResultCode Process::LoadFromMetadata(const FileSys::ProgramMetadata& metadata) {
     return handle_table.SetSize(capabilities.GetHandleTableSize());
 }
 
-void Process::Run(VAddr entry_point, s32 main_thread_priority, u64 stack_size) {
+void Process::Run(s32 main_thread_priority, u64 stack_size) {
     // The kernel always ensures that the given stack size is page aligned.
     main_thread_stack_size = Common::AlignUp(stack_size, Memory::PAGE_SIZE);
 
@@ -135,7 +132,7 @@ void Process::Run(VAddr entry_point, s32 main_thread_priority, u64 stack_size) {
     vm_manager.LogLayout();
     ChangeStatus(ProcessStatus::Running);
 
-    SetupMainThread(*this, kernel, entry_point, main_thread_priority);
+    SetupMainThread(*this, kernel, main_thread_priority);
 }
 
 void Process::PrepareForTermination() {
@@ -242,9 +239,6 @@ void Process::LoadModule(CodeSet module_, VAddr base_addr) {
     MapSegment(module_.DataSegment(), VMAPermission::ReadWrite, MemoryState::CodeData);
 
     code_memory_size += module_.memory.size();
-
-    // Clear instruction cache in CPU JIT
-    system.InvalidateCpuInstructionCaches();
 }
 
 Process::Process(Core::System& system)

src/core/hle/kernel/process.h

@@ -225,9 +225,12 @@ public:
     ResultCode LoadFromMetadata(const FileSys::ProgramMetadata& metadata);
 
     /**
-     * Applies address space changes and launches the process main thread.
+     * Starts the main application thread for this process.
+     *
+     * @param main_thread_priority The priority for the main thread.
+     * @param stack_size           The stack size for the main thread in bytes.
      */
-    void Run(VAddr entry_point, s32 main_thread_priority, u64 stack_size);
+    void Run(s32 main_thread_priority, u64 stack_size);
 
     /**
      * Prepares a process for termination by stopping all of its threads

src/core/hle/kernel/svc.cpp

@@ -1380,20 +1380,22 @@ static ResultCode CreateThread(Core::System& system, Handle* out_handle, VAddr e
         return ERR_INVALID_THREAD_PRIORITY;
     }
 
-    const std::string name = fmt::format("thread-{:X}", entry_point);
     auto& kernel = system.Kernel();
     CASCADE_RESULT(SharedPtr<Thread> thread,
-                   Thread::Create(kernel, name, entry_point, priority, arg, processor_id, stack_top,
+                   Thread::Create(kernel, "", entry_point, priority, arg, processor_id, stack_top,
                                   *current_process));
 
-    const auto new_guest_handle = current_process->GetHandleTable().Create(thread);
-    if (new_guest_handle.Failed()) {
+    const auto new_thread_handle = current_process->GetHandleTable().Create(thread);
+    if (new_thread_handle.Failed()) {
         LOG_ERROR(Kernel_SVC, "Failed to create handle with error=0x{:X}",
-                  new_guest_handle.Code().raw);
-        return new_guest_handle.Code();
+                  new_thread_handle.Code().raw);
+        return new_thread_handle.Code();
     }
-    thread->SetGuestHandle(*new_guest_handle);
-    *out_handle = *new_guest_handle;
+    *out_handle = *new_thread_handle;
+
+    // Set the thread name for debugging purposes.
+    thread->SetName(
+        fmt::format("thread[entry_point={:X}, handle={:X}]", entry_point, *new_thread_handle));
 
     system.CpuCore(thread->GetProcessorID()).PrepareReschedule();
 
@@ -2288,7 +2290,7 @@ static const FunctionDef SVC_Table[] = {
     {0x33, SvcWrap<GetThreadContext>, "GetThreadContext"},
     {0x34, SvcWrap<WaitForAddress>, "WaitForAddress"},
     {0x35, SvcWrap<SignalToAddress>, "SignalToAddress"},
-    {0x36, nullptr, "Unknown"},
+    {0x36, nullptr, "SynchronizePreemptionState"},
     {0x37, nullptr, "Unknown"},
     {0x38, nullptr, "Unknown"},
     {0x39, nullptr, "Unknown"},

src/core/hle/kernel/thread.h

@@ -102,6 +102,11 @@ public:
     std::string GetName() const override {
         return name;
     }
+
+    void SetName(std::string new_name) {
+        name = std::move(new_name);
+    }
+
     std::string GetTypeName() const override {
         return "Thread";
     }
@@ -339,10 +344,6 @@ public:
         arb_wait_address = address;
     }
 
-    void SetGuestHandle(Handle handle) {
-        guest_handle = handle;
-    }
-
     bool HasWakeupCallback() const {
         return wakeup_callback != nullptr;
     }
@@ -436,9 +437,6 @@ private:
     /// If waiting for an AddressArbiter, this is the address being waited on.
     VAddr arb_wait_address{0};
 
-    /// Handle used by guest emulated application to access this thread
-    Handle guest_handle = 0;
-
     /// Handle used as userdata to reference this object when inserting into the CoreTiming queue.
     Handle callback_handle = 0;
 

src/core/hle/kernel/wait_object.cpp

@@ -30,7 +30,7 @@ void WaitObject::RemoveWaitingThread(Thread* thread) {
         waiting_threads.erase(itr);
 }
 
-SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() {
+SharedPtr<Thread> WaitObject::GetHighestPriorityReadyThread() const {
     Thread* candidate = nullptr;
     u32 candidate_priority = THREADPRIO_LOWEST + 1;
 

src/core/hle/kernel/wait_object.h

@@ -54,7 +54,7 @@ public:
     void WakeupWaitingThread(SharedPtr<Thread> thread);
 
     /// Obtains the highest priority thread that is ready to run from this object's waiting list.
-    SharedPtr<Thread> GetHighestPriorityReadyThread();
+    SharedPtr<Thread> GetHighestPriorityReadyThread() const;
 
     /// Get a const reference to the waiting threads list for debug use
     const std::vector<SharedPtr<Thread>>& GetWaitingThreads() const;

src/core/loader/deconstructed_rom_directory.cpp

@@ -86,25 +86,29 @@ FileType AppLoader_DeconstructedRomDirectory::IdentifyType(const FileSys::Virtua
     return FileType::Error;
 }
 
-ResultStatus AppLoader_DeconstructedRomDirectory::Load(Kernel::Process& process) {
+AppLoader_DeconstructedRomDirectory::LoadResult AppLoader_DeconstructedRomDirectory::Load(
+    Kernel::Process& process) {
     if (is_loaded) {
-        return ResultStatus::ErrorAlreadyLoaded;
+        return {ResultStatus::ErrorAlreadyLoaded, {}};
     }
 
     if (dir == nullptr) {
-        if (file == nullptr)
-            return ResultStatus::ErrorNullFile;
+        if (file == nullptr) {
+            return {ResultStatus::ErrorNullFile, {}};
+        }
+
         dir = file->GetContainingDirectory();
     }
 
     // Read meta to determine title ID
     FileSys::VirtualFile npdm = dir->GetFile("main.npdm");
-    if (npdm == nullptr)
-        return ResultStatus::ErrorMissingNPDM;
+    if (npdm == nullptr) {
+        return {ResultStatus::ErrorMissingNPDM, {}};
+    }
 
-    ResultStatus result = metadata.Load(npdm);
+    const ResultStatus result = metadata.Load(npdm);
     if (result != ResultStatus::Success) {
-        return result;
+        return {result, {}};
     }
 
     if (override_update) {
@@ -114,23 +118,24 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(Kernel::Process& process)
 
     // Reread in case PatchExeFS affected the main.npdm
     npdm = dir->GetFile("main.npdm");
-    if (npdm == nullptr)
-        return ResultStatus::ErrorMissingNPDM;
+    if (npdm == nullptr) {
+        return {ResultStatus::ErrorMissingNPDM, {}};
+    }
 
-    ResultStatus result2 = metadata.Load(npdm);
+    const ResultStatus result2 = metadata.Load(npdm);
     if (result2 != ResultStatus::Success) {
-        return result2;
+        return {result2, {}};
     }
     metadata.Print();
 
     const FileSys::ProgramAddressSpaceType arch_bits{metadata.GetAddressSpaceType()};
     if (arch_bits == FileSys::ProgramAddressSpaceType::Is32Bit ||
         arch_bits == FileSys::ProgramAddressSpaceType::Is32BitNoMap) {
-        return ResultStatus::Error32BitISA;
+        return {ResultStatus::Error32BitISA, {}};
     }
 
     if (process.LoadFromMetadata(metadata).IsError()) {
-        return ResultStatus::ErrorUnableToParseKernelMetadata;
+        return {ResultStatus::ErrorUnableToParseKernelMetadata, {}};
     }
 
     const FileSys::PatchManager pm(metadata.GetTitleID());
@@ -150,7 +155,7 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(Kernel::Process& process)
         const auto tentative_next_load_addr =
             AppLoader_NSO::LoadModule(process, *module_file, load_addr, should_pass_arguments, pm);
         if (!tentative_next_load_addr) {
-            return ResultStatus::ErrorLoadingNSO;
+            return {ResultStatus::ErrorLoadingNSO, {}};
         }
 
         next_load_addr = *tentative_next_load_addr;
@@ -159,8 +164,6 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(Kernel::Process& process)
         GDBStub::RegisterModule(module, load_addr, next_load_addr - 1, false);
     }
 
-    process.Run(base_address, metadata.GetMainThreadPriority(), metadata.GetMainThreadStackSize());
-
     // Find the RomFS by searching for a ".romfs" file in this directory
     const auto& files = dir->GetFiles();
     const auto romfs_iter =
@@ -175,7 +178,8 @@ ResultStatus AppLoader_DeconstructedRomDirectory::Load(Kernel::Process& process)
     }
 
     is_loaded = true;
-    return ResultStatus::Success;
+    return {ResultStatus::Success,
+            LoadParameters{metadata.GetMainThreadPriority(), metadata.GetMainThreadStackSize()}};
 }
 
 ResultStatus AppLoader_DeconstructedRomDirectory::ReadRomFS(FileSys::VirtualFile& dir) {

src/core/loader/deconstructed_rom_directory.h

@@ -37,7 +37,7 @@ public:
         return IdentifyType(file);
     }
 
-    ResultStatus Load(Kernel::Process& process) override;
+    LoadResult Load(Kernel::Process& process) override;
 
     ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
     ResultStatus ReadIcon(std::vector<u8>& buffer) override;

src/core/loader/elf.cpp

@@ -382,13 +382,15 @@ FileType AppLoader_ELF::IdentifyType(const FileSys::VirtualFile& file) {
     return FileType::Error;
 }
 
-ResultStatus AppLoader_ELF::Load(Kernel::Process& process) {
-    if (is_loaded)
-        return ResultStatus::ErrorAlreadyLoaded;
+AppLoader_ELF::LoadResult AppLoader_ELF::Load(Kernel::Process& process) {
+    if (is_loaded) {
+        return {ResultStatus::ErrorAlreadyLoaded, {}};
+    }
 
     std::vector<u8> buffer = file->ReadAllBytes();
-    if (buffer.size() != file->GetSize())
-        return ResultStatus::ErrorIncorrectELFFileSize;
+    if (buffer.size() != file->GetSize()) {
+        return {ResultStatus::ErrorIncorrectELFFileSize, {}};
+    }
 
     const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress();
     ElfReader elf_reader(&buffer[0]);
@@ -396,10 +398,9 @@ ResultStatus AppLoader_ELF::Load(Kernel::Process& process) {
     const VAddr entry_point = codeset.entrypoint;
 
     process.LoadModule(std::move(codeset), entry_point);
-    process.Run(entry_point, 48, Memory::DEFAULT_STACK_SIZE);
 
     is_loaded = true;
-    return ResultStatus::Success;
+    return {ResultStatus::Success, LoadParameters{48, Memory::DEFAULT_STACK_SIZE}};
 }
 
 } // namespace Loader

src/core/loader/elf.h

@@ -26,7 +26,7 @@ public:
         return IdentifyType(file);
     }
 
-    ResultStatus Load(Kernel::Process& process) override;
+    LoadResult Load(Kernel::Process& process) override;
 };
 
 } // namespace Loader

src/core/loader/loader.h

@@ -131,6 +131,12 @@ std::ostream& operator<<(std::ostream& os, ResultStatus status);
 /// Interface for loading an application
 class AppLoader : NonCopyable {
 public:
+    struct LoadParameters {
+        s32 main_thread_priority;
+        u64 main_thread_stack_size;
+    };
+    using LoadResult = std::pair<ResultStatus, std::optional<LoadParameters>>;
+
     explicit AppLoader(FileSys::VirtualFile file);
     virtual ~AppLoader();
 
@@ -145,7 +151,7 @@ public:
      * @param process The newly created process.
      * @return The status result of the operation.
      */
-    virtual ResultStatus Load(Kernel::Process& process) = 0;
+    virtual LoadResult Load(Kernel::Process& process) = 0;
 
     /**
      * Loads the system mode that this application needs.

src/core/loader/nax.cpp

@@ -41,31 +41,37 @@ FileType AppLoader_NAX::GetFileType() const {
     return IdentifyTypeImpl(*nax);
 }
 
-ResultStatus AppLoader_NAX::Load(Kernel::Process& process) {
+AppLoader_NAX::LoadResult AppLoader_NAX::Load(Kernel::Process& process) {
     if (is_loaded) {
-        return ResultStatus::ErrorAlreadyLoaded;
+        return {ResultStatus::ErrorAlreadyLoaded, {}};
     }
 
-    if (nax->GetStatus() != ResultStatus::Success)
-        return nax->GetStatus();
+    const auto nax_status = nax->GetStatus();
+    if (nax_status != ResultStatus::Success) {
+        return {nax_status, {}};
+    }
 
     const auto nca = nax->AsNCA();
     if (nca == nullptr) {
-        if (!Core::Crypto::KeyManager::KeyFileExists(false))
-            return ResultStatus::ErrorMissingProductionKeyFile;
-        return ResultStatus::ErrorNAXInconvertibleToNCA;
+        if (!Core::Crypto::KeyManager::KeyFileExists(false)) {
+            return {ResultStatus::ErrorMissingProductionKeyFile, {}};
+        }
+
+        return {ResultStatus::ErrorNAXInconvertibleToNCA, {}};
     }
 
-    if (nca->GetStatus() != ResultStatus::Success)
-        return nca->GetStatus();
+    const auto nca_status = nca->GetStatus();
+    if (nca_status != ResultStatus::Success) {
+        return {nca_status, {}};
+    }
 
     const auto result = nca_loader->Load(process);
-    if (result != ResultStatus::Success)
+    if (result.first != ResultStatus::Success) {
         return result;
+    }
 
     is_loaded = true;
-
-    return ResultStatus::Success;
+    return result;
 }
 
 ResultStatus AppLoader_NAX::ReadRomFS(FileSys::VirtualFile& dir) {

src/core/loader/nax.h

@@ -33,7 +33,7 @@ public:
 
     FileType GetFileType() const override;
 
-    ResultStatus Load(Kernel::Process& process) override;
+    LoadResult Load(Kernel::Process& process) override;
 
     ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
     u64 ReadRomFSIVFCOffset() const override;

src/core/loader/nca.cpp

@@ -30,36 +30,38 @@ FileType AppLoader_NCA::IdentifyType(const FileSys::VirtualFile& file) {
     return FileType::Error;
 }
 
-ResultStatus AppLoader_NCA::Load(Kernel::Process& process) {
+AppLoader_NCA::LoadResult AppLoader_NCA::Load(Kernel::Process& process) {
     if (is_loaded) {
-        return ResultStatus::ErrorAlreadyLoaded;
+        return {ResultStatus::ErrorAlreadyLoaded, {}};
     }
 
     const auto result = nca->GetStatus();
     if (result != ResultStatus::Success) {
-        return result;
+        return {result, {}};
     }
 
-    if (nca->GetType() != FileSys::NCAContentType::Program)
-        return ResultStatus::ErrorNCANotProgram;
+    if (nca->GetType() != FileSys::NCAContentType::Program) {
+        return {ResultStatus::ErrorNCANotProgram, {}};
+    }
 
     const auto exefs = nca->GetExeFS();
-
-    if (exefs == nullptr)
-        return ResultStatus::ErrorNoExeFS;
+    if (exefs == nullptr) {
+        return {ResultStatus::ErrorNoExeFS, {}};
+    }
 
     directory_loader = std::make_unique<AppLoader_DeconstructedRomDirectory>(exefs, true);
 
     const auto load_result = directory_loader->Load(process);
-    if (load_result != ResultStatus::Success)
+    if (load_result.first != ResultStatus::Success) {
         return load_result;
+    }
 
-    if (nca->GetRomFS() != nullptr && nca->GetRomFS()->GetSize() > 0)
+    if (nca->GetRomFS() != nullptr && nca->GetRomFS()->GetSize() > 0) {
         Service::FileSystem::RegisterRomFS(std::make_unique<FileSys::RomFSFactory>(*this));
+    }
 
     is_loaded = true;
-
-    return ResultStatus::Success;
+    return load_result;
 }
 
 ResultStatus AppLoader_NCA::ReadRomFS(FileSys::VirtualFile& dir) {

src/core/loader/nca.h

@@ -33,7 +33,7 @@ public:
         return IdentifyType(file);
     }
 
-    ResultStatus Load(Kernel::Process& process) override;
+    LoadResult Load(Kernel::Process& process) override;
 
     ResultStatus ReadRomFS(FileSys::VirtualFile& dir) override;
     u64 ReadRomFSIVFCOffset() const override;

src/core/loader/nro.cpp

@@ -201,25 +201,25 @@ bool AppLoader_NRO::LoadNro(Kernel::Process& process, const FileSys::VfsFile& fi
     return LoadNroImpl(process, file.ReadAllBytes(), file.GetName(), load_base);
 }
 
-ResultStatus AppLoader_NRO::Load(Kernel::Process& process) {
+AppLoader_NRO::LoadResult AppLoader_NRO::Load(Kernel::Process& process) {
     if (is_loaded) {
-        return ResultStatus::ErrorAlreadyLoaded;
+        return {ResultStatus::ErrorAlreadyLoaded, {}};
     }
 
     // Load NRO
     const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress();
 
     if (!LoadNro(process, *file, base_address)) {
-        return ResultStatus::ErrorLoadingNRO;
+        return {ResultStatus::ErrorLoadingNRO, {}};
     }
 
-    if (romfs != nullptr)
+    if (romfs != nullptr) {
         Service::FileSystem::RegisterRomFS(std::make_unique<FileSys::RomFSFactory>(*this));
-
-    process.Run(base_address, Kernel::THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);
+    }
 
     is_loaded = true;
-    return ResultStatus::Success;
+    return {ResultStatus::Success,
+            LoadParameters{Kernel::THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE}};
 }
 
 ResultStatus AppLoader_NRO::ReadIcon(std::vector<u8>& buffer) {

src/core/loader/nro.h

@@ -37,7 +37,7 @@ public:
         return IdentifyType(file);
     }
 
-    ResultStatus Load(Kernel::Process& process) override;
+    LoadResult Load(Kernel::Process& process) override;
 
     ResultStatus ReadIcon(std::vector<u8>& buffer) override;
     ResultStatus ReadProgramId(u64& out_program_id) override;

src/core/loader/nso.cpp

@@ -169,22 +169,21 @@ std::optional<VAddr> AppLoader_NSO::LoadModule(Kernel::Process& process,
     return load_base + image_size;
 }
 
-ResultStatus AppLoader_NSO::Load(Kernel::Process& process) {
+AppLoader_NSO::LoadResult AppLoader_NSO::Load(Kernel::Process& process) {
     if (is_loaded) {
-        return ResultStatus::ErrorAlreadyLoaded;
+        return {ResultStatus::ErrorAlreadyLoaded, {}};
     }
 
     // Load module
     const VAddr base_address = process.VMManager().GetCodeRegionBaseAddress();
     if (!LoadModule(process, *file, base_address, true)) {
-        return ResultStatus::ErrorLoadingNSO;
+        return {ResultStatus::ErrorLoadingNSO, {}};
     }
     LOG_DEBUG(Loader, "loaded module {} @ 0x{:X}", file->GetName(), base_address);
 
-    process.Run(base_address, Kernel::THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE);
-
     is_loaded = true;
-    return ResultStatus::Success;
+    return {ResultStatus::Success,
+            LoadParameters{Kernel::THREADPRIO_DEFAULT, Memory::DEFAULT_STACK_SIZE}};
 }
 
 } // namespace Loader

src/core/loader/nso.h

@@ -84,7 +84,7 @@ public:
                                            VAddr load_base, bool should_pass_arguments,
                                            std::optional<FileSys::PatchManager> pm = {});
 
-    ResultStatus Load(Kernel::Process& process) override;
+    LoadResult Load(Kernel::Process& process) override;
 };
 
 } // namespace Loader

src/core/loader/nsp.cpp

@@ -72,37 +72,45 @@ FileType AppLoader_NSP::IdentifyType(const FileSys::VirtualFile& file) {
     return FileType::Error;
 }
 
-ResultStatus AppLoader_NSP::Load(Kernel::Process& process) {
+AppLoader_NSP::LoadResult AppLoader_NSP::Load(Kernel::Process& process) {
     if (is_loaded) {
-        return ResultStatus::ErrorAlreadyLoaded;
+        return {ResultStatus::ErrorAlreadyLoaded, {}};
     }
 
-    if (title_id == 0)
-        return ResultStatus::ErrorNSPMissingProgramNCA;
+    if (title_id == 0) {
+        return {ResultStatus::ErrorNSPMissingProgramNCA, {}};
+    }
 
-    if (nsp->GetStatus() != ResultStatus::Success)
-        return nsp->GetStatus();
+    const auto nsp_status = nsp->GetStatus();
+    if (nsp_status != ResultStatus::Success) {
+        return {nsp_status, {}};
+    }
 
-    if (nsp->GetProgramStatus(title_id) != ResultStatus::Success)
-        return nsp->GetProgramStatus(title_id);
+    const auto nsp_program_status = nsp->GetProgramStatus(title_id);
+    if (nsp_program_status != ResultStatus::Success) {
+        return {nsp_program_status, {}};
+    }
 
     if (nsp->GetNCA(title_id, FileSys::ContentRecordType::Program) == nullptr) {
-        if (!Core::Crypto::KeyManager::KeyFileExists(false))
-            return ResultStatus::ErrorMissingProductionKeyFile;
-        return ResultStatus::ErrorNSPMissingProgramNCA;
+        if (!Core::Crypto::KeyManager::KeyFileExists(false)) {
+            return {ResultStatus::ErrorMissingProductionKeyFile, {}};
+        }
+
+        return {ResultStatus::ErrorNSPMissingProgramNCA, {}};
     }
 
     const auto result = secondary_loader->Load(process);
-    if (result != ResultStatus::Success)
+    if (result.first != ResultStatus::Success) {
         return result;
+    }
 
     FileSys::VirtualFile update_raw;
-    if (ReadUpdateRaw(update_raw) == ResultStatus::Success && update_raw != nullptr)
+    if (ReadUpdateRaw(update_raw) == ResultStatus::Success && update_raw != nullptr) {
         Service::FileSystem::SetPackedUpdate(std::move(update_raw));
+    }
 
     is_loaded = true;
-
-    return ResultStatus::Success;
+    return result;
 }
 
 ResultStatus AppLoader_NSP::ReadRomFS(FileSys::VirtualFile& file) {

src/core/loader/nsp.h

@@ -35,7 +35,7 @@ public:
         return IdentifyType(file);
     }
 
-    ResultStatus Load(Kernel::Process& process) override;
+    LoadResult Load(Kernel::Process& process) override;
 
     ResultStatus ReadRomFS(FileSys::VirtualFile& file) override;
     u64 ReadRomFSIVFCOffset() const override;

src/core/loader/xci.cpp

@@ -48,31 +48,35 @@ FileType AppLoader_XCI::IdentifyType(const FileSys::VirtualFile& file) {
     return FileType::Error;
 }
 
-ResultStatus AppLoader_XCI::Load(Kernel::Process& process) {
+AppLoader_XCI::LoadResult AppLoader_XCI::Load(Kernel::Process& process) {
     if (is_loaded) {
-        return ResultStatus::ErrorAlreadyLoaded;
+        return {ResultStatus::ErrorAlreadyLoaded, {}};
     }
 
-    if (xci->GetStatus() != ResultStatus::Success)
-        return xci->GetStatus();
+    if (xci->GetStatus() != ResultStatus::Success) {
+        return {xci->GetStatus(), {}};
+    }
 
-    if (xci->GetProgramNCAStatus() != ResultStatus::Success)
-        return xci->GetProgramNCAStatus();
+    if (xci->GetProgramNCAStatus() != ResultStatus::Success) {
+        return {xci->GetProgramNCAStatus(), {}};
+    }
 
-    if (!xci->HasProgramNCA() && !Core::Crypto::KeyManager::KeyFileExists(false))
-        return ResultStatus::ErrorMissingProductionKeyFile;
+    if (!xci->HasProgramNCA() && !Core::Crypto::KeyManager::KeyFileExists(false)) {
+        return {ResultStatus::ErrorMissingProductionKeyFile, {}};
+    }
 
     const auto result = nca_loader->Load(process);
-    if (result != ResultStatus::Success)
+    if (result.first != ResultStatus::Success) {
         return result;
+    }
 
     FileSys::VirtualFile update_raw;
-    if (ReadUpdateRaw(update_raw) == ResultStatus::Success && update_raw != nullptr)
+    if (ReadUpdateRaw(update_raw) == ResultStatus::Success && update_raw != nullptr) {
         Service::FileSystem::SetPackedUpdate(std::move(update_raw));
+    }
 
     is_loaded = true;
-
-    return ResultStatus::Success;
+    return result;
 }
 
 ResultStatus AppLoader_XCI::ReadRomFS(FileSys::VirtualFile& file) {

src/core/loader/xci.h

@@ -35,7 +35,7 @@ public:
         return IdentifyType(file);
     }
 
-    ResultStatus Load(Kernel::Process& process) override;
+    LoadResult Load(Kernel::Process& process) override;
 
     ResultStatus ReadRomFS(FileSys::VirtualFile& file) override;
     u64 ReadRomFSIVFCOffset() const override;

src/core/memory.cpp

@@ -26,16 +26,16 @@ namespace Memory {
 
 static Common::PageTable* current_page_table = nullptr;
 
-void SetCurrentPageTable(Common::PageTable* page_table) {
-    current_page_table = page_table;
+void SetCurrentPageTable(Kernel::Process& process) {
+    current_page_table = &process.VMManager().page_table;
+
+    const std::size_t address_space_width = process.VMManager().GetAddressSpaceWidth();
 
     auto& system = Core::System::GetInstance();
-    if (system.IsPoweredOn()) {
-        system.ArmInterface(0).PageTableChanged();
-        system.ArmInterface(1).PageTableChanged();
-        system.ArmInterface(2).PageTableChanged();
-        system.ArmInterface(3).PageTableChanged();
-    }
+    system.ArmInterface(0).PageTableChanged(*current_page_table, address_space_width);
+    system.ArmInterface(1).PageTableChanged(*current_page_table, address_space_width);
+    system.ArmInterface(2).PageTableChanged(*current_page_table, address_space_width);
+    system.ArmInterface(3).PageTableChanged(*current_page_table, address_space_width);
 }
 
 static void MapPages(Common::PageTable& page_table, VAddr base, u64 size, u8* memory,

src/core/memory.h

@@ -40,8 +40,9 @@ enum : VAddr {
     KERNEL_REGION_END = KERNEL_REGION_VADDR + KERNEL_REGION_SIZE,
 };
 
-/// Changes the currently active page table.
-void SetCurrentPageTable(Common::PageTable* page_table);
+/// Changes the currently active page table to that of
+/// the given process instance.
+void SetCurrentPageTable(Kernel::Process& process);
 
 /// Determines if the given VAddr is valid for the specified process.
 bool IsValidVirtualAddress(const Kernel::Process& process, VAddr vaddr);

src/core/settings.cpp

@@ -90,6 +90,7 @@ void LogSettings() {
     LogSetting("Renderer_UseResolutionFactor", Settings::values.resolution_factor);
     LogSetting("Renderer_UseFrameLimit", Settings::values.use_frame_limit);
     LogSetting("Renderer_FrameLimit", Settings::values.frame_limit);
+    LogSetting("Renderer_UseCompatibilityProfile", Settings::values.use_compatibility_profile);
     LogSetting("Renderer_UseDiskShaderCache", Settings::values.use_disk_shader_cache);
     LogSetting("Renderer_UseAccurateGpuEmulation", Settings::values.use_accurate_gpu_emulation);
     LogSetting("Renderer_UseAsynchronousGpuEmulation",

src/core/settings.h

@@ -390,6 +390,7 @@ struct Values {
     float resolution_factor;
     bool use_frame_limit;
     u16 frame_limit;
+    bool use_compatibility_profile;
     bool use_disk_shader_cache;
     bool use_accurate_gpu_emulation;
     bool use_asynchronous_gpu_emulation;

src/video_core/CMakeLists.txt

@@ -46,6 +46,8 @@ add_library(video_core STATIC
     renderer_opengl/gl_rasterizer_cache.h
     renderer_opengl/gl_resource_manager.cpp
     renderer_opengl/gl_resource_manager.h
+    renderer_opengl/gl_sampler_cache.cpp
+    renderer_opengl/gl_sampler_cache.h
     renderer_opengl/gl_shader_cache.cpp
     renderer_opengl/gl_shader_cache.h
     renderer_opengl/gl_shader_decompiler.cpp
@@ -67,6 +69,8 @@ add_library(video_core STATIC
     renderer_opengl/renderer_opengl.h
     renderer_opengl/utils.cpp
     renderer_opengl/utils.h
+    sampler_cache.cpp
+    sampler_cache.h
     shader/decode/arithmetic.cpp
     shader/decode/arithmetic_immediate.cpp
     shader/decode/bfe.cpp

src/video_core/engines/maxwell_3d.cpp

@@ -482,19 +482,8 @@ std::vector<Texture::FullTextureInfo> Maxwell3D::GetStageTextures(Regs::ShaderSt
     return textures;
 }
 
-Texture::FullTextureInfo Maxwell3D::GetStageTexture(Regs::ShaderStage stage,
-                                                    std::size_t offset) const {
-    auto& shader = state.shader_stages[static_cast<std::size_t>(stage)];
-    auto& tex_info_buffer = shader.const_buffers[regs.tex_cb_index];
-    ASSERT(tex_info_buffer.enabled && tex_info_buffer.address != 0);
-
-    const GPUVAddr tex_info_address =
-        tex_info_buffer.address + offset * sizeof(Texture::TextureHandle);
-
-    ASSERT(tex_info_address < tex_info_buffer.address + tex_info_buffer.size);
-
-    const Texture::TextureHandle tex_handle{memory_manager.Read<u32>(tex_info_address)};
-
+Texture::FullTextureInfo Maxwell3D::GetTextureInfo(const Texture::TextureHandle tex_handle,
+                                                   std::size_t offset) const {
     Texture::FullTextureInfo tex_info{};
     tex_info.index = static_cast<u32>(offset);
 
@@ -511,6 +500,22 @@ Texture::FullTextureInfo Maxwell3D::GetStageTexture(Regs::ShaderStage stage,
     return tex_info;
 }
 
+Texture::FullTextureInfo Maxwell3D::GetStageTexture(Regs::ShaderStage stage,
+                                                    std::size_t offset) const {
+    const auto& shader = state.shader_stages[static_cast<std::size_t>(stage)];
+    const auto& tex_info_buffer = shader.const_buffers[regs.tex_cb_index];
+    ASSERT(tex_info_buffer.enabled && tex_info_buffer.address != 0);
+
+    const GPUVAddr tex_info_address =
+        tex_info_buffer.address + offset * sizeof(Texture::TextureHandle);
+
+    ASSERT(tex_info_address < tex_info_buffer.address + tex_info_buffer.size);
+
+    const Texture::TextureHandle tex_handle{memory_manager.Read<u32>(tex_info_address)};
+
+    return GetTextureInfo(tex_handle, offset);
+}
+
 u32 Maxwell3D::GetRegisterValue(u32 method) const {
     ASSERT_MSG(method < Regs::NUM_REGS, "Invalid Maxwell3D register");
     return regs.reg_array[method];
@@ -524,4 +529,12 @@ void Maxwell3D::ProcessClearBuffers() {
     rasterizer.Clear();
 }
 
+u32 Maxwell3D::AccessConstBuffer32(Regs::ShaderStage stage, u64 const_buffer, u64 offset) const {
+    const auto& shader_stage = state.shader_stages[static_cast<std::size_t>(stage)];
+    const auto& buffer = shader_stage.const_buffers[const_buffer];
+    u32 result;
+    std::memcpy(&result, memory_manager.GetPointer(buffer.address + offset), sizeof(u32));
+    return result;
+}
+
 } // namespace Tegra::Engines

src/video_core/engines/maxwell_3d.h

@@ -1131,12 +1131,18 @@ public:
     /// Write the value to the register identified by method.
     void CallMethod(const GPU::MethodCall& method_call);
 
+    /// Given a Texture Handle, returns the TSC and TIC entries.
+    Texture::FullTextureInfo GetTextureInfo(const Texture::TextureHandle tex_handle,
+                                            std::size_t offset) const;
+
     /// Returns a list of enabled textures for the specified shader stage.
     std::vector<Texture::FullTextureInfo> GetStageTextures(Regs::ShaderStage stage) const;
 
     /// Returns the texture information for a specific texture in a specific shader stage.
     Texture::FullTextureInfo GetStageTexture(Regs::ShaderStage stage, std::size_t offset) const;
 
+    u32 AccessConstBuffer32(Regs::ShaderStage stage, u64 const_buffer, u64 offset) const;
+
     /// Memory for macro code - it's undetermined how big this is, however 1MB is much larger than
     /// we've seen used.
     using MacroMemory = std::array<u32, 0x40000>;

src/video_core/engines/shader_bytecode.h

@@ -986,6 +986,38 @@ union Instruction {
         }
     } tex;
 
+    union {
+        BitField<28, 1, u64> array;
+        BitField<29, 2, TextureType> texture_type;
+        BitField<31, 4, u64> component_mask;
+        BitField<49, 1, u64> nodep_flag;
+        BitField<50, 1, u64> dc_flag;
+        BitField<36, 1, u64> aoffi_flag;
+        BitField<37, 3, TextureProcessMode> process_mode;
+
+        bool IsComponentEnabled(std::size_t component) const {
+            return ((1ULL << component) & component_mask) != 0;
+        }
+
+        TextureProcessMode GetTextureProcessMode() const {
+            return process_mode;
+        }
+
+        bool UsesMiscMode(TextureMiscMode mode) const {
+            switch (mode) {
+            case TextureMiscMode::DC:
+                return dc_flag != 0;
+            case TextureMiscMode::NODEP:
+                return nodep_flag != 0;
+            case TextureMiscMode::AOFFI:
+                return aoffi_flag != 0;
+            default:
+                break;
+            }
+            return false;
+        }
+    } tex_b;
+
     union {
         BitField<22, 6, TextureQueryType> query_type;
         BitField<31, 4, u64> component_mask;
@@ -1332,7 +1364,9 @@ public:
         LDG, // Load from global memory
         STG, // Store in global memory
         TEX,
+        TEX_B,  // Texture Load Bindless
         TXQ,    // Texture Query
+        TXQ_B,  // Texture Query Bindless
         TEXS,   // Texture Fetch with scalar/non-vec4 source/destinations
         TLDS,   // Texture Load with scalar/non-vec4 source/destinations
         TLD4,   // Texture Load 4
@@ -1600,7 +1634,9 @@ private:
             INST("1110111011010---", Id::LDG, Type::Memory, "LDG"),
             INST("1110111011011---", Id::STG, Type::Memory, "STG"),
             INST("110000----111---", Id::TEX, Type::Texture, "TEX"),
+            INST("1101111010111---", Id::TEX_B, Type::Texture, "TEX_B"),
             INST("1101111101001---", Id::TXQ, Type::Texture, "TXQ"),
+            INST("1101111101010---", Id::TXQ_B, Type::Texture, "TXQ_B"),
             INST("1101-00---------", Id::TEXS, Type::Texture, "TEXS"),
             INST("1101101---------", Id::TLDS, Type::Texture, "TLDS"),
             INST("110010----111---", Id::TLD4, Type::Texture, "TLD4"),

src/video_core/gpu.h

@@ -207,6 +207,11 @@ public:
         };
     } regs{};
 
+    /// Performs any additional setup necessary in order to begin GPU emulation.
+    /// This can be used to launch any necessary threads and register any necessary
+    /// core timing events.
+    virtual void Start() = 0;
+
     /// Push GPU command entries to be processed
     virtual void PushGPUEntries(Tegra::CommandList&& entries) = 0;
 

src/video_core/gpu_asynch.cpp

@@ -9,10 +9,14 @@
 namespace VideoCommon {
 
 GPUAsynch::GPUAsynch(Core::System& system, VideoCore::RendererBase& renderer)
-    : Tegra::GPU(system, renderer), gpu_thread{system, renderer, *dma_pusher} {}
+    : GPU(system, renderer), gpu_thread{system} {}
 
 GPUAsynch::~GPUAsynch() = default;
 
+void GPUAsynch::Start() {
+    gpu_thread.StartThread(renderer, *dma_pusher);
+}
+
 void GPUAsynch::PushGPUEntries(Tegra::CommandList&& entries) {
     gpu_thread.SubmitList(std::move(entries));
 }

src/video_core/gpu_asynch.h

@@ -13,16 +13,13 @@ class RendererBase;
 
 namespace VideoCommon {
 
-namespace GPUThread {
-class ThreadManager;
-} // namespace GPUThread
-
 /// Implementation of GPU interface that runs the GPU asynchronously
 class GPUAsynch : public Tegra::GPU {
 public:
     explicit GPUAsynch(Core::System& system, VideoCore::RendererBase& renderer);
     ~GPUAsynch() override;
 
+    void Start() override;
     void PushGPUEntries(Tegra::CommandList&& entries) override;
     void SwapBuffers(
         std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) override;

src/video_core/gpu_synch.cpp

@@ -8,10 +8,12 @@
 namespace VideoCommon {
 
 GPUSynch::GPUSynch(Core::System& system, VideoCore::RendererBase& renderer)
-    : Tegra::GPU(system, renderer) {}
+    : GPU(system, renderer) {}
 
 GPUSynch::~GPUSynch() = default;
 
+void GPUSynch::Start() {}
+
 void GPUSynch::PushGPUEntries(Tegra::CommandList&& entries) {
     dma_pusher->Push(std::move(entries));
     dma_pusher->DispatchCalls();

src/video_core/gpu_synch.h

@@ -18,6 +18,7 @@ public:
     explicit GPUSynch(Core::System& system, VideoCore::RendererBase& renderer);
     ~GPUSynch() override;
 
+    void Start() override;
     void PushGPUEntries(Tegra::CommandList&& entries) override;
     void SwapBuffers(
         std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) override;

src/video_core/gpu_thread.cpp

@@ -55,19 +55,24 @@ static void RunThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_p
     }
 }
 
-ThreadManager::ThreadManager(Core::System& system, VideoCore::RendererBase& renderer,
-                             Tegra::DmaPusher& dma_pusher)
-    : system{system}, thread{RunThread, std::ref(renderer), std::ref(dma_pusher), std::ref(state)} {
-    synchronization_event = system.CoreTiming().RegisterEvent(
-        "GPUThreadSynch", [this](u64 fence, s64) { state.WaitForSynchronization(fence); });
-}
+ThreadManager::ThreadManager(Core::System& system) : system{system} {}
 
 ThreadManager::~ThreadManager() {
+    if (!thread.joinable()) {
+        return;
+    }
+
     // Notify GPU thread that a shutdown is pending
     PushCommand(EndProcessingCommand());
     thread.join();
 }
 
+void ThreadManager::StartThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher) {
+    thread = std::thread{RunThread, std::ref(renderer), std::ref(dma_pusher), std::ref(state)};
+    synchronization_event = system.CoreTiming().RegisterEvent(
+        "GPUThreadSynch", [this](u64 fence, s64) { state.WaitForSynchronization(fence); });
+}
+
 void ThreadManager::SubmitList(Tegra::CommandList&& entries) {
     const u64 fence{PushCommand(SubmitListCommand(std::move(entries)))};
     const s64 synchronization_ticks{Core::Timing::usToCycles(9000)};

src/video_core/gpu_thread.h

@@ -138,10 +138,12 @@ struct SynchState final {
 /// Class used to manage the GPU thread
 class ThreadManager final {
 public:
-    explicit ThreadManager(Core::System& system, VideoCore::RendererBase& renderer,
-                           Tegra::DmaPusher& dma_pusher);
+    explicit ThreadManager(Core::System& system);
     ~ThreadManager();
 
+    /// Creates and starts the GPU thread.
+    void StartThread(VideoCore::RendererBase& renderer, Tegra::DmaPusher& dma_pusher);
+
     /// Push GPU command entries to be processed
     void SubmitList(Tegra::CommandList&& entries);
 

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -101,12 +101,6 @@ struct FramebufferCacheKey {
 RasterizerOpenGL::RasterizerOpenGL(Core::System& system, ScreenInfo& info)
     : res_cache{*this}, shader_cache{*this, system}, global_cache{*this}, system{system},
       screen_info{info}, buffer_cache(*this, STREAM_BUFFER_SIZE) {
-    // Create sampler objects
-    for (std::size_t i = 0; i < texture_samplers.size(); ++i) {
-        texture_samplers[i].Create();
-        state.texture_units[i].sampler = texture_samplers[i].sampler.handle;
-    }
-
     OpenGLState::ApplyDefaultState();
 
     shader_program_manager = std::make_unique<GLShader::ProgramManager>();
@@ -813,92 +807,6 @@ bool RasterizerOpenGL::AccelerateDisplay(const Tegra::FramebufferConfig& config,
     return true;
 }
 
-void RasterizerOpenGL::SamplerInfo::Create() {
-    sampler.Create();
-    mag_filter = Tegra::Texture::TextureFilter::Linear;
-    min_filter = Tegra::Texture::TextureFilter::Linear;
-    wrap_u = Tegra::Texture::WrapMode::Wrap;
-    wrap_v = Tegra::Texture::WrapMode::Wrap;
-    wrap_p = Tegra::Texture::WrapMode::Wrap;
-    use_depth_compare = false;
-    depth_compare_func = Tegra::Texture::DepthCompareFunc::Never;
-
-    // OpenGL's default is GL_LINEAR_MIPMAP_LINEAR
-    glSamplerParameteri(sampler.handle, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
-    glSamplerParameteri(sampler.handle, GL_TEXTURE_COMPARE_FUNC, GL_NEVER);
-
-    // Other attributes have correct defaults
-}
-
-void RasterizerOpenGL::SamplerInfo::SyncWithConfig(const Tegra::Texture::TSCEntry& config) {
-    const GLuint sampler_id = sampler.handle;
-    if (mag_filter != config.mag_filter) {
-        mag_filter = config.mag_filter;
-        glSamplerParameteri(
-            sampler_id, GL_TEXTURE_MAG_FILTER,
-            MaxwellToGL::TextureFilterMode(mag_filter, Tegra::Texture::TextureMipmapFilter::None));
-    }
-    if (min_filter != config.min_filter || mipmap_filter != config.mipmap_filter) {
-        min_filter = config.min_filter;
-        mipmap_filter = config.mipmap_filter;
-        glSamplerParameteri(sampler_id, GL_TEXTURE_MIN_FILTER,
-                            MaxwellToGL::TextureFilterMode(min_filter, mipmap_filter));
-    }
-
-    if (wrap_u != config.wrap_u) {
-        wrap_u = config.wrap_u;
-        glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_S, MaxwellToGL::WrapMode(wrap_u));
-    }
-    if (wrap_v != config.wrap_v) {
-        wrap_v = config.wrap_v;
-        glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_T, MaxwellToGL::WrapMode(wrap_v));
-    }
-    if (wrap_p != config.wrap_p) {
-        wrap_p = config.wrap_p;
-        glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_R, MaxwellToGL::WrapMode(wrap_p));
-    }
-
-    if (const bool enabled = config.depth_compare_enabled == 1; use_depth_compare != enabled) {
-        use_depth_compare = enabled;
-        glSamplerParameteri(sampler_id, GL_TEXTURE_COMPARE_MODE,
-                            use_depth_compare ? GL_COMPARE_REF_TO_TEXTURE : GL_NONE);
-    }
-
-    if (depth_compare_func != config.depth_compare_func) {
-        depth_compare_func = config.depth_compare_func;
-        glSamplerParameteri(sampler_id, GL_TEXTURE_COMPARE_FUNC,
-                            MaxwellToGL::DepthCompareFunc(depth_compare_func));
-    }
-
-    if (const auto new_border_color = config.GetBorderColor(); border_color != new_border_color) {
-        border_color = new_border_color;
-        glSamplerParameterfv(sampler_id, GL_TEXTURE_BORDER_COLOR, border_color.data());
-    }
-
-    if (const float anisotropic = config.GetMaxAnisotropy(); max_anisotropic != anisotropic) {
-        max_anisotropic = anisotropic;
-        if (GLAD_GL_ARB_texture_filter_anisotropic) {
-            glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_ANISOTROPY, max_anisotropic);
-        } else if (GLAD_GL_EXT_texture_filter_anisotropic) {
-            glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_ANISOTROPY_EXT, max_anisotropic);
-        }
-    }
-
-    if (const float min = config.GetMinLod(); min_lod != min) {
-        min_lod = min;
-        glSamplerParameterf(sampler_id, GL_TEXTURE_MIN_LOD, min_lod);
-    }
-    if (const float max = config.GetMaxLod(); max_lod != max) {
-        max_lod = max;
-        glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_LOD, max_lod);
-    }
-
-    if (const float bias = config.GetLodBias(); lod_bias != bias) {
-        lod_bias = bias;
-        glSamplerParameterf(sampler_id, GL_TEXTURE_LOD_BIAS, lod_bias);
-    }
-}
-
 void RasterizerOpenGL::SetupConstBuffers(Tegra::Engines::Maxwell3D::Regs::ShaderStage stage,
                                          const Shader& shader, GLuint program_handle,
                                          BaseBindings base_bindings) {
@@ -974,10 +882,18 @@ void RasterizerOpenGL::SetupTextures(Maxwell::ShaderStage stage, const Shader& s
 
     for (u32 bindpoint = 0; bindpoint < entries.size(); ++bindpoint) {
         const auto& entry = entries[bindpoint];
-        const auto texture = maxwell3d.GetStageTexture(stage, entry.GetOffset());
+        Tegra::Texture::FullTextureInfo texture;
+        if (entry.IsBindless()) {
+            const auto cbuf = entry.GetBindlessCBuf();
+            Tegra::Texture::TextureHandle tex_handle;
+            tex_handle.raw = maxwell3d.AccessConstBuffer32(stage, cbuf.first, cbuf.second);
+            texture = maxwell3d.GetTextureInfo(tex_handle, entry.GetOffset());
+        } else {
+            texture = maxwell3d.GetStageTexture(stage, entry.GetOffset());
+        }
         const u32 current_bindpoint = base_bindings.sampler + bindpoint;
 
-        texture_samplers[current_bindpoint].SyncWithConfig(texture.tsc);
+        state.texture_units[current_bindpoint].sampler = sampler_cache.GetSampler(texture.tsc);
 
         if (Surface surface = res_cache.GetTextureSurface(texture, entry); surface) {
             state.texture_units[current_bindpoint].texture =

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -25,6 +25,7 @@
 #include "video_core/renderer_opengl/gl_primitive_assembler.h"
 #include "video_core/renderer_opengl/gl_rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/renderer_opengl/gl_sampler_cache.h"
 #include "video_core/renderer_opengl/gl_shader_cache.h"
 #include "video_core/renderer_opengl/gl_shader_manager.h"
 #include "video_core/renderer_opengl/gl_state.h"
@@ -72,34 +73,6 @@ public:
                   "The maximum size of a constbuffer must be a multiple of the size of GLvec4");
 
 private:
-    class SamplerInfo {
-    public:
-        OGLSampler sampler;
-
-        /// Creates the sampler object, initializing its state so that it's in sync with the
-        /// SamplerInfo struct.
-        void Create();
-        /// Syncs the sampler object with the config, updating any necessary state.
-        void SyncWithConfig(const Tegra::Texture::TSCEntry& info);
-
-    private:
-        Tegra::Texture::TextureFilter mag_filter = Tegra::Texture::TextureFilter::Nearest;
-        Tegra::Texture::TextureFilter min_filter = Tegra::Texture::TextureFilter::Nearest;
-        Tegra::Texture::TextureMipmapFilter mipmap_filter =
-            Tegra::Texture::TextureMipmapFilter::None;
-        Tegra::Texture::WrapMode wrap_u = Tegra::Texture::WrapMode::ClampToEdge;
-        Tegra::Texture::WrapMode wrap_v = Tegra::Texture::WrapMode::ClampToEdge;
-        Tegra::Texture::WrapMode wrap_p = Tegra::Texture::WrapMode::ClampToEdge;
-        bool use_depth_compare = false;
-        Tegra::Texture::DepthCompareFunc depth_compare_func =
-            Tegra::Texture::DepthCompareFunc::Always;
-        GLvec4 border_color = {};
-        float min_lod = 0.0f;
-        float max_lod = 16.0f;
-        float lod_bias = 0.0f;
-        float max_anisotropic = 1.0f;
-    };
-
     struct FramebufferConfigState {
         bool using_color_fb{};
         bool using_depth_fb{};
@@ -204,6 +177,7 @@ private:
     RasterizerCacheOpenGL res_cache;
     ShaderCacheOpenGL shader_cache;
     GlobalRegionCacheOpenGL global_cache;
+    SamplerCacheOpenGL sampler_cache;
 
     Core::System& system;
 
@@ -219,8 +193,6 @@ private:
     FramebufferConfigState current_framebuffer_config_state;
     std::pair<bool, bool> current_depth_stencil_usage{};
 
-    std::array<SamplerInfo, Tegra::Engines::Maxwell3D::Regs::NumTextureSamplers> texture_samplers;
-
     static constexpr std::size_t STREAM_BUFFER_SIZE = 128 * 1024 * 1024;
     OGLBufferCache buffer_cache;
     PrimitiveAssembler primitive_assembler{buffer_cache};

src/video_core/renderer_opengl/gl_sampler_cache.cpp

@@ -0,0 +1,52 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/logging/log.h"
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/renderer_opengl/gl_sampler_cache.h"
+#include "video_core/renderer_opengl/maxwell_to_gl.h"
+
+namespace OpenGL {
+
+SamplerCacheOpenGL::SamplerCacheOpenGL() = default;
+
+SamplerCacheOpenGL::~SamplerCacheOpenGL() = default;
+
+OGLSampler SamplerCacheOpenGL::CreateSampler(const Tegra::Texture::TSCEntry& tsc) const {
+    OGLSampler sampler;
+    sampler.Create();
+
+    const GLuint sampler_id{sampler.handle};
+    glSamplerParameteri(
+        sampler_id, GL_TEXTURE_MAG_FILTER,
+        MaxwellToGL::TextureFilterMode(tsc.mag_filter, Tegra::Texture::TextureMipmapFilter::None));
+    glSamplerParameteri(sampler_id, GL_TEXTURE_MIN_FILTER,
+                        MaxwellToGL::TextureFilterMode(tsc.min_filter, tsc.mipmap_filter));
+    glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_S, MaxwellToGL::WrapMode(tsc.wrap_u));
+    glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_T, MaxwellToGL::WrapMode(tsc.wrap_v));
+    glSamplerParameteri(sampler_id, GL_TEXTURE_WRAP_R, MaxwellToGL::WrapMode(tsc.wrap_p));
+    glSamplerParameteri(sampler_id, GL_TEXTURE_COMPARE_MODE,
+                        tsc.depth_compare_enabled == 1 ? GL_COMPARE_REF_TO_TEXTURE : GL_NONE);
+    glSamplerParameteri(sampler_id, GL_TEXTURE_COMPARE_FUNC,
+                        MaxwellToGL::DepthCompareFunc(tsc.depth_compare_func));
+    glSamplerParameterfv(sampler_id, GL_TEXTURE_BORDER_COLOR, tsc.GetBorderColor().data());
+    glSamplerParameterf(sampler_id, GL_TEXTURE_MIN_LOD, tsc.GetMinLod());
+    glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_LOD, tsc.GetMaxLod());
+    glSamplerParameterf(sampler_id, GL_TEXTURE_LOD_BIAS, tsc.GetLodBias());
+    if (GLAD_GL_ARB_texture_filter_anisotropic) {
+        glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_ANISOTROPY, tsc.GetMaxAnisotropy());
+    } else if (GLAD_GL_EXT_texture_filter_anisotropic) {
+        glSamplerParameterf(sampler_id, GL_TEXTURE_MAX_ANISOTROPY_EXT, tsc.GetMaxAnisotropy());
+    } else if (tsc.GetMaxAnisotropy() != 1) {
+        LOG_WARNING(Render_OpenGL, "Anisotropy not supported by host GPU driver");
+    }
+
+    return sampler;
+}
+
+GLuint SamplerCacheOpenGL::ToSamplerType(const OGLSampler& sampler) const {
+    return sampler.handle;
+}
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_sampler_cache.h

@@ -0,0 +1,25 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <glad/glad.h>
+
+#include "video_core/renderer_opengl/gl_resource_manager.h"
+#include "video_core/sampler_cache.h"
+
+namespace OpenGL {
+
+class SamplerCacheOpenGL final : public VideoCommon::SamplerCache<GLuint, OGLSampler> {
+public:
+    explicit SamplerCacheOpenGL();
+    ~SamplerCacheOpenGL();
+
+protected:
+    OGLSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const;
+
+    GLuint ToSamplerType(const OGLSampler& sampler) const;
+};
+
+} // namespace OpenGL

src/video_core/renderer_opengl/gl_shader_decompiler.h

@@ -69,6 +69,7 @@ private:
 struct ShaderEntries {
     std::vector<ConstBufferEntry> const_buffers;
     std::vector<SamplerEntry> samplers;
+    std::vector<SamplerEntry> bindless_samplers;
     std::vector<GlobalMemoryEntry> global_memory_entries;
     std::array<bool, Maxwell::NumClipDistances> clip_distances{};
     std::size_t shader_length{};
@@ -79,4 +80,4 @@ std::string GetCommonDeclarations();
 ProgramResult Decompile(const VideoCommon::Shader::ShaderIR& ir, Maxwell::ShaderStage stage,
                         const std::string& suffix);
 
-} // namespace OpenGL::GLShader
+} // namespace OpenGL::GLShader

src/video_core/renderer_opengl/gl_shader_disk_cache.cpp

@@ -319,16 +319,19 @@ std::optional<ShaderDiskCacheDecompiled> ShaderDiskCacheOpenGL::LoadDecompiledEn
         u32 type{};
         u8 is_array{};
         u8 is_shadow{};
+        u8 is_bindless{};
         if (file.ReadBytes(&offset, sizeof(u64)) != sizeof(u64) ||
             file.ReadBytes(&index, sizeof(u64)) != sizeof(u64) ||
             file.ReadBytes(&type, sizeof(u32)) != sizeof(u32) ||
             file.ReadBytes(&is_array, sizeof(u8)) != sizeof(u8) ||
-            file.ReadBytes(&is_shadow, sizeof(u8)) != sizeof(u8)) {
+            file.ReadBytes(&is_shadow, sizeof(u8)) != sizeof(u8) ||
+            file.ReadBytes(&is_bindless, sizeof(u8)) != sizeof(u8)) {
             return {};
         }
-        entry.entries.samplers.emplace_back(
-            static_cast<std::size_t>(offset), static_cast<std::size_t>(index),
-            static_cast<Tegra::Shader::TextureType>(type), is_array != 0, is_shadow != 0);
+        entry.entries.samplers.emplace_back(static_cast<std::size_t>(offset),
+                                            static_cast<std::size_t>(index),
+                                            static_cast<Tegra::Shader::TextureType>(type),
+                                            is_array != 0, is_shadow != 0, is_bindless != 0);
     }
 
     u32 global_memory_count{};
@@ -393,7 +396,8 @@ bool ShaderDiskCacheOpenGL::SaveDecompiledFile(FileUtil::IOFile& file, u64 uniqu
             file.WriteObject(static_cast<u64>(sampler.GetIndex())) != 1 ||
             file.WriteObject(static_cast<u32>(sampler.GetType())) != 1 ||
             file.WriteObject(static_cast<u8>(sampler.IsArray() ? 1 : 0)) != 1 ||
-            file.WriteObject(static_cast<u8>(sampler.IsShadow() ? 1 : 0)) != 1) {
+            file.WriteObject(static_cast<u8>(sampler.IsShadow() ? 1 : 0)) != 1 ||
+            file.WriteObject(static_cast<u8>(sampler.IsBindless() ? 1 : 0)) != 1) {
             return false;
         }
     }

src/video_core/renderer_vulkan/vk_sampler_cache.cpp

@@ -7,7 +7,6 @@
 #include <unordered_map>
 
 #include "common/assert.h"
-#include "common/cityhash.h"
 #include "video_core/renderer_vulkan/declarations.h"
 #include "video_core/renderer_vulkan/maxwell_to_vk.h"
 #include "video_core/renderer_vulkan/vk_sampler_cache.h"
@@ -28,39 +27,20 @@ static std::optional<vk::BorderColor> TryConvertBorderColor(std::array<float, 4>
     }
 }
 
-std::size_t SamplerCacheKey::Hash() const {
-    static_assert(sizeof(raw) % sizeof(u64) == 0);
-    return static_cast<std::size_t>(
-        Common::CityHash64(reinterpret_cast<const char*>(raw.data()), sizeof(raw) / sizeof(u64)));
-}
-
-bool SamplerCacheKey::operator==(const SamplerCacheKey& rhs) const {
-    return raw == rhs.raw;
-}
-
 VKSamplerCache::VKSamplerCache(const VKDevice& device) : device{device} {}
 
 VKSamplerCache::~VKSamplerCache() = default;
 
-vk::Sampler VKSamplerCache::GetSampler(const Tegra::Texture::TSCEntry& tsc) {
-    const auto [entry, is_cache_miss] = cache.try_emplace(SamplerCacheKey{tsc});
-    auto& sampler = entry->second;
-    if (is_cache_miss) {
-        sampler = CreateSampler(tsc);
-    }
-    return *sampler;
-}
+UniqueSampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc) const {
+    const float max_anisotropy{tsc.GetMaxAnisotropy()};
+    const bool has_anisotropy{max_anisotropy > 1.0f};
 
-UniqueSampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc) {
-    const float max_anisotropy = tsc.GetMaxAnisotropy();
-    const bool has_anisotropy = max_anisotropy > 1.0f;
-
-    const auto border_color = tsc.GetBorderColor();
-    const auto vk_border_color = TryConvertBorderColor(border_color);
+    const auto border_color{tsc.GetBorderColor()};
+    const auto vk_border_color{TryConvertBorderColor(border_color)};
     UNIMPLEMENTED_IF_MSG(!vk_border_color, "Unimplemented border color {} {} {} {}",
                          border_color[0], border_color[1], border_color[2], border_color[3]);
 
-    constexpr bool unnormalized_coords = false;
+    constexpr bool unnormalized_coords{false};
 
     const vk::SamplerCreateInfo sampler_ci(
         {}, MaxwellToVK::Sampler::Filter(tsc.mag_filter),
@@ -73,9 +53,13 @@ UniqueSampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc)
         tsc.GetMaxLod(), vk_border_color.value_or(vk::BorderColor::eFloatTransparentBlack),
         unnormalized_coords);
 
-    const auto& dld = device.GetDispatchLoader();
-    const auto dev = device.GetLogical();
+    const auto& dld{device.GetDispatchLoader()};
+    const auto dev{device.GetLogical()};
     return dev.createSamplerUnique(sampler_ci, nullptr, dld);
 }
 
+vk::Sampler VKSamplerCache::ToSamplerType(const UniqueSampler& sampler) const {
+    return *sampler;
+}
+
 } // namespace Vulkan

src/video_core/renderer_vulkan/vk_sampler_cache.h

@@ -8,49 +8,25 @@
 
 #include "common/common_types.h"
 #include "video_core/renderer_vulkan/declarations.h"
+#include "video_core/sampler_cache.h"
 #include "video_core/textures/texture.h"
 
 namespace Vulkan {
 
 class VKDevice;
 
-struct SamplerCacheKey final : public Tegra::Texture::TSCEntry {
-    std::size_t Hash() const;
-
-    bool operator==(const SamplerCacheKey& rhs) const;
-
-    bool operator!=(const SamplerCacheKey& rhs) const {
-        return !operator==(rhs);
-    }
-};
-
-} // namespace Vulkan
-
-namespace std {
-
-template <>
-struct hash<Vulkan::SamplerCacheKey> {
-    std::size_t operator()(const Vulkan::SamplerCacheKey& k) const noexcept {
-        return k.Hash();
-    }
-};
-
-} // namespace std
-
-namespace Vulkan {
-
-class VKSamplerCache {
+class VKSamplerCache final : public VideoCommon::SamplerCache<vk::Sampler, UniqueSampler> {
 public:
     explicit VKSamplerCache(const VKDevice& device);
     ~VKSamplerCache();
 
-    vk::Sampler GetSampler(const Tegra::Texture::TSCEntry& tsc);
+protected:
+    UniqueSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc) const;
+
+    vk::Sampler ToSamplerType(const UniqueSampler& sampler) const;
 
 private:
-    UniqueSampler CreateSampler(const Tegra::Texture::TSCEntry& tsc);
-
     const VKDevice& device;
-    std::unordered_map<SamplerCacheKey, UniqueSampler> cache;
 };
 
 } // namespace Vulkan

src/video_core/sampler_cache.cpp

@@ -0,0 +1,21 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/cityhash.h"
+#include "common/common_types.h"
+#include "video_core/sampler_cache.h"
+
+namespace VideoCommon {
+
+std::size_t SamplerCacheKey::Hash() const {
+    static_assert(sizeof(raw) % sizeof(u64) == 0);
+    return static_cast<std::size_t>(
+        Common::CityHash64(reinterpret_cast<const char*>(raw.data()), sizeof(raw) / sizeof(u64)));
+}
+
+bool SamplerCacheKey::operator==(const SamplerCacheKey& rhs) const {
+    return raw == rhs.raw;
+}
+
+} // namespace VideoCommon

src/video_core/sampler_cache.h

@@ -0,0 +1,60 @@
+// Copyright 2019 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <cstddef>
+#include <unordered_map>
+
+#include "video_core/textures/texture.h"
+
+namespace VideoCommon {
+
+struct SamplerCacheKey final : public Tegra::Texture::TSCEntry {
+    std::size_t Hash() const;
+
+    bool operator==(const SamplerCacheKey& rhs) const;
+
+    bool operator!=(const SamplerCacheKey& rhs) const {
+        return !operator==(rhs);
+    }
+};
+
+} // namespace VideoCommon
+
+namespace std {
+
+template <>
+struct hash<VideoCommon::SamplerCacheKey> {
+    std::size_t operator()(const VideoCommon::SamplerCacheKey& k) const noexcept {
+        return k.Hash();
+    }
+};
+
+} // namespace std
+
+namespace VideoCommon {
+
+template <typename SamplerType, typename SamplerStorageType>
+class SamplerCache {
+public:
+    SamplerType GetSampler(const Tegra::Texture::TSCEntry& tsc) {
+        const auto [entry, is_cache_miss] = cache.try_emplace(SamplerCacheKey{tsc});
+        auto& sampler = entry->second;
+        if (is_cache_miss) {
+            sampler = CreateSampler(tsc);
+        }
+        return ToSamplerType(sampler);
+    }
+
+protected:
+    virtual SamplerStorageType CreateSampler(const Tegra::Texture::TSCEntry& tsc) const = 0;
+
+    virtual SamplerType ToSamplerType(const SamplerStorageType& sampler) const = 0;
+
+private:
+    std::unordered_map<SamplerCacheKey, SamplerStorageType> cache;
+};
+
+} // namespace VideoCommon

src/video_core/shader/decode/texture.cpp

@@ -40,7 +40,7 @@ static std::size_t GetCoordCount(TextureType texture_type) {
 u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
     const Instruction instr = {program_code[pc]};
     const auto opcode = OpCode::Decode(instr);
-
+    bool is_bindless = false;
     switch (opcode->get().GetId()) {
     case OpCode::Id::TEX: {
         if (instr.tex.UsesMiscMode(TextureMiscMode::NODEP)) {
@@ -54,7 +54,25 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         const auto process_mode = instr.tex.GetTextureProcessMode();
         WriteTexInstructionFloat(
             bb, instr,
-            GetTexCode(instr, texture_type, process_mode, depth_compare, is_array, is_aoffi));
+            GetTexCode(instr, texture_type, process_mode, depth_compare, is_array, is_aoffi, {}));
+        break;
+    }
+    case OpCode::Id::TEX_B: {
+        UNIMPLEMENTED_IF_MSG(instr.tex.UsesMiscMode(TextureMiscMode::AOFFI),
+                             "AOFFI is not implemented");
+
+        if (instr.tex.UsesMiscMode(TextureMiscMode::NODEP)) {
+            LOG_WARNING(HW_GPU, "TEX.NODEP implementation is incomplete");
+        }
+
+        const TextureType texture_type{instr.tex_b.texture_type};
+        const bool is_array = instr.tex_b.array != 0;
+        const bool is_aoffi = instr.tex.UsesMiscMode(TextureMiscMode::AOFFI);
+        const bool depth_compare = instr.tex_b.UsesMiscMode(TextureMiscMode::DC);
+        const auto process_mode = instr.tex_b.GetTextureProcessMode();
+        WriteTexInstructionFloat(bb, instr,
+                                 GetTexCode(instr, texture_type, process_mode, depth_compare,
+                                            is_array, is_aoffi, {instr.gpr20}));
         break;
     }
     case OpCode::Id::TEXS: {
@@ -134,6 +152,9 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         WriteTexsInstructionFloat(bb, instr, values);
         break;
     }
+    case OpCode::Id::TXQ_B:
+        is_bindless = true;
+        [[fallthrough]];
     case OpCode::Id::TXQ: {
         if (instr.txq.UsesMiscMode(TextureMiscMode::NODEP)) {
             LOG_WARNING(HW_GPU, "TXQ.NODEP implementation is incomplete");
@@ -143,7 +164,10 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         // Sadly, not all texture instructions specify the type of texture their sampler
         // uses. This must be fixed at a later instance.
         const auto& sampler =
-            GetSampler(instr.sampler, Tegra::Shader::TextureType::Texture2D, false, false);
+            is_bindless
+                ? GetBindlessSampler(instr.gpr8, Tegra::Shader::TextureType::Texture2D, false,
+                                     false)
+                : GetSampler(instr.sampler, Tegra::Shader::TextureType::Texture2D, false, false);
 
         u32 indexer = 0;
         switch (instr.txq.query_type) {
@@ -154,7 +178,8 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
                 }
                 MetaTexture meta{sampler, {}, {}, {}, {}, {}, {}, element};
                 const Node value =
-                    Operation(OperationCode::TextureQueryDimensions, meta, GetRegister(instr.gpr8));
+                    Operation(OperationCode::TextureQueryDimensions, meta,
+                              GetRegister(instr.gpr8.Value() + (is_bindless ? 1 : 0)));
                 SetTemporal(bb, indexer++, value);
             }
             for (u32 i = 0; i < indexer; ++i) {
@@ -168,6 +193,9 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
         }
         break;
     }
+    case OpCode::Id::TMML_B:
+        is_bindless = true;
+        [[fallthrough]];
     case OpCode::Id::TMML: {
         UNIMPLEMENTED_IF_MSG(instr.tmml.UsesMiscMode(Tegra::Shader::TextureMiscMode::NDV),
                              "NDV is not implemented");
@@ -178,7 +206,9 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
 
         auto texture_type = instr.tmml.texture_type.Value();
         const bool is_array = instr.tmml.array != 0;
-        const auto& sampler = GetSampler(instr.sampler, texture_type, is_array, false);
+        const auto& sampler = is_bindless
+                                  ? GetBindlessSampler(instr.gpr20, texture_type, is_array, false)
+                                  : GetSampler(instr.sampler, texture_type, is_array, false);
 
         std::vector<Node> coords;
 
@@ -199,17 +229,19 @@ u32 ShaderIR::DecodeTexture(NodeBlock& bb, u32 pc) {
             coords.push_back(GetRegister(instr.gpr8.Value() + 1));
             texture_type = TextureType::Texture2D;
         }
-
+        u32 indexer = 0;
         for (u32 element = 0; element < 2; ++element) {
+            if (!instr.tmml.IsComponentEnabled(element)) {
+                continue;
+            }
             auto params = coords;
             MetaTexture meta{sampler, {}, {}, {}, {}, {}, {}, element};
             const Node value = Operation(OperationCode::TextureQueryLod, meta, std::move(params));
-            SetTemporal(bb, element, value);
+            SetTemporal(bb, indexer++, value);
         }
-        for (u32 element = 0; element < 2; ++element) {
-            SetRegister(bb, instr.gpr0.Value() + element, GetTemporal(element));
+        for (u32 i = 0; i < indexer; ++i) {
+            SetRegister(bb, instr.gpr0.Value() + i, GetTemporal(i));
         }
-
         break;
     }
     case OpCode::Id::TLDS: {
@@ -254,6 +286,34 @@ const Sampler& ShaderIR::GetSampler(const Tegra::Shader::Sampler& sampler, Textu
     return *used_samplers.emplace(entry).first;
 }
 
+const Sampler& ShaderIR::GetBindlessSampler(const Tegra::Shader::Register& reg, TextureType type,
+                                            bool is_array, bool is_shadow) {
+    const Node sampler_register = GetRegister(reg);
+    const Node base_sampler =
+        TrackCbuf(sampler_register, global_code, static_cast<s64>(global_code.size()));
+    const auto cbuf = std::get_if<CbufNode>(base_sampler);
+    const auto cbuf_offset_imm = std::get_if<ImmediateNode>(cbuf->GetOffset());
+    ASSERT(cbuf_offset_imm != nullptr);
+    const auto cbuf_offset = cbuf_offset_imm->GetValue();
+    const auto cbuf_index = cbuf->GetIndex();
+    const u64 cbuf_key = (cbuf_index << 32) | cbuf_offset;
+
+    // If this sampler has already been used, return the existing mapping.
+    const auto itr =
+        std::find_if(used_samplers.begin(), used_samplers.end(),
+                     [&](const Sampler& entry) { return entry.GetOffset() == cbuf_key; });
+    if (itr != used_samplers.end()) {
+        ASSERT(itr->GetType() == type && itr->IsArray() == is_array &&
+               itr->IsShadow() == is_shadow);
+        return *itr;
+    }
+
+    // Otherwise create a new mapping for this sampler
+    const std::size_t next_index = used_samplers.size();
+    const Sampler entry{cbuf_index, cbuf_offset, next_index, type, is_array, is_shadow};
+    return *used_samplers.emplace(entry).first;
+}
+
 void ShaderIR::WriteTexInstructionFloat(NodeBlock& bb, Instruction instr, const Node4& components) {
     u32 dest_elem = 0;
     for (u32 elem = 0; elem < 4; ++elem) {
@@ -326,22 +386,27 @@ void ShaderIR::WriteTexsInstructionHalfFloat(NodeBlock& bb, Instruction instr,
 Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
                                TextureProcessMode process_mode, std::vector<Node> coords,
                                Node array, Node depth_compare, u32 bias_offset,
-                               std::vector<Node> aoffi) {
+                               std::vector<Node> aoffi,
+                               std::optional<Tegra::Shader::Register> bindless_reg) {
     const bool is_array = array;
     const bool is_shadow = depth_compare;
+    const bool is_bindless = bindless_reg.has_value();
 
     UNIMPLEMENTED_IF_MSG((texture_type == TextureType::Texture3D && (is_array || is_shadow)) ||
                              (texture_type == TextureType::TextureCube && is_array && is_shadow),
                          "This method is not supported.");
 
-    const auto& sampler = GetSampler(instr.sampler, texture_type, is_array, is_shadow);
+    const auto& sampler = is_bindless
+                              ? GetBindlessSampler(*bindless_reg, texture_type, is_array, is_shadow)
+                              : GetSampler(instr.sampler, texture_type, is_array, is_shadow);
 
     const bool lod_needed = process_mode == TextureProcessMode::LZ ||
                             process_mode == TextureProcessMode::LL ||
                             process_mode == TextureProcessMode::LLA;
 
-    // LOD selection (either via bias or explicit textureLod) not supported in GL for
-    // sampler2DArrayShadow and samplerCubeArrayShadow.
+    // LOD selection (either via bias or explicit textureLod) not
+    // supported in GL for sampler2DArrayShadow and
+    // samplerCubeArrayShadow.
     const bool gl_lod_supported =
         !((texture_type == Tegra::Shader::TextureType::Texture2D && is_array && is_shadow) ||
           (texture_type == Tegra::Shader::TextureType::TextureCube && is_array && is_shadow));
@@ -359,8 +424,9 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
             lod = Immediate(0.0f);
             break;
         case TextureProcessMode::LB:
-            // If present, lod or bias are always stored in the register indexed by the gpr20
-            // field with an offset depending on the usage of the other registers
+            // If present, lod or bias are always stored in the register
+            // indexed by the gpr20 field with an offset depending on the
+            // usage of the other registers
             bias = GetRegister(instr.gpr20.Value() + bias_offset);
             break;
         case TextureProcessMode::LL:
@@ -384,11 +450,18 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
 
 Node4 ShaderIR::GetTexCode(Instruction instr, TextureType texture_type,
                            TextureProcessMode process_mode, bool depth_compare, bool is_array,
-                           bool is_aoffi) {
+                           bool is_aoffi, std::optional<Tegra::Shader::Register> bindless_reg) {
     const bool lod_bias_enabled{
         (process_mode != TextureProcessMode::None && process_mode != TextureProcessMode::LZ)};
 
+    const bool is_bindless = bindless_reg.has_value();
+
     u64 parameter_register = instr.gpr20.Value();
+    if (is_bindless) {
+        ++parameter_register;
+    }
+
+    const u32 bias_lod_offset = (is_bindless ? 1 : 0);
     if (lod_bias_enabled) {
         ++parameter_register;
     }
@@ -423,7 +496,8 @@ Node4 ShaderIR::GetTexCode(Instruction instr, TextureType texture_type,
         dc = GetRegister(parameter_register++);
     }
 
-    return GetTextureCode(instr, texture_type, process_mode, coords, array, dc, 0, aoffi);
+    return GetTextureCode(instr, texture_type, process_mode, coords, array, dc, bias_lod_offset,
+                          aoffi, bindless_reg);
 }
 
 Node4 ShaderIR::GetTexsCode(Instruction instr, TextureType texture_type,
@@ -459,7 +533,8 @@ Node4 ShaderIR::GetTexsCode(Instruction instr, TextureType texture_type,
         dc = GetRegister(depth_register);
     }
 
-    return GetTextureCode(instr, texture_type, process_mode, coords, array, dc, bias_offset, {});
+    return GetTextureCode(instr, texture_type, process_mode, coords, array, dc, bias_offset, {},
+                          {});
 }
 
 Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool depth_compare,

src/video_core/shader/shader_ir.h

@@ -196,9 +196,23 @@ enum class ExitMethod {
 
 class Sampler {
 public:
+    // Use this constructor for bounded Samplers
     explicit Sampler(std::size_t offset, std::size_t index, Tegra::Shader::TextureType type,
                      bool is_array, bool is_shadow)
-        : offset{offset}, index{index}, type{type}, is_array{is_array}, is_shadow{is_shadow} {}
+        : offset{offset}, index{index}, type{type}, is_array{is_array}, is_shadow{is_shadow},
+          is_bindless{false} {}
+
+    // Use this constructor for bindless Samplers
+    explicit Sampler(u32 cbuf_index, u32 cbuf_offset, std::size_t index,
+                     Tegra::Shader::TextureType type, bool is_array, bool is_shadow)
+        : offset{(static_cast<u64>(cbuf_index) << 32) | cbuf_offset}, index{index}, type{type},
+          is_array{is_array}, is_shadow{is_shadow}, is_bindless{true} {}
+
+    // Use this only for serialization/deserialization
+    explicit Sampler(std::size_t offset, std::size_t index, Tegra::Shader::TextureType type,
+                     bool is_array, bool is_shadow, bool is_bindless)
+        : offset{offset}, index{index}, type{type}, is_array{is_array}, is_shadow{is_shadow},
+          is_bindless{is_bindless} {}
 
     std::size_t GetOffset() const {
         return offset;
@@ -220,6 +234,14 @@ public:
         return is_shadow;
     }
 
+    bool IsBindless() const {
+        return is_bindless;
+    }
+
+    std::pair<u32, u32> GetBindlessCBuf() const {
+        return {static_cast<u32>(offset >> 32), static_cast<u32>(offset)};
+    }
+
     bool operator<(const Sampler& rhs) const {
         return std::tie(offset, index, type, is_array, is_shadow) <
                std::tie(rhs.offset, rhs.index, rhs.type, rhs.is_array, rhs.is_shadow);
@@ -231,8 +253,9 @@ private:
     std::size_t offset{};
     std::size_t index{}; ///< Value used to index into the generated GLSL sampler array.
     Tegra::Shader::TextureType type{}; ///< The type used to sample this texture (Texture2D, etc)
-    bool is_array{};  ///< Whether the texture is being sampled as an array texture or not.
-    bool is_shadow{}; ///< Whether the texture is being sampled as a depth texture or not.
+    bool is_array{};    ///< Whether the texture is being sampled as an array texture or not.
+    bool is_shadow{};   ///< Whether the texture is being sampled as a depth texture or not.
+    bool is_bindless{}; ///< Whether this sampler belongs to a bindless texture or not.
 };
 
 class ConstBuffer {
@@ -735,6 +758,11 @@ private:
     const Sampler& GetSampler(const Tegra::Shader::Sampler& sampler,
                               Tegra::Shader::TextureType type, bool is_array, bool is_shadow);
 
+    // Accesses a texture sampler for a bindless texture.
+    const Sampler& GetBindlessSampler(const Tegra::Shader::Register& reg,
+                                      Tegra::Shader::TextureType type, bool is_array,
+                                      bool is_shadow);
+
     /// Extracts a sequence of bits from a node
     Node BitfieldExtract(Node value, u32 offset, u32 bits);
 
@@ -748,7 +776,8 @@ private:
 
     Node4 GetTexCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
                      Tegra::Shader::TextureProcessMode process_mode, bool depth_compare,
-                     bool is_array, bool is_aoffi);
+                     bool is_array, bool is_aoffi,
+                     std::optional<Tegra::Shader::Register> bindless_reg);
 
     Node4 GetTexsCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
                       Tegra::Shader::TextureProcessMode process_mode, bool depth_compare,
@@ -768,7 +797,8 @@ private:
 
     Node4 GetTextureCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
                          Tegra::Shader::TextureProcessMode process_mode, std::vector<Node> coords,
-                         Node array, Node depth_compare, u32 bias_offset, std::vector<Node> aoffi);
+                         Node array, Node depth_compare, u32 bias_offset, std::vector<Node> aoffi,
+                         std::optional<Tegra::Shader::Register> bindless_reg);
 
     Node GetVideoOperand(Node op, bool is_chunk, bool is_signed, Tegra::Shader::VideoType type,
                          u64 byte_height);

src/video_core/video_core.cpp

@@ -5,6 +5,8 @@
 #include <memory>
 #include "core/core.h"
 #include "core/settings.h"
+#include "video_core/gpu_asynch.h"
+#include "video_core/gpu_synch.h"
 #include "video_core/renderer_base.h"
 #include "video_core/renderer_opengl/renderer_opengl.h"
 #include "video_core/video_core.h"
@@ -16,6 +18,14 @@ std::unique_ptr<RendererBase> CreateRenderer(Core::Frontend::EmuWindow& emu_wind
     return std::make_unique<OpenGL::RendererOpenGL>(emu_window, system);
 }
 
+std::unique_ptr<Tegra::GPU> CreateGPU(Core::System& system) {
+    if (Settings::values.use_asynchronous_gpu_emulation) {
+        return std::make_unique<VideoCommon::GPUAsynch>(system, system.Renderer());
+    }
+
+    return std::make_unique<VideoCommon::GPUSynch>(system, system.Renderer());
+}
+
 u16 GetResolutionScaleFactor(const RendererBase& renderer) {
     return static_cast<u16>(
         Settings::values.resolution_factor

src/video_core/video_core.h

@@ -14,6 +14,10 @@ namespace Core::Frontend {
 class EmuWindow;
 }
 
+namespace Tegra {
+class GPU;
+}
+
 namespace VideoCore {
 
 class RendererBase;
@@ -27,6 +31,9 @@ class RendererBase;
 std::unique_ptr<RendererBase> CreateRenderer(Core::Frontend::EmuWindow& emu_window,
                                              Core::System& system);
 
+/// Creates an emulated GPU instance using the given system context.
+std::unique_ptr<Tegra::GPU> CreateGPU(Core::System& system);
+
 u16 GetResolutionScaleFactor(const RendererBase& renderer);
 
 } // namespace VideoCore

src/yuzu/bootmanager.cpp

@@ -377,7 +377,11 @@ void GRenderWindow::InitRenderTarget() {
     // WA_DontShowOnScreen, WA_DeleteOnClose
     QSurfaceFormat fmt;
     fmt.setVersion(4, 3);
-    fmt.setProfile(QSurfaceFormat::CoreProfile);
+    if (Settings::values.use_compatibility_profile) {
+        fmt.setProfile(QSurfaceFormat::CompatibilityProfile);
+    } else {
+        fmt.setProfile(QSurfaceFormat::CoreProfile);
+    }
     // TODO: expose a setting for buffer value (ie default/single/double/triple)
     fmt.setSwapBehavior(QSurfaceFormat::DefaultSwapBehavior);
     shared_context = std::make_unique<QOpenGLContext>();

src/yuzu/configuration/config.cpp

@@ -389,6 +389,8 @@ void Config::ReadValues() {
     Settings::values.resolution_factor = ReadSetting("resolution_factor", 1.0).toFloat();
     Settings::values.use_frame_limit = ReadSetting("use_frame_limit", true).toBool();
     Settings::values.frame_limit = ReadSetting("frame_limit", 100).toInt();
+    Settings::values.use_compatibility_profile =
+        ReadSetting("use_compatibility_profile", true).toBool();
     Settings::values.use_disk_shader_cache = ReadSetting("use_disk_shader_cache", true).toBool();
     Settings::values.use_accurate_gpu_emulation =
         ReadSetting("use_accurate_gpu_emulation", false).toBool();
@@ -661,6 +663,7 @@ void Config::SaveValues() {
     WriteSetting("resolution_factor", (double)Settings::values.resolution_factor, 1.0);
     WriteSetting("use_frame_limit", Settings::values.use_frame_limit, true);
     WriteSetting("frame_limit", Settings::values.frame_limit, 100);
+    WriteSetting("use_compatibility_profile", Settings::values.use_compatibility_profile, true);
     WriteSetting("use_disk_shader_cache", Settings::values.use_disk_shader_cache, true);
     WriteSetting("use_accurate_gpu_emulation", Settings::values.use_accurate_gpu_emulation, false);
     WriteSetting("use_asynchronous_gpu_emulation", Settings::values.use_asynchronous_gpu_emulation,

src/yuzu/configuration/configure_graphics.cpp

@@ -73,6 +73,7 @@ void ConfigureGraphics::setConfiguration() {
         static_cast<int>(FromResolutionFactor(Settings::values.resolution_factor)));
     ui->toggle_frame_limit->setChecked(Settings::values.use_frame_limit);
     ui->frame_limit->setValue(Settings::values.frame_limit);
+    ui->use_compatibility_profile->setChecked(Settings::values.use_compatibility_profile);
     ui->use_disk_shader_cache->setChecked(Settings::values.use_disk_shader_cache);
     ui->use_accurate_gpu_emulation->setChecked(Settings::values.use_accurate_gpu_emulation);
     ui->use_asynchronous_gpu_emulation->setEnabled(!Core::System::GetInstance().IsPoweredOn());
@@ -88,6 +89,7 @@ void ConfigureGraphics::applyConfiguration() {
         ToResolutionFactor(static_cast<Resolution>(ui->resolution_factor_combobox->currentIndex()));
     Settings::values.use_frame_limit = ui->toggle_frame_limit->isChecked();
     Settings::values.frame_limit = ui->frame_limit->value();
+    Settings::values.use_compatibility_profile = ui->use_compatibility_profile->isChecked();
     Settings::values.use_disk_shader_cache = ui->use_disk_shader_cache->isChecked();
     Settings::values.use_accurate_gpu_emulation = ui->use_accurate_gpu_emulation->isChecked();
     Settings::values.use_asynchronous_gpu_emulation =

src/yuzu/configuration/configure_graphics.ui

@@ -49,6 +49,13 @@
           </item>
          </layout>
         </item>
+        <item>
+         <widget class="QCheckBox" name="use_compatibility_profile">
+          <property name="text">
+           <string>Use OpenGL compatibility profile</string>
+          </property>
+         </widget>
+        </item>
         <item>
          <widget class="QCheckBox" name="use_disk_shader_cache">
           <property name="text">

src/yuzu_cmd/config.cpp

@@ -349,6 +349,8 @@ void Config::ReadValues() {
     Settings::values.use_frame_limit = sdl2_config->GetBoolean("Renderer", "use_frame_limit", true);
     Settings::values.frame_limit =
         static_cast<u16>(sdl2_config->GetInteger("Renderer", "frame_limit", 100));
+    Settings::values.use_compatibility_profile =
+        sdl2_config->GetBoolean("Renderer", "use_compatibility_profile", true);
     Settings::values.use_disk_shader_cache =
         sdl2_config->GetBoolean("Renderer", "use_disk_shader_cache", false);
     Settings::values.use_accurate_gpu_emulation =