maxwell_3d: Reduce FlushMMEInlineDraw logging to Trace

core: Fix clang-format errors.

src/core/core.cpp

@@ -112,8 +112,8 @@ FileSys::VirtualFile GetGameFileFromPath(const FileSys::VirtualFilesystem& vfs,
 }
 struct System::Impl {
     explicit Impl(System& system)
-        : kernel{system}, fs_controller{system}, cpu_core_manager{system},
-          applet_manager{system}, reporter{system} {}
+        : kernel{system}, fs_controller{system}, cpu_core_manager{system}, reporter{system},
+          applet_manager{system} {}
 
     Cpu& CurrentCpuCore() {
         return cpu_core_manager.GetCurrentCore();
@@ -240,22 +240,27 @@ struct System::Impl {
     }
 
     void Shutdown() {
-        // Log last frame performance stats
-        const auto perf_results = GetAndResetPerfStats();
-        telemetry_session->AddField(Telemetry::FieldType::Performance, "Shutdown_EmulationSpeed",
-                                    perf_results.emulation_speed * 100.0);
-        telemetry_session->AddField(Telemetry::FieldType::Performance, "Shutdown_Framerate",
-                                    perf_results.game_fps);
-        telemetry_session->AddField(Telemetry::FieldType::Performance, "Shutdown_Frametime",
-                                    perf_results.frametime * 1000.0);
-        telemetry_session->AddField(Telemetry::FieldType::Performance, "Mean_Frametime_MS",
-                                    perf_stats->GetMeanFrametime());
+        // Log last frame performance stats if game was loded
+        if (perf_stats) {
+            const auto perf_results = GetAndResetPerfStats();
+            telemetry_session->AddField(Telemetry::FieldType::Performance,
+                                        "Shutdown_EmulationSpeed",
+                                        perf_results.emulation_speed * 100.0);
+            telemetry_session->AddField(Telemetry::FieldType::Performance, "Shutdown_Framerate",
+                                        perf_results.game_fps);
+            telemetry_session->AddField(Telemetry::FieldType::Performance, "Shutdown_Frametime",
+                                        perf_results.frametime * 1000.0);
+            telemetry_session->AddField(Telemetry::FieldType::Performance, "Mean_Frametime_MS",
+                                        perf_stats->GetMeanFrametime());
+        }
 
         lm_manager.Flush();
 
         is_powered_on = false;
         exit_lock = false;
 
+        gpu_core->WaitIdle();
+
         // Shutdown emulation session
         renderer.reset();
         GDBStub::Shutdown();

src/core/hle/service/am/am.cpp

@@ -1140,8 +1140,9 @@ void IApplicationFunctions::PopLaunchParameter(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_AM, "called, kind={:08X}", static_cast<u8>(kind));
 
     if (kind == LaunchParameterKind::ApplicationSpecific && !launch_popped_application_specific) {
-        const auto backend = BCAT::CreateBackendFromSettings(
-            [this](u64 tid) { return system.GetFileSystemController().GetBCATDirectory(tid); });
+        const auto backend = BCAT::CreateBackendFromSettings(system, [this](u64 tid) {
+            return system.GetFileSystemController().GetBCATDirectory(tid);
+        });
         const auto build_id_full = system.GetCurrentProcessBuildID();
         u64 build_id{};
         std::memcpy(&build_id, build_id_full.data(), sizeof(u64));

src/core/hle/service/apm/controller.cpp

@@ -2,6 +2,10 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <algorithm>
+#include <array>
+#include <utility>
+
 #include "common/logging/log.h"
 #include "core/core_timing.h"
 #include "core/hle/service/apm/controller.h"
@@ -9,8 +13,7 @@
 
 namespace Service::APM {
 
-constexpr PerformanceConfiguration DEFAULT_PERFORMANCE_CONFIGURATION =
-    PerformanceConfiguration::Config7;
+constexpr auto DEFAULT_PERFORMANCE_CONFIGURATION = PerformanceConfiguration::Config7;
 
 Controller::Controller(Core::Timing::CoreTiming& core_timing)
     : core_timing{core_timing}, configs{
@@ -22,18 +25,35 @@ Controller::~Controller() = default;
 
 void Controller::SetPerformanceConfiguration(PerformanceMode mode,
                                              PerformanceConfiguration config) {
-    static const std::map<PerformanceConfiguration, u32> PCONFIG_TO_SPEED_MAP{
-        {PerformanceConfiguration::Config1, 1020},  {PerformanceConfiguration::Config2, 1020},
-        {PerformanceConfiguration::Config3, 1224},  {PerformanceConfiguration::Config4, 1020},
-        {PerformanceConfiguration::Config5, 1020},  {PerformanceConfiguration::Config6, 1224},
-        {PerformanceConfiguration::Config7, 1020},  {PerformanceConfiguration::Config8, 1020},
-        {PerformanceConfiguration::Config9, 1020},  {PerformanceConfiguration::Config10, 1020},
-        {PerformanceConfiguration::Config11, 1020}, {PerformanceConfiguration::Config12, 1020},
-        {PerformanceConfiguration::Config13, 1785}, {PerformanceConfiguration::Config14, 1785},
-        {PerformanceConfiguration::Config15, 1020}, {PerformanceConfiguration::Config16, 1020},
-    };
+    static constexpr std::array<std::pair<PerformanceConfiguration, u32>, 16> config_to_speed{{
+        {PerformanceConfiguration::Config1, 1020},
+        {PerformanceConfiguration::Config2, 1020},
+        {PerformanceConfiguration::Config3, 1224},
+        {PerformanceConfiguration::Config4, 1020},
+        {PerformanceConfiguration::Config5, 1020},
+        {PerformanceConfiguration::Config6, 1224},
+        {PerformanceConfiguration::Config7, 1020},
+        {PerformanceConfiguration::Config8, 1020},
+        {PerformanceConfiguration::Config9, 1020},
+        {PerformanceConfiguration::Config10, 1020},
+        {PerformanceConfiguration::Config11, 1020},
+        {PerformanceConfiguration::Config12, 1020},
+        {PerformanceConfiguration::Config13, 1785},
+        {PerformanceConfiguration::Config14, 1785},
+        {PerformanceConfiguration::Config15, 1020},
+        {PerformanceConfiguration::Config16, 1020},
+    }};
 
-    SetClockSpeed(PCONFIG_TO_SPEED_MAP.find(config)->second);
+    const auto iter = std::find_if(config_to_speed.cbegin(), config_to_speed.cend(),
+                                   [config](const auto& entry) { return entry.first == config; });
+
+    if (iter == config_to_speed.cend()) {
+        LOG_ERROR(Service_APM, "Invalid performance configuration value provided: {}",
+                  static_cast<u32>(config));
+        return;
+    }
+
+    SetClockSpeed(iter->second);
     configs.insert_or_assign(mode, config);
 }
 
@@ -48,7 +68,7 @@ void Controller::SetFromCpuBoostMode(CpuBoostMode mode) {
                                 BOOST_MODE_TO_CONFIG_MAP.at(static_cast<u32>(mode)));
 }
 
-PerformanceMode Controller::GetCurrentPerformanceMode() {
+PerformanceMode Controller::GetCurrentPerformanceMode() const {
     return Settings::values.use_docked_mode ? PerformanceMode::Docked : PerformanceMode::Handheld;
 }
 

src/core/hle/service/apm/controller.h

@@ -56,7 +56,7 @@ public:
     void SetPerformanceConfiguration(PerformanceMode mode, PerformanceConfiguration config);
     void SetFromCpuBoostMode(CpuBoostMode mode);
 
-    PerformanceMode GetCurrentPerformanceMode();
+    PerformanceMode GetCurrentPerformanceMode() const;
     PerformanceConfiguration GetCurrentPerformanceConfiguration(PerformanceMode mode);
 
 private:

src/core/hle/service/bcat/backend/backend.cpp

@@ -10,8 +10,8 @@
 
 namespace Service::BCAT {
 
-ProgressServiceBackend::ProgressServiceBackend(std::string_view event_name) {
-    auto& kernel{Core::System::GetInstance().Kernel()};
+ProgressServiceBackend::ProgressServiceBackend(Kernel::KernelCore& kernel,
+                                               std::string_view event_name) {
     event = Kernel::WritableEvent::CreateEventPair(
         kernel, Kernel::ResetType::Automatic,
         std::string("ProgressServiceBackend:UpdateEvent:").append(event_name));

src/core/hle/service/bcat/backend/backend.h

@@ -15,6 +15,14 @@
 #include "core/hle/kernel/writable_event.h"
 #include "core/hle/result.h"
 
+namespace Core {
+class System;
+}
+
+namespace Kernel {
+class KernelCore;
+}
+
 namespace Service::BCAT {
 
 struct DeliveryCacheProgressImpl;
@@ -88,7 +96,7 @@ public:
     void FinishDownload(ResultCode result);
 
 private:
-    explicit ProgressServiceBackend(std::string_view event_name);
+    explicit ProgressServiceBackend(Kernel::KernelCore& kernel, std::string_view event_name);
 
     Kernel::SharedPtr<Kernel::ReadableEvent> GetEvent() const;
     DeliveryCacheProgressImpl& GetImpl();
@@ -145,6 +153,6 @@ public:
     std::optional<std::vector<u8>> GetLaunchParameter(TitleIDVersion title) override;
 };
 
-std::unique_ptr<Backend> CreateBackendFromSettings(DirectoryGetter getter);
+std::unique_ptr<Backend> CreateBackendFromSettings(Core::System& system, DirectoryGetter getter);
 
 } // namespace Service::BCAT

src/core/hle/service/bcat/backend/boxcat.cpp

@@ -104,14 +104,15 @@ std::string GetZIPFilePath(u64 title_id) {
 
 // If the error is something the user should know about (build ID mismatch, bad client version),
 // display an error.
-void HandleDownloadDisplayResult(DownloadResult res) {
+void HandleDownloadDisplayResult(const AM::Applets::AppletManager& applet_manager,
+                                 DownloadResult res) {
     if (res == DownloadResult::Success || res == DownloadResult::NoResponse ||
         res == DownloadResult::GeneralWebError || res == DownloadResult::GeneralFSError ||
         res == DownloadResult::NoMatchTitleId || res == DownloadResult::InvalidContentType) {
         return;
     }
 
-    const auto& frontend{Core::System::GetInstance().GetAppletManager().GetAppletFrontendSet()};
+    const auto& frontend{applet_manager.GetAppletFrontendSet()};
     frontend.error->ShowCustomErrorText(
         ResultCode(-1), "There was an error while attempting to use Boxcat.",
         DOWNLOAD_RESULT_LOG_MESSAGES[static_cast<std::size_t>(res)], [] {});
@@ -264,12 +265,13 @@ private:
     u64 build_id;
 };
 
-Boxcat::Boxcat(DirectoryGetter getter) : Backend(std::move(getter)) {}
+Boxcat::Boxcat(AM::Applets::AppletManager& applet_manager_, DirectoryGetter getter)
+    : Backend(std::move(getter)), applet_manager{applet_manager_} {}
 
 Boxcat::~Boxcat() = default;
 
-void SynchronizeInternal(DirectoryGetter dir_getter, TitleIDVersion title,
-                         ProgressServiceBackend& progress,
+void SynchronizeInternal(AM::Applets::AppletManager& applet_manager, DirectoryGetter dir_getter,
+                         TitleIDVersion title, ProgressServiceBackend& progress,
                          std::optional<std::string> dir_name = {}) {
     progress.SetNeedHLELock(true);
 
@@ -295,7 +297,7 @@ void SynchronizeInternal(DirectoryGetter dir_getter, TitleIDVersion title,
             FileUtil::Delete(zip_path);
         }
 
-        HandleDownloadDisplayResult(res);
+        HandleDownloadDisplayResult(applet_manager, res);
         progress.FinishDownload(ERROR_GENERAL_BCAT_FAILURE);
         return;
     }
@@ -364,17 +366,24 @@ void SynchronizeInternal(DirectoryGetter dir_getter, TitleIDVersion title,
 
 bool Boxcat::Synchronize(TitleIDVersion title, ProgressServiceBackend& progress) {
     is_syncing.exchange(true);
-    std::thread([this, title, &progress] { SynchronizeInternal(dir_getter, title, progress); })
+
+    std::thread([this, title, &progress] {
+        SynchronizeInternal(applet_manager, dir_getter, title, progress);
+    })
         .detach();
+
     return true;
 }
 
 bool Boxcat::SynchronizeDirectory(TitleIDVersion title, std::string name,
                                   ProgressServiceBackend& progress) {
     is_syncing.exchange(true);
-    std::thread(
-        [this, title, name, &progress] { SynchronizeInternal(dir_getter, title, progress, name); })
+
+    std::thread([this, title, name, &progress] {
+        SynchronizeInternal(applet_manager, dir_getter, title, progress, name);
+    })
         .detach();
+
     return true;
 }
 
@@ -420,7 +429,7 @@ std::optional<std::vector<u8>> Boxcat::GetLaunchParameter(TitleIDVersion title)
                 FileUtil::Delete(path);
             }
 
-            HandleDownloadDisplayResult(res);
+            HandleDownloadDisplayResult(applet_manager, res);
             return std::nullopt;
         }
     }

src/core/hle/service/bcat/backend/boxcat.h

@@ -9,6 +9,10 @@
 #include <optional>
 #include "core/hle/service/bcat/backend/backend.h"
 
+namespace Service::AM::Applets {
+class AppletManager;
+}
+
 namespace Service::BCAT {
 
 struct EventStatus {
@@ -20,12 +24,13 @@ struct EventStatus {
 /// Boxcat is yuzu's custom backend implementation of Nintendo's BCAT service. It is free to use and
 /// doesn't require a switch or nintendo account. The content is controlled by the yuzu team.
 class Boxcat final : public Backend {
-    friend void SynchronizeInternal(DirectoryGetter dir_getter, TitleIDVersion title,
+    friend void SynchronizeInternal(AM::Applets::AppletManager& applet_manager,
+                                    DirectoryGetter dir_getter, TitleIDVersion title,
                                     ProgressServiceBackend& progress,
                                     std::optional<std::string> dir_name);
 
 public:
-    explicit Boxcat(DirectoryGetter getter);
+    explicit Boxcat(AM::Applets::AppletManager& applet_manager_, DirectoryGetter getter);
     ~Boxcat() override;
 
     bool Synchronize(TitleIDVersion title, ProgressServiceBackend& progress) override;
@@ -53,6 +58,7 @@ private:
 
     class Client;
     std::unique_ptr<Client> client;
+    AM::Applets::AppletManager& applet_manager;
 };
 
 } // namespace Service::BCAT

src/core/hle/service/bcat/module.cpp

@@ -125,7 +125,11 @@ private:
 class IBcatService final : public ServiceFramework<IBcatService> {
 public:
     explicit IBcatService(Core::System& system_, Backend& backend_)
-        : ServiceFramework("IBcatService"), system{system_}, backend{backend_} {
+        : ServiceFramework("IBcatService"), system{system_}, backend{backend_},
+          progress{{
+              ProgressServiceBackend{system_.Kernel(), "Normal"},
+              ProgressServiceBackend{system_.Kernel(), "Directory"},
+          }} {
         // clang-format off
         static const FunctionInfo functions[] = {
             {10100, &IBcatService::RequestSyncDeliveryCache, "RequestSyncDeliveryCache"},
@@ -249,10 +253,7 @@ private:
     Core::System& system;
     Backend& backend;
 
-    std::array<ProgressServiceBackend, static_cast<std::size_t>(SyncType::Count)> progress{
-        ProgressServiceBackend{"Normal"},
-        ProgressServiceBackend{"Directory"},
-    };
+    std::array<ProgressServiceBackend, static_cast<std::size_t>(SyncType::Count)> progress;
 };
 
 void Module::Interface::CreateBcatService(Kernel::HLERequestContext& ctx) {
@@ -557,12 +558,12 @@ void Module::Interface::CreateDeliveryCacheStorageServiceWithApplicationId(
     rb.PushIpcInterface<IDeliveryCacheStorageService>(fsc.GetBCATDirectory(title_id));
 }
 
-std::unique_ptr<Backend> CreateBackendFromSettings(DirectoryGetter getter) {
-    const auto backend = Settings::values.bcat_backend;
-
+std::unique_ptr<Backend> CreateBackendFromSettings([[maybe_unused]] Core::System& system,
+                                                   DirectoryGetter getter) {
 #ifdef YUZU_ENABLE_BOXCAT
-    if (backend == "boxcat")
-        return std::make_unique<Boxcat>(std::move(getter));
+    if (Settings::values.bcat_backend == "boxcat") {
+        return std::make_unique<Boxcat>(system.GetAppletManager(), std::move(getter));
+    }
 #endif
 
     return std::make_unique<NullBackend>(std::move(getter));
@@ -571,7 +572,8 @@ std::unique_ptr<Backend> CreateBackendFromSettings(DirectoryGetter getter) {
 Module::Interface::Interface(Core::System& system_, std::shared_ptr<Module> module_,
                              FileSystem::FileSystemController& fsc_, const char* name)
     : ServiceFramework(name), fsc{fsc_}, module{std::move(module_)},
-      backend{CreateBackendFromSettings([&fsc_](u64 tid) { return fsc_.GetBCATDirectory(tid); })},
+      backend{CreateBackendFromSettings(system_,
+                                        [&fsc_](u64 tid) { return fsc_.GetBCATDirectory(tid); })},
       system{system_} {}
 
 Module::Interface::~Interface() = default;

src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp

@@ -5,6 +5,7 @@
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "core/core.h"
+#include "core/core_timing.h"
 #include "core/hle/service/nvdrv/devices/nvdisp_disp0.h"
 #include "core/hle/service/nvdrv/devices/nvmap.h"
 #include "core/perf_stats.h"
@@ -38,7 +39,10 @@ void nvdisp_disp0::flip(u32 buffer_handle, u32 offset, u32 format, u32 width, u3
         transform, crop_rect};
 
     system.GetPerfStats().EndGameFrame();
+    system.GetPerfStats().EndSystemFrame();
     system.GPU().SwapBuffers(&framebuffer);
+    system.FrameLimiter().DoFrameLimiting(system.CoreTiming().GetGlobalTimeUs());
+    system.GetPerfStats().BeginSystemFrame();
 }
 
 } // namespace Service::Nvidia::Devices

src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp

@@ -63,16 +63,26 @@ u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>&
         return NvResult::BadParameter;
     }
 
+    u32 event_id = params.value & 0x00FF;
+
+    if (event_id >= MaxNvEvents) {
+        std::memcpy(output.data(), &params, sizeof(params));
+        return NvResult::BadParameter;
+    }
+
+    auto event = events_interface.events[event_id];
     auto& gpu = system.GPU();
     // This is mostly to take into account unimplemented features. As synced
     // gpu is always synced.
     if (!gpu.IsAsync()) {
+        event.writable->Signal();
         return NvResult::Success;
     }
     auto lock = gpu.LockSync();
     const u32 current_syncpoint_value = gpu.GetSyncpointValue(params.syncpt_id);
     const s32 diff = current_syncpoint_value - params.threshold;
     if (diff >= 0) {
+        event.writable->Signal();
         params.value = current_syncpoint_value;
         std::memcpy(output.data(), &params, sizeof(params));
         return NvResult::Success;
@@ -88,27 +98,6 @@ u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>&
         return NvResult::Timeout;
     }
 
-    u32 event_id;
-    if (is_async) {
-        event_id = params.value & 0x00FF;
-        if (event_id >= MaxNvEvents) {
-            std::memcpy(output.data(), &params, sizeof(params));
-            return NvResult::BadParameter;
-        }
-    } else {
-        if (ctrl.fresh_call) {
-            const auto result = events_interface.GetFreeEvent();
-            if (result) {
-                event_id = *result;
-            } else {
-                LOG_CRITICAL(Service_NVDRV, "No Free Events available!");
-                event_id = params.value & 0x00FF;
-            }
-        } else {
-            event_id = ctrl.event_id;
-        }
-    }
-
     EventState status = events_interface.status[event_id];
     if (event_id < MaxNvEvents || status == EventState::Free || status == EventState::Registered) {
         events_interface.SetEventStatus(event_id, EventState::Waiting);
@@ -120,7 +109,7 @@ u32 nvhost_ctrl::IocCtrlEventWait(const std::vector<u8>& input, std::vector<u8>&
             params.value = ((params.syncpt_id & 0xfff) << 16) | 0x10000000;
         }
         params.value |= event_id;
-        events_interface.events[event_id].writable->Clear();
+        event.writable->Clear();
         gpu.RegisterSyncptInterrupt(params.syncpt_id, target_value);
         if (!is_async && ctrl.fresh_call) {
             ctrl.must_delay = true;

src/core/hle/service/nvdrv/interface.cpp

@@ -134,7 +134,9 @@ void NVDRV::QueryEvent(Kernel::HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 3, 1};
     rb.Push(RESULT_SUCCESS);
     if (event_id < MaxNvEvents) {
-        rb.PushCopyObjects(nvdrv->GetEvent(event_id));
+        auto event = nvdrv->GetEvent(event_id);
+        event->Clear();
+        rb.PushCopyObjects(event);
         rb.Push<u32>(NvResult::Success);
     } else {
         rb.Push<u32>(0);

src/core/hle/service/nvdrv/nvdrv.cpp

@@ -40,8 +40,8 @@ Module::Module(Core::System& system) {
     auto& kernel = system.Kernel();
     for (u32 i = 0; i < MaxNvEvents; i++) {
         std::string event_label = fmt::format("NVDRV::NvEvent_{}", i);
-        events_interface.events[i] = Kernel::WritableEvent::CreateEventPair(
-            kernel, Kernel::ResetType::Automatic, event_label);
+        events_interface.events[i] =
+            Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Manual, event_label);
         events_interface.status[i] = EventState::Free;
         events_interface.registered[i] = false;
     }

src/core/hle/service/nvflinger/buffer_queue.cpp

@@ -14,8 +14,8 @@
 
 namespace Service::NVFlinger {
 
-BufferQueue::BufferQueue(u32 id, u64 layer_id) : id(id), layer_id(layer_id) {
-    auto& kernel = Core::System::GetInstance().Kernel();
+BufferQueue::BufferQueue(Kernel::KernelCore& kernel, u32 id, u64 layer_id)
+    : id(id), layer_id(layer_id) {
     buffer_wait_event = Kernel::WritableEvent::CreateEventPair(kernel, Kernel::ResetType::Manual,
                                                                "BufferQueue NativeHandle");
 }

src/core/hle/service/nvflinger/buffer_queue.h

@@ -15,6 +15,10 @@
 #include "core/hle/kernel/writable_event.h"
 #include "core/hle/service/nvdrv/nvdata.h"
 
+namespace Kernel {
+class KernelCore;
+}
+
 namespace Service::NVFlinger {
 
 struct IGBPBuffer {
@@ -44,7 +48,7 @@ public:
         NativeWindowFormat = 2,
     };
 
-    BufferQueue(u32 id, u64 layer_id);
+    explicit BufferQueue(Kernel::KernelCore& kernel, u32 id, u64 layer_id);
     ~BufferQueue();
 
     enum class BufferTransformFlags : u32 {

src/core/hle/service/nvflinger/nvflinger.cpp

@@ -83,7 +83,7 @@ std::optional<u64> NVFlinger::CreateLayer(u64 display_id) {
 
     const u64 layer_id = next_layer_id++;
     const u32 buffer_queue_id = next_buffer_queue_id++;
-    buffer_queues.emplace_back(buffer_queue_id, layer_id);
+    buffer_queues.emplace_back(system.Kernel(), buffer_queue_id, layer_id);
     display->CreateLayer(layer_id, buffer_queues.back());
     return layer_id;
 }
@@ -187,14 +187,18 @@ void NVFlinger::Compose() {
         MicroProfileFlip();
 
         if (!buffer) {
-            // There was no queued buffer to draw, render previous frame
-            system.GetPerfStats().EndGameFrame();
-            system.GPU().SwapBuffers({});
             continue;
         }
 
         const auto& igbp_buffer = buffer->get().igbp_buffer;
 
+        const auto& gpu = system.GPU();
+        const auto& multi_fence = buffer->get().multi_fence;
+        for (u32 fence_id = 0; fence_id < multi_fence.num_fences; fence_id++) {
+            const auto& fence = multi_fence.fences[fence_id];
+            gpu.WaitFence(fence.id, fence.value);
+        }
+
         // Now send the buffer to the GPU for drawing.
         // TODO(Subv): Support more than just disp0. The display device selection is probably based
         // on which display we're drawing (Default, Internal, External, etc)

src/core/memory/cheat_engine.cpp

@@ -22,7 +22,7 @@ constexpr u32 KEYPAD_BITMASK = 0x3FFFFFF;
 
 StandardVmCallbacks::StandardVmCallbacks(const Core::System& system,
                                          const CheatProcessMetadata& metadata)
-    : system(system), metadata(metadata) {}
+    : metadata(metadata), system(system) {}
 
 StandardVmCallbacks::~StandardVmCallbacks() = default;
 
@@ -176,9 +176,8 @@ std::vector<CheatEntry> TextCheatParser::Parse(const Core::System& system,
 
 CheatEngine::CheatEngine(Core::System& system, std::vector<CheatEntry> cheats,
                          const std::array<u8, 0x20>& build_id)
-    : system{system}, core_timing{system.CoreTiming()}, vm{std::make_unique<StandardVmCallbacks>(
-                                                            system, metadata)},
-      cheats(std::move(cheats)) {
+    : vm{std::make_unique<StandardVmCallbacks>(system, metadata)},
+      cheats(std::move(cheats)), core_timing{system.CoreTiming()}, system{system} {
     metadata.main_nso_build_id = build_id;
 }
 

src/core/memory/dmnt_cheat_vm.cpp

@@ -1133,8 +1133,8 @@ void DmntCheatVm::Execute(const CheatProcessMetadata& metadata) {
             case SaveRestoreRegisterOpType::ClearRegs:
             case SaveRestoreRegisterOpType::Restore:
             default:
-                src = registers.data();
-                dst = saved_values.data();
+                src = saved_values.data();
+                dst = registers.data();
                 break;
             }
             for (std::size_t i = 0; i < NumRegisters; i++) {

src/video_core/engines/maxwell_3d.cpp

@@ -101,7 +101,8 @@ void Maxwell3D::InitializeRegisterDefaults() {
 #define DIRTY_REGS_POS(field_name) (offsetof(Maxwell3D::DirtyRegs, field_name))
 
 void Maxwell3D::InitDirtySettings() {
-    const auto set_block = [this](const u32 start, const u32 range, const u8 position) {
+    const auto set_block = [this](const std::size_t start, const std::size_t range,
+                                  const u8 position) {
         const auto start_itr = dirty_pointers.begin() + start;
         const auto end_itr = start_itr + range;
         std::fill(start_itr, end_itr, position);
@@ -478,7 +479,7 @@ void Maxwell3D::CallMethodFromMME(const GPU::MethodCall& method_call) {
 }
 
 void Maxwell3D::FlushMMEInlineDraw() {
-    LOG_DEBUG(HW_GPU, "called, topology={}, count={}", static_cast<u32>(regs.draw.topology.Value()),
+    LOG_TRACE(HW_GPU, "called, topology={}, count={}", static_cast<u32>(regs.draw.topology.Value()),
               regs.vertex_buffer.count);
     ASSERT_MSG(!(regs.index_array.count && regs.vertex_buffer.count), "Both indexed and direct?");
     ASSERT(mme_draw.instance_count == mme_draw.gl_end_count);

src/video_core/gpu.cpp

@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include "common/assert.h"
+#include "common/microprofile.h"
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/memory.h"
@@ -17,6 +18,8 @@
 
 namespace Tegra {
 
+MICROPROFILE_DEFINE(GPU_wait, "GPU", "Wait for the GPU", MP_RGB(128, 128, 192));
+
 GPU::GPU(Core::System& system, VideoCore::RendererBase& renderer, bool is_async)
     : system{system}, renderer{renderer}, is_async{is_async} {
     auto& rasterizer{renderer.Rasterizer()};
@@ -63,6 +66,16 @@ const DmaPusher& GPU::DmaPusher() const {
     return *dma_pusher;
 }
 
+void GPU::WaitFence(u32 syncpoint_id, u32 value) const {
+    // Synced GPU, is always in sync
+    if (!is_async) {
+        return;
+    }
+    MICROPROFILE_SCOPE(GPU_wait);
+    while (syncpoints[syncpoint_id].load(std::memory_order_relaxed) < value) {
+    }
+}
+
 void GPU::IncrementSyncPoint(const u32 syncpoint_id) {
     syncpoints[syncpoint_id]++;
     std::lock_guard lock{sync_mutex};
@@ -326,7 +339,7 @@ void GPU::ProcessSemaphoreTriggerMethod() {
         block.sequence = regs.semaphore_sequence;
         // TODO(Kmather73): Generate a real GPU timestamp and write it here instead of
         // CoreTiming
-        block.timestamp = Core::System::GetInstance().CoreTiming().GetTicks();
+        block.timestamp = system.CoreTiming().GetTicks();
         memory_manager->WriteBlock(regs.semaphore_address.SemaphoreAddress(), &block,
                                    sizeof(block));
     } else {

src/video_core/gpu.h

@@ -177,6 +177,12 @@ public:
     /// Returns a reference to the GPU DMA pusher.
     Tegra::DmaPusher& DmaPusher();
 
+    // Waits for the GPU to finish working
+    virtual void WaitIdle() const = 0;
+
+    /// Allows the CPU/NvFlinger to wait on the GPU before presenting a frame.
+    void WaitFence(u32 syncpoint_id, u32 value) const;
+
     void IncrementSyncPoint(u32 syncpoint_id);
 
     u32 GetSyncpointValue(u32 syncpoint_id) const;

src/video_core/gpu_asynch.cpp

@@ -44,4 +44,8 @@ void GPUAsynch::TriggerCpuInterrupt(const u32 syncpoint_id, const u32 value) con
     interrupt_manager.GPUInterruptSyncpt(syncpoint_id, value);
 }
 
+void GPUAsynch::WaitIdle() const {
+    gpu_thread.WaitIdle();
+}
+
 } // namespace VideoCommon

src/video_core/gpu_asynch.h

@@ -25,6 +25,7 @@ public:
     void FlushRegion(CacheAddr addr, u64 size) override;
     void InvalidateRegion(CacheAddr addr, u64 size) override;
     void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;
+    void WaitIdle() const override;
 
 protected:
     void TriggerCpuInterrupt(u32 syncpoint_id, u32 value) const override;

src/video_core/gpu_synch.h

@@ -24,6 +24,7 @@ public:
     void FlushRegion(CacheAddr addr, u64 size) override;
     void InvalidateRegion(CacheAddr addr, u64 size) override;
     void FlushAndInvalidateRegion(CacheAddr addr, u64 size) override;
+    void WaitIdle() const override {}
 
 protected:
     void TriggerCpuInterrupt([[maybe_unused]] u32 syncpoint_id,

src/video_core/gpu_thread.cpp

@@ -5,8 +5,6 @@
 #include "common/assert.h"
 #include "common/microprofile.h"
 #include "core/core.h"
-#include "core/core_timing.h"
-#include "core/core_timing_util.h"
 #include "core/frontend/scope_acquire_window_context.h"
 #include "video_core/dma_pusher.h"
 #include "video_core/gpu.h"
@@ -68,14 +66,10 @@ ThreadManager::~ThreadManager() {
 
 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(std::chrono::microseconds{9000})};
-    system.CoreTiming().ScheduleEvent(synchronization_ticks, synchronization_event, fence);
+    PushCommand(SubmitListCommand(std::move(entries)));
 }
 
 void ThreadManager::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
@@ -96,16 +90,15 @@ void ThreadManager::FlushAndInvalidateRegion(CacheAddr addr, u64 size) {
     InvalidateRegion(addr, size);
 }
 
+void ThreadManager::WaitIdle() const {
+    while (state.last_fence > state.signaled_fence.load(std::memory_order_relaxed)) {
+    }
+}
+
 u64 ThreadManager::PushCommand(CommandData&& command_data) {
     const u64 fence{++state.last_fence};
     state.queue.Push(CommandDataContainer(std::move(command_data), fence));
     return fence;
 }
 
-MICROPROFILE_DEFINE(GPU_wait, "GPU", "Wait for the GPU", MP_RGB(128, 128, 192));
-void SynchState::WaitForSynchronization(u64 fence) {
-    while (signaled_fence.load() < fence)
-        ;
-}
-
 } // namespace VideoCommon::GPUThread

src/video_core/gpu_thread.h

@@ -21,9 +21,6 @@ class DmaPusher;
 
 namespace Core {
 class System;
-namespace Timing {
-struct EventType;
-} // namespace Timing
 } // namespace Core
 
 namespace VideoCommon::GPUThread {
@@ -89,8 +86,6 @@ struct CommandDataContainer {
 struct SynchState final {
     std::atomic_bool is_running{true};
 
-    void WaitForSynchronization(u64 fence);
-
     using CommandQueue = Common::SPSCQueue<CommandDataContainer>;
     CommandQueue queue;
     u64 last_fence{};
@@ -121,6 +116,9 @@ public:
     /// Notify rasterizer that any caches of the specified region should be flushed and invalidated
     void FlushAndInvalidateRegion(CacheAddr addr, u64 size);
 
+    // Wait until the gpu thread is idle.
+    void WaitIdle() const;
+
 private:
     /// Pushes a command to be executed by the GPU thread
     u64 PushCommand(CommandData&& command_data);
@@ -128,7 +126,6 @@ private:
 private:
     SynchState state;
     Core::System& system;
-    Core::Timing::EventType* synchronization_event{};
     std::thread thread;
     std::thread::id thread_id;
 };

src/video_core/macro_interpreter.cpp

@@ -11,6 +11,77 @@
 MICROPROFILE_DEFINE(MacroInterp, "GPU", "Execute macro interpreter", MP_RGB(128, 128, 192));
 
 namespace Tegra {
+namespace {
+enum class Operation : u32 {
+    ALU = 0,
+    AddImmediate = 1,
+    ExtractInsert = 2,
+    ExtractShiftLeftImmediate = 3,
+    ExtractShiftLeftRegister = 4,
+    Read = 5,
+    Unused = 6, // This operation doesn't seem to be a valid encoding.
+    Branch = 7,
+};
+} // Anonymous namespace
+
+enum class MacroInterpreter::ALUOperation : u32 {
+    Add = 0,
+    AddWithCarry = 1,
+    Subtract = 2,
+    SubtractWithBorrow = 3,
+    // Operations 4-7 don't seem to be valid encodings.
+    Xor = 8,
+    Or = 9,
+    And = 10,
+    AndNot = 11,
+    Nand = 12
+};
+
+enum class MacroInterpreter::ResultOperation : u32 {
+    IgnoreAndFetch = 0,
+    Move = 1,
+    MoveAndSetMethod = 2,
+    FetchAndSend = 3,
+    MoveAndSend = 4,
+    FetchAndSetMethod = 5,
+    MoveAndSetMethodFetchAndSend = 6,
+    MoveAndSetMethodSend = 7
+};
+
+enum class MacroInterpreter::BranchCondition : u32 {
+    Zero = 0,
+    NotZero = 1,
+};
+
+union MacroInterpreter::Opcode {
+    u32 raw;
+    BitField<0, 3, Operation> operation;
+    BitField<4, 3, ResultOperation> result_operation;
+    BitField<4, 1, BranchCondition> branch_condition;
+    // If set on a branch, then the branch doesn't have a delay slot.
+    BitField<5, 1, u32> branch_annul;
+    BitField<7, 1, u32> is_exit;
+    BitField<8, 3, u32> dst;
+    BitField<11, 3, u32> src_a;
+    BitField<14, 3, u32> src_b;
+    // The signed immediate overlaps the second source operand and the alu operation.
+    BitField<14, 18, s32> immediate;
+
+    BitField<17, 5, ALUOperation> alu_operation;
+
+    // Bitfield instructions data
+    BitField<17, 5, u32> bf_src_bit;
+    BitField<22, 5, u32> bf_size;
+    BitField<27, 5, u32> bf_dst_bit;
+
+    u32 GetBitfieldMask() const {
+        return (1 << bf_size) - 1;
+    }
+
+    s32 GetBranchTarget() const {
+        return static_cast<s32>(immediate * sizeof(u32));
+    }
+};
 
 MacroInterpreter::MacroInterpreter(Engines::Maxwell3D& maxwell3d) : maxwell3d(maxwell3d) {}
 

src/video_core/macro_interpreter.h

@@ -6,7 +6,6 @@
 
 #include <array>
 #include <optional>
-#include <vector>
 
 #include "common/bit_field.h"
 #include "common/common_types.h"
@@ -28,75 +27,11 @@ public:
     void Execute(u32 offset, std::size_t num_parameters, const u32* parameters);
 
 private:
-    enum class Operation : u32 {
-        ALU = 0,
-        AddImmediate = 1,
-        ExtractInsert = 2,
-        ExtractShiftLeftImmediate = 3,
-        ExtractShiftLeftRegister = 4,
-        Read = 5,
-        Unused = 6, // This operation doesn't seem to be a valid encoding.
-        Branch = 7,
-    };
+    enum class ALUOperation : u32;
+    enum class BranchCondition : u32;
+    enum class ResultOperation : u32;
 
-    enum class ALUOperation : u32 {
-        Add = 0,
-        AddWithCarry = 1,
-        Subtract = 2,
-        SubtractWithBorrow = 3,
-        // Operations 4-7 don't seem to be valid encodings.
-        Xor = 8,
-        Or = 9,
-        And = 10,
-        AndNot = 11,
-        Nand = 12
-    };
-
-    enum class ResultOperation : u32 {
-        IgnoreAndFetch = 0,
-        Move = 1,
-        MoveAndSetMethod = 2,
-        FetchAndSend = 3,
-        MoveAndSend = 4,
-        FetchAndSetMethod = 5,
-        MoveAndSetMethodFetchAndSend = 6,
-        MoveAndSetMethodSend = 7
-    };
-
-    enum class BranchCondition : u32 {
-        Zero = 0,
-        NotZero = 1,
-    };
-
-    union Opcode {
-        u32 raw;
-        BitField<0, 3, Operation> operation;
-        BitField<4, 3, ResultOperation> result_operation;
-        BitField<4, 1, BranchCondition> branch_condition;
-        BitField<5, 1, u32>
-            branch_annul; // If set on a branch, then the branch doesn't have a delay slot.
-        BitField<7, 1, u32> is_exit;
-        BitField<8, 3, u32> dst;
-        BitField<11, 3, u32> src_a;
-        BitField<14, 3, u32> src_b;
-        // The signed immediate overlaps the second source operand and the alu operation.
-        BitField<14, 18, s32> immediate;
-
-        BitField<17, 5, ALUOperation> alu_operation;
-
-        // Bitfield instructions data
-        BitField<17, 5, u32> bf_src_bit;
-        BitField<22, 5, u32> bf_size;
-        BitField<27, 5, u32> bf_dst_bit;
-
-        u32 GetBitfieldMask() const {
-            return (1 << bf_size) - 1;
-        }
-
-        s32 GetBranchTarget() const {
-            return static_cast<s32>(immediate * sizeof(u32));
-        }
-    };
+    union Opcode;
 
     union MethodAddress {
         u32 raw;
@@ -149,9 +84,10 @@ private:
 
     Engines::Maxwell3D& maxwell3d;
 
-    u32 pc; ///< Current program counter
-    std::optional<u32>
-        delayed_pc; ///< Program counter to execute at after the delay slot is executed.
+    /// Current program counter
+    u32 pc;
+    /// Program counter to execute at after the delay slot is executed.
+    std::optional<u32> delayed_pc;
 
     static constexpr std::size_t NumMacroRegisters = 8;
 

src/video_core/morton.cpp

@@ -93,6 +93,7 @@ static constexpr ConversionArray morton_to_linear_fns = {
     MortonCopy<true, PixelFormat::DXT23_SRGB>,
     MortonCopy<true, PixelFormat::DXT45_SRGB>,
     MortonCopy<true, PixelFormat::BC7U_SRGB>,
+    MortonCopy<true, PixelFormat::R4G4B4A4U>,
     MortonCopy<true, PixelFormat::ASTC_2D_4X4_SRGB>,
     MortonCopy<true, PixelFormat::ASTC_2D_8X8_SRGB>,
     MortonCopy<true, PixelFormat::ASTC_2D_8X5_SRGB>,
@@ -101,6 +102,16 @@ static constexpr ConversionArray morton_to_linear_fns = {
     MortonCopy<true, PixelFormat::ASTC_2D_5X5_SRGB>,
     MortonCopy<true, PixelFormat::ASTC_2D_10X8>,
     MortonCopy<true, PixelFormat::ASTC_2D_10X8_SRGB>,
+    MortonCopy<true, PixelFormat::ASTC_2D_6X6>,
+    MortonCopy<true, PixelFormat::ASTC_2D_6X6_SRGB>,
+    MortonCopy<true, PixelFormat::ASTC_2D_10X10>,
+    MortonCopy<true, PixelFormat::ASTC_2D_10X10_SRGB>,
+    MortonCopy<true, PixelFormat::ASTC_2D_12X12>,
+    MortonCopy<true, PixelFormat::ASTC_2D_12X12_SRGB>,
+    MortonCopy<true, PixelFormat::ASTC_2D_8X6>,
+    MortonCopy<true, PixelFormat::ASTC_2D_8X6_SRGB>,
+    MortonCopy<true, PixelFormat::ASTC_2D_6X5>,
+    MortonCopy<true, PixelFormat::ASTC_2D_6X5_SRGB>,
     MortonCopy<true, PixelFormat::Z32F>,
     MortonCopy<true, PixelFormat::Z16>,
     MortonCopy<true, PixelFormat::Z24S8>,
@@ -162,6 +173,17 @@ static constexpr ConversionArray linear_to_morton_fns = {
     MortonCopy<false, PixelFormat::DXT23_SRGB>,
     MortonCopy<false, PixelFormat::DXT45_SRGB>,
     MortonCopy<false, PixelFormat::BC7U_SRGB>,
+    MortonCopy<false, PixelFormat::R4G4B4A4U>,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
+    nullptr,
     nullptr,
     nullptr,
     nullptr,

src/video_core/renderer_opengl/gl_rasterizer.cpp

@@ -348,6 +348,7 @@ static constexpr auto RangeFromInterval(Map& map, const Interval& interval) {
 }
 
 void RasterizerOpenGL::UpdatePagesCachedCount(VAddr addr, u64 size, int delta) {
+    std::lock_guard lock{pages_mutex};
     const u64 page_start{addr >> Memory::PAGE_BITS};
     const u64 page_end{(addr + size + Memory::PAGE_SIZE - 1) >> Memory::PAGE_BITS};
 

src/video_core/renderer_opengl/gl_rasterizer.h

@@ -9,6 +9,7 @@
 #include <cstddef>
 #include <map>
 #include <memory>
+#include <mutex>
 #include <optional>
 #include <tuple>
 #include <utility>
@@ -230,6 +231,8 @@ private:
 
     using CachedPageMap = boost::icl::interval_map<u64, int>;
     CachedPageMap cached_pages;
+
+    std::mutex pages_mutex;
 };
 
 } // namespace OpenGL

src/video_core/renderer_opengl/gl_texture_cache.cpp

@@ -111,7 +111,8 @@ constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> tex_format
     {GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
      true}, // DXT45_SRGB
     {GL_COMPRESSED_SRGB_ALPHA_BPTC_UNORM, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8, ComponentType::UNorm,
-     true},                                                                    // BC7U_SRGB
+     true},                                                                          // BC7U_SRGB
+    {GL_RGBA4, GL_RGBA, GL_UNSIGNED_SHORT_4_4_4_4_REV, ComponentType::UNorm, false}, // R4G4B4A4U
     {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_4X4_SRGB
     {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X8_SRGB
     {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X5_SRGB
@@ -120,6 +121,16 @@ constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> tex_format
     {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_5X5_SRGB
     {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_10X8
     {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_10X8_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_6X6
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_6X6_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_10X10
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_10X10_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_12X12
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_12X12_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_8X6
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_8X6_SRGB
+    {GL_RGBA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false},        // ASTC_2D_6X5
+    {GL_SRGB8_ALPHA8, GL_RGBA, GL_UNSIGNED_BYTE, ComponentType::UNorm, false}, // ASTC_2D_6X5_SRGB
 
     // Depth formats
     {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -102,8 +102,6 @@ RendererOpenGL::RendererOpenGL(Core::Frontend::EmuWindow& emu_window, Core::Syst
 RendererOpenGL::~RendererOpenGL() = default;
 
 void RendererOpenGL::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
-    system.GetPerfStats().EndSystemFrame();
-
     // Maintain the rasterizer's state as a priority
     OpenGLState prev_state = OpenGLState::GetCurState();
     state.AllDirty();
@@ -135,9 +133,6 @@ void RendererOpenGL::SwapBuffers(const Tegra::FramebufferConfig* framebuffer) {
 
     render_window.PollEvents();
 
-    system.FrameLimiter().DoFrameLimiting(system.CoreTiming().GetGlobalTimeUs());
-    system.GetPerfStats().BeginSystemFrame();
-
     // Restore the rasterizer state
     prev_state.AllDirty();
     prev_state.Apply();

src/video_core/renderer_vulkan/vk_shader_decompiler.cpp

@@ -1648,32 +1648,32 @@ class ExprDecompiler {
 public:
     explicit ExprDecompiler(SPIRVDecompiler& decomp) : decomp{decomp} {}
 
-    Id operator()(VideoCommon::Shader::ExprAnd& expr) {
+    Id operator()(const ExprAnd& expr) {
         const Id type_def = decomp.GetTypeDefinition(Type::Bool);
         const Id op1 = Visit(expr.operand1);
         const Id op2 = Visit(expr.operand2);
         return decomp.Emit(decomp.OpLogicalAnd(type_def, op1, op2));
     }
 
-    Id operator()(VideoCommon::Shader::ExprOr& expr) {
+    Id operator()(const ExprOr& expr) {
         const Id type_def = decomp.GetTypeDefinition(Type::Bool);
         const Id op1 = Visit(expr.operand1);
         const Id op2 = Visit(expr.operand2);
         return decomp.Emit(decomp.OpLogicalOr(type_def, op1, op2));
     }
 
-    Id operator()(VideoCommon::Shader::ExprNot& expr) {
+    Id operator()(const ExprNot& expr) {
         const Id type_def = decomp.GetTypeDefinition(Type::Bool);
         const Id op1 = Visit(expr.operand1);
         return decomp.Emit(decomp.OpLogicalNot(type_def, op1));
     }
 
-    Id operator()(VideoCommon::Shader::ExprPredicate& expr) {
+    Id operator()(const ExprPredicate& expr) {
         const auto pred = static_cast<Tegra::Shader::Pred>(expr.predicate);
         return decomp.Emit(decomp.OpLoad(decomp.t_bool, decomp.predicates.at(pred)));
     }
 
-    Id operator()(VideoCommon::Shader::ExprCondCode& expr) {
+    Id operator()(const ExprCondCode& expr) {
         const Node cc = decomp.ir.GetConditionCode(expr.cc);
         Id target;
 
@@ -1693,15 +1693,15 @@ public:
         return decomp.Emit(decomp.OpLoad(decomp.t_bool, target));
     }
 
-    Id operator()(VideoCommon::Shader::ExprVar& expr) {
+    Id operator()(const ExprVar& expr) {
         return decomp.Emit(decomp.OpLoad(decomp.t_bool, decomp.flow_variables.at(expr.var_index)));
     }
 
-    Id operator()(VideoCommon::Shader::ExprBoolean& expr) {
+    Id operator()(const ExprBoolean& expr) {
         return expr.value ? decomp.v_true : decomp.v_false;
     }
 
-    Id Visit(VideoCommon::Shader::Expr& node) {
+    Id Visit(const Expr& node) {
         return std::visit(*this, *node);
     }
 
@@ -1713,7 +1713,7 @@ class ASTDecompiler {
 public:
     explicit ASTDecompiler(SPIRVDecompiler& decomp) : decomp{decomp} {}
 
-    void operator()(VideoCommon::Shader::ASTProgram& ast) {
+    void operator()(const ASTProgram& ast) {
         ASTNode current = ast.nodes.GetFirst();
         while (current) {
             Visit(current);
@@ -1721,7 +1721,7 @@ public:
         }
     }
 
-    void operator()(VideoCommon::Shader::ASTIfThen& ast) {
+    void operator()(const ASTIfThen& ast) {
         ExprDecompiler expr_parser{decomp};
         const Id condition = expr_parser.Visit(ast.condition);
         const Id then_label = decomp.OpLabel();
@@ -1738,33 +1738,33 @@ public:
         decomp.Emit(endif_label);
     }
 
-    void operator()(VideoCommon::Shader::ASTIfElse& ast) {
+    void operator()([[maybe_unused]] const ASTIfElse& ast) {
         UNREACHABLE();
     }
 
-    void operator()(VideoCommon::Shader::ASTBlockEncoded& ast) {
+    void operator()([[maybe_unused]] const ASTBlockEncoded& ast) {
         UNREACHABLE();
     }
 
-    void operator()(VideoCommon::Shader::ASTBlockDecoded& ast) {
+    void operator()(const ASTBlockDecoded& ast) {
         decomp.VisitBasicBlock(ast.nodes);
     }
 
-    void operator()(VideoCommon::Shader::ASTVarSet& ast) {
+    void operator()(const ASTVarSet& ast) {
         ExprDecompiler expr_parser{decomp};
         const Id condition = expr_parser.Visit(ast.condition);
         decomp.Emit(decomp.OpStore(decomp.flow_variables.at(ast.index), condition));
     }
 
-    void operator()(VideoCommon::Shader::ASTLabel& ast) {
+    void operator()([[maybe_unused]] const ASTLabel& ast) {
         // Do nothing
     }
 
-    void operator()(VideoCommon::Shader::ASTGoto& ast) {
+    void operator()([[maybe_unused]] const ASTGoto& ast) {
         UNREACHABLE();
     }
 
-    void operator()(VideoCommon::Shader::ASTDoWhile& ast) {
+    void operator()(const ASTDoWhile& ast) {
         const Id loop_label = decomp.OpLabel();
         const Id endloop_label = decomp.OpLabel();
         const Id loop_start_block = decomp.OpLabel();
@@ -1787,7 +1787,7 @@ public:
         decomp.Emit(endloop_label);
     }
 
-    void operator()(VideoCommon::Shader::ASTReturn& ast) {
+    void operator()(const ASTReturn& ast) {
         if (!VideoCommon::Shader::ExprIsTrue(ast.condition)) {
             ExprDecompiler expr_parser{decomp};
             const Id condition = expr_parser.Visit(ast.condition);
@@ -1817,7 +1817,7 @@ public:
         }
     }
 
-    void operator()(VideoCommon::Shader::ASTBreak& ast) {
+    void operator()(const ASTBreak& ast) {
         if (!VideoCommon::Shader::ExprIsTrue(ast.condition)) {
             ExprDecompiler expr_parser{decomp};
             const Id condition = expr_parser.Visit(ast.condition);
@@ -1837,7 +1837,7 @@ public:
         }
     }
 
-    void Visit(VideoCommon::Shader::ASTNode& node) {
+    void Visit(const ASTNode& node) {
         std::visit(*this, *node->GetInnerData());
     }
 
@@ -1853,9 +1853,11 @@ void SPIRVDecompiler::DecompileAST() {
         Name(id, fmt::format("flow_var_{}", i));
         flow_variables.emplace(i, AddGlobalVariable(id));
     }
+
+    const ASTNode program = ir.GetASTProgram();
     ASTDecompiler decompiler{*this};
-    VideoCommon::Shader::ASTNode program = ir.GetASTProgram();
     decompiler.Visit(program);
+
     const Id next_block = OpLabel();
     Emit(OpBranch(next_block));
     Emit(next_block);

src/video_core/shader/ast.cpp

@@ -3,6 +3,9 @@
 // Refer to the license.txt file included.
 
 #include <string>
+#include <string_view>
+
+#include <fmt/format.h>
 
 #include "common/assert.h"
 #include "common/common_types.h"
@@ -229,7 +232,8 @@ public:
         return inner;
     }
 
-    std::string inner{};
+private:
+    std::string inner;
 };
 
 class ASTPrinter {
@@ -249,7 +253,7 @@ public:
     void operator()(const ASTIfThen& ast) {
         ExprPrinter expr_parser{};
         std::visit(expr_parser, *ast.condition);
-        inner += Ident() + "if (" + expr_parser.GetResult() + ") {\n";
+        inner += fmt::format("{}if ({}) {{\n", Indent(), expr_parser.GetResult());
         scope++;
         ASTNode current = ast.nodes.GetFirst();
         while (current) {
@@ -257,11 +261,13 @@ public:
             current = current->GetNext();
         }
         scope--;
-        inner += Ident() + "}\n";
+        inner += fmt::format("{}}}\n", Indent());
     }
 
     void operator()(const ASTIfElse& ast) {
-        inner += Ident() + "else {\n";
+        inner += Indent();
+        inner += "else {\n";
+
         scope++;
         ASTNode current = ast.nodes.GetFirst();
         while (current) {
@@ -269,40 +275,41 @@ public:
             current = current->GetNext();
         }
         scope--;
-        inner += Ident() + "}\n";
+
+        inner += Indent();
+        inner += "}\n";
     }
 
     void operator()(const ASTBlockEncoded& ast) {
-        inner += Ident() + "Block(" + std::to_string(ast.start) + ", " + std::to_string(ast.end) +
-                 ");\n";
+        inner += fmt::format("{}Block({}, {});\n", Indent(), ast.start, ast.end);
     }
 
-    void operator()(const ASTBlockDecoded& ast) {
-        inner += Ident() + "Block;\n";
+    void operator()([[maybe_unused]] const ASTBlockDecoded& ast) {
+        inner += Indent();
+        inner += "Block;\n";
     }
 
     void operator()(const ASTVarSet& ast) {
         ExprPrinter expr_parser{};
         std::visit(expr_parser, *ast.condition);
-        inner +=
-            Ident() + "V" + std::to_string(ast.index) + " := " + expr_parser.GetResult() + ";\n";
+        inner += fmt::format("{}V{} := {};\n", Indent(), ast.index, expr_parser.GetResult());
     }
 
     void operator()(const ASTLabel& ast) {
-        inner += "Label_" + std::to_string(ast.index) + ":\n";
+        inner += fmt::format("Label_{}:\n", ast.index);
     }
 
     void operator()(const ASTGoto& ast) {
         ExprPrinter expr_parser{};
         std::visit(expr_parser, *ast.condition);
-        inner += Ident() + "(" + expr_parser.GetResult() + ") -> goto Label_" +
-                 std::to_string(ast.label) + ";\n";
+        inner +=
+            fmt::format("{}({}) -> goto Label_{};\n", Indent(), expr_parser.GetResult(), ast.label);
     }
 
     void operator()(const ASTDoWhile& ast) {
         ExprPrinter expr_parser{};
         std::visit(expr_parser, *ast.condition);
-        inner += Ident() + "do {\n";
+        inner += fmt::format("{}do {{\n", Indent());
         scope++;
         ASTNode current = ast.nodes.GetFirst();
         while (current) {
@@ -310,32 +317,23 @@ public:
             current = current->GetNext();
         }
         scope--;
-        inner += Ident() + "} while (" + expr_parser.GetResult() + ");\n";
+        inner += fmt::format("{}}} while ({});\n", Indent(), expr_parser.GetResult());
     }
 
     void operator()(const ASTReturn& ast) {
         ExprPrinter expr_parser{};
         std::visit(expr_parser, *ast.condition);
-        inner += Ident() + "(" + expr_parser.GetResult() + ") -> " +
-                 (ast.kills ? "discard" : "exit") + ";\n";
+        inner += fmt::format("{}({}) -> {};\n", Indent(), expr_parser.GetResult(),
+                             ast.kills ? "discard" : "exit");
     }
 
     void operator()(const ASTBreak& ast) {
         ExprPrinter expr_parser{};
         std::visit(expr_parser, *ast.condition);
-        inner += Ident() + "(" + expr_parser.GetResult() + ") -> break;\n";
+        inner += fmt::format("{}({}) -> break;\n", Indent(), expr_parser.GetResult());
     }
 
-    std::string& Ident() {
-        if (memo_scope == scope) {
-            return tabs_memo;
-        }
-        tabs_memo = tabs.substr(0, scope * 2);
-        memo_scope = scope;
-        return tabs_memo;
-    }
-
-    void Visit(ASTNode& node) {
+    void Visit(const ASTNode& node) {
         std::visit(*this, *node->GetInnerData());
     }
 
@@ -344,16 +342,29 @@ public:
     }
 
 private:
+    std::string_view Indent() {
+        if (space_segment_scope == scope) {
+            return space_segment;
+        }
+
+        // Ensure that we don't exceed our view.
+        ASSERT(scope * 2 < spaces.size());
+
+        space_segment = spaces.substr(0, scope * 2);
+        space_segment_scope = scope;
+        return space_segment;
+    }
+
     std::string inner{};
+    std::string_view space_segment;
+
     u32 scope{};
+    u32 space_segment_scope{};
 
-    std::string tabs_memo{};
-    u32 memo_scope{};
-
-    static constexpr std::string_view tabs{"                                    "};
+    static constexpr std::string_view spaces{"                                    "};
 };
 
-std::string ASTManager::Print() {
+std::string ASTManager::Print() const {
     ASTPrinter printer{};
     printer.Visit(main_node);
     return printer.GetResult();
@@ -549,13 +560,13 @@ bool ASTManager::DirectlyRelated(const ASTNode& first, const ASTNode& second) co
     return min->GetParent() == max->GetParent();
 }
 
-void ASTManager::ShowCurrentState(std::string_view state) {
+void ASTManager::ShowCurrentState(std::string_view state) const {
     LOG_CRITICAL(HW_GPU, "\nState {}:\n\n{}\n", state, Print());
     SanityCheck();
 }
 
-void ASTManager::SanityCheck() {
-    for (auto& label : labels) {
+void ASTManager::SanityCheck() const {
+    for (const auto& label : labels) {
         if (!label->GetParent()) {
             LOG_CRITICAL(HW_GPU, "Sanity Check Failed");
         }

src/video_core/shader/ast.h

@@ -328,13 +328,13 @@ public:
 
     void InsertReturn(Expr condition, bool kills);
 
-    std::string Print();
+    std::string Print() const;
 
     void Decompile();
 
-    void ShowCurrentState(std::string_view state);
+    void ShowCurrentState(std::string_view state) const;
 
-    void SanityCheck();
+    void SanityCheck() const;
 
     void Clear();
 

src/video_core/shader/control_flow.cpp

@@ -473,8 +473,8 @@ void DecompileShader(CFGRebuildState& state) {
     state.manager->Decompile();
 }
 
-std::unique_ptr<ShaderCharacteristics> ScanFlow(const ProgramCode& program_code, u32 program_size,
-                                                u32 start_address,
+std::unique_ptr<ShaderCharacteristics> ScanFlow(const ProgramCode& program_code,
+                                                std::size_t program_size, u32 start_address,
                                                 const CompilerSettings& settings) {
     auto result_out = std::make_unique<ShaderCharacteristics>();
     if (settings.depth == CompileDepth::BruteForce) {

src/video_core/shader/control_flow.h

@@ -76,8 +76,8 @@ struct ShaderCharacteristics {
     CompilerSettings settings{};
 };
 
-std::unique_ptr<ShaderCharacteristics> ScanFlow(const ProgramCode& program_code, u32 program_size,
-                                                u32 start_address,
+std::unique_ptr<ShaderCharacteristics> ScanFlow(const ProgramCode& program_code,
+                                                std::size_t program_size, u32 start_address,
                                                 const CompilerSettings& settings);
 
 } // namespace VideoCommon::Shader

src/video_core/surface.cpp

@@ -212,6 +212,14 @@ PixelFormat PixelFormatFromTextureFormat(Tegra::Texture::TextureFormat format,
             break;
         }
         break;
+    case Tegra::Texture::TextureFormat::A4B4G4R4:
+        switch (component_type) {
+        case Tegra::Texture::ComponentType::UNORM:
+            return PixelFormat::R4G4B4A4U;
+        default:
+            break;
+        }
+        break;
     case Tegra::Texture::TextureFormat::R8:
         switch (component_type) {
         case Tegra::Texture::ComponentType::UNORM:
@@ -252,6 +260,7 @@ PixelFormat PixelFormatFromTextureFormat(Tegra::Texture::TextureFormat format,
         default:
             break;
         }
+        break;
     case Tegra::Texture::TextureFormat::R32_G32_B32_A32:
         switch (component_type) {
         case Tegra::Texture::ComponentType::FLOAT:
@@ -350,6 +359,16 @@ PixelFormat PixelFormatFromTextureFormat(Tegra::Texture::TextureFormat format,
         return is_srgb ? PixelFormat::ASTC_2D_8X5_SRGB : PixelFormat::ASTC_2D_8X5;
     case Tegra::Texture::TextureFormat::ASTC_2D_10X8:
         return is_srgb ? PixelFormat::ASTC_2D_10X8_SRGB : PixelFormat::ASTC_2D_10X8;
+    case Tegra::Texture::TextureFormat::ASTC_2D_6X6:
+        return is_srgb ? PixelFormat::ASTC_2D_6X6_SRGB : PixelFormat::ASTC_2D_6X6;
+    case Tegra::Texture::TextureFormat::ASTC_2D_10X10:
+        return is_srgb ? PixelFormat::ASTC_2D_10X10_SRGB : PixelFormat::ASTC_2D_10X10;
+    case Tegra::Texture::TextureFormat::ASTC_2D_12X12:
+        return is_srgb ? PixelFormat::ASTC_2D_12X12_SRGB : PixelFormat::ASTC_2D_12X12;
+    case Tegra::Texture::TextureFormat::ASTC_2D_8X6:
+        return is_srgb ? PixelFormat::ASTC_2D_8X6_SRGB : PixelFormat::ASTC_2D_8X6;
+    case Tegra::Texture::TextureFormat::ASTC_2D_6X5:
+        return is_srgb ? PixelFormat::ASTC_2D_6X5_SRGB : PixelFormat::ASTC_2D_6X5;
     case Tegra::Texture::TextureFormat::R16_G16:
         switch (component_type) {
         case Tegra::Texture::ComponentType::FLOAT:
@@ -510,6 +529,16 @@ bool IsPixelFormatASTC(PixelFormat format) {
     case PixelFormat::ASTC_2D_8X5_SRGB:
     case PixelFormat::ASTC_2D_10X8:
     case PixelFormat::ASTC_2D_10X8_SRGB:
+    case PixelFormat::ASTC_2D_6X6:
+    case PixelFormat::ASTC_2D_6X6_SRGB:
+    case PixelFormat::ASTC_2D_10X10:
+    case PixelFormat::ASTC_2D_10X10_SRGB:
+    case PixelFormat::ASTC_2D_12X12:
+    case PixelFormat::ASTC_2D_12X12_SRGB:
+    case PixelFormat::ASTC_2D_8X6:
+    case PixelFormat::ASTC_2D_8X6_SRGB:
+    case PixelFormat::ASTC_2D_6X5:
+    case PixelFormat::ASTC_2D_6X5_SRGB:
         return true;
     default:
         return false;
@@ -530,6 +559,11 @@ bool IsPixelFormatSRGB(PixelFormat format) {
     case PixelFormat::ASTC_2D_5X4_SRGB:
     case PixelFormat::ASTC_2D_5X5_SRGB:
     case PixelFormat::ASTC_2D_10X8_SRGB:
+    case PixelFormat::ASTC_2D_6X6_SRGB:
+    case PixelFormat::ASTC_2D_10X10_SRGB:
+    case PixelFormat::ASTC_2D_12X12_SRGB:
+    case PixelFormat::ASTC_2D_8X6_SRGB:
+    case PixelFormat::ASTC_2D_6X5_SRGB:
         return true;
     default:
         return false;

src/video_core/surface.h

@@ -67,27 +67,38 @@ enum class PixelFormat {
     DXT23_SRGB = 49,
     DXT45_SRGB = 50,
     BC7U_SRGB = 51,
-    ASTC_2D_4X4_SRGB = 52,
-    ASTC_2D_8X8_SRGB = 53,
-    ASTC_2D_8X5_SRGB = 54,
-    ASTC_2D_5X4_SRGB = 55,
-    ASTC_2D_5X5 = 56,
-    ASTC_2D_5X5_SRGB = 57,
-    ASTC_2D_10X8 = 58,
-    ASTC_2D_10X8_SRGB = 59,
+    R4G4B4A4U = 52,
+    ASTC_2D_4X4_SRGB = 53,
+    ASTC_2D_8X8_SRGB = 54,
+    ASTC_2D_8X5_SRGB = 55,
+    ASTC_2D_5X4_SRGB = 56,
+    ASTC_2D_5X5 = 57,
+    ASTC_2D_5X5_SRGB = 58,
+    ASTC_2D_10X8 = 59,
+    ASTC_2D_10X8_SRGB = 60,
+    ASTC_2D_6X6 = 61,
+    ASTC_2D_6X6_SRGB = 62,
+    ASTC_2D_10X10 = 63,
+    ASTC_2D_10X10_SRGB = 64,
+    ASTC_2D_12X12 = 65,
+    ASTC_2D_12X12_SRGB = 66,
+    ASTC_2D_8X6 = 67,
+    ASTC_2D_8X6_SRGB = 68,
+    ASTC_2D_6X5 = 69,
+    ASTC_2D_6X5_SRGB = 70,
 
     MaxColorFormat,
 
     // Depth formats
-    Z32F = 60,
-    Z16 = 61,
+    Z32F = 71,
+    Z16 = 72,
 
     MaxDepthFormat,
 
     // DepthStencil formats
-    Z24S8 = 62,
-    S8Z24 = 63,
-    Z32FS8 = 64,
+    Z24S8 = 73,
+    S8Z24 = 74,
+    Z32FS8 = 75,
 
     MaxDepthStencilFormat,
 
@@ -177,6 +188,7 @@ constexpr std::array<u32, MaxPixelFormat> compression_factor_shift_table = {{
     2, // DXT23_SRGB
     2, // DXT45_SRGB
     2, // BC7U_SRGB
+    0, // R4G4B4A4U
     2, // ASTC_2D_4X4_SRGB
     2, // ASTC_2D_8X8_SRGB
     2, // ASTC_2D_8X5_SRGB
@@ -185,6 +197,16 @@ constexpr std::array<u32, MaxPixelFormat> compression_factor_shift_table = {{
     2, // ASTC_2D_5X5_SRGB
     2, // ASTC_2D_10X8
     2, // ASTC_2D_10X8_SRGB
+    2, // ASTC_2D_6X6
+    2, // ASTC_2D_6X6_SRGB
+    2, // ASTC_2D_10X10
+    2, // ASTC_2D_10X10_SRGB
+    2, // ASTC_2D_12X12
+    2, // ASTC_2D_12X12_SRGB
+    2, // ASTC_2D_8X6
+    2, // ASTC_2D_8X6_SRGB
+    2, // ASTC_2D_6X5
+    2, // ASTC_2D_6X5_SRGB
     0, // Z32F
     0, // Z16
     0, // Z24S8
@@ -261,6 +283,7 @@ constexpr std::array<u32, MaxPixelFormat> block_width_table = {{
     4,  // DXT23_SRGB
     4,  // DXT45_SRGB
     4,  // BC7U_SRGB
+    1,  // R4G4B4A4U
     4,  // ASTC_2D_4X4_SRGB
     8,  // ASTC_2D_8X8_SRGB
     8,  // ASTC_2D_8X5_SRGB
@@ -269,6 +292,16 @@ constexpr std::array<u32, MaxPixelFormat> block_width_table = {{
     5,  // ASTC_2D_5X5_SRGB
     10, // ASTC_2D_10X8
     10, // ASTC_2D_10X8_SRGB
+    6,  // ASTC_2D_6X6
+    6,  // ASTC_2D_6X6_SRGB
+    10, // ASTC_2D_10X10
+    10, // ASTC_2D_10X10_SRGB
+    12, // ASTC_2D_12X12
+    12, // ASTC_2D_12X12_SRGB
+    8,  // ASTC_2D_8X6
+    8,  // ASTC_2D_8X6_SRGB
+    6,  // ASTC_2D_6X5
+    6,  // ASTC_2D_6X5_SRGB
     1,  // Z32F
     1,  // Z16
     1,  // Z24S8
@@ -285,71 +318,82 @@ static constexpr u32 GetDefaultBlockWidth(PixelFormat format) {
 }
 
 constexpr std::array<u32, MaxPixelFormat> block_height_table = {{
-    1, // ABGR8U
-    1, // ABGR8S
-    1, // ABGR8UI
-    1, // B5G6R5U
-    1, // A2B10G10R10U
-    1, // A1B5G5R5U
-    1, // R8U
-    1, // R8UI
-    1, // RGBA16F
-    1, // RGBA16U
-    1, // RGBA16UI
-    1, // R11FG11FB10F
-    1, // RGBA32UI
-    4, // DXT1
-    4, // DXT23
-    4, // DXT45
-    4, // DXN1
-    4, // DXN2UNORM
-    4, // DXN2SNORM
-    4, // BC7U
-    4, // BC6H_UF16
-    4, // BC6H_SF16
-    4, // ASTC_2D_4X4
-    1, // BGRA8
-    1, // RGBA32F
-    1, // RG32F
-    1, // R32F
-    1, // R16F
-    1, // R16U
-    1, // R16S
-    1, // R16UI
-    1, // R16I
-    1, // RG16
-    1, // RG16F
-    1, // RG16UI
-    1, // RG16I
-    1, // RG16S
-    1, // RGB32F
-    1, // RGBA8_SRGB
-    1, // RG8U
-    1, // RG8S
-    1, // RG32UI
-    1, // RGBX16F
-    1, // R32UI
-    8, // ASTC_2D_8X8
-    5, // ASTC_2D_8X5
-    4, // ASTC_2D_5X4
-    1, // BGRA8_SRGB
-    4, // DXT1_SRGB
-    4, // DXT23_SRGB
-    4, // DXT45_SRGB
-    4, // BC7U_SRGB
-    4, // ASTC_2D_4X4_SRGB
-    8, // ASTC_2D_8X8_SRGB
-    5, // ASTC_2D_8X5_SRGB
-    4, // ASTC_2D_5X4_SRGB
-    5, // ASTC_2D_5X5
-    5, // ASTC_2D_5X5_SRGB
-    8, // ASTC_2D_10X8
-    8, // ASTC_2D_10X8_SRGB
-    1, // Z32F
-    1, // Z16
-    1, // Z24S8
-    1, // S8Z24
-    1, // Z32FS8
+    1,  // ABGR8U
+    1,  // ABGR8S
+    1,  // ABGR8UI
+    1,  // B5G6R5U
+    1,  // A2B10G10R10U
+    1,  // A1B5G5R5U
+    1,  // R8U
+    1,  // R8UI
+    1,  // RGBA16F
+    1,  // RGBA16U
+    1,  // RGBA16UI
+    1,  // R11FG11FB10F
+    1,  // RGBA32UI
+    4,  // DXT1
+    4,  // DXT23
+    4,  // DXT45
+    4,  // DXN1
+    4,  // DXN2UNORM
+    4,  // DXN2SNORM
+    4,  // BC7U
+    4,  // BC6H_UF16
+    4,  // BC6H_SF16
+    4,  // ASTC_2D_4X4
+    1,  // BGRA8
+    1,  // RGBA32F
+    1,  // RG32F
+    1,  // R32F
+    1,  // R16F
+    1,  // R16U
+    1,  // R16S
+    1,  // R16UI
+    1,  // R16I
+    1,  // RG16
+    1,  // RG16F
+    1,  // RG16UI
+    1,  // RG16I
+    1,  // RG16S
+    1,  // RGB32F
+    1,  // RGBA8_SRGB
+    1,  // RG8U
+    1,  // RG8S
+    1,  // RG32UI
+    1,  // RGBX16F
+    1,  // R32UI
+    8,  // ASTC_2D_8X8
+    5,  // ASTC_2D_8X5
+    4,  // ASTC_2D_5X4
+    1,  // BGRA8_SRGB
+    4,  // DXT1_SRGB
+    4,  // DXT23_SRGB
+    4,  // DXT45_SRGB
+    4,  // BC7U_SRGB
+    1,  // R4G4B4A4U
+    4,  // ASTC_2D_4X4_SRGB
+    8,  // ASTC_2D_8X8_SRGB
+    5,  // ASTC_2D_8X5_SRGB
+    4,  // ASTC_2D_5X4_SRGB
+    5,  // ASTC_2D_5X5
+    5,  // ASTC_2D_5X5_SRGB
+    8,  // ASTC_2D_10X8
+    8,  // ASTC_2D_10X8_SRGB
+    6,  // ASTC_2D_6X6
+    6,  // ASTC_2D_6X6_SRGB
+    10, // ASTC_2D_10X10
+    10, // ASTC_2D_10X10_SRGB
+    12, // ASTC_2D_12X12
+    12, // ASTC_2D_12X12_SRGB
+    6,  // ASTC_2D_8X6
+    6,  // ASTC_2D_8X6_SRGB
+    5,  // ASTC_2D_6X5
+    5,  // ASTC_2D_6X5_SRGB
+    1,  // Z32F
+    1,  // Z16
+    1,  // Z24S8
+    1,  // S8Z24
+    1,  // Z32FS8
 }};
 
 static constexpr u32 GetDefaultBlockHeight(PixelFormat format) {
@@ -413,6 +457,7 @@ constexpr std::array<u32, MaxPixelFormat> bpp_table = {{
     128, // DXT23_SRGB
     128, // DXT45_SRGB
     128, // BC7U
+    16,  // R4G4B4A4U
     128, // ASTC_2D_4X4_SRGB
     128, // ASTC_2D_8X8_SRGB
     128, // ASTC_2D_8X5_SRGB
@@ -421,6 +466,16 @@ constexpr std::array<u32, MaxPixelFormat> bpp_table = {{
     128, // ASTC_2D_5X5_SRGB
     128, // ASTC_2D_10X8
     128, // ASTC_2D_10X8_SRGB
+    128, // ASTC_2D_6X6
+    128, // ASTC_2D_6X6_SRGB
+    128, // ASTC_2D_10X10
+    128, // ASTC_2D_10X10_SRGB
+    128, // ASTC_2D_12X12
+    128, // ASTC_2D_12X12_SRGB
+    128, // ASTC_2D_8X6
+    128, // ASTC_2D_8X6_SRGB
+    128, // ASTC_2D_6X5
+    128, // ASTC_2D_6X5_SRGB
     32,  // Z32F
     16,  // Z16
     32,  // Z24S8
@@ -504,6 +559,7 @@ constexpr std::array<SurfaceCompression, MaxPixelFormat> compression_type_table
     SurfaceCompression::Compressed, // DXT23_SRGB
     SurfaceCompression::Compressed, // DXT45_SRGB
     SurfaceCompression::Compressed, // BC7U_SRGB
+    SurfaceCompression::None,       // R4G4B4A4U
     SurfaceCompression::Converted,  // ASTC_2D_4X4_SRGB
     SurfaceCompression::Converted,  // ASTC_2D_8X8_SRGB
     SurfaceCompression::Converted,  // ASTC_2D_8X5_SRGB
@@ -512,6 +568,16 @@ constexpr std::array<SurfaceCompression, MaxPixelFormat> compression_type_table
     SurfaceCompression::Converted,  // ASTC_2D_5X5_SRGB
     SurfaceCompression::Converted,  // ASTC_2D_10X8
     SurfaceCompression::Converted,  // ASTC_2D_10X8_SRGB
+    SurfaceCompression::Converted,  // ASTC_2D_6X6
+    SurfaceCompression::Converted,  // ASTC_2D_6X6_SRGB
+    SurfaceCompression::Converted,  // ASTC_2D_10X10
+    SurfaceCompression::Converted,  // ASTC_2D_10X10_SRGB
+    SurfaceCompression::Converted,  // ASTC_2D_12X12
+    SurfaceCompression::Converted,  // ASTC_2D_12X12_SRGB
+    SurfaceCompression::Converted,  // ASTC_2D_8X6
+    SurfaceCompression::Converted,  // ASTC_2D_8X6_SRGB
+    SurfaceCompression::Converted,  // ASTC_2D_6X5
+    SurfaceCompression::Converted,  // ASTC_2D_6X5_SRGB
     SurfaceCompression::None,       // Z32F
     SurfaceCompression::None,       // Z16
     SurfaceCompression::None,       // Z24S8

src/video_core/texture_cache/texture_cache.h

@@ -62,10 +62,10 @@ public:
         }
     }
 
-    /***
-     * `Guard` guarantees that rendertargets don't unregister themselves if the
+    /**
+     * Guarantees that rendertargets don't unregister themselves if the
      * collide. Protection is currently only done on 3D slices.
-     ***/
+     */
     void GuardRenderTargets(bool new_guard) {
         guard_render_targets = new_guard;
     }
@@ -287,7 +287,7 @@ protected:
                            const Tegra::Engines::Fermi2D::Config& copy_config) = 0;
 
     // Depending on the backend, a buffer copy can be slow as it means deoptimizing the texture
-    // and reading it from a sepparate buffer.
+    // and reading it from a separate buffer.
     virtual void BufferCopy(TSurface& src_surface, TSurface& dst_surface) = 0;
 
     void ManageRenderTargetUnregister(TSurface& surface) {
@@ -386,12 +386,13 @@ private:
     };
 
     /**
-     * `PickStrategy` takes care of selecting a proper strategy to deal with a texture recycle.
-     * @param overlaps, the overlapping surfaces registered in the cache.
-     * @param params, the paremeters on the new surface.
-     * @param gpu_addr, the starting address of the new surface.
-     * @param untopological, tells the recycler that the texture has no way to match the overlaps
-     * due to topological reasons.
+     * Takes care of selecting a proper strategy to deal with a texture recycle.
+     *
+     * @param overlaps      The overlapping surfaces registered in the cache.
+     * @param params        The parameters on the new surface.
+     * @param gpu_addr      The starting address of the new surface.
+     * @param untopological Indicates to the recycler that the texture has no way
+     *                      to match the overlaps due to topological reasons.
      **/
     RecycleStrategy PickStrategy(std::vector<TSurface>& overlaps, const SurfaceParams& params,
                                  const GPUVAddr gpu_addr, const MatchTopologyResult untopological) {
@@ -402,7 +403,7 @@ private:
         if (params.block_depth > 1 || params.target == SurfaceTarget::Texture3D) {
             return RecycleStrategy::Flush;
         }
-        for (auto s : overlaps) {
+        for (const auto& s : overlaps) {
             const auto& s_params = s->GetSurfaceParams();
             if (s_params.block_depth > 1 || s_params.target == SurfaceTarget::Texture3D) {
                 return RecycleStrategy::Flush;
@@ -419,16 +420,19 @@ private:
     }
 
     /**
-     *  `RecycleSurface` es a method we use to decide what to do with textures we can't resolve in
-     *the cache It has 2 implemented strategies: Ignore and Flush. Ignore just unregisters all the
-     *overlaps and loads the new texture. Flush, flushes all the overlaps into memory and loads the
-     *new surface from that data.
-     * @param overlaps, the overlapping surfaces registered in the cache.
-     * @param params, the paremeters on the new surface.
-     * @param gpu_addr, the starting address of the new surface.
-     * @param preserve_contents, tells if the new surface should be loaded from meory or left blank
-     * @param untopological, tells the recycler that the texture has no way to match the overlaps
-     * due to topological reasons.
+     * Used to decide what to do with textures we can't resolve in the cache It has 2 implemented
+     * strategies: Ignore and Flush.
+     *
+     * - Ignore: Just unregisters all the overlaps and loads the new texture.
+     * - Flush: Flushes all the overlaps into memory and loads the new surface from that data.
+     *
+     * @param overlaps          The overlapping surfaces registered in the cache.
+     * @param params            The parameters for the new surface.
+     * @param gpu_addr          The starting address of the new surface.
+     * @param preserve_contents Indicates that the new surface should be loaded from memory or left
+     *                          blank.
+     * @param untopological     Indicates to the recycler that the texture has no way to match the
+     *                          overlaps due to topological reasons.
      **/
     std::pair<TSurface, TView> RecycleSurface(std::vector<TSurface>& overlaps,
                                               const SurfaceParams& params, const GPUVAddr gpu_addr,
@@ -465,10 +469,12 @@ private:
     }
 
     /**
-     * `RebuildSurface` this method takes a single surface and recreates into another that
-     * may differ in format, target or width alingment.
-     * @param current_surface, the registered surface in the cache which we want to convert.
-     * @param params, the new surface params which we'll use to recreate the surface.
+     * Takes a single surface and recreates into another that may differ in
+     * format, target or width alignment.
+     *
+     * @param current_surface The registered surface in the cache which we want to convert.
+     * @param params          The new surface params which we'll use to recreate the surface.
+     * @param is_render       Whether or not the surface is a render target.
      **/
     std::pair<TSurface, TView> RebuildSurface(TSurface current_surface, const SurfaceParams& params,
                                               bool is_render) {
@@ -502,12 +508,14 @@ private:
     }
 
     /**
-     * `ManageStructuralMatch` this method takes a single surface and checks with the new surface's
-     * params if it's an exact match, we return the main view of the registered surface. If it's
-     * formats don't match, we rebuild the surface. We call this last method a `Mirage`. If formats
+     * Takes a single surface and checks with the new surface's params if it's an exact
+     * match, we return the main view of the registered surface. If its formats don't
+     * match, we rebuild the surface. We call this last method a `Mirage`. If formats
      * match but the targets don't, we create an overview View of the registered surface.
-     * @param current_surface, the registered surface in the cache which we want to convert.
-     * @param params, the new surface params which we want to check.
+     *
+     * @param current_surface The registered surface in the cache which we want to convert.
+     * @param params          The new surface params which we want to check.
+     * @param is_render       Whether or not the surface is a render target.
      **/
     std::pair<TSurface, TView> ManageStructuralMatch(TSurface current_surface,
                                                      const SurfaceParams& params, bool is_render) {
@@ -529,13 +537,14 @@ private:
     }
 
     /**
-     * `TryReconstructSurface` unlike `RebuildSurface` where we know the registered surface
-     * matches the candidate in some way, we got no guarantess here. We try to see if the overlaps
-     * are sublayers/mipmaps of the new surface, if they all match we end up recreating a surface
-     * for them, else we return nothing.
-     * @param overlaps, the overlapping surfaces registered in the cache.
-     * @param params, the paremeters on the new surface.
-     * @param gpu_addr, the starting address of the new surface.
+     * Unlike RebuildSurface where we know whether or not registered surfaces match the candidate
+     * in some way, we have no guarantees here. We try to see if the overlaps are sublayers/mipmaps
+     * of the new surface, if they all match we end up recreating a surface for them,
+     * else we return nothing.
+     *
+     * @param overlaps The overlapping surfaces registered in the cache.
+     * @param params   The parameters on the new surface.
+     * @param gpu_addr The starting address of the new surface.
      **/
     std::optional<std::pair<TSurface, TView>> TryReconstructSurface(std::vector<TSurface>& overlaps,
                                                                     const SurfaceParams& params,
@@ -575,7 +584,7 @@ private:
         } else if (Settings::values.use_accurate_gpu_emulation && passed_tests != overlaps.size()) {
             return {};
         }
-        for (auto surface : overlaps) {
+        for (const auto& surface : overlaps) {
             Unregister(surface);
         }
         new_surface->MarkAsModified(modified, Tick());
@@ -584,19 +593,27 @@ private:
     }
 
     /**
-     * `GetSurface` gets the starting address and parameters of a candidate surface and tries
-     * to find a matching surface within the cache. This is done in 3 big steps. The first is to
-     * check the 1st Level Cache in order to find an exact match, if we fail, we move to step 2.
-     * Step 2 is checking if there are any overlaps at all, if none, we just load the texture from
-     * memory else we move to step 3. Step 3 consists on figuring the relationship between the
-     * candidate texture and the overlaps. We divide the scenarios depending if there's 1 or many
-     * overlaps. If there's many, we just try to reconstruct a new surface out of them based on the
-     * candidate's parameters, if we fail, we recycle. When there's only 1 overlap then we have to
-     * check if the candidate is a view (layer/mipmap) of the overlap or if the registered surface
-     * is a mipmap/layer of the candidate. In this last case we reconstruct a new surface.
-     * @param gpu_addr, the starting address of the candidate surface.
-     * @param params, the paremeters on the candidate surface.
-     * @param preserve_contents, tells if the new surface should be loaded from meory or left blank.
+     * Gets the starting address and parameters of a candidate surface and tries
+     * to find a matching surface within the cache. This is done in 3 big steps:
+     *
+     * 1. Check the 1st Level Cache in order to find an exact match, if we fail, we move to step 2.
+     *
+     * 2. Check if there are any overlaps at all, if there are none, we just load the texture from
+     *    memory else we move to step 3.
+     *
+     * 3. Consists of figuring out the relationship between the candidate texture and the
+     *    overlaps. We divide the scenarios depending if there's 1 or many overlaps. If
+     *    there's many, we just try to reconstruct a new surface out of them based on the
+     *    candidate's parameters, if we fail, we recycle. When there's only 1 overlap then we
+     *    have to check if the candidate is a view (layer/mipmap) of the overlap or if the
+     *    registered surface is a mipmap/layer of the candidate. In this last case we reconstruct
+     *    a new surface.
+     *
+     * @param gpu_addr          The starting address of the candidate surface.
+     * @param params            The parameters on the candidate surface.
+     * @param preserve_contents Indicates that the new surface should be loaded from memory or
+     *                          left blank.
+     * @param is_render         Whether or not the surface is a render target.
      **/
     std::pair<TSurface, TView> GetSurface(const GPUVAddr gpu_addr, const SurfaceParams& params,
                                           bool preserve_contents, bool is_render) {
@@ -651,7 +668,7 @@ private:
         // Step 3
         // Now we need to figure the relationship between the texture and its overlaps
         // we do a topological test to ensure we can find some relationship. If it fails
-        // inmediatly recycle the texture
+        // immediately recycle the texture
         for (const auto& surface : overlaps) {
             const auto topological_result = surface->MatchesTopology(params);
             if (topological_result != MatchTopologyResult::FullMatch) {
@@ -720,12 +737,13 @@ private:
     }
 
     /**
-     * `DeduceSurface` gets the starting address and parameters of a candidate surface and tries
-     * to find a matching surface within the cache that's similar to it. If there are many textures
+     * Gets the starting address and parameters of a candidate surface and tries to find a
+     * matching surface within the cache that's similar to it. If there are many textures
      * or the texture found if entirely incompatible, it will fail. If no texture is found, the
      * blit will be unsuccessful.
-     * @param gpu_addr, the starting address of the candidate surface.
-     * @param params, the paremeters on the candidate surface.
+     *
+     * @param gpu_addr The starting address of the candidate surface.
+     * @param params   The parameters on the candidate surface.
      **/
     Deduction DeduceSurface(const GPUVAddr gpu_addr, const SurfaceParams& params) {
         const auto host_ptr{system.GPU().MemoryManager().GetPointer(gpu_addr)};
@@ -777,11 +795,14 @@ private:
     }
 
     /**
-     * `DeduceBestBlit` gets the a source and destination starting address and parameters,
+     * Gets the a source and destination starting address and parameters,
      * and tries to deduce if they are supposed to be depth textures. If so, their
      * parameters are modified and fixed into so.
-     * @param gpu_addr, the starting address of the candidate surface.
-     * @param params, the parameters on the candidate surface.
+     *
+     * @param src_params   The parameters of the candidate surface.
+     * @param dst_params   The parameters of the destination surface.
+     * @param src_gpu_addr The starting address of the candidate surface.
+     * @param dst_gpu_addr The starting address of the destination surface.
      **/
     void DeduceBestBlit(SurfaceParams& src_params, SurfaceParams& dst_params,
                         const GPUVAddr src_gpu_addr, const GPUVAddr dst_gpu_addr) {