27 files changed, 164 insertions, 120 deletions

diff --git a/src/core/core.cpp b/src/core/core.cpp
index 4d0ac72a5..ddc767e30 100644
--- a/src/core/core.cpp
+++ b/src/core/core.cpp
@@ -256,6 +256,8 @@ struct System::Impl {
         is_powered_on = false;
         exit_lock = false;
 
+        gpu_core->WaitIdle();
+
         // Shutdown emulation session
         renderer.reset();
         GDBStub::Shutdown();
diff --git a/src/core/hle/service/am/am.cpp b/src/core/hle/service/am/am.cpp
index 941ebc93a..3a32d5b41 100644
--- a/src/core/hle/service/am/am.cpp
+++ b/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));
diff --git a/src/core/hle/service/bcat/backend/backend.cpp b/src/core/hle/service/bcat/backend/backend.cpp
index 9d6946bc5..b86fda29a 100644
--- a/src/core/hle/service/bcat/backend/backend.cpp
+++ b/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));
diff --git a/src/core/hle/service/bcat/backend/backend.h b/src/core/hle/service/bcat/backend/backend.h
index 51dbd3316..ea4b16ad0 100644
--- a/src/core/hle/service/bcat/backend/backend.h
+++ b/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
diff --git a/src/core/hle/service/bcat/backend/boxcat.cpp b/src/core/hle/service/bcat/backend/boxcat.cpp
index 64022982b..918159e11 100644
--- a/src/core/hle/service/bcat/backend/boxcat.cpp
+++ b/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;
         }
     }
diff --git a/src/core/hle/service/bcat/backend/boxcat.h b/src/core/hle/service/bcat/backend/boxcat.h
index 601151189..d65b42e58 100644
--- a/src/core/hle/service/bcat/backend/boxcat.h
+++ b/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
diff --git a/src/core/hle/service/bcat/module.cpp b/src/core/hle/service/bcat/module.cpp
index 4e4aa758b..6d9d1527d 100644
--- a/src/core/hle/service/bcat/module.cpp
+++ b/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;
diff --git a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp b/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp
index f764388bc..3f7b8e670 100644
--- a/src/core/hle/service/nvdrv/devices/nvdisp_disp0.cpp
+++ b/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
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp b/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp
index eb88fee1b..b27ee0502 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl.cpp
+++ b/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;
diff --git a/src/core/hle/service/nvdrv/interface.cpp b/src/core/hle/service/nvdrv/interface.cpp
index 5e0c23602..68d139cfb 100644
--- a/src/core/hle/service/nvdrv/interface.cpp
+++ b/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);
diff --git a/src/core/hle/service/nvdrv/nvdrv.cpp b/src/core/hle/service/nvdrv/nvdrv.cpp
index 307a7e928..7bfb99e34 100644
--- a/src/core/hle/service/nvdrv/nvdrv.cpp
+++ b/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;
     }
diff --git a/src/core/hle/service/nvflinger/buffer_queue.cpp b/src/core/hle/service/nvflinger/buffer_queue.cpp
index e1a07d3ee..55b68eb0c 100644
--- a/src/core/hle/service/nvflinger/buffer_queue.cpp
+++ b/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");
 }
diff --git a/src/core/hle/service/nvflinger/buffer_queue.h b/src/core/hle/service/nvflinger/buffer_queue.h
index 356bedb81..8f9b18547 100644
--- a/src/core/hle/service/nvflinger/buffer_queue.h
+++ b/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 {
diff --git a/src/core/hle/service/nvflinger/nvflinger.cpp b/src/core/hle/service/nvflinger/nvflinger.cpp
index 2e4d707b9..cc9522aad 100644
--- a/src/core/hle/service/nvflinger/nvflinger.cpp
+++ b/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)
diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp
index 76cfe8107..095660115 100644
--- a/src/video_core/gpu.cpp
+++ b/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 {
diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h
index 29fa8e95b..dbca19f35 100644
--- a/src/video_core/gpu.h
+++ b/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;
diff --git a/src/video_core/gpu_asynch.cpp b/src/video_core/gpu_asynch.cpp
index f2a3a390e..04222d060 100644
--- a/src/video_core/gpu_asynch.cpp
+++ b/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
diff --git a/src/video_core/gpu_asynch.h b/src/video_core/gpu_asynch.h
index a12f9bac4..1241ade1d 100644
--- a/src/video_core/gpu_asynch.h
+++ b/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;
diff --git a/src/video_core/gpu_synch.h b/src/video_core/gpu_synch.h
index 5eb1c461c..c71baee89 100644
--- a/src/video_core/gpu_synch.h
+++ b/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,
diff --git a/src/video_core/gpu_thread.cpp b/src/video_core/gpu_thread.cpp
index 5f039e4fd..758a37f14 100644
--- a/src/video_core/gpu_thread.cpp
+++ b/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
diff --git a/src/video_core/gpu_thread.h b/src/video_core/gpu_thread.h
index 3ae0ec9f3..08dc96bb3 100644
--- a/src/video_core/gpu_thread.h
+++ b/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;
 };
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index a85f730a8..cbcf81414 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/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};
 
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.h b/src/video_core/renderer_opengl/gl_rasterizer.h
index 9c10ebda3..c24a02d71 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.h
+++ b/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
diff --git a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
index 6a610a3bc..a3524a6a9 100644
--- a/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
+++ b/src/video_core/renderer_opengl/gl_shader_decompiler.cpp
@@ -1148,7 +1148,7 @@ private:
         for (const auto& variant : extras) {
             if (const auto argument = std::get_if<TextureArgument>(&variant)) {
                 expr += GenerateTextureArgument(*argument);
-            } else if (std::get_if<TextureAoffi>(&variant)) {
+            } else if (std::holds_alternative<TextureAoffi>(variant)) {
                 expr += GenerateTextureAoffi(meta->aoffi);
             } else {
                 UNREACHABLE();
@@ -1158,8 +1158,8 @@ private:
         return expr + ')';
     }
 
-    std::string GenerateTextureArgument(TextureArgument argument) {
-        const auto [type, operand] = argument;
+    std::string GenerateTextureArgument(const TextureArgument& argument) {
+        const auto& [type, operand] = argument;
         if (operand == nullptr) {
             return {};
         }
@@ -1235,7 +1235,7 @@ private:
 
     std::string BuildImageValues(Operation operation) {
         constexpr std::array constructors{"uint", "uvec2", "uvec3", "uvec4"};
-        const auto meta{std::get<MetaImage>(operation.GetMeta())};
+        const auto& meta{std::get<MetaImage>(operation.GetMeta())};
 
         const std::size_t values_count{meta.values.size()};
         std::string expr = fmt::format("{}(", constructors.at(values_count - 1));
@@ -1780,14 +1780,14 @@ private:
             return {"0", Type::Int};
         }
 
-        const auto meta{std::get<MetaImage>(operation.GetMeta())};
+        const auto& meta{std::get<MetaImage>(operation.GetMeta())};
         return {fmt::format("imageLoad({}, {}){}", GetImage(meta.image),
                             BuildIntegerCoordinates(operation), GetSwizzle(meta.element)),
                 Type::Uint};
     }
 
     Expression ImageStore(Operation operation) {
-        const auto meta{std::get<MetaImage>(operation.GetMeta())};
+        const auto& meta{std::get<MetaImage>(operation.GetMeta())};
         code.AddLine("imageStore({}, {}, {});", GetImage(meta.image),
                      BuildIntegerCoordinates(operation), BuildImageValues(operation));
         return {};
@@ -1795,7 +1795,7 @@ private:
 
     template <const std::string_view& opname>
     Expression AtomicImage(Operation operation) {
-        const auto meta{std::get<MetaImage>(operation.GetMeta())};
+        const auto& meta{std::get<MetaImage>(operation.GetMeta())};
         ASSERT(meta.values.size() == 1);
 
         return {fmt::format("imageAtomic{}({}, {}, {})", opname, GetImage(meta.image),
@@ -2246,7 +2246,7 @@ private:
         code.AddLine("#ifdef SAMPLER_{}_IS_BUFFER", sampler.GetIndex());
     }
 
-    std::string GetDeclarationWithSuffix(u32 index, const std::string& name) const {
+    std::string GetDeclarationWithSuffix(u32 index, std::string_view name) const {
         return fmt::format("{}_{}_{}", name, index, suffix);
     }
 
@@ -2271,17 +2271,15 @@ private:
     ShaderWriter code;
 };
 
-static constexpr std::string_view flow_var = "flow_var_";
-
 std::string GetFlowVariable(u32 i) {
-    return fmt::format("{}{}", flow_var, i);
+    return fmt::format("flow_var_{}", i);
 }
 
 class ExprDecompiler {
 public:
     explicit ExprDecompiler(GLSLDecompiler& decomp) : decomp{decomp} {}
 
-    void operator()(VideoCommon::Shader::ExprAnd& expr) {
+    void operator()(const ExprAnd& expr) {
         inner += "( ";
         std::visit(*this, *expr.operand1);
         inner += " && ";
@@ -2289,7 +2287,7 @@ public:
         inner += ')';
     }
 
-    void operator()(VideoCommon::Shader::ExprOr& expr) {
+    void operator()(const ExprOr& expr) {
         inner += "( ";
         std::visit(*this, *expr.operand1);
         inner += " || ";
@@ -2297,17 +2295,17 @@ public:
         inner += ')';
     }
 
-    void operator()(VideoCommon::Shader::ExprNot& expr) {
+    void operator()(const ExprNot& expr) {
         inner += '!';
         std::visit(*this, *expr.operand1);
     }
 
-    void operator()(VideoCommon::Shader::ExprPredicate& expr) {
+    void operator()(const ExprPredicate& expr) {
         const auto pred = static_cast<Tegra::Shader::Pred>(expr.predicate);
         inner += decomp.GetPredicate(pred);
     }
 
-    void operator()(VideoCommon::Shader::ExprCondCode& expr) {
+    void operator()(const ExprCondCode& expr) {
         const Node cc = decomp.ir.GetConditionCode(expr.cc);
         std::string target;
 
@@ -2329,15 +2327,15 @@ public:
         inner += target;
     }
 
-    void operator()(VideoCommon::Shader::ExprVar& expr) {
+    void operator()(const ExprVar& expr) {
         inner += GetFlowVariable(expr.var_index);
     }
 
-    void operator()(VideoCommon::Shader::ExprBoolean& expr) {
+    void operator()(const ExprBoolean& expr) {
         inner += expr.value ? "true" : "false";
     }
 
-    std::string& GetResult() {
+    const std::string& GetResult() const {
         return inner;
     }
 
@@ -2350,7 +2348,7 @@ class ASTDecompiler {
 public:
     explicit ASTDecompiler(GLSLDecompiler& decomp) : decomp{decomp} {}
 
-    void operator()(VideoCommon::Shader::ASTProgram& ast) {
+    void operator()(const ASTProgram& ast) {
         ASTNode current = ast.nodes.GetFirst();
         while (current) {
             Visit(current);
@@ -2358,7 +2356,7 @@ public:
         }
     }
 
-    void operator()(VideoCommon::Shader::ASTIfThen& ast) {
+    void operator()(const ASTIfThen& ast) {
         ExprDecompiler expr_parser{decomp};
         std::visit(expr_parser, *ast.condition);
         decomp.code.AddLine("if ({}) {{", expr_parser.GetResult());
@@ -2372,7 +2370,7 @@ public:
         decomp.code.AddLine("}}");
     }
 
-    void operator()(VideoCommon::Shader::ASTIfElse& ast) {
+    void operator()(const ASTIfElse& ast) {
         decomp.code.AddLine("else {{");
         decomp.code.scope++;
         ASTNode current = ast.nodes.GetFirst();
@@ -2384,29 +2382,29 @@ public:
         decomp.code.AddLine("}}");
     }
 
-    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.VisitBlock(ast.nodes);
     }
 
-    void operator()(VideoCommon::Shader::ASTVarSet& ast) {
+    void operator()(const ASTVarSet& ast) {
         ExprDecompiler expr_parser{decomp};
         std::visit(expr_parser, *ast.condition);
         decomp.code.AddLine("{} = {};", GetFlowVariable(ast.index), expr_parser.GetResult());
     }
 
-    void operator()(VideoCommon::Shader::ASTLabel& ast) {
+    void operator()(const ASTLabel& ast) {
         decomp.code.AddLine("// Label_{}:", ast.index);
     }
 
-    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) {
         ExprDecompiler expr_parser{decomp};
         std::visit(expr_parser, *ast.condition);
         decomp.code.AddLine("do {{");
@@ -2420,7 +2418,7 @@ public:
         decomp.code.AddLine("}} while({});", expr_parser.GetResult());
     }
 
-    void operator()(VideoCommon::Shader::ASTReturn& ast) {
+    void operator()(const ASTReturn& ast) {
         const bool is_true = VideoCommon::Shader::ExprIsTrue(ast.condition);
         if (!is_true) {
             ExprDecompiler expr_parser{decomp};
@@ -2440,7 +2438,7 @@ public:
         }
     }
 
-    void operator()(VideoCommon::Shader::ASTBreak& ast) {
+    void operator()(const ASTBreak& ast) {
         const bool is_true = VideoCommon::Shader::ExprIsTrue(ast.condition);
         if (!is_true) {
             ExprDecompiler expr_parser{decomp};
@@ -2455,7 +2453,7 @@ public:
         }
     }
 
-    void Visit(VideoCommon::Shader::ASTNode& node) {
+    void Visit(const ASTNode& node) {
         std::visit(*this, *node->GetInnerData());
     }
 
@@ -2468,9 +2466,9 @@ void GLSLDecompiler::DecompileAST() {
     for (u32 i = 0; i < num_flow_variables; i++) {
         code.AddLine("bool {} = false;", GetFlowVariable(i));
     }
+
     ASTDecompiler decompiler{*this};
-    VideoCommon::Shader::ASTNode program = ir.GetASTProgram();
-    decompiler.Visit(program);
+    decompiler.Visit(ir.GetASTProgram());
 }
 
 } // Anonymous namespace
diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp
index 1e6ef66ab..4bbd17b12 100644
--- a/src/video_core/renderer_opengl/renderer_opengl.cpp
+++ b/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();
diff --git a/src/video_core/shader/node.h b/src/video_core/shader/node.h
index 338bab17c..447fb5c1d 100644
--- a/src/video_core/shader/node.h
+++ b/src/video_core/shader/node.h
@@ -410,7 +410,7 @@ public:
     explicit OperationNode(OperationCode code) : OperationNode(code, Meta{}) {}
 
     explicit OperationNode(OperationCode code, Meta meta)
-        : OperationNode(code, meta, std::vector<Node>{}) {}
+        : OperationNode(code, std::move(meta), std::vector<Node>{}) {}
 
     explicit OperationNode(OperationCode code, std::vector<Node> operands)
         : OperationNode(code, Meta{}, std::move(operands)) {}
diff --git a/src/video_core/surface.cpp b/src/video_core/surface.cpp
index 250afc6d6..ef6b3592e 100644
--- a/src/video_core/surface.cpp
+++ b/src/video_core/surface.cpp
@@ -252,6 +252,7 @@ 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: