Add Configure Custome Keys Directory to File Menu

Addressing Issue #1226, this should allow moving the Keys directory to a
different location if desired.

src/audio_core/stream.cpp

@@ -37,7 +37,7 @@ Stream::Stream(u32 sample_rate, Format format, ReleaseCallback&& release_callbac
     : sample_rate{sample_rate}, format{format}, release_callback{std::move(release_callback)},
       sink_stream{sink_stream}, name{std::move(name_)} {
 
-    release_event = CoreTiming::RegisterEvent(
+    release_event = Core::Timing::RegisterEvent(
         name, [this](u64 userdata, int cycles_late) { ReleaseActiveBuffer(); });
 }
 
@@ -57,7 +57,7 @@ Stream::State Stream::GetState() const {
 
 s64 Stream::GetBufferReleaseCycles(const Buffer& buffer) const {
     const std::size_t num_samples{buffer.GetSamples().size() / GetNumChannels()};
-    return CoreTiming::usToCycles((static_cast<u64>(num_samples) * 1000000) / sample_rate);
+    return Core::Timing::usToCycles((static_cast<u64>(num_samples) * 1000000) / sample_rate);
 }
 
 static void VolumeAdjustSamples(std::vector<s16>& samples) {
@@ -99,7 +99,8 @@ void Stream::PlayNextBuffer() {
 
     sink_stream.EnqueueSamples(GetNumChannels(), active_buffer->GetSamples());
 
-    CoreTiming::ScheduleEventThreadsafe(GetBufferReleaseCycles(*active_buffer), release_event, {});
+    Core::Timing::ScheduleEventThreadsafe(GetBufferReleaseCycles(*active_buffer), release_event,
+                                          {});
 }
 
 void Stream::ReleaseActiveBuffer() {

src/audio_core/stream.h

@@ -13,7 +13,7 @@
 #include "audio_core/buffer.h"
 #include "common/common_types.h"
 
-namespace CoreTiming {
+namespace Core::Timing {
 struct EventType;
 }
 
@@ -91,16 +91,16 @@ private:
     /// Gets the number of core cycles when the specified buffer will be released
     s64 GetBufferReleaseCycles(const Buffer& buffer) const;
 
-    u32 sample_rate;                        ///< Sample rate of the stream
-    Format format;                          ///< Format of the stream
-    ReleaseCallback release_callback;       ///< Buffer release callback for the stream
-    State state{State::Stopped};            ///< Playback state of the stream
-    CoreTiming::EventType* release_event{}; ///< Core timing release event for the stream
-    BufferPtr active_buffer;                ///< Actively playing buffer in the stream
-    std::queue<BufferPtr> queued_buffers;   ///< Buffers queued to be played in the stream
-    std::queue<BufferPtr> released_buffers; ///< Buffers recently released from the stream
-    SinkStream& sink_stream;                ///< Output sink for the stream
-    std::string name;                       ///< Name of the stream, must be unique
+    u32 sample_rate;                          ///< Sample rate of the stream
+    Format format;                            ///< Format of the stream
+    ReleaseCallback release_callback;         ///< Buffer release callback for the stream
+    State state{State::Stopped};              ///< Playback state of the stream
+    Core::Timing::EventType* release_event{}; ///< Core timing release event for the stream
+    BufferPtr active_buffer;                  ///< Actively playing buffer in the stream
+    std::queue<BufferPtr> queued_buffers;     ///< Buffers queued to be played in the stream
+    std::queue<BufferPtr> released_buffers;   ///< Buffers recently released from the stream
+    SinkStream& sink_stream;                  ///< Output sink for the stream
+    std::string name;                         ///< Name of the stream, must be unique
 };
 
 using StreamPtr = std::shared_ptr<Stream>;

src/core/arm/dynarmic/arm_dynarmic.cpp

@@ -112,14 +112,14 @@ public:
         // Always execute at least one tick.
         amortized_ticks = std::max<u64>(amortized_ticks, 1);
 
-        CoreTiming::AddTicks(amortized_ticks);
+        Timing::AddTicks(amortized_ticks);
         num_interpreted_instructions = 0;
     }
     u64 GetTicksRemaining() override {
-        return std::max(CoreTiming::GetDowncount(), 0);
+        return std::max(Timing::GetDowncount(), 0);
     }
     u64 GetCNTPCT() override {
-        return CoreTiming::GetTicks();
+        return Timing::GetTicks();
     }
 
     ARM_Dynarmic& parent;

src/core/arm/unicorn/arm_unicorn.cpp

@@ -177,7 +177,7 @@ void ARM_Unicorn::Run() {
     if (GDBStub::IsServerEnabled()) {
         ExecuteInstructions(std::max(4000000, 0));
     } else {
-        ExecuteInstructions(std::max(CoreTiming::GetDowncount(), 0));
+        ExecuteInstructions(std::max(Timing::GetDowncount(), 0));
     }
 }
 
@@ -190,7 +190,7 @@ MICROPROFILE_DEFINE(ARM_Jit_Unicorn, "ARM JIT", "Unicorn", MP_RGB(255, 64, 64));
 void ARM_Unicorn::ExecuteInstructions(int num_instructions) {
     MICROPROFILE_SCOPE(ARM_Jit_Unicorn);
     CHECKED(uc_emu_start(uc, GetPC(), 1ULL << 63, 0, num_instructions));
-    CoreTiming::AddTicks(num_instructions);
+    Timing::AddTicks(num_instructions);
     if (GDBStub::IsServerEnabled()) {
         if (last_bkpt_hit) {
             uc_reg_write(uc, UC_ARM64_REG_PC, &last_bkpt.address);

src/core/core.cpp

@@ -94,7 +94,7 @@ struct System::Impl {
     ResultStatus Init(System& system, Frontend::EmuWindow& emu_window) {
         LOG_DEBUG(HW_Memory, "initialized OK");
 
-        CoreTiming::Init();
+        Timing::Init();
         kernel.Initialize();
 
         const auto current_time = std::chrono::duration_cast<std::chrono::seconds>(
@@ -205,7 +205,7 @@ struct System::Impl {
 
         // Shutdown kernel and core timing
         kernel.Shutdown();
-        CoreTiming::Shutdown();
+        Timing::Shutdown();
 
         // Close app loader
         app_loader.reset();
@@ -232,7 +232,7 @@ struct System::Impl {
     }
 
     PerfStatsResults GetAndResetPerfStats() {
-        return perf_stats.GetAndResetStats(CoreTiming::GetGlobalTimeUs());
+        return perf_stats.GetAndResetStats(Timing::GetGlobalTimeUs());
     }
 
     Kernel::KernelCore kernel;

src/core/core_cpu.cpp

@@ -93,14 +93,14 @@ void Cpu::RunLoop(bool tight_loop) {
 
         if (IsMainCore()) {
             // TODO(Subv): Only let CoreTiming idle if all 4 cores are idling.
-            CoreTiming::Idle();
-            CoreTiming::Advance();
+            Timing::Idle();
+            Timing::Advance();
         }
 
         PrepareReschedule();
     } else {
         if (IsMainCore()) {
-            CoreTiming::Advance();
+            Timing::Advance();
         }
 
         if (tight_loop) {

src/core/core_timing.cpp

@@ -15,7 +15,7 @@
 #include "common/threadsafe_queue.h"
 #include "core/core_timing_util.h"
 
-namespace CoreTiming {
+namespace Core::Timing {
 
 static s64 global_timer;
 static int slice_length;
@@ -242,4 +242,4 @@ int GetDowncount() {
     return downcount;
 }
 
-} // namespace CoreTiming
+} // namespace Core::Timing

src/core/core_timing.h

@@ -22,7 +22,7 @@
 #include <string>
 #include "common/common_types.h"
 
-namespace CoreTiming {
+namespace Core::Timing {
 
 struct EventType;
 
@@ -92,4 +92,4 @@ std::chrono::microseconds GetGlobalTimeUs();
 
 int GetDowncount();
 
-} // namespace CoreTiming
+} // namespace Core::Timing

src/core/core_timing_util.cpp

@@ -8,7 +8,7 @@
 #include <limits>
 #include "common/logging/log.h"
 
-namespace CoreTiming {
+namespace Core::Timing {
 
 constexpr u64 MAX_VALUE_TO_MULTIPLY = std::numeric_limits<s64>::max() / BASE_CLOCK_RATE;
 
@@ -60,4 +60,4 @@ s64 nsToCycles(u64 ns) {
     return (BASE_CLOCK_RATE * static_cast<s64>(ns)) / 1000000000;
 }
 
-} // namespace CoreTiming
+} // namespace Core::Timing

src/core/core_timing_util.h

@@ -6,7 +6,7 @@
 
 #include "common/common_types.h"
 
-namespace CoreTiming {
+namespace Core::Timing {
 
 // The below clock rate is based on Switch's clockspeed being widely known as 1.020GHz
 // The exact value used is of course unverified.
@@ -61,4 +61,4 @@ inline u64 cyclesToMs(s64 cycles) {
     return cycles * 1000 / BASE_CLOCK_RATE;
 }
 
-} // namespace CoreTiming
+} // namespace Core::Timing

src/core/cpu_core_manager.cpp

@@ -37,10 +37,10 @@ void CpuCoreManager::Initialize(System& system) {
         return;
     }
 
-    for (std::size_t index = 1; index < core_threads.size(); index++) {
+    for (std::size_t index = 0; index < core_threads.size(); ++index) {
         core_threads[index] = std::make_unique<std::thread>(RunCpuCore, std::cref(system),
-                                                            std::ref(*cores[index]));
-        thread_to_cpu[core_threads[index]->get_id()] = cores[index].get();
+                                                            std::ref(*cores[index + 1]));
+        thread_to_cpu[core_threads[index]->get_id()] = cores[index + 1].get();
     }
 }
 

src/core/hle/kernel/kernel.cpp

@@ -124,7 +124,7 @@ struct KernelCore::Impl {
 
     void InitializeThreads() {
         thread_wakeup_event_type =
-            CoreTiming::RegisterEvent("ThreadWakeupCallback", ThreadWakeupCallback);
+            Core::Timing::RegisterEvent("ThreadWakeupCallback", ThreadWakeupCallback);
     }
 
     std::atomic<u32> next_object_id{0};
@@ -137,7 +137,7 @@ struct KernelCore::Impl {
 
     SharedPtr<ResourceLimit> system_resource_limit;
 
-    CoreTiming::EventType* thread_wakeup_event_type = nullptr;
+    Core::Timing::EventType* thread_wakeup_event_type = nullptr;
     // TODO(yuriks): This can be removed if Thread objects are explicitly pooled in the future,
     // allowing us to simply use a pool index or similar.
     Kernel::HandleTable thread_wakeup_callback_handle_table;
@@ -213,7 +213,7 @@ u64 KernelCore::CreateNewProcessID() {
     return impl->next_process_id++;
 }
 
-CoreTiming::EventType* KernelCore::ThreadWakeupCallbackEventType() const {
+Core::Timing::EventType* KernelCore::ThreadWakeupCallbackEventType() const {
     return impl->thread_wakeup_event_type;
 }
 

src/core/hle/kernel/kernel.h

@@ -11,7 +11,7 @@
 template <typename T>
 class ResultVal;
 
-namespace CoreTiming {
+namespace Core::Timing {
 struct EventType;
 }
 
@@ -89,7 +89,7 @@ private:
     u64 CreateNewThreadID();
 
     /// Retrieves the event type used for thread wakeup callbacks.
-    CoreTiming::EventType* ThreadWakeupCallbackEventType() const;
+    Core::Timing::EventType* ThreadWakeupCallbackEventType() const;
 
     /// Provides a reference to the thread wakeup callback handle table.
     Kernel::HandleTable& ThreadWakeupCallbackHandleTable();

src/core/hle/kernel/scheduler.cpp

@@ -111,7 +111,7 @@ void Scheduler::SwitchContext(Thread* new_thread) {
 
 void Scheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) {
     const u64 prev_switch_ticks = last_context_switch_time;
-    const u64 most_recent_switch_ticks = CoreTiming::GetTicks();
+    const u64 most_recent_switch_ticks = Core::Timing::GetTicks();
     const u64 update_ticks = most_recent_switch_ticks - prev_switch_ticks;
 
     if (thread != nullptr) {

src/core/hle/kernel/svc.cpp

@@ -927,9 +927,9 @@ static ResultCode GetInfo(u64* result, u64 info_id, u64 handle, u64 info_sub_id)
         if (same_thread && info_sub_id == 0xFFFFFFFFFFFFFFFF) {
             const u64 thread_ticks = current_thread->GetTotalCPUTimeTicks();
 
-            out_ticks = thread_ticks + (CoreTiming::GetTicks() - prev_ctx_ticks);
+            out_ticks = thread_ticks + (Core::Timing::GetTicks() - prev_ctx_ticks);
         } else if (same_thread && info_sub_id == system.CurrentCoreIndex()) {
-            out_ticks = CoreTiming::GetTicks() - prev_ctx_ticks;
+            out_ticks = Core::Timing::GetTicks() - prev_ctx_ticks;
         }
 
         *result = out_ticks;
@@ -1546,10 +1546,10 @@ static ResultCode SignalToAddress(VAddr address, u32 type, s32 value, s32 num_to
 static u64 GetSystemTick() {
     LOG_TRACE(Kernel_SVC, "called");
 
-    const u64 result{CoreTiming::GetTicks()};
+    const u64 result{Core::Timing::GetTicks()};
 
     // Advance time to defeat dumb games that busy-wait for the frame to end.
-    CoreTiming::AddTicks(400);
+    Core::Timing::AddTicks(400);
 
     return result;
 }

src/core/hle/kernel/thread.cpp

@@ -43,7 +43,7 @@ Thread::~Thread() = default;
 
 void Thread::Stop() {
     // Cancel any outstanding wakeup events for this thread
-    CoreTiming::UnscheduleEvent(kernel.ThreadWakeupCallbackEventType(), callback_handle);
+    Core::Timing::UnscheduleEvent(kernel.ThreadWakeupCallbackEventType(), callback_handle);
     kernel.ThreadWakeupCallbackHandleTable().Close(callback_handle);
     callback_handle = 0;
 
@@ -85,12 +85,13 @@ void Thread::WakeAfterDelay(s64 nanoseconds) {
 
     // This function might be called from any thread so we have to be cautious and use the
     // thread-safe version of ScheduleEvent.
-    CoreTiming::ScheduleEventThreadsafe(CoreTiming::nsToCycles(nanoseconds),
-                                        kernel.ThreadWakeupCallbackEventType(), callback_handle);
+    Core::Timing::ScheduleEventThreadsafe(Core::Timing::nsToCycles(nanoseconds),
+                                          kernel.ThreadWakeupCallbackEventType(), callback_handle);
 }
 
 void Thread::CancelWakeupTimer() {
-    CoreTiming::UnscheduleEventThreadsafe(kernel.ThreadWakeupCallbackEventType(), callback_handle);
+    Core::Timing::UnscheduleEventThreadsafe(kernel.ThreadWakeupCallbackEventType(),
+                                            callback_handle);
 }
 
 static std::optional<s32> GetNextProcessorId(u64 mask) {
@@ -197,7 +198,7 @@ ResultVal<SharedPtr<Thread>> Thread::Create(KernelCore& kernel, std::string name
     thread->stack_top = stack_top;
     thread->tpidr_el0 = 0;
     thread->nominal_priority = thread->current_priority = priority;
-    thread->last_running_ticks = CoreTiming::GetTicks();
+    thread->last_running_ticks = Core::Timing::GetTicks();
     thread->processor_id = processor_id;
     thread->ideal_core = processor_id;
     thread->affinity_mask = 1ULL << processor_id;
@@ -257,7 +258,7 @@ void Thread::SetStatus(ThreadStatus new_status) {
     }
 
     if (status == ThreadStatus::Running) {
-        last_running_ticks = CoreTiming::GetTicks();
+        last_running_ticks = Core::Timing::GetTicks();
     }
 
     status = new_status;

src/core/hle/service/hid/controllers/debug_pad.cpp

@@ -22,7 +22,7 @@ void Controller_DebugPad::OnInit() {}
 void Controller_DebugPad::OnRelease() {}
 
 void Controller_DebugPad::OnUpdate(u8* data, std::size_t size) {
-    shared_memory.header.timestamp = CoreTiming::GetTicks();
+    shared_memory.header.timestamp = Core::Timing::GetTicks();
     shared_memory.header.total_entry_count = 17;
 
     if (!IsControllerActivated()) {

src/core/hle/service/hid/controllers/gesture.cpp

@@ -18,7 +18,7 @@ void Controller_Gesture::OnInit() {}
 void Controller_Gesture::OnRelease() {}
 
 void Controller_Gesture::OnUpdate(u8* data, std::size_t size) {
-    shared_memory.header.timestamp = CoreTiming::GetTicks();
+    shared_memory.header.timestamp = Core::Timing::GetTicks();
     shared_memory.header.total_entry_count = 17;
 
     if (!IsControllerActivated()) {

src/core/hle/service/hid/controllers/keyboard.cpp

@@ -20,7 +20,7 @@ void Controller_Keyboard::OnInit() {}
 void Controller_Keyboard::OnRelease() {}
 
 void Controller_Keyboard::OnUpdate(u8* data, std::size_t size) {
-    shared_memory.header.timestamp = CoreTiming::GetTicks();
+    shared_memory.header.timestamp = Core::Timing::GetTicks();
     shared_memory.header.total_entry_count = 17;
 
     if (!IsControllerActivated()) {

src/core/hle/service/hid/controllers/mouse.cpp

@@ -18,7 +18,7 @@ void Controller_Mouse::OnInit() {}
 void Controller_Mouse::OnRelease() {}
 
 void Controller_Mouse::OnUpdate(u8* data, std::size_t size) {
-    shared_memory.header.timestamp = CoreTiming::GetTicks();
+    shared_memory.header.timestamp = Core::Timing::GetTicks();
     shared_memory.header.total_entry_count = 17;
 
     if (!IsControllerActivated()) {

src/core/hle/service/hid/controllers/npad.cpp

@@ -308,7 +308,7 @@ void Controller_NPad::OnUpdate(u8* data, std::size_t data_len) {
             const auto& last_entry =
                 main_controller->npad[main_controller->common.last_entry_index];
 
-            main_controller->common.timestamp = CoreTiming::GetTicks();
+            main_controller->common.timestamp = Core::Timing::GetTicks();
             main_controller->common.last_entry_index =
                 (main_controller->common.last_entry_index + 1) % 17;
 

src/core/hle/service/hid/controllers/stubbed.cpp

@@ -22,7 +22,7 @@ void Controller_Stubbed::OnUpdate(u8* data, std::size_t size) {
     }
 
     CommonHeader header{};
-    header.timestamp = CoreTiming::GetTicks();
+    header.timestamp = Core::Timing::GetTicks();
     header.total_entry_count = 17;
     header.entry_count = 0;
     header.last_entry_index = 0;

src/core/hle/service/hid/controllers/touchscreen.cpp

@@ -21,7 +21,7 @@ void Controller_Touchscreen::OnInit() {}
 void Controller_Touchscreen::OnRelease() {}
 
 void Controller_Touchscreen::OnUpdate(u8* data, std::size_t size) {
-    shared_memory.header.timestamp = CoreTiming::GetTicks();
+    shared_memory.header.timestamp = Core::Timing::GetTicks();
     shared_memory.header.total_entry_count = 17;
 
     if (!IsControllerActivated()) {
@@ -48,7 +48,7 @@ void Controller_Touchscreen::OnUpdate(u8* data, std::size_t size) {
         touch_entry.diameter_x = Settings::values.touchscreen.diameter_x;
         touch_entry.diameter_y = Settings::values.touchscreen.diameter_y;
         touch_entry.rotation_angle = Settings::values.touchscreen.rotation_angle;
-        const u64 tick = CoreTiming::GetTicks();
+        const u64 tick = Core::Timing::GetTicks();
         touch_entry.delta_time = tick - last_touch;
         last_touch = tick;
         touch_entry.finger = Settings::values.touchscreen.finger;

src/core/hle/service/hid/controllers/xpad.cpp

@@ -19,7 +19,7 @@ void Controller_XPad::OnRelease() {}
 
 void Controller_XPad::OnUpdate(u8* data, std::size_t size) {
     for (auto& xpad_entry : shared_memory.shared_memory_entries) {
-        xpad_entry.header.timestamp = CoreTiming::GetTicks();
+        xpad_entry.header.timestamp = Core::Timing::GetTicks();
         xpad_entry.header.total_entry_count = 17;
 
         if (!IsControllerActivated()) {

src/core/hle/service/hid/hid.cpp

@@ -36,9 +36,9 @@ namespace Service::HID {
 
 // Updating period for each HID device.
 // TODO(ogniK): Find actual polling rate of hid
-constexpr u64 pad_update_ticks = CoreTiming::BASE_CLOCK_RATE / 66;
-constexpr u64 accelerometer_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
-constexpr u64 gyroscope_update_ticks = CoreTiming::BASE_CLOCK_RATE / 100;
+constexpr u64 pad_update_ticks = Core::Timing::BASE_CLOCK_RATE / 66;
+constexpr u64 accelerometer_update_ticks = Core::Timing::BASE_CLOCK_RATE / 100;
+constexpr u64 gyroscope_update_ticks = Core::Timing::BASE_CLOCK_RATE / 100;
 constexpr std::size_t SHARED_MEMORY_SIZE = 0x40000;
 
 IAppletResource::IAppletResource() : ServiceFramework("IAppletResource") {
@@ -73,14 +73,13 @@ IAppletResource::IAppletResource() : ServiceFramework("IAppletResource") {
     GetController<Controller_Stubbed>(HidController::Unknown3).SetCommonHeaderOffset(0x5000);
 
     // Register update callbacks
-    pad_update_event =
-        CoreTiming::RegisterEvent("HID::UpdatePadCallback", [this](u64 userdata, int cycles_late) {
-            UpdateControllers(userdata, cycles_late);
-        });
+    pad_update_event = Core::Timing::RegisterEvent(
+        "HID::UpdatePadCallback",
+        [this](u64 userdata, int cycles_late) { UpdateControllers(userdata, cycles_late); });
 
     // TODO(shinyquagsire23): Other update callbacks? (accel, gyro?)
 
-    CoreTiming::ScheduleEvent(pad_update_ticks, pad_update_event);
+    Core::Timing::ScheduleEvent(pad_update_ticks, pad_update_event);
 
     ReloadInputDevices();
 }
@@ -94,7 +93,7 @@ void IAppletResource::DeactivateController(HidController controller) {
 }
 
 IAppletResource ::~IAppletResource() {
-    CoreTiming::UnscheduleEvent(pad_update_event, 0);
+    Core::Timing::UnscheduleEvent(pad_update_event, 0);
 }
 
 void IAppletResource::GetSharedMemoryHandle(Kernel::HLERequestContext& ctx) {
@@ -114,7 +113,7 @@ void IAppletResource::UpdateControllers(u64 userdata, int cycles_late) {
         controller->OnUpdate(shared_mem->GetPointer(), SHARED_MEMORY_SIZE);
     }
 
-    CoreTiming::ScheduleEvent(pad_update_ticks - cycles_late, pad_update_event);
+    Core::Timing::ScheduleEvent(pad_update_ticks - cycles_late, pad_update_event);
 }
 
 class IActiveVibrationDeviceList final : public ServiceFramework<IActiveVibrationDeviceList> {

src/core/hle/service/hid/hid.h

@@ -7,7 +7,7 @@
 #include "controllers/controller_base.h"
 #include "core/hle/service/service.h"
 
-namespace CoreTiming {
+namespace Core::Timing {
 struct EventType;
 }
 
@@ -66,7 +66,7 @@ private:
 
     Kernel::SharedPtr<Kernel::SharedMemory> shared_mem;
 
-    CoreTiming::EventType* pad_update_event;
+    Core::Timing::EventType* pad_update_event;
 
     std::array<std::unique_ptr<ControllerBase>, static_cast<size_t>(HidController::MaxControllers)>
         controllers{};

src/core/hle/service/hid/irs.cpp

@@ -98,7 +98,7 @@ void IRS::GetImageTransferProcessorState(Kernel::HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 5};
     rb.Push(RESULT_SUCCESS);
-    rb.PushRaw<u64>(CoreTiming::GetTicks());
+    rb.PushRaw<u64>(Core::Timing::GetTicks());
     rb.PushRaw<u32>(0);
 }
 

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

@@ -184,7 +184,7 @@ u32 nvhost_ctrl_gpu::GetGpuTime(const std::vector<u8>& input, std::vector<u8>& o
 
     IoctlGetGpuTime params{};
     std::memcpy(&params, input.data(), input.size());
-    params.gpu_time = CoreTiming::cyclesToNs(CoreTiming::GetTicks());
+    params.gpu_time = Core::Timing::cyclesToNs(Core::Timing::GetTicks());
     std::memcpy(output.data(), &params, output.size());
     return 0;
 }

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

@@ -13,10 +13,6 @@
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/writable_event.h"
 
-namespace CoreTiming {
-struct EventType;
-}
-
 namespace Service::NVFlinger {
 
 struct IGBPBuffer {

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

@@ -25,21 +25,21 @@
 namespace Service::NVFlinger {
 
 constexpr std::size_t SCREEN_REFRESH_RATE = 60;
-constexpr u64 frame_ticks = static_cast<u64>(CoreTiming::BASE_CLOCK_RATE / SCREEN_REFRESH_RATE);
+constexpr u64 frame_ticks = static_cast<u64>(Core::Timing::BASE_CLOCK_RATE / SCREEN_REFRESH_RATE);
 
 NVFlinger::NVFlinger() {
     // Schedule the screen composition events
     composition_event =
-        CoreTiming::RegisterEvent("ScreenComposition", [this](u64 userdata, int cycles_late) {
+        Core::Timing::RegisterEvent("ScreenComposition", [this](u64 userdata, int cycles_late) {
             Compose();
-            CoreTiming::ScheduleEvent(frame_ticks - cycles_late, composition_event);
+            Core::Timing::ScheduleEvent(frame_ticks - cycles_late, composition_event);
         });
 
-    CoreTiming::ScheduleEvent(frame_ticks, composition_event);
+    Core::Timing::ScheduleEvent(frame_ticks, composition_event);
 }
 
 NVFlinger::~NVFlinger() {
-    CoreTiming::UnscheduleEvent(composition_event, 0);
+    Core::Timing::UnscheduleEvent(composition_event, 0);
 }
 
 void NVFlinger::SetNVDrvInstance(std::shared_ptr<Nvidia::Module> instance) {

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

@@ -14,7 +14,7 @@
 #include "common/common_types.h"
 #include "core/hle/kernel/object.h"
 
-namespace CoreTiming {
+namespace Core::Timing {
 struct EventType;
 }
 
@@ -115,8 +115,8 @@ private:
     /// layers.
     u32 next_buffer_queue_id = 1;
 
-    /// CoreTiming event that handles screen composition.
-    CoreTiming::EventType* composition_event;
+    /// Event that handles screen composition.
+    Core::Timing::EventType* composition_event;
 };
 
 } // namespace Service::NVFlinger

src/core/hle/service/time/time.cpp

@@ -106,8 +106,8 @@ private:
     void GetCurrentTimePoint(Kernel::HLERequestContext& ctx) {
         LOG_DEBUG(Service_Time, "called");
 
-        SteadyClockTimePoint steady_clock_time_point{
-            CoreTiming::cyclesToMs(CoreTiming::GetTicks()) / 1000};
+        const SteadyClockTimePoint steady_clock_time_point{
+            Core::Timing::cyclesToMs(Core::Timing::GetTicks()) / 1000};
         IPC::ResponseBuilder rb{ctx, (sizeof(SteadyClockTimePoint) / 4) + 2};
         rb.Push(RESULT_SUCCESS);
         rb.PushRaw(steady_clock_time_point);
@@ -282,7 +282,7 @@ void Module::Interface::GetClockSnapshot(Kernel::HLERequestContext& ctx) {
     }
 
     const SteadyClockTimePoint steady_clock_time_point{
-        CoreTiming::cyclesToMs(CoreTiming::GetTicks()) / 1000, {}};
+        Core::Timing::cyclesToMs(Core::Timing::GetTicks()) / 1000, {}};
 
     CalendarTime calendar_time{};
     calendar_time.year = tm->tm_year + 1900;

src/tests/core/core_timing.cpp

@@ -31,10 +31,10 @@ void CallbackTemplate(u64 userdata, s64 cycles_late) {
 class ScopeInit final {
 public:
     ScopeInit() {
-        CoreTiming::Init();
+        Core::Timing::Init();
     }
     ~ScopeInit() {
-        CoreTiming::Shutdown();
+        Core::Timing::Shutdown();
     }
 };
 
@@ -44,37 +44,37 @@ static void AdvanceAndCheck(u32 idx, int downcount, int expected_lateness = 0,
     expected_callback = CB_IDS[idx];
     lateness = expected_lateness;
 
-    CoreTiming::AddTicks(CoreTiming::GetDowncount() -
-                         cpu_downcount); // Pretend we executed X cycles of instructions.
-    CoreTiming::Advance();
+    // Pretend we executed X cycles of instructions.
+    Core::Timing::AddTicks(Core::Timing::GetDowncount() - cpu_downcount);
+    Core::Timing::Advance();
 
     REQUIRE(decltype(callbacks_ran_flags)().set(idx) == callbacks_ran_flags);
-    REQUIRE(downcount == CoreTiming::GetDowncount());
+    REQUIRE(downcount == Core::Timing::GetDowncount());
 }
 
 TEST_CASE("CoreTiming[BasicOrder]", "[core]") {
     ScopeInit guard;
 
-    CoreTiming::EventType* cb_a = CoreTiming::RegisterEvent("callbackA", CallbackTemplate<0>);
-    CoreTiming::EventType* cb_b = CoreTiming::RegisterEvent("callbackB", CallbackTemplate<1>);
-    CoreTiming::EventType* cb_c = CoreTiming::RegisterEvent("callbackC", CallbackTemplate<2>);
-    CoreTiming::EventType* cb_d = CoreTiming::RegisterEvent("callbackD", CallbackTemplate<3>);
-    CoreTiming::EventType* cb_e = CoreTiming::RegisterEvent("callbackE", CallbackTemplate<4>);
+    Core::Timing::EventType* cb_a = Core::Timing::RegisterEvent("callbackA", CallbackTemplate<0>);
+    Core::Timing::EventType* cb_b = Core::Timing::RegisterEvent("callbackB", CallbackTemplate<1>);
+    Core::Timing::EventType* cb_c = Core::Timing::RegisterEvent("callbackC", CallbackTemplate<2>);
+    Core::Timing::EventType* cb_d = Core::Timing::RegisterEvent("callbackD", CallbackTemplate<3>);
+    Core::Timing::EventType* cb_e = Core::Timing::RegisterEvent("callbackE", CallbackTemplate<4>);
 
     // Enter slice 0
-    CoreTiming::Advance();
+    Core::Timing::Advance();
 
     // D -> B -> C -> A -> E
-    CoreTiming::ScheduleEvent(1000, cb_a, CB_IDS[0]);
-    REQUIRE(1000 == CoreTiming::GetDowncount());
-    CoreTiming::ScheduleEvent(500, cb_b, CB_IDS[1]);
-    REQUIRE(500 == CoreTiming::GetDowncount());
-    CoreTiming::ScheduleEvent(800, cb_c, CB_IDS[2]);
-    REQUIRE(500 == CoreTiming::GetDowncount());
-    CoreTiming::ScheduleEvent(100, cb_d, CB_IDS[3]);
-    REQUIRE(100 == CoreTiming::GetDowncount());
-    CoreTiming::ScheduleEvent(1200, cb_e, CB_IDS[4]);
-    REQUIRE(100 == CoreTiming::GetDowncount());
+    Core::Timing::ScheduleEvent(1000, cb_a, CB_IDS[0]);
+    REQUIRE(1000 == Core::Timing::GetDowncount());
+    Core::Timing::ScheduleEvent(500, cb_b, CB_IDS[1]);
+    REQUIRE(500 == Core::Timing::GetDowncount());
+    Core::Timing::ScheduleEvent(800, cb_c, CB_IDS[2]);
+    REQUIRE(500 == Core::Timing::GetDowncount());
+    Core::Timing::ScheduleEvent(100, cb_d, CB_IDS[3]);
+    REQUIRE(100 == Core::Timing::GetDowncount());
+    Core::Timing::ScheduleEvent(1200, cb_e, CB_IDS[4]);
+    REQUIRE(100 == Core::Timing::GetDowncount());
 
     AdvanceAndCheck(3, 400);
     AdvanceAndCheck(1, 300);
@@ -86,36 +86,36 @@ TEST_CASE("CoreTiming[BasicOrder]", "[core]") {
 TEST_CASE("CoreTiming[Threadsave]", "[core]") {
     ScopeInit guard;
 
-    CoreTiming::EventType* cb_a = CoreTiming::RegisterEvent("callbackA", CallbackTemplate<0>);
-    CoreTiming::EventType* cb_b = CoreTiming::RegisterEvent("callbackB", CallbackTemplate<1>);
-    CoreTiming::EventType* cb_c = CoreTiming::RegisterEvent("callbackC", CallbackTemplate<2>);
-    CoreTiming::EventType* cb_d = CoreTiming::RegisterEvent("callbackD", CallbackTemplate<3>);
-    CoreTiming::EventType* cb_e = CoreTiming::RegisterEvent("callbackE", CallbackTemplate<4>);
+    Core::Timing::EventType* cb_a = Core::Timing::RegisterEvent("callbackA", CallbackTemplate<0>);
+    Core::Timing::EventType* cb_b = Core::Timing::RegisterEvent("callbackB", CallbackTemplate<1>);
+    Core::Timing::EventType* cb_c = Core::Timing::RegisterEvent("callbackC", CallbackTemplate<2>);
+    Core::Timing::EventType* cb_d = Core::Timing::RegisterEvent("callbackD", CallbackTemplate<3>);
+    Core::Timing::EventType* cb_e = Core::Timing::RegisterEvent("callbackE", CallbackTemplate<4>);
 
     // Enter slice 0
-    CoreTiming::Advance();
+    Core::Timing::Advance();
 
     // D -> B -> C -> A -> E
-    CoreTiming::ScheduleEventThreadsafe(1000, cb_a, CB_IDS[0]);
+    Core::Timing::ScheduleEventThreadsafe(1000, cb_a, CB_IDS[0]);
     // Manually force since ScheduleEventThreadsafe doesn't call it
-    CoreTiming::ForceExceptionCheck(1000);
-    REQUIRE(1000 == CoreTiming::GetDowncount());
-    CoreTiming::ScheduleEventThreadsafe(500, cb_b, CB_IDS[1]);
+    Core::Timing::ForceExceptionCheck(1000);
+    REQUIRE(1000 == Core::Timing::GetDowncount());
+    Core::Timing::ScheduleEventThreadsafe(500, cb_b, CB_IDS[1]);
     // Manually force since ScheduleEventThreadsafe doesn't call it
-    CoreTiming::ForceExceptionCheck(500);
-    REQUIRE(500 == CoreTiming::GetDowncount());
-    CoreTiming::ScheduleEventThreadsafe(800, cb_c, CB_IDS[2]);
+    Core::Timing::ForceExceptionCheck(500);
+    REQUIRE(500 == Core::Timing::GetDowncount());
+    Core::Timing::ScheduleEventThreadsafe(800, cb_c, CB_IDS[2]);
     // Manually force since ScheduleEventThreadsafe doesn't call it
-    CoreTiming::ForceExceptionCheck(800);
-    REQUIRE(500 == CoreTiming::GetDowncount());
-    CoreTiming::ScheduleEventThreadsafe(100, cb_d, CB_IDS[3]);
+    Core::Timing::ForceExceptionCheck(800);
+    REQUIRE(500 == Core::Timing::GetDowncount());
+    Core::Timing::ScheduleEventThreadsafe(100, cb_d, CB_IDS[3]);
     // Manually force since ScheduleEventThreadsafe doesn't call it
-    CoreTiming::ForceExceptionCheck(100);
-    REQUIRE(100 == CoreTiming::GetDowncount());
-    CoreTiming::ScheduleEventThreadsafe(1200, cb_e, CB_IDS[4]);
+    Core::Timing::ForceExceptionCheck(100);
+    REQUIRE(100 == Core::Timing::GetDowncount());
+    Core::Timing::ScheduleEventThreadsafe(1200, cb_e, CB_IDS[4]);
     // Manually force since ScheduleEventThreadsafe doesn't call it
-    CoreTiming::ForceExceptionCheck(1200);
-    REQUIRE(100 == CoreTiming::GetDowncount());
+    Core::Timing::ForceExceptionCheck(1200);
+    REQUIRE(100 == Core::Timing::GetDowncount());
 
     AdvanceAndCheck(3, 400);
     AdvanceAndCheck(1, 300);
@@ -143,42 +143,42 @@ TEST_CASE("CoreTiming[SharedSlot]", "[core]") {
 
     ScopeInit guard;
 
-    CoreTiming::EventType* cb_a = CoreTiming::RegisterEvent("callbackA", FifoCallback<0>);
-    CoreTiming::EventType* cb_b = CoreTiming::RegisterEvent("callbackB", FifoCallback<1>);
-    CoreTiming::EventType* cb_c = CoreTiming::RegisterEvent("callbackC", FifoCallback<2>);
-    CoreTiming::EventType* cb_d = CoreTiming::RegisterEvent("callbackD", FifoCallback<3>);
-    CoreTiming::EventType* cb_e = CoreTiming::RegisterEvent("callbackE", FifoCallback<4>);
+    Core::Timing::EventType* cb_a = Core::Timing::RegisterEvent("callbackA", FifoCallback<0>);
+    Core::Timing::EventType* cb_b = Core::Timing::RegisterEvent("callbackB", FifoCallback<1>);
+    Core::Timing::EventType* cb_c = Core::Timing::RegisterEvent("callbackC", FifoCallback<2>);
+    Core::Timing::EventType* cb_d = Core::Timing::RegisterEvent("callbackD", FifoCallback<3>);
+    Core::Timing::EventType* cb_e = Core::Timing::RegisterEvent("callbackE", FifoCallback<4>);
 
-    CoreTiming::ScheduleEvent(1000, cb_a, CB_IDS[0]);
-    CoreTiming::ScheduleEvent(1000, cb_b, CB_IDS[1]);
-    CoreTiming::ScheduleEvent(1000, cb_c, CB_IDS[2]);
-    CoreTiming::ScheduleEvent(1000, cb_d, CB_IDS[3]);
-    CoreTiming::ScheduleEvent(1000, cb_e, CB_IDS[4]);
+    Core::Timing::ScheduleEvent(1000, cb_a, CB_IDS[0]);
+    Core::Timing::ScheduleEvent(1000, cb_b, CB_IDS[1]);
+    Core::Timing::ScheduleEvent(1000, cb_c, CB_IDS[2]);
+    Core::Timing::ScheduleEvent(1000, cb_d, CB_IDS[3]);
+    Core::Timing::ScheduleEvent(1000, cb_e, CB_IDS[4]);
 
     // Enter slice 0
-    CoreTiming::Advance();
-    REQUIRE(1000 == CoreTiming::GetDowncount());
+    Core::Timing::Advance();
+    REQUIRE(1000 == Core::Timing::GetDowncount());
 
     callbacks_ran_flags = 0;
     counter = 0;
     lateness = 0;
-    CoreTiming::AddTicks(CoreTiming::GetDowncount());
-    CoreTiming::Advance();
-    REQUIRE(MAX_SLICE_LENGTH == CoreTiming::GetDowncount());
+    Core::Timing::AddTicks(Core::Timing::GetDowncount());
+    Core::Timing::Advance();
+    REQUIRE(MAX_SLICE_LENGTH == Core::Timing::GetDowncount());
     REQUIRE(0x1FULL == callbacks_ran_flags.to_ullong());
 }
 
-TEST_CASE("CoreTiming[PredictableLateness]", "[core]") {
+TEST_CASE("Core::Timing[PredictableLateness]", "[core]") {
     ScopeInit guard;
 
-    CoreTiming::EventType* cb_a = CoreTiming::RegisterEvent("callbackA", CallbackTemplate<0>);
-    CoreTiming::EventType* cb_b = CoreTiming::RegisterEvent("callbackB", CallbackTemplate<1>);
+    Core::Timing::EventType* cb_a = Core::Timing::RegisterEvent("callbackA", CallbackTemplate<0>);
+    Core::Timing::EventType* cb_b = Core::Timing::RegisterEvent("callbackB", CallbackTemplate<1>);
 
     // Enter slice 0
-    CoreTiming::Advance();
+    Core::Timing::Advance();
 
-    CoreTiming::ScheduleEvent(100, cb_a, CB_IDS[0]);
-    CoreTiming::ScheduleEvent(200, cb_b, CB_IDS[1]);
+    Core::Timing::ScheduleEvent(100, cb_a, CB_IDS[0]);
+    Core::Timing::ScheduleEvent(200, cb_b, CB_IDS[1]);
 
     AdvanceAndCheck(0, 90, 10, -10); // (100 - 10)
     AdvanceAndCheck(1, MAX_SLICE_LENGTH, 50, -50);
@@ -192,9 +192,10 @@ static void RescheduleCallback(u64 userdata, s64 cycles_late) {
     REQUIRE(reschedules >= 0);
     REQUIRE(lateness == cycles_late);
 
-    if (reschedules > 0)
-        CoreTiming::ScheduleEvent(1000, reinterpret_cast<CoreTiming::EventType*>(userdata),
-                                  userdata);
+    if (reschedules > 0) {
+        Core::Timing::ScheduleEvent(1000, reinterpret_cast<Core::Timing::EventType*>(userdata),
+                                    userdata);
+    }
 }
 } // namespace ChainSchedulingTest
 
@@ -203,35 +204,35 @@ TEST_CASE("CoreTiming[ChainScheduling]", "[core]") {
 
     ScopeInit guard;
 
-    CoreTiming::EventType* cb_a = CoreTiming::RegisterEvent("callbackA", CallbackTemplate<0>);
-    CoreTiming::EventType* cb_b = CoreTiming::RegisterEvent("callbackB", CallbackTemplate<1>);
-    CoreTiming::EventType* cb_c = CoreTiming::RegisterEvent("callbackC", CallbackTemplate<2>);
-    CoreTiming::EventType* cb_rs =
-        CoreTiming::RegisterEvent("callbackReschedule", RescheduleCallback);
+    Core::Timing::EventType* cb_a = Core::Timing::RegisterEvent("callbackA", CallbackTemplate<0>);
+    Core::Timing::EventType* cb_b = Core::Timing::RegisterEvent("callbackB", CallbackTemplate<1>);
+    Core::Timing::EventType* cb_c = Core::Timing::RegisterEvent("callbackC", CallbackTemplate<2>);
+    Core::Timing::EventType* cb_rs =
+        Core::Timing::RegisterEvent("callbackReschedule", RescheduleCallback);
 
     // Enter slice 0
-    CoreTiming::Advance();
+    Core::Timing::Advance();
 
-    CoreTiming::ScheduleEvent(800, cb_a, CB_IDS[0]);
-    CoreTiming::ScheduleEvent(1000, cb_b, CB_IDS[1]);
-    CoreTiming::ScheduleEvent(2200, cb_c, CB_IDS[2]);
-    CoreTiming::ScheduleEvent(1000, cb_rs, reinterpret_cast<u64>(cb_rs));
-    REQUIRE(800 == CoreTiming::GetDowncount());
+    Core::Timing::ScheduleEvent(800, cb_a, CB_IDS[0]);
+    Core::Timing::ScheduleEvent(1000, cb_b, CB_IDS[1]);
+    Core::Timing::ScheduleEvent(2200, cb_c, CB_IDS[2]);
+    Core::Timing::ScheduleEvent(1000, cb_rs, reinterpret_cast<u64>(cb_rs));
+    REQUIRE(800 == Core::Timing::GetDowncount());
 
     reschedules = 3;
     AdvanceAndCheck(0, 200);  // cb_a
     AdvanceAndCheck(1, 1000); // cb_b, cb_rs
     REQUIRE(2 == reschedules);
 
-    CoreTiming::AddTicks(CoreTiming::GetDowncount());
-    CoreTiming::Advance(); // cb_rs
+    Core::Timing::AddTicks(Core::Timing::GetDowncount());
+    Core::Timing::Advance(); // cb_rs
     REQUIRE(1 == reschedules);
-    REQUIRE(200 == CoreTiming::GetDowncount());
+    REQUIRE(200 == Core::Timing::GetDowncount());
 
     AdvanceAndCheck(2, 800); // cb_c
 
-    CoreTiming::AddTicks(CoreTiming::GetDowncount());
-    CoreTiming::Advance(); // cb_rs
+    Core::Timing::AddTicks(Core::Timing::GetDowncount());
+    Core::Timing::Advance(); // cb_rs
     REQUIRE(0 == reschedules);
-    REQUIRE(MAX_SLICE_LENGTH == CoreTiming::GetDowncount());
+    REQUIRE(MAX_SLICE_LENGTH == Core::Timing::GetDowncount());
 }

src/video_core/CMakeLists.txt

@@ -5,12 +5,12 @@ add_library(video_core STATIC
     debug_utils/debug_utils.h
     engines/fermi_2d.cpp
     engines/fermi_2d.h
+    engines/kepler_compute.cpp
+    engines/kepler_compute.h
     engines/kepler_memory.cpp
     engines/kepler_memory.h
     engines/maxwell_3d.cpp
     engines/maxwell_3d.h
-    engines/maxwell_compute.cpp
-    engines/maxwell_compute.h
     engines/maxwell_dma.cpp
     engines/maxwell_dma.h
     engines/shader_bytecode.h

src/video_core/engines/kepler_compute.cpp

@@ -0,0 +1,34 @@
+// Copyright 2018 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/logging/log.h"
+#include "core/core.h"
+#include "core/memory.h"
+#include "video_core/engines/kepler_compute.h"
+#include "video_core/memory_manager.h"
+
+namespace Tegra::Engines {
+
+KeplerCompute::KeplerCompute(MemoryManager& memory_manager) : memory_manager{memory_manager} {}
+
+KeplerCompute::~KeplerCompute() = default;
+
+void KeplerCompute::CallMethod(const GPU::MethodCall& method_call) {
+    ASSERT_MSG(method_call.method < Regs::NUM_REGS,
+               "Invalid KeplerCompute register, increase the size of the Regs structure");
+
+    regs.reg_array[method_call.method] = method_call.argument;
+
+    switch (method_call.method) {
+    case KEPLER_COMPUTE_REG_INDEX(launch):
+        // Abort execution since compute shaders can be used to alter game memory (e.g. CUDA
+        // kernels)
+        UNREACHABLE_MSG("Compute shaders are not implemented");
+        break;
+    default:
+        break;
+    }
+}
+
+} // namespace Tegra::Engines

src/video_core/engines/kepler_compute.h

@@ -10,47 +10,48 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "video_core/gpu.h"
+#include "video_core/memory_manager.h"
 
 namespace Tegra::Engines {
 
-#define MAXWELL_COMPUTE_REG_INDEX(field_name)                                                      \
-    (offsetof(Tegra::Engines::MaxwellCompute::Regs, field_name) / sizeof(u32))
+#define KEPLER_COMPUTE_REG_INDEX(field_name)                                                       \
+    (offsetof(Tegra::Engines::KeplerCompute::Regs, field_name) / sizeof(u32))
 
-class MaxwellCompute final {
+class KeplerCompute final {
 public:
-    MaxwellCompute() = default;
-    ~MaxwellCompute() = default;
+    explicit KeplerCompute(MemoryManager& memory_manager);
+    ~KeplerCompute();
+
+    static constexpr std::size_t NumConstBuffers = 8;
 
     struct Regs {
         static constexpr std::size_t NUM_REGS = 0xCF8;
 
         union {
             struct {
-                INSERT_PADDING_WORDS(0x281);
+                INSERT_PADDING_WORDS(0xAF);
 
-                union {
-                    u32 compute_end;
-                    BitField<0, 1, u32> unknown;
-                } compute;
+                u32 launch;
 
-                INSERT_PADDING_WORDS(0xA76);
+                INSERT_PADDING_WORDS(0xC48);
             };
             std::array<u32, NUM_REGS> reg_array;
         };
     } regs{};
-
     static_assert(sizeof(Regs) == Regs::NUM_REGS * sizeof(u32),
-                  "MaxwellCompute Regs has wrong size");
+                  "KeplerCompute Regs has wrong size");
+
+    MemoryManager& memory_manager;
 
     /// Write the value to the register identified by method.
     void CallMethod(const GPU::MethodCall& method_call);
 };
 
 #define ASSERT_REG_POSITION(field_name, position)                                                  \
-    static_assert(offsetof(MaxwellCompute::Regs, field_name) == position * 4,                      \
+    static_assert(offsetof(KeplerCompute::Regs, field_name) == position * 4,                       \
                   "Field " #field_name " has invalid position")
 
-ASSERT_REG_POSITION(compute, 0x281);
+ASSERT_REG_POSITION(launch, 0xAF);
 
 #undef ASSERT_REG_POSITION
 

src/video_core/engines/maxwell_3d.cpp

@@ -317,7 +317,7 @@ void Maxwell3D::ProcessQueryGet() {
             LongQueryResult query_result{};
             query_result.value = result;
             // TODO(Subv): Generate a real GPU timestamp and write it here instead of CoreTiming
-            query_result.timestamp = CoreTiming::GetTicks();
+            query_result.timestamp = Core::Timing::GetTicks();
             Memory::WriteBlock(*address, &query_result, sizeof(query_result));
         }
         dirty_flags.OnMemoryWrite();

src/video_core/engines/maxwell_compute.cpp

@@ -1,28 +0,0 @@
-// Copyright 2018 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "common/logging/log.h"
-#include "core/core.h"
-#include "video_core/engines/maxwell_compute.h"
-
-namespace Tegra::Engines {
-
-void MaxwellCompute::CallMethod(const GPU::MethodCall& method_call) {
-    ASSERT_MSG(method_call.method < Regs::NUM_REGS,
-               "Invalid MaxwellCompute register, increase the size of the Regs structure");
-
-    regs.reg_array[method_call.method] = method_call.argument;
-
-    switch (method_call.method) {
-    case MAXWELL_COMPUTE_REG_INDEX(compute): {
-        LOG_CRITICAL(HW_GPU, "Compute shaders are not implemented");
-        UNREACHABLE();
-        break;
-    }
-    default:
-        break;
-    }
-}
-
-} // namespace Tegra::Engines

src/video_core/engines/shader_bytecode.h

@@ -186,7 +186,7 @@ enum class SubOp : u64 {
 };
 
 enum class F2iRoundingOp : u64 {
-    None = 0,
+    RoundEven = 0,
     Floor = 1,
     Ceil = 2,
     Trunc = 3,

src/video_core/gpu.cpp

@@ -6,9 +6,9 @@
 #include "core/core_timing.h"
 #include "core/memory.h"
 #include "video_core/engines/fermi_2d.h"
+#include "video_core/engines/kepler_compute.h"
 #include "video_core/engines/kepler_memory.h"
 #include "video_core/engines/maxwell_3d.h"
-#include "video_core/engines/maxwell_compute.h"
 #include "video_core/engines/maxwell_dma.h"
 #include "video_core/gpu.h"
 #include "video_core/rasterizer_interface.h"
@@ -18,6 +18,7 @@ namespace Tegra {
 u32 FramebufferConfig::BytesPerPixel(PixelFormat format) {
     switch (format) {
     case PixelFormat::ABGR8:
+    case PixelFormat::BGRA8:
         return 4;
     default:
         return 4;
@@ -31,7 +32,7 @@ GPU::GPU(VideoCore::RasterizerInterface& rasterizer) {
     dma_pusher = std::make_unique<Tegra::DmaPusher>(*this);
     maxwell_3d = std::make_unique<Engines::Maxwell3D>(rasterizer, *memory_manager);
     fermi_2d = std::make_unique<Engines::Fermi2D>(rasterizer, *memory_manager);
-    maxwell_compute = std::make_unique<Engines::MaxwellCompute>();
+    kepler_compute = std::make_unique<Engines::KeplerCompute>(*memory_manager);
     maxwell_dma = std::make_unique<Engines::MaxwellDMA>(rasterizer, *memory_manager);
     kepler_memory = std::make_unique<Engines::KeplerMemory>(rasterizer, *memory_manager);
 }
@@ -245,8 +246,8 @@ void GPU::CallEngineMethod(const MethodCall& method_call) {
     case EngineID::MAXWELL_B:
         maxwell_3d->CallMethod(method_call);
         break;
-    case EngineID::MAXWELL_COMPUTE_B:
-        maxwell_compute->CallMethod(method_call);
+    case EngineID::KEPLER_COMPUTE_B:
+        kepler_compute->CallMethod(method_call);
         break;
     case EngineID::MAXWELL_DMA_COPY_A:
         maxwell_dma->CallMethod(method_call);
@@ -282,7 +283,7 @@ void GPU::ProcessSemaphoreTriggerMethod() {
         block.sequence = regs.semaphore_sequence;
         // TODO(Kmather73): Generate a real GPU timestamp and write it here instead of
         // CoreTiming
-        block.timestamp = CoreTiming::GetTicks();
+        block.timestamp = Core::Timing::GetTicks();
         Memory::WriteBlock(*address, &block, sizeof(block));
     } else {
         const auto address =

src/video_core/gpu.h

@@ -80,6 +80,7 @@ class DebugContext;
 struct FramebufferConfig {
     enum class PixelFormat : u32 {
         ABGR8 = 1,
+        BGRA8 = 5,
     };
 
     /**
@@ -102,15 +103,15 @@ struct FramebufferConfig {
 namespace Engines {
 class Fermi2D;
 class Maxwell3D;
-class MaxwellCompute;
 class MaxwellDMA;
+class KeplerCompute;
 class KeplerMemory;
 } // namespace Engines
 
 enum class EngineID {
     FERMI_TWOD_A = 0x902D, // 2D Engine
     MAXWELL_B = 0xB197,    // 3D Engine
-    MAXWELL_COMPUTE_B = 0xB1C0,
+    KEPLER_COMPUTE_B = 0xB1C0,
     KEPLER_INLINE_TO_MEMORY_B = 0xA140,
     MAXWELL_DMA_COPY_A = 0xB0B5,
 };
@@ -208,7 +209,7 @@ private:
     /// 2D engine
     std::unique_ptr<Engines::Fermi2D> fermi_2d;
     /// Compute engine
-    std::unique_ptr<Engines::MaxwellCompute> maxwell_compute;
+    std::unique_ptr<Engines::KeplerCompute> kepler_compute;
     /// DMA engine
     std::unique_ptr<Engines::MaxwellDMA> maxwell_dma;
     /// Inline memory engine

src/video_core/renderer_opengl/gl_shader_decompiler.cpp

@@ -719,45 +719,51 @@ private:
         constexpr std::array<const char*, 4> coord_constructors = {"float", "vec2", "vec3", "vec4"};
 
         const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
-        const auto count = static_cast<u32>(operation.GetOperandsCount());
         ASSERT(meta);
 
+        const auto count = static_cast<u32>(operation.GetOperandsCount());
+        const bool has_array = meta->sampler.IsArray();
+        const bool has_shadow = meta->sampler.IsShadow();
+
         std::string expr = func;
         expr += '(';
         expr += GetSampler(meta->sampler);
         expr += ", ";
 
-        expr += coord_constructors[meta->coords_count - 1];
+        expr += coord_constructors.at(count + (has_array ? 1 : 0) + (has_shadow ? 1 : 0) - 1);
         expr += '(';
         for (u32 i = 0; i < count; ++i) {
-            const bool is_extra = i >= meta->coords_count;
-            const bool is_array = i == meta->array_index;
+            expr += Visit(operation[i]);
 
-            std::string operand = [&]() {
-                if (is_extra && is_extra_int) {
-                    if (const auto immediate = std::get_if<ImmediateNode>(operation[i])) {
-                        return std::to_string(static_cast<s32>(immediate->GetValue()));
-                    } else {
-                        return "ftoi(" + Visit(operation[i]) + ')';
-                    }
-                } else {
-                    return Visit(operation[i]);
-                }
-            }();
-            if (is_array) {
-                ASSERT(!is_extra);
-                operand = "float(ftoi(" + operand + "))";
-            }
-
-            expr += operand;
-
-            if (i + 1 == meta->coords_count) {
-                expr += ')';
-            }
-            if (i + 1 < count) {
+            const u32 next = i + 1;
+            if (next < count || has_array || has_shadow)
                 expr += ", ";
+        }
+        if (has_array) {
+            expr += "float(ftoi(" + Visit(meta->array) + "))";
+        }
+        if (has_shadow) {
+            if (has_array)
+                expr += ", ";
+            expr += Visit(meta->depth_compare);
+        }
+        expr += ')';
+
+        for (const Node extra : meta->extras) {
+            expr += ", ";
+            if (is_extra_int) {
+                if (const auto immediate = std::get_if<ImmediateNode>(extra)) {
+                    // Inline the string as an immediate integer in GLSL (some extra arguments are
+                    // required to be constant)
+                    expr += std::to_string(static_cast<s32>(immediate->GetValue()));
+                } else {
+                    expr += "ftoi(" + Visit(extra) + ')';
+                }
+            } else {
+                expr += Visit(extra);
             }
         }
+
         expr += ')';
         return expr;
     }
@@ -1134,7 +1140,7 @@ private:
                                   Type::HalfFloat);
     }
 
-    std::string F4Texture(Operation operation) {
+    std::string Texture(Operation operation) {
         const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
         ASSERT(meta);
 
@@ -1145,7 +1151,7 @@ private:
         return expr + GetSwizzle(meta->element);
     }
 
-    std::string F4TextureLod(Operation operation) {
+    std::string TextureLod(Operation operation) {
         const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
         ASSERT(meta);
 
@@ -1156,7 +1162,7 @@ private:
         return expr + GetSwizzle(meta->element);
     }
 
-    std::string F4TextureGather(Operation operation) {
+    std::string TextureGather(Operation operation) {
         const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
         ASSERT(meta);
 
@@ -1164,7 +1170,7 @@ private:
                GetSwizzle(meta->element);
     }
 
-    std::string F4TextureQueryDimensions(Operation operation) {
+    std::string TextureQueryDimensions(Operation operation) {
         const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
         ASSERT(meta);
 
@@ -1184,7 +1190,7 @@ private:
         return "0";
     }
 
-    std::string F4TextureQueryLod(Operation operation) {
+    std::string TextureQueryLod(Operation operation) {
         const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
         ASSERT(meta);
 
@@ -1195,29 +1201,32 @@ private:
         return "0";
     }
 
-    std::string F4TexelFetch(Operation operation) {
+    std::string TexelFetch(Operation operation) {
         constexpr std::array<const char*, 4> constructors = {"int", "ivec2", "ivec3", "ivec4"};
         const auto meta = std::get_if<MetaTexture>(&operation.GetMeta());
-        const auto count = static_cast<u32>(operation.GetOperandsCount());
         ASSERT(meta);
+        UNIMPLEMENTED_IF(meta->sampler.IsArray());
+        UNIMPLEMENTED_IF(!meta->extras.empty());
+
+        const auto count = static_cast<u32>(operation.GetOperandsCount());
 
         std::string expr = "texelFetch(";
         expr += GetSampler(meta->sampler);
         expr += ", ";
 
-        expr += constructors[meta->coords_count - 1];
+        expr += constructors.at(count - 1);
         expr += '(';
         for (u32 i = 0; i < count; ++i) {
             expr += VisitOperand(operation, i, Type::Int);
 
-            if (i + 1 == meta->coords_count) {
+            const u32 next = i + 1;
+            if (next == count)
                 expr += ')';
-            }
-            if (i + 1 < count) {
+            if (next < count)
                 expr += ", ";
-            }
         }
         expr += ')';
+
         return expr + GetSwizzle(meta->element);
     }
 
@@ -1454,12 +1463,12 @@ private:
         &GLSLDecompiler::Logical2HNotEqual,
         &GLSLDecompiler::Logical2HGreaterEqual,
 
-        &GLSLDecompiler::F4Texture,
-        &GLSLDecompiler::F4TextureLod,
-        &GLSLDecompiler::F4TextureGather,
-        &GLSLDecompiler::F4TextureQueryDimensions,
-        &GLSLDecompiler::F4TextureQueryLod,
-        &GLSLDecompiler::F4TexelFetch,
+        &GLSLDecompiler::Texture,
+        &GLSLDecompiler::TextureLod,
+        &GLSLDecompiler::TextureGather,
+        &GLSLDecompiler::TextureQueryDimensions,
+        &GLSLDecompiler::TextureQueryLod,
+        &GLSLDecompiler::TexelFetch,
 
         &GLSLDecompiler::Branch,
         &GLSLDecompiler::PushFlowStack,

src/video_core/renderer_opengl/renderer_opengl.cpp

@@ -107,7 +107,7 @@ RendererOpenGL::~RendererOpenGL() = default;
 void RendererOpenGL::SwapBuffers(
     std::optional<std::reference_wrapper<const Tegra::FramebufferConfig>> framebuffer) {
 
-    Core::System::GetInstance().GetPerfStats().EndSystemFrame();
+    system.GetPerfStats().EndSystemFrame();
 
     // Maintain the rasterizer's state as a priority
     OpenGLState prev_state = OpenGLState::GetCurState();
@@ -137,8 +137,8 @@ void RendererOpenGL::SwapBuffers(
 
     render_window.PollEvents();
 
-    Core::System::GetInstance().FrameLimiter().DoFrameLimiting(CoreTiming::GetGlobalTimeUs());
-    Core::System::GetInstance().GetPerfStats().BeginSystemFrame();
+    system.FrameLimiter().DoFrameLimiting(Core::Timing::GetGlobalTimeUs());
+    system.GetPerfStats().BeginSystemFrame();
 
     // Restore the rasterizer state
     prev_state.Apply();

src/video_core/shader/decode/arithmetic_integer.cpp

@@ -41,7 +41,7 @@ u32 ShaderIR::DecodeArithmeticInteger(NodeBlock& bb, u32 pc) {
 
         const Node value = Operation(OperationCode::IAdd, PRECISE, op_a, op_b);
 
-        SetInternalFlagsFromInteger(bb, value, instr.op_32.generates_cc);
+        SetInternalFlagsFromInteger(bb, value, instr.generates_cc);
         SetRegister(bb, instr.gpr0, value);
         break;
     }
@@ -284,4 +284,4 @@ void ShaderIR::WriteLop3Instruction(NodeBlock& bb, Register dest, Node op_a, Nod
     SetRegister(bb, dest, value);
 }
 
-} // namespace VideoCommon::Shader
+} // namespace VideoCommon::Shader

src/video_core/shader/decode/conversion.cpp

@@ -118,8 +118,8 @@ u32 ShaderIR::DecodeConversion(NodeBlock& bb, u32 pc) {
 
         value = [&]() {
             switch (instr.conversion.f2i.rounding) {
-            case Tegra::Shader::F2iRoundingOp::None:
-                return value;
+            case Tegra::Shader::F2iRoundingOp::RoundEven:
+                return Operation(OperationCode::FRoundEven, PRECISE, value);
             case Tegra::Shader::F2iRoundingOp::Floor:
                 return Operation(OperationCode::FFloor, PRECISE, value);
             case Tegra::Shader::F2iRoundingOp::Ceil:
@@ -146,4 +146,4 @@ u32 ShaderIR::DecodeConversion(NodeBlock& bb, u32 pc) {
     return pc;
 }
 
-} // namespace VideoCommon::Shader
+} // namespace VideoCommon::Shader

src/video_core/shader/decode/memory.cpp

@@ -306,7 +306,6 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
     case OpCode::Id::TLD4S: {
         UNIMPLEMENTED_IF_MSG(instr.tld4s.UsesMiscMode(TextureMiscMode::AOFFI),
                              "AOFFI is not implemented");
-
         if (instr.tld4s.UsesMiscMode(TextureMiscMode::NODEP)) {
             LOG_WARNING(HW_GPU, "TLD4S.NODEP implementation is incomplete");
         }
@@ -315,9 +314,8 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
         const Node op_a = GetRegister(instr.gpr8);
         const Node op_b = GetRegister(instr.gpr20);
 
-        std::vector<Node> coords;
-
         // TODO(Subv): Figure out how the sampler type is encoded in the TLD4S instruction.
+        std::vector<Node> coords;
         if (depth_compare) {
             // Note: TLD4S coordinate encoding works just like TEXS's
             const Node op_y = GetRegister(instr.gpr8.Value() + 1);
@@ -328,18 +326,17 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
             coords.push_back(op_a);
             coords.push_back(op_b);
         }
-        const auto num_coords = static_cast<u32>(coords.size());
-        coords.push_back(Immediate(static_cast<u32>(instr.tld4s.component)));
+        std::vector<Node> extras;
+        extras.push_back(Immediate(static_cast<u32>(instr.tld4s.component)));
 
         const auto& sampler =
             GetSampler(instr.sampler, TextureType::Texture2D, false, depth_compare);
 
         Node4 values;
         for (u32 element = 0; element < values.size(); ++element) {
-            auto params = coords;
-            MetaTexture meta{sampler, element, num_coords};
-            values[element] =
-                Operation(OperationCode::F4TextureGather, std::move(meta), std::move(params));
+            auto coords_copy = coords;
+            MetaTexture meta{sampler, {}, {}, extras, element};
+            values[element] = Operation(OperationCode::TextureGather, meta, std::move(coords_copy));
         }
 
         WriteTexsInstructionFloat(bb, instr, values);
@@ -360,12 +357,13 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
         switch (instr.txq.query_type) {
         case Tegra::Shader::TextureQueryType::Dimension: {
             for (u32 element = 0; element < 4; ++element) {
-                if (instr.txq.IsComponentEnabled(element)) {
-                    MetaTexture meta{sampler, element};
-                    const Node value = Operation(OperationCode::F4TextureQueryDimensions,
-                                                 std::move(meta), GetRegister(instr.gpr8));
-                    SetTemporal(bb, indexer++, value);
+                if (!instr.txq.IsComponentEnabled(element)) {
+                    continue;
                 }
+                MetaTexture meta{sampler, {}, {}, {}, element};
+                const Node value =
+                    Operation(OperationCode::TextureQueryDimensions, meta, GetRegister(instr.gpr8));
+                SetTemporal(bb, indexer++, value);
             }
             for (u32 i = 0; i < indexer; ++i) {
                 SetRegister(bb, instr.gpr0.Value() + i, GetTemporal(i));
@@ -412,9 +410,8 @@ u32 ShaderIR::DecodeMemory(NodeBlock& bb, u32 pc) {
 
         for (u32 element = 0; element < 2; ++element) {
             auto params = coords;
-            MetaTexture meta_texture{sampler, element, static_cast<u32>(coords.size())};
-            const Node value =
-                Operation(OperationCode::F4TextureQueryLod, meta_texture, std::move(params));
+            MetaTexture meta{sampler, {}, {}, {}, element};
+            const Node value = Operation(OperationCode::TextureQueryLod, meta, std::move(params));
             SetTemporal(bb, element, value);
         }
         for (u32 element = 0; element < 2; ++element) {
@@ -535,15 +532,16 @@ void ShaderIR::WriteTexsInstructionHalfFloat(NodeBlock& bb, Instruction instr,
 }
 
 Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
-                               TextureProcessMode process_mode, bool depth_compare, bool is_array,
-                               std::size_t array_offset, std::size_t bias_offset,
-                               std::vector<Node>&& coords) {
-    UNIMPLEMENTED_IF_MSG(
-        (texture_type == TextureType::Texture3D && (is_array || depth_compare)) ||
-            (texture_type == TextureType::TextureCube && is_array && depth_compare),
-        "This method is not supported.");
+                               TextureProcessMode process_mode, std::vector<Node> coords,
+                               Node array, Node depth_compare, u32 bias_offset) {
+    const bool is_array = array;
+    const bool is_shadow = depth_compare;
 
-    const auto& sampler = GetSampler(instr.sampler, texture_type, is_array, depth_compare);
+    UNIMPLEMENTED_IF_MSG((texture_type == TextureType::Texture3D && (is_array || is_shadow)) ||
+                             (texture_type == TextureType::TextureCube && is_array && is_shadow),
+                         "This method is not supported.");
+
+    const auto& sampler = GetSampler(instr.sampler, texture_type, is_array, is_shadow);
 
     const bool lod_needed = process_mode == TextureProcessMode::LZ ||
                             process_mode == TextureProcessMode::LL ||
@@ -552,35 +550,30 @@ Node4 ShaderIR::GetTextureCode(Instruction instr, TextureType texture_type,
     // LOD selection (either via bias or explicit textureLod) not supported in GL for
     // sampler2DArrayShadow and samplerCubeArrayShadow.
     const bool gl_lod_supported =
-        !((texture_type == Tegra::Shader::TextureType::Texture2D && is_array && depth_compare) ||
-          (texture_type == Tegra::Shader::TextureType::TextureCube && is_array && depth_compare));
+        !((texture_type == Tegra::Shader::TextureType::Texture2D && is_array && is_shadow) ||
+          (texture_type == Tegra::Shader::TextureType::TextureCube && is_array && is_shadow));
 
     const OperationCode read_method =
-        lod_needed && gl_lod_supported ? OperationCode::F4TextureLod : OperationCode::F4Texture;
+        lod_needed && gl_lod_supported ? OperationCode::TextureLod : OperationCode::Texture;
 
     UNIMPLEMENTED_IF(process_mode != TextureProcessMode::None && !gl_lod_supported);
 
-    std::optional<u32> array_offset_value;
-    if (is_array)
-        array_offset_value = static_cast<u32>(array_offset);
-
-    const auto coords_count = static_cast<u32>(coords.size());
-
+    std::vector<Node> extras;
     if (process_mode != TextureProcessMode::None && gl_lod_supported) {
         if (process_mode == TextureProcessMode::LZ) {
-            coords.push_back(Immediate(0.0f));
+            extras.push_back(Immediate(0.0f));
         } else {
             // If present, lod or bias are always stored in the register indexed by the gpr20
             // field with an offset depending on the usage of the other registers
-            coords.push_back(GetRegister(instr.gpr20.Value() + bias_offset));
+            extras.push_back(GetRegister(instr.gpr20.Value() + bias_offset));
         }
     }
 
     Node4 values;
     for (u32 element = 0; element < values.size(); ++element) {
-        auto params = coords;
-        MetaTexture meta{sampler, element, coords_count, array_offset_value};
-        values[element] = Operation(read_method, std::move(meta), std::move(params));
+        auto copy_coords = coords;
+        MetaTexture meta{sampler, array, depth_compare, extras, element};
+        values[element] = Operation(read_method, meta, std::move(copy_coords));
     }
 
     return values;
@@ -602,28 +595,22 @@ Node4 ShaderIR::GetTexCode(Instruction instr, TextureType texture_type,
     for (std::size_t i = 0; i < coord_count; ++i) {
         coords.push_back(GetRegister(coord_register + i));
     }
-    // 1D.DC in opengl the 2nd component is ignored.
+    // 1D.DC in OpenGL the 2nd component is ignored.
     if (depth_compare && !is_array && texture_type == TextureType::Texture1D) {
         coords.push_back(Immediate(0.0f));
     }
-    std::size_t array_offset{};
-    if (is_array) {
-        array_offset = coords.size();
-        coords.push_back(GetRegister(array_register));
-    }
+
+    const Node array = is_array ? GetRegister(array_register) : nullptr;
+
+    Node dc{};
     if (depth_compare) {
-        // Depth is always stored in the register signaled by gpr20
-        // or in the next register if lod or bias are used
+        // Depth is always stored in the register signaled by gpr20 or in the next register if lod
+        // or bias are used
         const u64 depth_register = instr.gpr20.Value() + (lod_bias_enabled ? 1 : 0);
-        coords.push_back(GetRegister(depth_register));
-    }
-    // Fill ignored coordinates
-    while (coords.size() < total_coord_count) {
-        coords.push_back(Immediate(0));
+        dc = GetRegister(depth_register);
     }
 
-    return GetTextureCode(instr, texture_type, process_mode, depth_compare, is_array, array_offset,
-                          0, std::move(coords));
+    return GetTextureCode(instr, texture_type, process_mode, coords, array, dc, 0);
 }
 
 Node4 ShaderIR::GetTexsCode(Instruction instr, TextureType texture_type,
@@ -641,6 +628,7 @@ Node4 ShaderIR::GetTexsCode(Instruction instr, TextureType texture_type,
         (is_array || !(lod_bias_enabled || depth_compare) || (coord_count > 2))
             ? static_cast<u64>(instr.gpr20.Value())
             : coord_register + 1;
+    const u32 bias_offset = coord_count > 2 ? 1 : 0;
 
     std::vector<Node> coords;
     for (std::size_t i = 0; i < coord_count; ++i) {
@@ -648,24 +636,17 @@ Node4 ShaderIR::GetTexsCode(Instruction instr, TextureType texture_type,
         coords.push_back(GetRegister(last ? last_coord_register : coord_register + i));
     }
 
-    std::size_t array_offset{};
-    if (is_array) {
-        array_offset = coords.size();
-        coords.push_back(GetRegister(array_register));
-    }
+    const Node array = is_array ? GetRegister(array_register) : nullptr;
+
+    Node dc{};
     if (depth_compare) {
-        // Depth is always stored in the register signaled by gpr20
-        // or in the next register if lod or bias are used
+        // Depth is always stored in the register signaled by gpr20 or in the next register if lod
+        // or bias are used
         const u64 depth_register = instr.gpr20.Value() + (lod_bias_enabled ? 1 : 0);
-        coords.push_back(GetRegister(depth_register));
-    }
-    // Fill ignored coordinates
-    while (coords.size() < total_coord_count) {
-        coords.push_back(Immediate(0));
+        dc = GetRegister(depth_register);
     }
 
-    return GetTextureCode(instr, texture_type, process_mode, depth_compare, is_array, array_offset,
-                          (coord_count > 2 ? 1 : 0), std::move(coords));
+    return GetTextureCode(instr, texture_type, process_mode, coords, array, dc, bias_offset);
 }
 
 Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool depth_compare,
@@ -680,24 +661,16 @@ Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool de
     const u64 coord_register = array_register + (is_array ? 1 : 0);
 
     std::vector<Node> coords;
-
-    for (size_t i = 0; i < coord_count; ++i) {
+    for (size_t i = 0; i < coord_count; ++i)
         coords.push_back(GetRegister(coord_register + i));
-    }
-    std::optional<u32> array_offset;
-    if (is_array) {
-        array_offset = static_cast<u32>(coords.size());
-        coords.push_back(GetRegister(array_register));
-    }
 
     const auto& sampler = GetSampler(instr.sampler, texture_type, is_array, depth_compare);
 
     Node4 values;
     for (u32 element = 0; element < values.size(); ++element) {
-        auto params = coords;
-        MetaTexture meta{sampler, element, static_cast<u32>(coords.size()), array_offset};
-        values[element] =
-            Operation(OperationCode::F4TextureGather, std::move(meta), std::move(params));
+        auto coords_copy = coords;
+        MetaTexture meta{sampler, GetRegister(array_register), {}, {}, element};
+        values[element] = Operation(OperationCode::TextureGather, meta, std::move(coords_copy));
     }
 
     return values;
@@ -705,7 +678,6 @@ Node4 ShaderIR::GetTld4Code(Instruction instr, TextureType texture_type, bool de
 
 Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is_array) {
     const std::size_t type_coord_count = GetCoordCount(texture_type);
-    const std::size_t total_coord_count = type_coord_count + (is_array ? 1 : 0);
     const bool lod_enabled = instr.tlds.GetTextureProcessMode() == TextureProcessMode::LL;
 
     // If enabled arrays index is always stored in the gpr8 field
@@ -719,33 +691,22 @@ Node4 ShaderIR::GetTldsCode(Instruction instr, TextureType texture_type, bool is
             : coord_register + 1;
 
     std::vector<Node> coords;
-
     for (std::size_t i = 0; i < type_coord_count; ++i) {
         const bool last = (i == (type_coord_count - 1)) && (type_coord_count > 1);
         coords.push_back(GetRegister(last ? last_coord_register : coord_register + i));
     }
-    std::optional<u32> array_offset;
-    if (is_array) {
-        array_offset = static_cast<u32>(coords.size());
-        coords.push_back(GetRegister(array_register));
-    }
-    const auto coords_count = static_cast<u32>(coords.size());
 
-    if (lod_enabled) {
-        // When lod is used always is in grp20
-        coords.push_back(GetRegister(instr.gpr20));
-    } else {
-        coords.push_back(Immediate(0));
-    }
+    const Node array = is_array ? GetRegister(array_register) : nullptr;
+    // When lod is used always is in gpr20
+    const Node lod = lod_enabled ? GetRegister(instr.gpr20) : Immediate(0);
 
     const auto& sampler = GetSampler(instr.sampler, texture_type, is_array, false);
 
     Node4 values;
     for (u32 element = 0; element < values.size(); ++element) {
-        auto params = coords;
-        MetaTexture meta{sampler, element, coords_count, array_offset};
-        values[element] =
-            Operation(OperationCode::F4TexelFetch, std::move(meta), std::move(params));
+        auto coords_copy = coords;
+        MetaTexture meta{sampler, array, {}, {lod}, element};
+        values[element] = Operation(OperationCode::TexelFetch, meta, std::move(coords_copy));
     }
     return values;
 }

src/video_core/shader/shader_ir.h

@@ -156,12 +156,12 @@ enum class OperationCode {
     Logical2HNotEqual,     /// (MetaHalfArithmetic, f16vec2 a, f16vec2) -> bool2
     Logical2HGreaterEqual, /// (MetaHalfArithmetic, f16vec2 a, f16vec2) -> bool2
 
-    F4Texture,                /// (MetaTexture, float[N] coords, float[M] params) -> float4
-    F4TextureLod,             /// (MetaTexture, float[N] coords, float[M] params) -> float4
-    F4TextureGather,          /// (MetaTexture, float[N] coords, float[M] params) -> float4
-    F4TextureQueryDimensions, /// (MetaTexture, float a) -> float4
-    F4TextureQueryLod,        /// (MetaTexture, float[N] coords) -> float4
-    F4TexelFetch,             /// (MetaTexture, int[N], int) -> float4
+    Texture,                /// (MetaTexture, float[N] coords) -> float4
+    TextureLod,             /// (MetaTexture, float[N] coords) -> float4
+    TextureGather,          /// (MetaTexture, float[N] coords) -> float4
+    TextureQueryDimensions, /// (MetaTexture, float a) -> float4
+    TextureQueryLod,        /// (MetaTexture, float[N] coords) -> float4
+    TexelFetch,             /// (MetaTexture, int[N], int) -> float4
 
     Branch,        /// (uint branch_target) -> void
     PushFlowStack, /// (uint branch_target) -> void
@@ -288,9 +288,10 @@ struct MetaHalfArithmetic {
 
 struct MetaTexture {
     const Sampler& sampler;
+    Node array{};
+    Node depth_compare{};
+    std::vector<Node> extras;
     u32 element{};
-    u32 coords_count{};
-    std::optional<u32> array_index;
 };
 
 constexpr MetaArithmetic PRECISE = {true};
@@ -754,9 +755,8 @@ private:
         bool lod_bias_enabled, std::size_t max_coords, std::size_t max_inputs);
 
     Node4 GetTextureCode(Tegra::Shader::Instruction instr, Tegra::Shader::TextureType texture_type,
-                         Tegra::Shader::TextureProcessMode process_mode, bool depth_compare,
-                         bool is_array, std::size_t array_offset, std::size_t bias_offset,
-                         std::vector<Node>&& coords);
+                         Tegra::Shader::TextureProcessMode process_mode, std::vector<Node> coords,
+                         Node array, Node depth_compare, u32 bias_offset);
 
     Node GetVideoOperand(Node op, bool is_chunk, bool is_signed, Tegra::Shader::VideoType type,
                          u64 byte_height);

src/video_core/surface.cpp

@@ -426,6 +426,8 @@ PixelFormat PixelFormatFromGPUPixelFormat(Tegra::FramebufferConfig::PixelFormat
     switch (format) {
     case Tegra::FramebufferConfig::PixelFormat::ABGR8:
         return PixelFormat::ABGR8U;
+    case Tegra::FramebufferConfig::PixelFormat::BGRA8:
+        return PixelFormat::BGRA8;
     default:
         LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
         UNREACHABLE();

src/yuzu/configuration/config.cpp

@@ -405,6 +405,13 @@ void Config::ReadValues() {
                     QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir)))
             .toString()
             .toStdString());
+    FileUtil::GetUserPath(
+        FileUtil::UserPath::KeysDir,
+        qt_config
+            ->value("keys_directory",
+                    QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::KeysDir)))
+            .toString()
+            .toStdString());
     qt_config->endGroup();
 
     qt_config->beginGroup("Core");
@@ -653,6 +660,8 @@ void Config::SaveValues() {
                         QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::NANDDir)));
     qt_config->setValue("sdmc_directory",
                         QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::SDMCDir)));
+    qt_config->setValue("keys_directory",
+                        QString::fromStdString(FileUtil::GetUserPath(FileUtil::UserPath::KeysDir)));
     qt_config->endGroup();
 
     qt_config->beginGroup("System");

src/yuzu/main.cpp

@@ -659,6 +659,8 @@ void GMainWindow::ConnectMenuEvents() {
             [this] { OnMenuSelectEmulatedDirectory(EmulatedDirectoryTarget::NAND); });
     connect(ui.action_Select_SDMC_Directory, &QAction::triggered, this,
             [this] { OnMenuSelectEmulatedDirectory(EmulatedDirectoryTarget::SDMC); });
+    connect(ui.action_Select_Keys_Directory, &QAction::triggered, this,
+            [this] { OnMenuSelectEmulatedDirectory(EmulatedDirectoryTarget::Keys); });
     connect(ui.action_Exit, &QAction::triggered, this, &QMainWindow::close);
     connect(ui.action_Load_Amiibo, &QAction::triggered, this, &GMainWindow::OnLoadAmiibo);
 
@@ -1447,12 +1449,27 @@ void GMainWindow::OnMenuSelectGameListRoot() {
 }
 
 void GMainWindow::OnMenuSelectEmulatedDirectory(EmulatedDirectoryTarget target) {
+    const char* directory_name;
+    FileUtil::UserPath path;
+    switch (target) {
+    case EmulatedDirectoryTarget::SDMC:
+        directory_name = "SD Card";
+        path = FileUtil::UserPath::SDMCDir;
+        break;
+    case EmulatedDirectoryTarget::NAND:
+        directory_name = "NAND";
+        path = FileUtil::UserPath::NANDDir;
+        break;
+    default:
+        directory_name = "keys";
+        path = FileUtil::UserPath::KeysDir;
+    }
     const auto res = QMessageBox::information(
         this, tr("Changing Emulated Directory"),
         tr("You are about to change the emulated %1 directory of the system. Please note "
            "that this does not also move the contents of the previous directory to the "
            "new one and you will have to do that yourself.")
-            .arg(target == EmulatedDirectoryTarget::SDMC ? tr("SD card") : tr("NAND")),
+            .arg(tr(directory_name)),
         QMessageBox::StandardButtons{QMessageBox::Ok, QMessageBox::Cancel});
 
     if (res == QMessageBox::Cancel)
@@ -1460,9 +1477,7 @@ void GMainWindow::OnMenuSelectEmulatedDirectory(EmulatedDirectoryTarget target)
 
     QString dir_path = QFileDialog::getExistingDirectory(this, tr("Select Directory"));
     if (!dir_path.isEmpty()) {
-        FileUtil::GetUserPath(target == EmulatedDirectoryTarget::SDMC ? FileUtil::UserPath::SDMCDir
-                                                                      : FileUtil::UserPath::NANDDir,
-                              dir_path.toStdString());
+        FileUtil::GetUserPath(path, dir_path.toStdString());
         Service::FileSystem::CreateFactories(*vfs);
         game_list->PopulateAsync(UISettings::values.gamedir, UISettings::values.gamedir_deepscan);
     }

src/yuzu/main.h

@@ -48,6 +48,7 @@ class DebugContext;
 enum class EmulatedDirectoryTarget {
     NAND,
     SDMC,
+    Keys,
 };
 
 enum class ReinitializeKeyBehavior {

src/yuzu/main.ui

@@ -67,6 +67,7 @@
     <addaction name="separator"/>
     <addaction name="action_Select_NAND_Directory"/>
     <addaction name="action_Select_SDMC_Directory"/>
+    <addaction name="action_Select_Keys_Directory"/>
     <addaction name="separator"/>
     <addaction name="action_Load_Amiibo"/>
     <addaction name="separator"/>
@@ -246,6 +247,14 @@
     <string>Selects a folder to use as the root of the emulated SD card</string>
    </property>
   </action>
+   <action name="action_Select_Keys_Directory">
+     <property name="text">
+       <string>Select Keys Directory...</string>
+     </property>
+     <property name="toolTip">
+       <string>Selects a folder to use as the root of the keys</string>
+     </property>
+   </action>
   <action name="action_Fullscreen">
    <property name="checkable">
     <bool>true</bool>