116 files changed, 3971 insertions, 2293 deletions

diff --git a/externals/dynarmic b/externals/dynarmic
-Subproject e7166e8ba74d7b9c85e87afc0aaf667e7e84cfe
+Subproject 4f967387c07365b7ea35d2fa3e19b7df8872a09
diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt
index 3a57356ab..1e977e8a8 100644
--- a/src/CMakeLists.txt
+++ b/src/CMakeLists.txt
@@ -62,6 +62,10 @@ else()
         -Wno-unused-parameter
     )
 
+    if (ARCHITECTURE_x86_64)
+        add_compile_options("-mcx16")
+    endif()
+
     if (APPLE AND CMAKE_CXX_COMPILER_ID STREQUAL Clang)
         add_compile_options("-stdlib=libc++")
     endif()
diff --git a/src/audio_core/stream.cpp b/src/audio_core/stream.cpp
index ca7cfb030..dfc4805d9 100644
--- a/src/audio_core/stream.cpp
+++ b/src/audio_core/stream.cpp
@@ -59,11 +59,20 @@ Stream::State Stream::GetState() const {
     return state;
 }
 
-s64 Stream::GetBufferReleaseCycles(const Buffer& buffer) const {
+s64 Stream::GetBufferReleaseNS(const Buffer& buffer) const {
     const std::size_t num_samples{buffer.GetSamples().size() / GetNumChannels()};
-    const auto us =
-        std::chrono::microseconds((static_cast<u64>(num_samples) * 1000000) / sample_rate);
-    return Core::Timing::usToCycles(us);
+    const auto ns =
+        std::chrono::nanoseconds((static_cast<u64>(num_samples) * 1000000000ULL) / sample_rate);
+    return ns.count();
+}
+
+s64 Stream::GetBufferReleaseNSHostTiming(const Buffer& buffer) const {
+    const std::size_t num_samples{buffer.GetSamples().size() / GetNumChannels()};
+    /// DSP signals before playing the last sample, in HLE we emulate this in this way
+    s64 base_samples = std::max<s64>(static_cast<s64>(num_samples) - 1, 0);
+    const auto ns =
+        std::chrono::nanoseconds((static_cast<u64>(base_samples) * 1000000000ULL) / sample_rate);
+    return ns.count();
 }
 
 static void VolumeAdjustSamples(std::vector<s16>& samples, float game_volume) {
@@ -105,7 +114,11 @@ void Stream::PlayNextBuffer() {
 
     sink_stream.EnqueueSamples(GetNumChannels(), active_buffer->GetSamples());
 
-    core_timing.ScheduleEvent(GetBufferReleaseCycles(*active_buffer), release_event, {});
+    if (core_timing.IsHostTiming()) {
+        core_timing.ScheduleEvent(GetBufferReleaseNSHostTiming(*active_buffer), release_event, {});
+    } else {
+        core_timing.ScheduleEvent(GetBufferReleaseNS(*active_buffer), release_event, {});
+    }
 }
 
 void Stream::ReleaseActiveBuffer() {
diff --git a/src/audio_core/stream.h b/src/audio_core/stream.h
index 1708a4d98..e309d60fe 100644
--- a/src/audio_core/stream.h
+++ b/src/audio_core/stream.h
@@ -96,7 +96,10 @@ private:
     void ReleaseActiveBuffer();
 
     /// Gets the number of core cycles when the specified buffer will be released
-    s64 GetBufferReleaseCycles(const Buffer& buffer) const;
+    s64 GetBufferReleaseNS(const Buffer& buffer) const;
+
+    /// Gets the number of core cycles when the specified buffer will be released
+    s64 GetBufferReleaseNSHostTiming(const Buffer& buffer) const;
 
     u32 sample_rate;                  ///< Sample rate of the stream
     Format format;                    ///< Format of the stream
diff --git a/src/common/CMakeLists.txt b/src/common/CMakeLists.txt
index 3cc17d0e9..d120c8d3d 100644
--- a/src/common/CMakeLists.txt
+++ b/src/common/CMakeLists.txt
@@ -98,6 +98,8 @@ add_library(common STATIC
     algorithm.h
     alignment.h
     assert.h
+    atomic_ops.cpp
+    atomic_ops.h
     detached_tasks.cpp
     detached_tasks.h
     bit_field.h
diff --git a/src/common/atomic_ops.cpp b/src/common/atomic_ops.cpp
new file mode 100644
index 000000000..1098e21ff
--- /dev/null
+++ b/src/common/atomic_ops.cpp
@@ -0,0 +1,70 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cstring>
+
+#include "common/atomic_ops.h"
+
+#if _MSC_VER
+#include <intrin.h>
+#endif
+
+namespace Common {
+
+#if _MSC_VER
+
+bool AtomicCompareAndSwap(u8 volatile* pointer, u8 value, u8 expected) {
+    u8 result = _InterlockedCompareExchange8((char*)pointer, value, expected);
+    return result == expected;
+}
+
+bool AtomicCompareAndSwap(u16 volatile* pointer, u16 value, u16 expected) {
+    u16 result = _InterlockedCompareExchange16((short*)pointer, value, expected);
+    return result == expected;
+}
+
+bool AtomicCompareAndSwap(u32 volatile* pointer, u32 value, u32 expected) {
+    u32 result = _InterlockedCompareExchange((long*)pointer, value, expected);
+    return result == expected;
+}
+
+bool AtomicCompareAndSwap(u64 volatile* pointer, u64 value, u64 expected) {
+    u64 result = _InterlockedCompareExchange64((__int64*)pointer, value, expected);
+    return result == expected;
+}
+
+bool AtomicCompareAndSwap(u64 volatile* pointer, u128 value, u128 expected) {
+    return _InterlockedCompareExchange128((__int64*)pointer, value[1], value[0],
+                                          (__int64*)expected.data()) != 0;
+}
+
+#else
+
+bool AtomicCompareAndSwap(u8 volatile* pointer, u8 value, u8 expected) {
+    return __sync_bool_compare_and_swap(pointer, expected, value);
+}
+
+bool AtomicCompareAndSwap(u16 volatile* pointer, u16 value, u16 expected) {
+    return __sync_bool_compare_and_swap(pointer, expected, value);
+}
+
+bool AtomicCompareAndSwap(u32 volatile* pointer, u32 value, u32 expected) {
+    return __sync_bool_compare_and_swap(pointer, expected, value);
+}
+
+bool AtomicCompareAndSwap(u64 volatile* pointer, u64 value, u64 expected) {
+    return __sync_bool_compare_and_swap(pointer, expected, value);
+}
+
+bool AtomicCompareAndSwap(u64 volatile* pointer, u128 value, u128 expected) {
+    unsigned __int128 value_a;
+    unsigned __int128 expected_a;
+    std::memcpy(&value_a, value.data(), sizeof(u128));
+    std::memcpy(&expected_a, expected.data(), sizeof(u128));
+    return __sync_bool_compare_and_swap((unsigned __int128*)pointer, expected_a, value_a);
+}
+
+#endif
+
+} // namespace Common
diff --git a/src/common/atomic_ops.h b/src/common/atomic_ops.h
new file mode 100644
index 000000000..e6181d521
--- /dev/null
+++ b/src/common/atomic_ops.h
@@ -0,0 +1,17 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace Common {
+
+bool AtomicCompareAndSwap(u8 volatile* pointer, u8 value, u8 expected);
+bool AtomicCompareAndSwap(u16 volatile* pointer, u16 value, u16 expected);
+bool AtomicCompareAndSwap(u32 volatile* pointer, u32 value, u32 expected);
+bool AtomicCompareAndSwap(u64 volatile* pointer, u64 value, u64 expected);
+bool AtomicCompareAndSwap(u64 volatile* pointer, u128 value, u128 expected);
+
+} // namespace Common
diff --git a/src/common/fiber.cpp b/src/common/fiber.cpp
index f97ad433b..1c1d09ccb 100644
--- a/src/common/fiber.cpp
+++ b/src/common/fiber.cpp
@@ -54,9 +54,7 @@ Fiber::Fiber(std::function<void(void*)>&& entry_point_func, void* start_paramete
     impl->handle = CreateFiber(default_stack_size, &FiberStartFunc, this);
 }
 
-Fiber::Fiber() {
-    impl = std::make_unique<FiberImpl>();
-}
+Fiber::Fiber() : impl{std::make_unique<FiberImpl>()} {}
 
 Fiber::~Fiber() {
     if (released) {
@@ -116,8 +114,8 @@ std::shared_ptr<Fiber> Fiber::ThreadToFiber() {
 
 struct Fiber::FiberImpl {
     alignas(64) std::array<u8, default_stack_size> stack;
-    u8* stack_limit;
     alignas(64) std::array<u8, default_stack_size> rewind_stack;
+    u8* stack_limit;
     u8* rewind_stack_limit;
     boost::context::detail::fcontext_t context;
     boost::context::detail::fcontext_t rewind_context;
@@ -168,9 +166,7 @@ void Fiber::SetRewindPoint(std::function<void(void*)>&& rewind_func, void* start
     rewind_parameter = start_parameter;
 }
 
-Fiber::Fiber() {
-    impl = std::make_unique<FiberImpl>();
-}
+Fiber::Fiber() : impl{std::make_unique<FiberImpl>()} {}
 
 Fiber::~Fiber() {
     if (released) {
diff --git a/src/common/spin_lock.cpp b/src/common/spin_lock.cpp
index c7b46aac6..c1524220f 100644
--- a/src/common/spin_lock.cpp
+++ b/src/common/spin_lock.cpp
@@ -20,7 +20,7 @@
 
 namespace {
 
-void thread_pause() {
+void ThreadPause() {
 #if __x86_64__
     _mm_pause();
 #elif __aarch64__ && _MSC_VER
@@ -30,13 +30,13 @@ void thread_pause() {
 #endif
 }
 
-} // namespace
+} // Anonymous namespace
 
 namespace Common {
 
 void SpinLock::lock() {
     while (lck.test_and_set(std::memory_order_acquire)) {
-        thread_pause();
+        ThreadPause();
     }
 }
 
diff --git a/src/common/spin_lock.h b/src/common/spin_lock.h
index 70282a961..1df5528c4 100644
--- a/src/common/spin_lock.h
+++ b/src/common/spin_lock.h
@@ -8,6 +8,11 @@
 
 namespace Common {
 
+/**
+ * SpinLock class
+ * a lock similar to mutex that forces a thread to spin wait instead calling the
+ * supervisor. Should be used on short sequences of code.
+ */
 class SpinLock {
 public:
     void lock();
diff --git a/src/common/thread.cpp b/src/common/thread.cpp
index 0cd2d10bf..8e5935e6a 100644
--- a/src/common/thread.cpp
+++ b/src/common/thread.cpp
@@ -25,6 +25,52 @@
 
 namespace Common {
 
+#ifdef _WIN32
+
+void SetCurrentThreadPriority(ThreadPriority new_priority) {
+    auto handle = GetCurrentThread();
+    int windows_priority = 0;
+    switch (new_priority) {
+    case ThreadPriority::Low:
+        windows_priority = THREAD_PRIORITY_BELOW_NORMAL;
+        break;
+    case ThreadPriority::Normal:
+        windows_priority = THREAD_PRIORITY_NORMAL;
+        break;
+    case ThreadPriority::High:
+        windows_priority = THREAD_PRIORITY_ABOVE_NORMAL;
+        break;
+    case ThreadPriority::VeryHigh:
+        windows_priority = THREAD_PRIORITY_HIGHEST;
+        break;
+    default:
+        windows_priority = THREAD_PRIORITY_NORMAL;
+        break;
+    }
+    SetThreadPriority(handle, windows_priority);
+}
+
+#else
+
+void SetCurrentThreadPriority(ThreadPriority new_priority) {
+    pthread_t this_thread = pthread_self();
+
+    s32 max_prio = sched_get_priority_max(SCHED_OTHER);
+    s32 min_prio = sched_get_priority_min(SCHED_OTHER);
+    u32 level = static_cast<u32>(new_priority) + 1;
+
+    struct sched_param params;
+    if (max_prio > min_prio) {
+        params.sched_priority = min_prio + ((max_prio - min_prio) * level) / 4;
+    } else {
+        params.sched_priority = min_prio - ((min_prio - max_prio) * level) / 4;
+    }
+
+    pthread_setschedparam(this_thread, SCHED_OTHER, &params);
+}
+
+#endif
+
 #ifdef _MSC_VER
 
 // Sets the debugger-visible name of the current thread.
@@ -70,6 +116,12 @@ void SetCurrentThreadName(const char* name) {
 }
 #endif
 
+#if defined(_WIN32)
+void SetCurrentThreadName(const char* name) {
+    // Do Nothing on MingW
+}
+#endif
+
 #endif
 
 } // namespace Common
diff --git a/src/common/thread.h b/src/common/thread.h
index 127cc7e23..52b359413 100644
--- a/src/common/thread.h
+++ b/src/common/thread.h
@@ -86,6 +86,15 @@ private:
     std::size_t generation = 0; // Incremented once each time the barrier is used
 };
 
+enum class ThreadPriority : u32 {
+    Low = 0,
+    Normal = 1,
+    High = 2,
+    VeryHigh = 3,
+};
+
+void SetCurrentThreadPriority(ThreadPriority new_priority);
+
 void SetCurrentThreadName(const char* name);
 
 } // namespace Common
diff --git a/src/common/wall_clock.cpp b/src/common/wall_clock.cpp
index d4d35f4e7..3afbdb898 100644
--- a/src/common/wall_clock.cpp
+++ b/src/common/wall_clock.cpp
@@ -53,6 +53,10 @@ public:
         return Common::Divide128On32(temporary, 1000000000).first;
     }
 
+    void Pause(bool is_paused) override {
+        // Do nothing in this clock type.
+    }
+
 private:
     base_time_point start_time;
 };
@@ -64,12 +68,7 @@ std::unique_ptr<WallClock> CreateBestMatchingClock(u32 emulated_cpu_frequency,
     const auto& caps = GetCPUCaps();
     u64 rtsc_frequency = 0;
     if (caps.invariant_tsc) {
-        if (caps.base_frequency != 0) {
-            rtsc_frequency = static_cast<u64>(caps.base_frequency) * 1000000U;
-        }
-        if (rtsc_frequency == 0) {
-            rtsc_frequency = EstimateRDTSCFrequency();
-        }
+        rtsc_frequency = EstimateRDTSCFrequency();
     }
     if (rtsc_frequency == 0) {
         return std::make_unique<StandardWallClock>(emulated_cpu_frequency,
diff --git a/src/common/wall_clock.h b/src/common/wall_clock.h
index ed284cf50..367d72134 100644
--- a/src/common/wall_clock.h
+++ b/src/common/wall_clock.h
@@ -28,6 +28,8 @@ public:
     /// Returns current wall time in emulated cpu cycles
     virtual u64 GetCPUCycles() = 0;
 
+    virtual void Pause(bool is_paused) = 0;
+
     /// Tells if the wall clock, uses the host CPU's hardware clock
     bool IsNative() const {
         return is_native;
diff --git a/src/common/x64/native_clock.cpp b/src/common/x64/native_clock.cpp
index 26d4d0ba6..424b39b1f 100644
--- a/src/common/x64/native_clock.cpp
+++ b/src/common/x64/native_clock.cpp
@@ -3,6 +3,7 @@
 // Refer to the license.txt file included.
 
 #include <chrono>
+#include <mutex>
 #include <thread>
 
 #ifdef _MSC_VER
@@ -52,7 +53,7 @@ NativeClock::NativeClock(u64 emulated_cpu_frequency, u64 emulated_clock_frequenc
 }
 
 u64 NativeClock::GetRTSC() {
-    rtsc_serialize.lock();
+    std::scoped_lock scope{rtsc_serialize};
     _mm_mfence();
     const u64 current_measure = __rdtsc();
     u64 diff = current_measure - last_measure;
@@ -61,8 +62,15 @@ u64 NativeClock::GetRTSC() {
         last_measure = current_measure;
     }
     accumulated_ticks += diff;
-    rtsc_serialize.unlock();
-    return accumulated_ticks;
+    /// The clock cannot be more precise than the guest timer, remove the lower bits
+    return accumulated_ticks & inaccuracy_mask;
+}
+
+void NativeClock::Pause(bool is_paused) {
+    if (!is_paused) {
+        _mm_mfence();
+        last_measure = __rdtsc();
+    }
 }
 
 std::chrono::nanoseconds NativeClock::GetTimeNS() {
diff --git a/src/common/x64/native_clock.h b/src/common/x64/native_clock.h
index b58cf9f5a..891a3bbfd 100644
--- a/src/common/x64/native_clock.h
+++ b/src/common/x64/native_clock.h
@@ -26,9 +26,16 @@ public:
 
     u64 GetCPUCycles() override;
 
+    void Pause(bool is_paused) override;
+
 private:
     u64 GetRTSC();
 
+    /// value used to reduce the native clocks accuracy as some apss rely on
+    /// undefined behavior where the level of accuracy in the clock shouldn't
+    /// be higher.
+    static constexpr u64 inaccuracy_mask = ~(0x400 - 1);
+
     SpinLock rtsc_serialize{};
     u64 last_measure{};
     u64 accumulated_ticks{};
diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt
index efbad628f..f87d67db5 100644
--- a/src/core/CMakeLists.txt
+++ b/src/core/CMakeLists.txt
@@ -7,6 +7,16 @@ endif()
 add_library(core STATIC
     arm/arm_interface.h
     arm/arm_interface.cpp
+    arm/cpu_interrupt_handler.cpp
+    arm/cpu_interrupt_handler.h
+    arm/dynarmic/arm_dynarmic_32.cpp
+    arm/dynarmic/arm_dynarmic_32.h
+    arm/dynarmic/arm_dynarmic_64.cpp
+    arm/dynarmic/arm_dynarmic_64.h
+    arm/dynarmic/arm_dynarmic_cp15.cpp
+    arm/dynarmic/arm_dynarmic_cp15.h
+    arm/dynarmic/arm_exclusive_monitor.cpp
+    arm/dynarmic/arm_exclusive_monitor.h
     arm/exclusive_monitor.cpp
     arm/exclusive_monitor.h
     arm/unicorn/arm_unicorn.cpp
@@ -15,8 +25,6 @@ add_library(core STATIC
     constants.h
     core.cpp
     core.h
-    core_manager.cpp
-    core_manager.h
     core_timing.cpp
     core_timing.h
     core_timing_util.cpp
@@ -547,8 +555,6 @@ add_library(core STATIC
     hle/service/vi/vi_u.h
     hle/service/wlan/wlan.cpp
     hle/service/wlan/wlan.h
-    host_timing.cpp
-    host_timing.h
     loader/deconstructed_rom_directory.cpp
     loader/deconstructed_rom_directory.h
     loader/elf.cpp
diff --git a/src/core/arm/arm_interface.cpp b/src/core/arm/arm_interface.cpp
index d079a1bc8..d2295ed90 100644
--- a/src/core/arm/arm_interface.cpp
+++ b/src/core/arm/arm_interface.cpp
@@ -139,6 +139,63 @@ std::optional<std::string> GetSymbolName(const Symbols& symbols, VAddr func_addr
 
 constexpr u64 SEGMENT_BASE = 0x7100000000ull;
 
+std::vector<ARM_Interface::BacktraceEntry> ARM_Interface::GetBacktraceFromContext(
+    System& system, const ThreadContext64& ctx) {
+    std::vector<BacktraceEntry> out;
+    auto& memory = system.Memory();
+
+    auto fp = ctx.cpu_registers[29];
+    auto lr = ctx.cpu_registers[30];
+    while (true) {
+        out.push_back({"", 0, lr, 0});
+        if (!fp) {
+            break;
+        }
+        lr = memory.Read64(fp + 8) - 4;
+        fp = memory.Read64(fp);
+    }
+
+    std::map<VAddr, std::string> modules;
+    auto& loader{system.GetAppLoader()};
+    if (loader.ReadNSOModules(modules) != Loader::ResultStatus::Success) {
+        return {};
+    }
+
+    std::map<std::string, Symbols> symbols;
+    for (const auto& module : modules) {
+        symbols.insert_or_assign(module.second, GetSymbols(module.first, memory));
+    }
+
+    for (auto& entry : out) {
+        VAddr base = 0;
+        for (auto iter = modules.rbegin(); iter != modules.rend(); ++iter) {
+            const auto& module{*iter};
+            if (entry.original_address >= module.first) {
+                entry.module = module.second;
+                base = module.first;
+                break;
+            }
+        }
+
+        entry.offset = entry.original_address - base;
+        entry.address = SEGMENT_BASE + entry.offset;
+
+        if (entry.module.empty())
+            entry.module = "unknown";
+
+        const auto symbol_set = symbols.find(entry.module);
+        if (symbol_set != symbols.end()) {
+            const auto symbol = GetSymbolName(symbol_set->second, entry.offset);
+            if (symbol.has_value()) {
+                // TODO(DarkLordZach): Add demangling of symbol names.
+                entry.name = *symbol;
+            }
+        }
+    }
+
+    return out;
+}
+
 std::vector<ARM_Interface::BacktraceEntry> ARM_Interface::GetBacktrace() const {
     std::vector<BacktraceEntry> out;
     auto& memory = system.Memory();
diff --git a/src/core/arm/arm_interface.h b/src/core/arm/arm_interface.h
index cb2e640e2..1f24051e4 100644
--- a/src/core/arm/arm_interface.h
+++ b/src/core/arm/arm_interface.h
@@ -7,6 +7,7 @@
 #include <array>
 #include <vector>
 #include "common/common_types.h"
+#include "core/hardware_properties.h"
 
 namespace Common {
 struct PageTable;
@@ -18,25 +19,29 @@ enum class VMAPermission : u8;
 
 namespace Core {
 class System;
+class CPUInterruptHandler;
+
+using CPUInterrupts = std::array<CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>;
 
 /// Generic ARMv8 CPU interface
 class ARM_Interface : NonCopyable {
 public:
-    explicit ARM_Interface(System& system_) : system{system_} {}
+    explicit ARM_Interface(System& system_, CPUInterrupts& interrupt_handlers, bool uses_wall_clock)
+        : system{system_}, interrupt_handlers{interrupt_handlers}, uses_wall_clock{
+                                                                       uses_wall_clock} {}
     virtual ~ARM_Interface() = default;
 
     struct ThreadContext32 {
         std::array<u32, 16> cpu_registers{};
+        std::array<u32, 64> extension_registers{};
         u32 cpsr{};
-        std::array<u8, 4> padding{};
-        std::array<u64, 32> fprs{};
         u32 fpscr{};
         u32 fpexc{};
         u32 tpidr{};
     };
     // Internally within the kernel, it expects the AArch32 version of the
     // thread context to be 344 bytes in size.
-    static_assert(sizeof(ThreadContext32) == 0x158);
+    static_assert(sizeof(ThreadContext32) == 0x150);
 
     struct ThreadContext64 {
         std::array<u64, 31> cpu_registers{};
@@ -143,6 +148,8 @@ public:
      */
     virtual void SetTPIDR_EL0(u64 value) = 0;
 
+    virtual void ChangeProcessorID(std::size_t new_core_id) = 0;
+
     virtual void SaveContext(ThreadContext32& ctx) = 0;
     virtual void SaveContext(ThreadContext64& ctx) = 0;
     virtual void LoadContext(const ThreadContext32& ctx) = 0;
@@ -162,6 +169,9 @@ public:
         std::string name;
     };
 
+    static std::vector<BacktraceEntry> GetBacktraceFromContext(System& system,
+                                                               const ThreadContext64& ctx);
+
     std::vector<BacktraceEntry> GetBacktrace() const;
 
     /// fp (= r29) points to the last frame record.
@@ -175,6 +185,8 @@ public:
 protected:
     /// System context that this ARM interface is running under.
     System& system;
+    CPUInterrupts& interrupt_handlers;
+    bool uses_wall_clock;
 };
 
 } // namespace Core
diff --git a/src/core/arm/cpu_interrupt_handler.cpp b/src/core/arm/cpu_interrupt_handler.cpp
new file mode 100644
index 000000000..2f1a1a269
--- /dev/null
+++ b/src/core/arm/cpu_interrupt_handler.cpp
@@ -0,0 +1,29 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/thread.h"
+#include "core/arm/cpu_interrupt_handler.h"
+
+namespace Core {
+
+CPUInterruptHandler::CPUInterruptHandler() : is_interrupted{} {
+    interrupt_event = std::make_unique<Common::Event>();
+}
+
+CPUInterruptHandler::~CPUInterruptHandler() = default;
+
+void CPUInterruptHandler::SetInterrupt(bool is_interrupted_) {
+    if (is_interrupted_) {
+        interrupt_event->Set();
+    }
+    this->is_interrupted = is_interrupted_;
+}
+
+void CPUInterruptHandler::AwaitInterrupt() {
+    interrupt_event->Wait();
+}
+
+} // namespace Core
diff --git a/src/core/arm/cpu_interrupt_handler.h b/src/core/arm/cpu_interrupt_handler.h
new file mode 100644
index 000000000..3d062d326
--- /dev/null
+++ b/src/core/arm/cpu_interrupt_handler.h
@@ -0,0 +1,39 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+
+namespace Common {
+class Event;
+}
+
+namespace Core {
+
+class CPUInterruptHandler {
+public:
+    CPUInterruptHandler();
+    ~CPUInterruptHandler();
+
+    CPUInterruptHandler(const CPUInterruptHandler&) = delete;
+    CPUInterruptHandler& operator=(const CPUInterruptHandler&) = delete;
+
+    CPUInterruptHandler(CPUInterruptHandler&&) = default;
+    CPUInterruptHandler& operator=(CPUInterruptHandler&&) = default;
+
+    bool IsInterrupted() const {
+        return is_interrupted;
+    }
+
+    void SetInterrupt(bool is_interrupted);
+
+    void AwaitInterrupt();
+
+private:
+    bool is_interrupted{};
+    std::unique_ptr<Common::Event> interrupt_event;
+};
+
+} // namespace Core
diff --git a/src/core/arm/dynarmic/arm_dynarmic_32.cpp b/src/core/arm/dynarmic/arm_dynarmic_32.cpp
index 4c8663d03..0d4ab95b7 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_32.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic_32.cpp
@@ -7,15 +7,17 @@
 #include <dynarmic/A32/a32.h>
 #include <dynarmic/A32/config.h>
 #include <dynarmic/A32/context.h>
-#include "common/microprofile.h"
+#include "common/logging/log.h"
+#include "common/page_table.h"
+#include "core/arm/cpu_interrupt_handler.h"
 #include "core/arm/dynarmic/arm_dynarmic_32.h"
-#include "core/arm/dynarmic/arm_dynarmic_64.h"
 #include "core/arm/dynarmic/arm_dynarmic_cp15.h"
+#include "core/arm/dynarmic/arm_exclusive_monitor.h"
 #include "core/core.h"
-#include "core/core_manager.h"
 #include "core/core_timing.h"
 #include "core/hle/kernel/svc.h"
 #include "core/memory.h"
+#include "core/settings.h"
 
 namespace Core {
 
@@ -49,6 +51,19 @@ public:
         parent.system.Memory().Write64(vaddr, value);
     }
 
+    bool MemoryWriteExclusive8(u32 vaddr, u8 value, u8 expected) override {
+        return parent.system.Memory().WriteExclusive8(vaddr, value, expected);
+    }
+    bool MemoryWriteExclusive16(u32 vaddr, u16 value, u16 expected) override {
+        return parent.system.Memory().WriteExclusive16(vaddr, value, expected);
+    }
+    bool MemoryWriteExclusive32(u32 vaddr, u32 value, u32 expected) override {
+        return parent.system.Memory().WriteExclusive32(vaddr, value, expected);
+    }
+    bool MemoryWriteExclusive64(u32 vaddr, u64 value, u64 expected) override {
+        return parent.system.Memory().WriteExclusive64(vaddr, value, expected);
+    }
+
     void InterpreterFallback(u32 pc, std::size_t num_instructions) override {
         UNIMPLEMENTED_MSG("This should never happen, pc = {:08X}, code = {:08X}", pc,
                           MemoryReadCode(pc));
@@ -72,24 +87,36 @@ public:
     }
 
     void AddTicks(u64 ticks) override {
+        if (parent.uses_wall_clock) {
+            return;
+        }
         // Divide the number of ticks by the amount of CPU cores. TODO(Subv): This yields only a
         // rough approximation of the amount of executed ticks in the system, it may be thrown off
         // if not all cores are doing a similar amount of work. Instead of doing this, we should
         // device a way so that timing is consistent across all cores without increasing the ticks 4
         // times.
-        u64 amortized_ticks = (ticks - num_interpreted_instructions) / Core::NUM_CPU_CORES;
+        u64 amortized_ticks =
+            (ticks - num_interpreted_instructions) / Core::Hardware::NUM_CPU_CORES;
         // Always execute at least one tick.
         amortized_ticks = std::max<u64>(amortized_ticks, 1);
 
         parent.system.CoreTiming().AddTicks(amortized_ticks);
         num_interpreted_instructions = 0;
     }
+
     u64 GetTicksRemaining() override {
-        return std::max(parent.system.CoreTiming().GetDowncount(), {});
+        if (parent.uses_wall_clock) {
+            if (!parent.interrupt_handlers[parent.core_index].IsInterrupted()) {
+                return minimum_run_cycles;
+            }
+            return 0U;
+        }
+        return std::max<s64>(parent.system.CoreTiming().GetDowncount(), 0);
     }
 
     ARM_Dynarmic_32& parent;
     std::size_t num_interpreted_instructions{};
+    static constexpr u64 minimum_run_cycles = 1000U;
 };
 
 std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable& page_table,
@@ -100,13 +127,31 @@ std::shared_ptr<Dynarmic::A32::Jit> ARM_Dynarmic_32::MakeJit(Common::PageTable&
     // config.page_table = &page_table.pointers;
     config.coprocessors[15] = cp15;
     config.define_unpredictable_behaviour = true;
+    static constexpr std::size_t PAGE_BITS = 12;
+    static constexpr std::size_t NUM_PAGE_TABLE_ENTRIES = 1 << (32 - PAGE_BITS);
+    config.page_table = reinterpret_cast<std::array<std::uint8_t*, NUM_PAGE_TABLE_ENTRIES>*>(
+        page_table.pointers.data());
+    config.absolute_offset_page_table = true;
+    config.detect_misaligned_access_via_page_table = 16 | 32 | 64 | 128;
+    config.only_detect_misalignment_via_page_table_on_page_boundary = true;
+
+    // Multi-process state
+    config.processor_id = core_index;
+    config.global_monitor = &exclusive_monitor.monitor;
+
+    // Timing
+    config.wall_clock_cntpct = uses_wall_clock;
+
+    // Optimizations
+    if (Settings::values.disable_cpu_opt) {
+        config.enable_optimizations = false;
+        config.enable_fast_dispatch = false;
+    }
+
     return std::make_unique<Dynarmic::A32::Jit>(config);
 }
 
-MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_32, "ARM JIT", "Dynarmic", MP_RGB(255, 64, 64));
-
 void ARM_Dynarmic_32::Run() {
-    MICROPROFILE_SCOPE(ARM_Jit_Dynarmic_32);
     jit->Run();
 }
 
@@ -114,9 +159,11 @@ void ARM_Dynarmic_32::Step() {
     jit->Step();
 }
 
-ARM_Dynarmic_32::ARM_Dynarmic_32(System& system, ExclusiveMonitor& exclusive_monitor,
+ARM_Dynarmic_32::ARM_Dynarmic_32(System& system, CPUInterrupts& interrupt_handlers,
+                                 bool uses_wall_clock, ExclusiveMonitor& exclusive_monitor,
                                  std::size_t core_index)
-    : ARM_Interface{system}, cb(std::make_unique<DynarmicCallbacks32>(*this)),
+    : ARM_Interface{system, interrupt_handlers, uses_wall_clock},
+      cb(std::make_unique<DynarmicCallbacks32>(*this)),
       cp15(std::make_shared<DynarmicCP15>(*this)), core_index{core_index},
       exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}
 
@@ -168,17 +215,25 @@ void ARM_Dynarmic_32::SetTPIDR_EL0(u64 value) {
     cp15->uprw = static_cast<u32>(value);
 }
 
+void ARM_Dynarmic_32::ChangeProcessorID(std::size_t new_core_id) {
+    jit->ChangeProcessorID(new_core_id);
+}
+
 void ARM_Dynarmic_32::SaveContext(ThreadContext32& ctx) {
     Dynarmic::A32::Context context;
     jit->SaveContext(context);
     ctx.cpu_registers = context.Regs();
+    ctx.extension_registers = context.ExtRegs();
     ctx.cpsr = context.Cpsr();
+    ctx.fpscr = context.Fpscr();
 }
 
 void ARM_Dynarmic_32::LoadContext(const ThreadContext32& ctx) {
     Dynarmic::A32::Context context;
     context.Regs() = ctx.cpu_registers;
+    context.ExtRegs() = ctx.extension_registers;
     context.SetCpsr(ctx.cpsr);
+    context.SetFpscr(ctx.fpscr);
     jit->LoadContext(context);
 }
 
@@ -187,10 +242,15 @@ void ARM_Dynarmic_32::PrepareReschedule() {
 }
 
 void ARM_Dynarmic_32::ClearInstructionCache() {
+    if (!jit) {
+        return;
+    }
     jit->ClearCache();
 }
 
-void ARM_Dynarmic_32::ClearExclusiveState() {}
+void ARM_Dynarmic_32::ClearExclusiveState() {
+    jit->ClearExclusiveState();
+}
 
 void ARM_Dynarmic_32::PageTableChanged(Common::PageTable& page_table,
                                        std::size_t new_address_space_size_in_bits) {
diff --git a/src/core/arm/dynarmic/arm_dynarmic_32.h b/src/core/arm/dynarmic/arm_dynarmic_32.h
index e5b92d7bb..2bab31b92 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_32.h
+++ b/src/core/arm/dynarmic/arm_dynarmic_32.h
@@ -9,7 +9,7 @@
 
 #include <dynarmic/A32/a32.h>
 #include <dynarmic/A64/a64.h>
-#include <dynarmic/A64/exclusive_monitor.h>
+#include <dynarmic/exclusive_monitor.h>
 #include "common/common_types.h"
 #include "common/hash.h"
 #include "core/arm/arm_interface.h"
@@ -21,6 +21,7 @@ class Memory;
 
 namespace Core {
 
+class CPUInterruptHandler;
 class DynarmicCallbacks32;
 class DynarmicCP15;
 class DynarmicExclusiveMonitor;
@@ -28,7 +29,8 @@ class System;
 
 class ARM_Dynarmic_32 final : public ARM_Interface {
 public:
-    ARM_Dynarmic_32(System& system, ExclusiveMonitor& exclusive_monitor, std::size_t core_index);
+    ARM_Dynarmic_32(System& system, CPUInterrupts& interrupt_handlers, bool uses_wall_clock,
+                    ExclusiveMonitor& exclusive_monitor, std::size_t core_index);
     ~ARM_Dynarmic_32() override;
 
     void SetPC(u64 pc) override;
@@ -45,6 +47,7 @@ public:
     void SetTlsAddress(VAddr address) override;
     void SetTPIDR_EL0(u64 value) override;
     u64 GetTPIDR_EL0() const override;
+    void ChangeProcessorID(std::size_t new_core_id) override;
 
     void SaveContext(ThreadContext32& ctx) override;
     void SaveContext(ThreadContext64& ctx) override {}
diff --git a/src/core/arm/dynarmic/arm_dynarmic_64.cpp b/src/core/arm/dynarmic/arm_dynarmic_64.cpp
index 5f5e36d94..790981034 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_64.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic_64.cpp
@@ -7,11 +7,11 @@
 #include <dynarmic/A64/a64.h>
 #include <dynarmic/A64/config.h>
 #include "common/logging/log.h"
-#include "common/microprofile.h"
 #include "common/page_table.h"
+#include "core/arm/cpu_interrupt_handler.h"
 #include "core/arm/dynarmic/arm_dynarmic_64.h"
+#include "core/arm/dynarmic/arm_exclusive_monitor.h"
 #include "core/core.h"
-#include "core/core_manager.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
 #include "core/gdbstub/gdbstub.h"
@@ -65,6 +65,22 @@ public:
         memory.Write64(vaddr + 8, value[1]);
     }
 
+    bool MemoryWriteExclusive8(u64 vaddr, std::uint8_t value, std::uint8_t expected) override {
+        return parent.system.Memory().WriteExclusive8(vaddr, value, expected);
+    }
+    bool MemoryWriteExclusive16(u64 vaddr, std::uint16_t value, std::uint16_t expected) override {
+        return parent.system.Memory().WriteExclusive16(vaddr, value, expected);
+    }
+    bool MemoryWriteExclusive32(u64 vaddr, std::uint32_t value, std::uint32_t expected) override {
+        return parent.system.Memory().WriteExclusive32(vaddr, value, expected);
+    }
+    bool MemoryWriteExclusive64(u64 vaddr, std::uint64_t value, std::uint64_t expected) override {
+        return parent.system.Memory().WriteExclusive64(vaddr, value, expected);
+    }
+    bool MemoryWriteExclusive128(u64 vaddr, Vector value, Vector expected) override {
+        return parent.system.Memory().WriteExclusive128(vaddr, value, expected);
+    }
+
     void InterpreterFallback(u64 pc, std::size_t num_instructions) override {
         LOG_INFO(Core_ARM, "Unicorn fallback @ 0x{:X} for {} instructions (instr = {:08X})", pc,
                  num_instructions, MemoryReadCode(pc));
@@ -108,29 +124,42 @@ public:
     }
 
     void AddTicks(u64 ticks) override {
+        if (parent.uses_wall_clock) {
+            return;
+        }
         // Divide the number of ticks by the amount of CPU cores. TODO(Subv): This yields only a
         // rough approximation of the amount of executed ticks in the system, it may be thrown off
         // if not all cores are doing a similar amount of work. Instead of doing this, we should
         // device a way so that timing is consistent across all cores without increasing the ticks 4
         // times.
-        u64 amortized_ticks = (ticks - num_interpreted_instructions) / Core::NUM_CPU_CORES;
+        u64 amortized_ticks =
+            (ticks - num_interpreted_instructions) / Core::Hardware::NUM_CPU_CORES;
         // Always execute at least one tick.
         amortized_ticks = std::max<u64>(amortized_ticks, 1);
 
         parent.system.CoreTiming().AddTicks(amortized_ticks);
         num_interpreted_instructions = 0;
     }
+
     u64 GetTicksRemaining() override {
-        return std::max(parent.system.CoreTiming().GetDowncount(), s64{0});
+        if (parent.uses_wall_clock) {
+            if (!parent.interrupt_handlers[parent.core_index].IsInterrupted()) {
+                return minimum_run_cycles;
+            }
+            return 0U;
+        }
+        return std::max<s64>(parent.system.CoreTiming().GetDowncount(), 0);
     }
+
     u64 GetCNTPCT() override {
-        return Timing::CpuCyclesToClockCycles(parent.system.CoreTiming().GetTicks());
+        return parent.system.CoreTiming().GetClockTicks();
     }
 
     ARM_Dynarmic_64& parent;
     std::size_t num_interpreted_instructions = 0;
     u64 tpidrro_el0 = 0;
     u64 tpidr_el0 = 0;
+    static constexpr u64 minimum_run_cycles = 1000U;
 };
 
 std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable& page_table,
@@ -168,14 +197,13 @@ std::shared_ptr<Dynarmic::A64::Jit> ARM_Dynarmic_64::MakeJit(Common::PageTable&
         config.enable_fast_dispatch = false;
     }
 
+    // Timing
+    config.wall_clock_cntpct = uses_wall_clock;
+
     return std::make_shared<Dynarmic::A64::Jit>(config);
 }
 
-MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_64, "ARM JIT", "Dynarmic", MP_RGB(255, 64, 64));
-
 void ARM_Dynarmic_64::Run() {
-    MICROPROFILE_SCOPE(ARM_Jit_Dynarmic_64);
-
     jit->Run();
 }
 
@@ -183,11 +211,16 @@ void ARM_Dynarmic_64::Step() {
     cb->InterpreterFallback(jit->GetPC(), 1);
 }
 
-ARM_Dynarmic_64::ARM_Dynarmic_64(System& system, ExclusiveMonitor& exclusive_monitor,
+ARM_Dynarmic_64::ARM_Dynarmic_64(System& system, CPUInterrupts& interrupt_handlers,
+                                 bool uses_wall_clock, ExclusiveMonitor& exclusive_monitor,
                                  std::size_t core_index)
-    : ARM_Interface{system}, cb(std::make_unique<DynarmicCallbacks64>(*this)),
-      inner_unicorn{system, ARM_Unicorn::Arch::AArch64}, core_index{core_index},
-      exclusive_monitor{dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}
+    : ARM_Interface{system, interrupt_handlers, uses_wall_clock},
+      cb(std::make_unique<DynarmicCallbacks64>(*this)), inner_unicorn{system, interrupt_handlers,
+                                                                      uses_wall_clock,
+                                                                      ARM_Unicorn::Arch::AArch64,
+                                                                      core_index},
+      core_index{core_index}, exclusive_monitor{
+                                  dynamic_cast<DynarmicExclusiveMonitor&>(exclusive_monitor)} {}
 
 ARM_Dynarmic_64::~ARM_Dynarmic_64() = default;
 
@@ -239,6 +272,10 @@ void ARM_Dynarmic_64::SetTPIDR_EL0(u64 value) {
     cb->tpidr_el0 = value;
 }
 
+void ARM_Dynarmic_64::ChangeProcessorID(std::size_t new_core_id) {
+    jit->ChangeProcessorID(new_core_id);
+}
+
 void ARM_Dynarmic_64::SaveContext(ThreadContext64& ctx) {
     ctx.cpu_registers = jit->GetRegisters();
     ctx.sp = jit->GetSP();
@@ -266,6 +303,9 @@ void ARM_Dynarmic_64::PrepareReschedule() {
 }
 
 void ARM_Dynarmic_64::ClearInstructionCache() {
+    if (!jit) {
+        return;
+    }
     jit->ClearCache();
 }
 
@@ -285,44 +325,4 @@ void ARM_Dynarmic_64::PageTableChanged(Common::PageTable& page_table,
     jit_cache.emplace(key, jit);
 }
 
-DynarmicExclusiveMonitor::DynarmicExclusiveMonitor(Memory::Memory& memory, std::size_t core_count)
-    : monitor(core_count), memory{memory} {}
-
-DynarmicExclusiveMonitor::~DynarmicExclusiveMonitor() = default;
-
-void DynarmicExclusiveMonitor::SetExclusive(std::size_t core_index, VAddr addr) {
-    // Size doesn't actually matter.
-    monitor.Mark(core_index, addr, 16);
-}
-
-void DynarmicExclusiveMonitor::ClearExclusive() {
-    monitor.Clear();
-}
-
-bool DynarmicExclusiveMonitor::ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) {
-    return monitor.DoExclusiveOperation(core_index, vaddr, 1, [&] { memory.Write8(vaddr, value); });
-}
-
-bool DynarmicExclusiveMonitor::ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) {
-    return monitor.DoExclusiveOperation(core_index, vaddr, 2,
-                                        [&] { memory.Write16(vaddr, value); });
-}
-
-bool DynarmicExclusiveMonitor::ExclusiveWrite32(std::size_t core_index, VAddr vaddr, u32 value) {
-    return monitor.DoExclusiveOperation(core_index, vaddr, 4,
-                                        [&] { memory.Write32(vaddr, value); });
-}
-
-bool DynarmicExclusiveMonitor::ExclusiveWrite64(std::size_t core_index, VAddr vaddr, u64 value) {
-    return monitor.DoExclusiveOperation(core_index, vaddr, 8,
-                                        [&] { memory.Write64(vaddr, value); });
-}
-
-bool DynarmicExclusiveMonitor::ExclusiveWrite128(std::size_t core_index, VAddr vaddr, u128 value) {
-    return monitor.DoExclusiveOperation(core_index, vaddr, 16, [&] {
-        memory.Write64(vaddr + 0, value[0]);
-        memory.Write64(vaddr + 8, value[1]);
-    });
-}
-
 } // namespace Core
diff --git a/src/core/arm/dynarmic/arm_dynarmic_64.h b/src/core/arm/dynarmic/arm_dynarmic_64.h
index 647cecaf0..403c55961 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_64.h
+++ b/src/core/arm/dynarmic/arm_dynarmic_64.h
@@ -8,7 +8,6 @@
 #include <unordered_map>
 
 #include <dynarmic/A64/a64.h>
-#include <dynarmic/A64/exclusive_monitor.h>
 #include "common/common_types.h"
 #include "common/hash.h"
 #include "core/arm/arm_interface.h"
@@ -22,12 +21,14 @@ class Memory;
 namespace Core {
 
 class DynarmicCallbacks64;
+class CPUInterruptHandler;
 class DynarmicExclusiveMonitor;
 class System;
 
 class ARM_Dynarmic_64 final : public ARM_Interface {
 public:
-    ARM_Dynarmic_64(System& system, ExclusiveMonitor& exclusive_monitor, std::size_t core_index);
+    ARM_Dynarmic_64(System& system, CPUInterrupts& interrupt_handlers, bool uses_wall_clock,
+                    ExclusiveMonitor& exclusive_monitor, std::size_t core_index);
     ~ARM_Dynarmic_64() override;
 
     void SetPC(u64 pc) override;
@@ -44,6 +45,7 @@ public:
     void SetTlsAddress(VAddr address) override;
     void SetTPIDR_EL0(u64 value) override;
     u64 GetTPIDR_EL0() const override;
+    void ChangeProcessorID(std::size_t new_core_id) override;
 
     void SaveContext(ThreadContext32& ctx) override {}
     void SaveContext(ThreadContext64& ctx) override;
@@ -75,24 +77,4 @@ private:
     DynarmicExclusiveMonitor& exclusive_monitor;
 };
 
-class DynarmicExclusiveMonitor final : public ExclusiveMonitor {
-public:
-    explicit DynarmicExclusiveMonitor(Memory::Memory& memory, std::size_t core_count);
-    ~DynarmicExclusiveMonitor() override;
-
-    void SetExclusive(std::size_t core_index, VAddr addr) override;
-    void ClearExclusive() override;
-
-    bool ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) override;
-    bool ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) override;
-    bool ExclusiveWrite32(std::size_t core_index, VAddr vaddr, u32 value) override;
-    bool ExclusiveWrite64(std::size_t core_index, VAddr vaddr, u64 value) override;
-    bool ExclusiveWrite128(std::size_t core_index, VAddr vaddr, u128 value) override;
-
-private:
-    friend class ARM_Dynarmic_64;
-    Dynarmic::A64::ExclusiveMonitor monitor;
-    Core::Memory::Memory& memory;
-};
-
 } // namespace Core
diff --git a/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp b/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp
index d43e4dd70..54556e0f9 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic_cp15.cpp
@@ -97,7 +97,7 @@ CallbackOrAccessTwoWords DynarmicCP15::CompileGetTwoWords(bool two, unsigned opc
         const auto callback = static_cast<u64 (*)(Dynarmic::A32::Jit*, void*, u32, u32)>(
             [](Dynarmic::A32::Jit*, void* arg, u32, u32) -> u64 {
                 ARM_Dynarmic_32& parent = *(ARM_Dynarmic_32*)arg;
-                return Timing::CpuCyclesToClockCycles(parent.system.CoreTiming().GetTicks());
+                return parent.system.CoreTiming().GetClockTicks();
             });
         return Dynarmic::A32::Coprocessor::Callback{callback, (void*)&parent};
     }
diff --git a/src/core/arm/dynarmic/arm_exclusive_monitor.cpp b/src/core/arm/dynarmic/arm_exclusive_monitor.cpp
new file mode 100644
index 000000000..4e209f6a5
--- /dev/null
+++ b/src/core/arm/dynarmic/arm_exclusive_monitor.cpp
@@ -0,0 +1,76 @@
+// Copyright 2018 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <cinttypes>
+#include <memory>
+#include "core/arm/dynarmic/arm_exclusive_monitor.h"
+#include "core/memory.h"
+
+namespace Core {
+
+DynarmicExclusiveMonitor::DynarmicExclusiveMonitor(Memory::Memory& memory, std::size_t core_count)
+    : monitor(core_count), memory{memory} {}
+
+DynarmicExclusiveMonitor::~DynarmicExclusiveMonitor() = default;
+
+u8 DynarmicExclusiveMonitor::ExclusiveRead8(std::size_t core_index, VAddr addr) {
+    return monitor.ReadAndMark<u8>(core_index, addr, [&]() -> u8 { return memory.Read8(addr); });
+}
+
+u16 DynarmicExclusiveMonitor::ExclusiveRead16(std::size_t core_index, VAddr addr) {
+    return monitor.ReadAndMark<u16>(core_index, addr, [&]() -> u16 { return memory.Read16(addr); });
+}
+
+u32 DynarmicExclusiveMonitor::ExclusiveRead32(std::size_t core_index, VAddr addr) {
+    return monitor.ReadAndMark<u32>(core_index, addr, [&]() -> u32 { return memory.Read32(addr); });
+}
+
+u64 DynarmicExclusiveMonitor::ExclusiveRead64(std::size_t core_index, VAddr addr) {
+    return monitor.ReadAndMark<u64>(core_index, addr, [&]() -> u64 { return memory.Read64(addr); });
+}
+
+u128 DynarmicExclusiveMonitor::ExclusiveRead128(std::size_t core_index, VAddr addr) {
+    return monitor.ReadAndMark<u128>(core_index, addr, [&]() -> u128 {
+        u128 result;
+        result[0] = memory.Read64(addr);
+        result[1] = memory.Read64(addr + 8);
+        return result;
+    });
+}
+
+void DynarmicExclusiveMonitor::ClearExclusive() {
+    monitor.Clear();
+}
+
+bool DynarmicExclusiveMonitor::ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) {
+    return monitor.DoExclusiveOperation<u8>(core_index, vaddr, [&](u8 expected) -> bool {
+        return memory.WriteExclusive8(vaddr, value, expected);
+    });
+}
+
+bool DynarmicExclusiveMonitor::ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) {
+    return monitor.DoExclusiveOperation<u16>(core_index, vaddr, [&](u16 expected) -> bool {
+        return memory.WriteExclusive16(vaddr, value, expected);
+    });
+}
+
+bool DynarmicExclusiveMonitor::ExclusiveWrite32(std::size_t core_index, VAddr vaddr, u32 value) {
+    return monitor.DoExclusiveOperation<u32>(core_index, vaddr, [&](u32 expected) -> bool {
+        return memory.WriteExclusive32(vaddr, value, expected);
+    });
+}
+
+bool DynarmicExclusiveMonitor::ExclusiveWrite64(std::size_t core_index, VAddr vaddr, u64 value) {
+    return monitor.DoExclusiveOperation<u64>(core_index, vaddr, [&](u64 expected) -> bool {
+        return memory.WriteExclusive64(vaddr, value, expected);
+    });
+}
+
+bool DynarmicExclusiveMonitor::ExclusiveWrite128(std::size_t core_index, VAddr vaddr, u128 value) {
+    return monitor.DoExclusiveOperation<u128>(core_index, vaddr, [&](u128 expected) -> bool {
+        return memory.WriteExclusive128(vaddr, value, expected);
+    });
+}
+
+} // namespace Core
diff --git a/src/core/arm/dynarmic/arm_exclusive_monitor.h b/src/core/arm/dynarmic/arm_exclusive_monitor.h
new file mode 100644
index 000000000..964f4a55d
--- /dev/null
+++ b/src/core/arm/dynarmic/arm_exclusive_monitor.h
@@ -0,0 +1,48 @@
+// Copyright 2020 yuzu emulator team
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <memory>
+#include <unordered_map>
+
+#include <dynarmic/exclusive_monitor.h>
+
+#include "common/common_types.h"
+#include "core/arm/dynarmic/arm_dynarmic_32.h"
+#include "core/arm/dynarmic/arm_dynarmic_64.h"
+#include "core/arm/exclusive_monitor.h"
+
+namespace Core::Memory {
+class Memory;
+}
+
+namespace Core {
+
+class DynarmicExclusiveMonitor final : public ExclusiveMonitor {
+public:
+    explicit DynarmicExclusiveMonitor(Memory::Memory& memory, std::size_t core_count);
+    ~DynarmicExclusiveMonitor() override;
+
+    u8 ExclusiveRead8(std::size_t core_index, VAddr addr) override;
+    u16 ExclusiveRead16(std::size_t core_index, VAddr addr) override;
+    u32 ExclusiveRead32(std::size_t core_index, VAddr addr) override;
+    u64 ExclusiveRead64(std::size_t core_index, VAddr addr) override;
+    u128 ExclusiveRead128(std::size_t core_index, VAddr addr) override;
+    void ClearExclusive() override;
+
+    bool ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) override;
+    bool ExclusiveWrite16(std::size_t core_index, VAddr vaddr, u16 value) override;
+    bool ExclusiveWrite32(std::size_t core_index, VAddr vaddr, u32 value) override;
+    bool ExclusiveWrite64(std::size_t core_index, VAddr vaddr, u64 value) override;
+    bool ExclusiveWrite128(std::size_t core_index, VAddr vaddr, u128 value) override;
+
+private:
+    friend class ARM_Dynarmic_32;
+    friend class ARM_Dynarmic_64;
+    Dynarmic::ExclusiveMonitor monitor;
+    Core::Memory::Memory& memory;
+};
+
+} // namespace Core
diff --git a/src/core/arm/exclusive_monitor.cpp b/src/core/arm/exclusive_monitor.cpp
index b32401e0b..d8cba369d 100644
--- a/src/core/arm/exclusive_monitor.cpp
+++ b/src/core/arm/exclusive_monitor.cpp
@@ -3,7 +3,7 @@
 // Refer to the license.txt file included.
 
 #ifdef ARCHITECTURE_x86_64
-#include "core/arm/dynarmic/arm_dynarmic_64.h"
+#include "core/arm/dynarmic/arm_exclusive_monitor.h"
 #endif
 #include "core/arm/exclusive_monitor.h"
 #include "core/memory.h"
diff --git a/src/core/arm/exclusive_monitor.h b/src/core/arm/exclusive_monitor.h
index ccd73b80f..62f6e6023 100644
--- a/src/core/arm/exclusive_monitor.h
+++ b/src/core/arm/exclusive_monitor.h
@@ -18,7 +18,11 @@ class ExclusiveMonitor {
 public:
     virtual ~ExclusiveMonitor();
 
-    virtual void SetExclusive(std::size_t core_index, VAddr addr) = 0;
+    virtual u8 ExclusiveRead8(std::size_t core_index, VAddr addr) = 0;
+    virtual u16 ExclusiveRead16(std::size_t core_index, VAddr addr) = 0;
+    virtual u32 ExclusiveRead32(std::size_t core_index, VAddr addr) = 0;
+    virtual u64 ExclusiveRead64(std::size_t core_index, VAddr addr) = 0;
+    virtual u128 ExclusiveRead128(std::size_t core_index, VAddr addr) = 0;
     virtual void ClearExclusive() = 0;
 
     virtual bool ExclusiveWrite8(std::size_t core_index, VAddr vaddr, u8 value) = 0;
diff --git a/src/core/arm/unicorn/arm_unicorn.cpp b/src/core/arm/unicorn/arm_unicorn.cpp
index e40e9626a..1df3f3ed1 100644
--- a/src/core/arm/unicorn/arm_unicorn.cpp
+++ b/src/core/arm/unicorn/arm_unicorn.cpp
@@ -6,6 +6,7 @@
 #include <unicorn/arm64.h>
 #include "common/assert.h"
 #include "common/microprofile.h"
+#include "core/arm/cpu_interrupt_handler.h"
 #include "core/arm/unicorn/arm_unicorn.h"
 #include "core/core.h"
 #include "core/core_timing.h"
@@ -62,7 +63,9 @@ static bool UnmappedMemoryHook(uc_engine* uc, uc_mem_type type, u64 addr, int si
     return false;
 }
 
-ARM_Unicorn::ARM_Unicorn(System& system, Arch architecture) : ARM_Interface{system} {
+ARM_Unicorn::ARM_Unicorn(System& system, CPUInterrupts& interrupt_handlers, bool uses_wall_clock,
+                         Arch architecture, std::size_t core_index)
+    : ARM_Interface{system, interrupt_handlers, uses_wall_clock}, core_index{core_index} {
     const auto arch = architecture == Arch::AArch32 ? UC_ARCH_ARM : UC_ARCH_ARM64;
     CHECKED(uc_open(arch, UC_MODE_ARM, &uc));
 
@@ -156,12 +159,20 @@ void ARM_Unicorn::SetTPIDR_EL0(u64 value) {
     CHECKED(uc_reg_write(uc, UC_ARM64_REG_TPIDR_EL0, &value));
 }
 
+void ARM_Unicorn::ChangeProcessorID(std::size_t new_core_id) {
+    core_index = new_core_id;
+}
+
 void ARM_Unicorn::Run() {
     if (GDBStub::IsServerEnabled()) {
         ExecuteInstructions(std::max(4000000U, 0U));
     } else {
-        ExecuteInstructions(
-            std::max(std::size_t(system.CoreTiming().GetDowncount()), std::size_t{0}));
+        while (true) {
+            if (interrupt_handlers[core_index].IsInterrupted()) {
+                return;
+            }
+            ExecuteInstructions(10);
+        }
     }
 }
 
@@ -183,8 +194,6 @@ void ARM_Unicorn::ExecuteInstructions(std::size_t num_instructions) {
                            UC_PROT_READ | UC_PROT_WRITE | UC_PROT_EXEC, page_buffer.data()));
     CHECKED(uc_emu_start(uc, GetPC(), 1ULL << 63, 0, num_instructions));
     CHECKED(uc_mem_unmap(uc, map_addr, page_buffer.size()));
-
-    system.CoreTiming().AddTicks(num_instructions);
     if (GDBStub::IsServerEnabled()) {
         if (last_bkpt_hit && last_bkpt.type == GDBStub::BreakpointType::Execute) {
             uc_reg_write(uc, UC_ARM64_REG_PC, &last_bkpt.address);
diff --git a/src/core/arm/unicorn/arm_unicorn.h b/src/core/arm/unicorn/arm_unicorn.h
index 725c65085..810aff311 100644
--- a/src/core/arm/unicorn/arm_unicorn.h
+++ b/src/core/arm/unicorn/arm_unicorn.h
@@ -20,7 +20,8 @@ public:
         AArch64, // 64-bit ARM
     };
 
-    explicit ARM_Unicorn(System& system, Arch architecture);
+    explicit ARM_Unicorn(System& system, CPUInterrupts& interrupt_handlers, bool uses_wall_clock,
+                         Arch architecture, std::size_t core_index);
     ~ARM_Unicorn() override;
 
     void SetPC(u64 pc) override;
@@ -35,6 +36,7 @@ public:
     void SetTlsAddress(VAddr address) override;
     void SetTPIDR_EL0(u64 value) override;
     u64 GetTPIDR_EL0() const override;
+    void ChangeProcessorID(std::size_t new_core_id) override;
     void PrepareReschedule() override;
     void ClearExclusiveState() override;
     void ExecuteInstructions(std::size_t num_instructions);
@@ -55,6 +57,7 @@ private:
     uc_engine* uc{};
     GDBStub::BreakpointAddress last_bkpt{};
     bool last_bkpt_hit = false;
+    std::size_t core_index;
 };
 
 } // namespace Core
diff --git a/src/core/core.cpp b/src/core/core.cpp
index f9f8a3000..1a243c515 100644
--- a/src/core/core.cpp
+++ b/src/core/core.cpp
@@ -8,10 +8,10 @@
 
 #include "common/file_util.h"
 #include "common/logging/log.h"
+#include "common/microprofile.h"
 #include "common/string_util.h"
 #include "core/arm/exclusive_monitor.h"
 #include "core/core.h"
-#include "core/core_manager.h"
 #include "core/core_timing.h"
 #include "core/cpu_manager.h"
 #include "core/device_memory.h"
@@ -51,6 +51,11 @@
 #include "video_core/renderer_base.h"
 #include "video_core/video_core.h"
 
+MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_CPU0, "ARM JIT", "Dynarmic CPU 0", MP_RGB(255, 64, 64));
+MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_CPU1, "ARM JIT", "Dynarmic CPU 1", MP_RGB(255, 64, 64));
+MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_CPU2, "ARM JIT", "Dynarmic CPU 2", MP_RGB(255, 64, 64));
+MICROPROFILE_DEFINE(ARM_Jit_Dynarmic_CPU3, "ARM JIT", "Dynarmic CPU 3", MP_RGB(255, 64, 64));
+
 namespace Core {
 
 namespace {
@@ -117,23 +122,22 @@ struct System::Impl {
         : kernel{system}, fs_controller{system}, memory{system},
           cpu_manager{system}, reporter{system}, applet_manager{system} {}
 
-    CoreManager& CurrentCoreManager() {
-        return cpu_manager.GetCurrentCoreManager();
-    }
+    ResultStatus Run() {
+        status = ResultStatus::Success;
 
-    Kernel::PhysicalCore& CurrentPhysicalCore() {
-        const auto index = cpu_manager.GetActiveCoreIndex();
-        return kernel.PhysicalCore(index);
-    }
+        kernel.Suspend(false);
+        core_timing.SyncPause(false);
+        cpu_manager.Pause(false);
 
-    Kernel::PhysicalCore& GetPhysicalCore(std::size_t index) {
-        return kernel.PhysicalCore(index);
+        return status;
     }
 
-    ResultStatus RunLoop(bool tight_loop) {
+    ResultStatus Pause() {
         status = ResultStatus::Success;
 
-        cpu_manager.RunLoop(tight_loop);
+        core_timing.SyncPause(true);
+        kernel.Suspend(true);
+        cpu_manager.Pause(true);
 
         return status;
     }
@@ -143,7 +147,15 @@ struct System::Impl {
 
         device_memory = std::make_unique<Core::DeviceMemory>(system);
 
-        core_timing.Initialize();
+        is_multicore = Settings::values.use_multi_core;
+        is_async_gpu = is_multicore || Settings::values.use_asynchronous_gpu_emulation;
+
+        kernel.SetMulticore(is_multicore);
+        cpu_manager.SetMulticore(is_multicore);
+        cpu_manager.SetAsyncGpu(is_async_gpu);
+        core_timing.SetMulticore(is_multicore);
+
+        core_timing.Initialize([&system]() { system.RegisterHostThread(); });
         kernel.Initialize();
         cpu_manager.Initialize();
 
@@ -180,6 +192,11 @@ struct System::Impl {
         is_powered_on = true;
         exit_lock = false;
 
+        microprofile_dynarmic[0] = MICROPROFILE_TOKEN(ARM_Jit_Dynarmic_CPU0);
+        microprofile_dynarmic[1] = MICROPROFILE_TOKEN(ARM_Jit_Dynarmic_CPU1);
+        microprofile_dynarmic[2] = MICROPROFILE_TOKEN(ARM_Jit_Dynarmic_CPU2);
+        microprofile_dynarmic[3] = MICROPROFILE_TOKEN(ARM_Jit_Dynarmic_CPU3);
+
         LOG_DEBUG(Core, "Initialized OK");
 
         return ResultStatus::Success;
@@ -277,8 +294,6 @@ struct System::Impl {
         service_manager.reset();
         cheat_engine.reset();
         telemetry_session.reset();
-        perf_stats.reset();
-        gpu_core.reset();
         device_memory.reset();
 
         // Close all CPU/threading state
@@ -290,6 +305,8 @@ struct System::Impl {
 
         // Close app loader
         app_loader.reset();
+        gpu_core.reset();
+        perf_stats.reset();
 
         // Clear all applets
         applet_manager.ClearAll();
@@ -382,25 +399,35 @@ struct System::Impl {
 
     std::unique_ptr<Core::PerfStats> perf_stats;
     Core::FrameLimiter frame_limiter;
+
+    bool is_multicore{};
+    bool is_async_gpu{};
+
+    std::array<u64, Core::Hardware::NUM_CPU_CORES> dynarmic_ticks{};
+    std::array<MicroProfileToken, Core::Hardware::NUM_CPU_CORES> microprofile_dynarmic{};
 };
 
 System::System() : impl{std::make_unique<Impl>(*this)} {}
 System::~System() = default;
 
-CoreManager& System::CurrentCoreManager() {
-    return impl->CurrentCoreManager();
+CpuManager& System::GetCpuManager() {
+    return impl->cpu_manager;
+}
+
+const CpuManager& System::GetCpuManager() const {
+    return impl->cpu_manager;
 }
 
-const CoreManager& System::CurrentCoreManager() const {
-    return impl->CurrentCoreManager();
+System::ResultStatus System::Run() {
+    return impl->Run();
 }
 
-System::ResultStatus System::RunLoop(bool tight_loop) {
-    return impl->RunLoop(tight_loop);
+System::ResultStatus System::Pause() {
+    return impl->Pause();
 }
 
 System::ResultStatus System::SingleStep() {
-    return RunLoop(false);
+    return ResultStatus::Success;
 }
 
 void System::InvalidateCpuInstructionCaches() {
@@ -416,7 +443,7 @@ bool System::IsPoweredOn() const {
 }
 
 void System::PrepareReschedule() {
-    impl->CurrentPhysicalCore().Stop();
+    // Deprecated, does nothing, kept for backward compatibility.
 }
 
 void System::PrepareReschedule(const u32 core_index) {
@@ -436,31 +463,41 @@ const TelemetrySession& System::TelemetrySession() const {
 }
 
 ARM_Interface& System::CurrentArmInterface() {
-    return impl->CurrentPhysicalCore().ArmInterface();
+    return impl->kernel.CurrentScheduler().GetCurrentThread()->ArmInterface();
 }
 
 const ARM_Interface& System::CurrentArmInterface() const {
-    return impl->CurrentPhysicalCore().ArmInterface();
+    return impl->kernel.CurrentScheduler().GetCurrentThread()->ArmInterface();
 }
 
 std::size_t System::CurrentCoreIndex() const {
-    return impl->cpu_manager.GetActiveCoreIndex();
+    std::size_t core = impl->kernel.GetCurrentHostThreadID();
+    ASSERT(core < Core::Hardware::NUM_CPU_CORES);
+    return core;
 }
 
 Kernel::Scheduler& System::CurrentScheduler() {
-    return impl->CurrentPhysicalCore().Scheduler();
+    return impl->kernel.CurrentScheduler();
 }
 
 const Kernel::Scheduler& System::CurrentScheduler() const {
-    return impl->CurrentPhysicalCore().Scheduler();
+    return impl->kernel.CurrentScheduler();
+}
+
+Kernel::PhysicalCore& System::CurrentPhysicalCore() {
+    return impl->kernel.CurrentPhysicalCore();
+}
+
+const Kernel::PhysicalCore& System::CurrentPhysicalCore() const {
+    return impl->kernel.CurrentPhysicalCore();
 }
 
 Kernel::Scheduler& System::Scheduler(std::size_t core_index) {
-    return impl->GetPhysicalCore(core_index).Scheduler();
+    return impl->kernel.Scheduler(core_index);
 }
 
 const Kernel::Scheduler& System::Scheduler(std::size_t core_index) const {
-    return impl->GetPhysicalCore(core_index).Scheduler();
+    return impl->kernel.Scheduler(core_index);
 }
 
 /// Gets the global scheduler
@@ -490,20 +527,15 @@ const Kernel::Process* System::CurrentProcess() const {
 }
 
 ARM_Interface& System::ArmInterface(std::size_t core_index) {
-    return impl->GetPhysicalCore(core_index).ArmInterface();
+    auto* thread = impl->kernel.Scheduler(core_index).GetCurrentThread();
+    ASSERT(thread && !thread->IsHLEThread());
+    return thread->ArmInterface();
 }
 
 const ARM_Interface& System::ArmInterface(std::size_t core_index) const {
-    return impl->GetPhysicalCore(core_index).ArmInterface();
-}
-
-CoreManager& System::GetCoreManager(std::size_t core_index) {
-    return impl->cpu_manager.GetCoreManager(core_index);
-}
-
-const CoreManager& System::GetCoreManager(std::size_t core_index) const {
-    ASSERT(core_index < NUM_CPU_CORES);
-    return impl->cpu_manager.GetCoreManager(core_index);
+    auto* thread = impl->kernel.Scheduler(core_index).GetCurrentThread();
+    ASSERT(thread && !thread->IsHLEThread());
+    return thread->ArmInterface();
 }
 
 ExclusiveMonitor& System::Monitor() {
@@ -722,4 +754,18 @@ void System::RegisterHostThread() {
     impl->kernel.RegisterHostThread();
 }
 
+void System::EnterDynarmicProfile() {
+    std::size_t core = impl->kernel.GetCurrentHostThreadID();
+    impl->dynarmic_ticks[core] = MicroProfileEnter(impl->microprofile_dynarmic[core]);
+}
+
+void System::ExitDynarmicProfile() {
+    std::size_t core = impl->kernel.GetCurrentHostThreadID();
+    MicroProfileLeave(impl->microprofile_dynarmic[core], impl->dynarmic_ticks[core]);
+}
+
+bool System::IsMulticore() const {
+    return impl->is_multicore;
+}
+
 } // namespace Core
diff --git a/src/core/core.h b/src/core/core.h
index acc53d6a1..5c6cfbffe 100644
--- a/src/core/core.h
+++ b/src/core/core.h
@@ -27,6 +27,7 @@ class VfsFilesystem;
 namespace Kernel {
 class GlobalScheduler;
 class KernelCore;
+class PhysicalCore;
 class Process;
 class Scheduler;
 } // namespace Kernel
@@ -90,7 +91,7 @@ class InterruptManager;
 namespace Core {
 
 class ARM_Interface;
-class CoreManager;
+class CpuManager;
 class DeviceMemory;
 class ExclusiveMonitor;
 class FrameLimiter;
@@ -136,16 +137,16 @@ public:
     };
 
     /**
-     * Run the core CPU loop
-     * This function runs the core for the specified number of CPU instructions before trying to
-     * update hardware. This is much faster than SingleStep (and should be equivalent), as the CPU
-     * is not required to do a full dispatch with each instruction. NOTE: the number of instructions
-     * requested is not guaranteed to run, as this will be interrupted preemptively if a hardware
-     * update is requested (e.g. on a thread switch).
-     * @param tight_loop If false, the CPU single-steps.
-     * @return Result status, indicating whether or not the operation succeeded.
+     * Run the OS and Application
+     * This function will start emulation and run the relevant devices
+     */
+    ResultStatus Run();
+
+    /**
+     * Pause the OS and Application
+     * This function will pause emulation and stop the relevant devices
      */
-    ResultStatus RunLoop(bool tight_loop = true);
+    ResultStatus Pause();
 
     /**
      * Step the CPU one instruction
@@ -209,17 +210,21 @@ public:
     /// Gets the scheduler for the CPU core that is currently running
     const Kernel::Scheduler& CurrentScheduler() const;
 
+    /// Gets the physical core for the CPU core that is currently running
+    Kernel::PhysicalCore& CurrentPhysicalCore();
+
+    /// Gets the physical core for the CPU core that is currently running
+    const Kernel::PhysicalCore& CurrentPhysicalCore() const;
+
     /// Gets a reference to an ARM interface for the CPU core with the specified index
     ARM_Interface& ArmInterface(std::size_t core_index);
 
     /// Gets a const reference to an ARM interface from the CPU core with the specified index
     const ARM_Interface& ArmInterface(std::size_t core_index) const;
 
-    /// Gets a CPU interface to the CPU core with the specified index
-    CoreManager& GetCoreManager(std::size_t core_index);
+    CpuManager& GetCpuManager();
 
-    /// Gets a CPU interface to the CPU core with the specified index
-    const CoreManager& GetCoreManager(std::size_t core_index) const;
+    const CpuManager& GetCpuManager() const;
 
     /// Gets a reference to the exclusive monitor
     ExclusiveMonitor& Monitor();
@@ -370,14 +375,17 @@ public:
     /// Register a host thread as an auxiliary thread.
     void RegisterHostThread();
 
-private:
-    System();
+    /// Enter Dynarmic Microprofile
+    void EnterDynarmicProfile();
+
+    /// Exit Dynarmic Microprofile
+    void ExitDynarmicProfile();
 
-    /// Returns the currently running CPU core
-    CoreManager& CurrentCoreManager();
+    /// Tells if system is running on multicore.
+    bool IsMulticore() const;
 
-    /// Returns the currently running CPU core
-    const CoreManager& CurrentCoreManager() const;
+private:
+    System();
 
     /**
      * Initialize the emulated system.
diff --git a/src/core/core_manager.cpp b/src/core/core_manager.cpp
deleted file mode 100644
index b6b797c80..000000000
--- a/src/core/core_manager.cpp
+++ /dev/null
@@ -1,67 +0,0 @@
-// Copyright 2018 yuzu emulator team
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <condition_variable>
-#include <mutex>
-
-#include "common/logging/log.h"
-#include "core/arm/exclusive_monitor.h"
-#include "core/arm/unicorn/arm_unicorn.h"
-#include "core/core.h"
-#include "core/core_manager.h"
-#include "core/core_timing.h"
-#include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/physical_core.h"
-#include "core/hle/kernel/scheduler.h"
-#include "core/hle/kernel/thread.h"
-#include "core/hle/lock.h"
-#include "core/settings.h"
-
-namespace Core {
-
-CoreManager::CoreManager(System& system, std::size_t core_index)
-    : global_scheduler{system.GlobalScheduler()}, physical_core{system.Kernel().PhysicalCore(
-                                                      core_index)},
-      core_timing{system.CoreTiming()}, core_index{core_index} {}
-
-CoreManager::~CoreManager() = default;
-
-void CoreManager::RunLoop(bool tight_loop) {
-    Reschedule();
-
-    // If we don't have a currently active thread then don't execute instructions,
-    // instead advance to the next event and try to yield to the next thread
-    if (Kernel::GetCurrentThread() == nullptr) {
-        LOG_TRACE(Core, "Core-{} idling", core_index);
-        core_timing.Idle();
-    } else {
-        if (tight_loop) {
-            physical_core.Run();
-        } else {
-            physical_core.Step();
-        }
-    }
-    core_timing.Advance();
-
-    Reschedule();
-}
-
-void CoreManager::SingleStep() {
-    return RunLoop(false);
-}
-
-void CoreManager::PrepareReschedule() {
-    physical_core.Stop();
-}
-
-void CoreManager::Reschedule() {
-    // Lock the global kernel mutex when we manipulate the HLE state
-    std::lock_guard lock(HLE::g_hle_lock);
-
-    global_scheduler.SelectThread(core_index);
-
-    physical_core.Scheduler().TryDoContextSwitch();
-}
-
-} // namespace Core
diff --git a/src/core/core_manager.h b/src/core/core_manager.h
deleted file mode 100644
index d525de00a..000000000
--- a/src/core/core_manager.h
+++ /dev/null
@@ -1,63 +0,0 @@
-// Copyright 2018 yuzu emulator team
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <atomic>
-#include <cstddef>
-#include <memory>
-#include "common/common_types.h"
-
-namespace Kernel {
-class GlobalScheduler;
-class PhysicalCore;
-} // namespace Kernel
-
-namespace Core {
-class System;
-}
-
-namespace Core::Timing {
-class CoreTiming;
-}
-
-namespace Core::Memory {
-class Memory;
-}
-
-namespace Core {
-
-constexpr unsigned NUM_CPU_CORES{4};
-
-class CoreManager {
-public:
-    CoreManager(System& system, std::size_t core_index);
-    ~CoreManager();
-
-    void RunLoop(bool tight_loop = true);
-
-    void SingleStep();
-
-    void PrepareReschedule();
-
-    bool IsMainCore() const {
-        return core_index == 0;
-    }
-
-    std::size_t CoreIndex() const {
-        return core_index;
-    }
-
-private:
-    void Reschedule();
-
-    Kernel::GlobalScheduler& global_scheduler;
-    Kernel::PhysicalCore& physical_core;
-    Timing::CoreTiming& core_timing;
-
-    std::atomic<bool> reschedule_pending = false;
-    std::size_t core_index;
-};
-
-} // namespace Core
diff --git a/src/core/core_timing.cpp b/src/core/core_timing.cpp
index 46d4178c4..5c83c41a4 100644
--- a/src/core/core_timing.cpp
+++ b/src/core/core_timing.cpp
@@ -1,29 +1,27 @@
-// Copyright 2008 Dolphin Emulator Project / 2017 Citra Emulator Project
-// Licensed under GPLv2+
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#include "core/core_timing.h"
-
 #include <algorithm>
 #include <mutex>
 #include <string>
 #include <tuple>
 
 #include "common/assert.h"
-#include "common/thread.h"
+#include "common/microprofile.h"
+#include "core/core_timing.h"
 #include "core/core_timing_util.h"
-#include "core/hardware_properties.h"
 
 namespace Core::Timing {
 
-constexpr int MAX_SLICE_LENGTH = 10000;
+constexpr u64 MAX_SLICE_LENGTH = 4000;
 
 std::shared_ptr<EventType> CreateEvent(std::string name, TimedCallback&& callback) {
     return std::make_shared<EventType>(std::move(callback), std::move(name));
 }
 
 struct CoreTiming::Event {
-    s64 time;
+    u64 time;
     u64 fifo_order;
     u64 userdata;
     std::weak_ptr<EventType> type;
@@ -39,51 +37,90 @@ struct CoreTiming::Event {
     }
 };
 
-CoreTiming::CoreTiming() = default;
-CoreTiming::~CoreTiming() = default;
+CoreTiming::CoreTiming() {
+    clock =
+        Common::CreateBestMatchingClock(Core::Hardware::BASE_CLOCK_RATE, Core::Hardware::CNTFREQ);
+}
 
-void CoreTiming::Initialize() {
-    downcounts.fill(MAX_SLICE_LENGTH);
-    time_slice.fill(MAX_SLICE_LENGTH);
-    slice_length = MAX_SLICE_LENGTH;
-    global_timer = 0;
-    idled_cycles = 0;
-    current_context = 0;
+CoreTiming::~CoreTiming() = default;
 
-    // The time between CoreTiming being initialized and the first call to Advance() is considered
-    // the slice boundary between slice -1 and slice 0. Dispatcher loops must call Advance() before
-    // executing the first cycle of each slice to prepare the slice length and downcount for
-    // that slice.
-    is_global_timer_sane = true;
+void CoreTiming::ThreadEntry(CoreTiming& instance) {
+    constexpr char name[] = "yuzu:HostTiming";
+    MicroProfileOnThreadCreate(name);
+    Common::SetCurrentThreadName(name);
+    Common::SetCurrentThreadPriority(Common::ThreadPriority::VeryHigh);
+    instance.on_thread_init();
+    instance.ThreadLoop();
+}
 
+void CoreTiming::Initialize(std::function<void(void)>&& on_thread_init_) {
+    on_thread_init = std::move(on_thread_init_);
     event_fifo_id = 0;
-
+    shutting_down = false;
+    ticks = 0;
     const auto empty_timed_callback = [](u64, s64) {};
     ev_lost = CreateEvent("_lost_event", empty_timed_callback);
+    if (is_multicore) {
+        timer_thread = std::make_unique<std::thread>(ThreadEntry, std::ref(*this));
+    }
 }
 
 void CoreTiming::Shutdown() {
+    paused = true;
+    shutting_down = true;
+    pause_event.Set();
+    event.Set();
+    if (timer_thread) {
+        timer_thread->join();
+    }
     ClearPendingEvents();
+    timer_thread.reset();
+    has_started = false;
 }
 
-void CoreTiming::ScheduleEvent(s64 cycles_into_future, const std::shared_ptr<EventType>& event_type,
-                               u64 userdata) {
-    std::lock_guard guard{inner_mutex};
-    const s64 timeout = GetTicks() + cycles_into_future;
+void CoreTiming::Pause(bool is_paused) {
+    paused = is_paused;
+    pause_event.Set();
+}
 
-    // If this event needs to be scheduled before the next advance(), force one early
-    if (!is_global_timer_sane) {
-        ForceExceptionCheck(cycles_into_future);
+void CoreTiming::SyncPause(bool is_paused) {
+    if (is_paused == paused && paused_set == paused) {
+        return;
+    }
+    Pause(is_paused);
+    if (timer_thread) {
+        if (!is_paused) {
+            pause_event.Set();
+        }
+        event.Set();
+        while (paused_set != is_paused)
+            ;
     }
+}
 
-    event_queue.emplace_back(Event{timeout, event_fifo_id++, userdata, event_type});
+bool CoreTiming::IsRunning() const {
+    return !paused_set;
+}
 
-    std::push_heap(event_queue.begin(), event_queue.end(), std::greater<>());
+bool CoreTiming::HasPendingEvents() const {
+    return !(wait_set && event_queue.empty());
 }
 
-void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u64 userdata) {
-    std::lock_guard guard{inner_mutex};
+void CoreTiming::ScheduleEvent(s64 ns_into_future, const std::shared_ptr<EventType>& event_type,
+                               u64 userdata) {
+    {
+        std::scoped_lock scope{basic_lock};
+        const u64 timeout = static_cast<u64>(GetGlobalTimeNs().count() + ns_into_future);
+
+        event_queue.emplace_back(Event{timeout, event_fifo_id++, userdata, event_type});
 
+        std::push_heap(event_queue.begin(), event_queue.end(), std::greater<>());
+    }
+    event.Set();
+}
+
+void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u64 userdata) {
+    std::scoped_lock scope{basic_lock};
     const auto itr = std::remove_if(event_queue.begin(), event_queue.end(), [&](const Event& e) {
         return e.type.lock().get() == event_type.get() && e.userdata == userdata;
     });
@@ -95,21 +132,39 @@ void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u
     }
 }
 
-u64 CoreTiming::GetTicks() const {
-    u64 ticks = static_cast<u64>(global_timer);
-    if (!is_global_timer_sane) {
-        ticks += accumulated_ticks;
+void CoreTiming::AddTicks(u64 ticks) {
+    this->ticks += ticks;
+    downcount -= ticks;
+}
+
+void CoreTiming::Idle() {
+    if (!event_queue.empty()) {
+        const u64 next_event_time = event_queue.front().time;
+        const u64 next_ticks = nsToCycles(std::chrono::nanoseconds(next_event_time)) + 10U;
+        if (next_ticks > ticks) {
+            ticks = next_ticks;
+        }
+        return;
     }
-    return ticks;
+    ticks += 1000U;
 }
 
-u64 CoreTiming::GetIdleTicks() const {
-    return static_cast<u64>(idled_cycles);
+void CoreTiming::ResetTicks() {
+    downcount = MAX_SLICE_LENGTH;
 }
 
-void CoreTiming::AddTicks(u64 ticks) {
-    accumulated_ticks += ticks;
-    downcounts[current_context] -= static_cast<s64>(ticks);
+u64 CoreTiming::GetCPUTicks() const {
+    if (is_multicore) {
+        return clock->GetCPUCycles();
+    }
+    return ticks;
+}
+
+u64 CoreTiming::GetClockTicks() const {
+    if (is_multicore) {
+        return clock->GetClockCycles();
+    }
+    return CpuCyclesToClockCycles(ticks);
 }
 
 void CoreTiming::ClearPendingEvents() {
@@ -117,7 +172,7 @@ void CoreTiming::ClearPendingEvents() {
 }
 
 void CoreTiming::RemoveEvent(const std::shared_ptr<EventType>& event_type) {
-    std::lock_guard guard{inner_mutex};
+    basic_lock.lock();
 
     const auto itr = std::remove_if(event_queue.begin(), event_queue.end(), [&](const Event& e) {
         return e.type.lock().get() == event_type.get();
@@ -128,99 +183,72 @@ void CoreTiming::RemoveEvent(const std::shared_ptr<EventType>& event_type) {
         event_queue.erase(itr, event_queue.end());
         std::make_heap(event_queue.begin(), event_queue.end(), std::greater<>());
     }
+    basic_lock.unlock();
 }
 
-void CoreTiming::ForceExceptionCheck(s64 cycles) {
-    cycles = std::max<s64>(0, cycles);
-    if (downcounts[current_context] <= cycles) {
-        return;
-    }
-
-    // downcount is always (much) smaller than MAX_INT so we can safely cast cycles to an int
-    // here. Account for cycles already executed by adjusting the g.slice_length
-    downcounts[current_context] = static_cast<int>(cycles);
-}
-
-std::optional<u64> CoreTiming::NextAvailableCore(const s64 needed_ticks) const {
-    const u64 original_context = current_context;
-    u64 next_context = (original_context + 1) % num_cpu_cores;
-    while (next_context != original_context) {
-        if (time_slice[next_context] >= needed_ticks) {
-            return {next_context};
-        } else if (time_slice[next_context] >= 0) {
-            return std::nullopt;
-        }
-        next_context = (next_context + 1) % num_cpu_cores;
-    }
-    return std::nullopt;
-}
-
-void CoreTiming::Advance() {
-    std::unique_lock<std::mutex> guard(inner_mutex);
-
-    const u64 cycles_executed = accumulated_ticks;
-    time_slice[current_context] = std::max<s64>(0, time_slice[current_context] - accumulated_ticks);
-    global_timer += cycles_executed;
-
-    is_global_timer_sane = true;
+std::optional<s64> CoreTiming::Advance() {
+    std::scoped_lock advance_scope{advance_lock};
+    std::scoped_lock basic_scope{basic_lock};
+    global_timer = GetGlobalTimeNs().count();
 
     while (!event_queue.empty() && event_queue.front().time <= global_timer) {
         Event evt = std::move(event_queue.front());
         std::pop_heap(event_queue.begin(), event_queue.end(), std::greater<>());
         event_queue.pop_back();
-        inner_mutex.unlock();
+        basic_lock.unlock();
 
         if (auto event_type{evt.type.lock()}) {
             event_type->callback(evt.userdata, global_timer - evt.time);
         }
 
-        inner_mutex.lock();
+        basic_lock.lock();
+        global_timer = GetGlobalTimeNs().count();
     }
 
-    is_global_timer_sane = false;
-
-    // Still events left (scheduled in the future)
     if (!event_queue.empty()) {
-        const s64 needed_ticks =
-            std::min<s64>(event_queue.front().time - global_timer, MAX_SLICE_LENGTH);
-        const auto next_core = NextAvailableCore(needed_ticks);
-        if (next_core) {
-            downcounts[*next_core] = needed_ticks;
-        }
+        const s64 next_time = event_queue.front().time - global_timer;
+        return next_time;
+    } else {
+        return std::nullopt;
     }
-
-    accumulated_ticks = 0;
-
-    downcounts[current_context] = time_slice[current_context];
 }
 
-void CoreTiming::ResetRun() {
-    downcounts.fill(MAX_SLICE_LENGTH);
-    time_slice.fill(MAX_SLICE_LENGTH);
-    current_context = 0;
-    // Still events left (scheduled in the future)
-    if (!event_queue.empty()) {
-        const s64 needed_ticks =
-            std::min<s64>(event_queue.front().time - global_timer, MAX_SLICE_LENGTH);
-        downcounts[current_context] = needed_ticks;
+void CoreTiming::ThreadLoop() {
+    has_started = true;
+    while (!shutting_down) {
+        while (!paused) {
+            paused_set = false;
+            const auto next_time = Advance();
+            if (next_time) {
+                if (*next_time > 0) {
+                    std::chrono::nanoseconds next_time_ns = std::chrono::nanoseconds(*next_time);
+                    event.WaitFor(next_time_ns);
+                }
+            } else {
+                wait_set = true;
+                event.Wait();
+            }
+            wait_set = false;
+        }
+        paused_set = true;
+        clock->Pause(true);
+        pause_event.Wait();
+        clock->Pause(false);
     }
-
-    is_global_timer_sane = false;
-    accumulated_ticks = 0;
 }
 
-void CoreTiming::Idle() {
-    accumulated_ticks += downcounts[current_context];
-    idled_cycles += downcounts[current_context];
-    downcounts[current_context] = 0;
+std::chrono::nanoseconds CoreTiming::GetGlobalTimeNs() const {
+    if (is_multicore) {
+        return clock->GetTimeNS();
+    }
+    return CyclesToNs(ticks);
 }
 
 std::chrono::microseconds CoreTiming::GetGlobalTimeUs() const {
-    return std::chrono::microseconds{GetTicks() * 1000000 / Hardware::BASE_CLOCK_RATE};
-}
-
-s64 CoreTiming::GetDowncount() const {
-    return downcounts[current_context];
+    if (is_multicore) {
+        return clock->GetTimeUS();
+    }
+    return CyclesToUs(ticks);
 }
 
 } // namespace Core::Timing
diff --git a/src/core/core_timing.h b/src/core/core_timing.h
index d50f4eb8a..72faaab64 100644
--- a/src/core/core_timing.h
+++ b/src/core/core_timing.h
@@ -1,19 +1,25 @@
-// Copyright 2008 Dolphin Emulator Project / 2017 Citra Emulator Project
-// Licensed under GPLv2+
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
 #pragma once
 
+#include <atomic>
 #include <chrono>
 #include <functional>
 #include <memory>
 #include <mutex>
 #include <optional>
 #include <string>
+#include <thread>
 #include <vector>
 
 #include "common/common_types.h"
+#include "common/spin_lock.h"
+#include "common/thread.h"
 #include "common/threadsafe_queue.h"
+#include "common/wall_clock.h"
+#include "core/hardware_properties.h"
 
 namespace Core::Timing {
 
@@ -56,16 +62,40 @@ public:
 
     /// CoreTiming begins at the boundary of timing slice -1. An initial call to Advance() is
     /// required to end slice - 1 and start slice 0 before the first cycle of code is executed.
-    void Initialize();
+    void Initialize(std::function<void(void)>&& on_thread_init_);
 
     /// Tears down all timing related functionality.
     void Shutdown();
 
-    /// After the first Advance, the slice lengths and the downcount will be reduced whenever an
-    /// event is scheduled earlier than the current values.
-    ///
-    /// Scheduling from a callback will not update the downcount until the Advance() completes.
-    void ScheduleEvent(s64 cycles_into_future, const std::shared_ptr<EventType>& event_type,
+    /// Sets if emulation is multicore or single core, must be set before Initialize
+    void SetMulticore(bool is_multicore) {
+        this->is_multicore = is_multicore;
+    }
+
+    /// Check if it's using host timing.
+    bool IsHostTiming() const {
+        return is_multicore;
+    }
+
+    /// Pauses/Unpauses the execution of the timer thread.
+    void Pause(bool is_paused);
+
+    /// Pauses/Unpauses the execution of the timer thread and waits until paused.
+    void SyncPause(bool is_paused);
+
+    /// Checks if core timing is running.
+    bool IsRunning() const;
+
+    /// Checks if the timer thread has started.
+    bool HasStarted() const {
+        return has_started;
+    }
+
+    /// Checks if there are any pending time events.
+    bool HasPendingEvents() const;
+
+    /// Schedules an event in core timing
+    void ScheduleEvent(s64 ns_into_future, const std::shared_ptr<EventType>& event_type,
                        u64 userdata = 0);
 
     void UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u64 userdata);
@@ -73,41 +103,30 @@ public:
     /// We only permit one event of each type in the queue at a time.
     void RemoveEvent(const std::shared_ptr<EventType>& event_type);
 
-    void ForceExceptionCheck(s64 cycles);
-
-    /// This should only be called from the emu thread, if you are calling it any other thread,
-    /// you are doing something evil
-    u64 GetTicks() const;
-
-    u64 GetIdleTicks() const;
-
     void AddTicks(u64 ticks);
 
-    /// Advance must be called at the beginning of dispatcher loops, not the end. Advance() ends
-    /// the previous timing slice and begins the next one, you must Advance from the previous
-    /// slice to the current one before executing any cycles. CoreTiming starts in slice -1 so an
-    /// Advance() is required to initialize the slice length before the first cycle of emulated
-    /// instructions is executed.
-    void Advance();
+    void ResetTicks();
 
-    /// Pretend that the main CPU has executed enough cycles to reach the next event.
     void Idle();
 
-    std::chrono::microseconds GetGlobalTimeUs() const;
+    s64 GetDowncount() const {
+        return downcount;
+    }
 
-    void ResetRun();
+    /// Returns current time in emulated CPU cycles
+    u64 GetCPUTicks() const;
 
-    s64 GetDowncount() const;
+    /// Returns current time in emulated in Clock cycles
+    u64 GetClockTicks() const;
 
-    void SwitchContext(u64 new_context) {
-        current_context = new_context;
-    }
+    /// Returns current time in microseconds.
+    std::chrono::microseconds GetGlobalTimeUs() const;
 
-    bool CanCurrentContextRun() const {
-        return time_slice[current_context] > 0;
-    }
+    /// Returns current time in nanoseconds.
+    std::chrono::nanoseconds GetGlobalTimeNs() const;
 
-    std::optional<u64> NextAvailableCore(const s64 needed_ticks) const;
+    /// Checks for events manually and returns time in nanoseconds for next event, threadsafe.
+    std::optional<s64> Advance();
 
 private:
     struct Event;
@@ -115,21 +134,14 @@ private:
     /// Clear all pending events. This should ONLY be done on exit.
     void ClearPendingEvents();
 
-    static constexpr u64 num_cpu_cores = 4;
+    static void ThreadEntry(CoreTiming& instance);
+    void ThreadLoop();
 
-    s64 global_timer = 0;
-    s64 idled_cycles = 0;
-    s64 slice_length = 0;
-    u64 accumulated_ticks = 0;
-    std::array<s64, num_cpu_cores> downcounts{};
-    // Slice of time assigned to each core per run.
-    std::array<s64, num_cpu_cores> time_slice{};
-    u64 current_context = 0;
+    std::unique_ptr<Common::WallClock> clock;
 
-    // Are we in a function that has been called from Advance()
-    // If events are scheduled from a function that gets called from Advance(),
-    // don't change slice_length and downcount.
-    bool is_global_timer_sane = false;
+    u64 global_timer = 0;
+
+    std::chrono::nanoseconds start_point;
 
     // The queue is a min-heap using std::make_heap/push_heap/pop_heap.
     // We don't use std::priority_queue because we need to be able to serialize, unserialize and
@@ -139,8 +151,23 @@ private:
     u64 event_fifo_id = 0;
 
     std::shared_ptr<EventType> ev_lost;
-
-    std::mutex inner_mutex;
+    Common::Event event{};
+    Common::Event pause_event{};
+    Common::SpinLock basic_lock{};
+    Common::SpinLock advance_lock{};
+    std::unique_ptr<std::thread> timer_thread;
+    std::atomic<bool> paused{};
+    std::atomic<bool> paused_set{};
+    std::atomic<bool> wait_set{};
+    std::atomic<bool> shutting_down{};
+    std::atomic<bool> has_started{};
+    std::function<void(void)> on_thread_init{};
+
+    bool is_multicore{};
+
+    /// Cycle timing
+    u64 ticks{};
+    s64 downcount{};
 };
 
 /// Creates a core timing event with the given name and callback.
diff --git a/src/core/core_timing_util.cpp b/src/core/core_timing_util.cpp
index be34b26fe..aefc63663 100644
--- a/src/core/core_timing_util.cpp
+++ b/src/core/core_timing_util.cpp
@@ -38,15 +38,8 @@ s64 usToCycles(std::chrono::microseconds us) {
 }
 
 s64 nsToCycles(std::chrono::nanoseconds ns) {
-    if (static_cast<u64>(ns.count() / 1000000000) > MAX_VALUE_TO_MULTIPLY) {
-        LOG_ERROR(Core_Timing, "Integer overflow, use max value");
-        return std::numeric_limits<s64>::max();
-    }
-    if (static_cast<u64>(ns.count()) > MAX_VALUE_TO_MULTIPLY) {
-        LOG_DEBUG(Core_Timing, "Time very big, do rounding");
-        return Hardware::BASE_CLOCK_RATE * (ns.count() / 1000000000);
-    }
-    return (Hardware::BASE_CLOCK_RATE * ns.count()) / 1000000000;
+    const u128 temporal = Common::Multiply64Into128(ns.count(), Hardware::BASE_CLOCK_RATE);
+    return Common::Divide128On32(temporal, static_cast<u32>(1000000000)).first;
 }
 
 u64 msToClockCycles(std::chrono::milliseconds ns) {
@@ -69,4 +62,22 @@ u64 CpuCyclesToClockCycles(u64 ticks) {
     return Common::Divide128On32(temporal, static_cast<u32>(Hardware::BASE_CLOCK_RATE)).first;
 }
 
+std::chrono::milliseconds CyclesToMs(s64 cycles) {
+    const u128 temporal = Common::Multiply64Into128(cycles, 1000);
+    u64 ms = Common::Divide128On32(temporal, static_cast<u32>(Hardware::BASE_CLOCK_RATE)).first;
+    return std::chrono::milliseconds(ms);
+}
+
+std::chrono::nanoseconds CyclesToNs(s64 cycles) {
+    const u128 temporal = Common::Multiply64Into128(cycles, 1000000000);
+    u64 ns = Common::Divide128On32(temporal, static_cast<u32>(Hardware::BASE_CLOCK_RATE)).first;
+    return std::chrono::nanoseconds(ns);
+}
+
+std::chrono::microseconds CyclesToUs(s64 cycles) {
+    const u128 temporal = Common::Multiply64Into128(cycles, 1000000);
+    u64 us = Common::Divide128On32(temporal, static_cast<u32>(Hardware::BASE_CLOCK_RATE)).first;
+    return std::chrono::microseconds(us);
+}
+
 } // namespace Core::Timing
diff --git a/src/core/core_timing_util.h b/src/core/core_timing_util.h
index b3c58447d..2ed979e14 100644
--- a/src/core/core_timing_util.h
+++ b/src/core/core_timing_util.h
@@ -16,18 +16,9 @@ s64 nsToCycles(std::chrono::nanoseconds ns);
 u64 msToClockCycles(std::chrono::milliseconds ns);
 u64 usToClockCycles(std::chrono::microseconds ns);
 u64 nsToClockCycles(std::chrono::nanoseconds ns);
-
-inline std::chrono::milliseconds CyclesToMs(s64 cycles) {
-    return std::chrono::milliseconds(cycles * 1000 / Hardware::BASE_CLOCK_RATE);
-}
-
-inline std::chrono::nanoseconds CyclesToNs(s64 cycles) {
-    return std::chrono::nanoseconds(cycles * 1000000000 / Hardware::BASE_CLOCK_RATE);
-}
-
-inline std::chrono::microseconds CyclesToUs(s64 cycles) {
-    return std::chrono::microseconds(cycles * 1000000 / Hardware::BASE_CLOCK_RATE);
-}
+std::chrono::milliseconds CyclesToMs(s64 cycles);
+std::chrono::nanoseconds CyclesToNs(s64 cycles);
+std::chrono::microseconds CyclesToUs(s64 cycles);
 
 u64 CpuCyclesToClockCycles(u64 ticks);
 
diff --git a/src/core/cpu_manager.cpp b/src/core/cpu_manager.cpp
index 70ddbdcca..32afcf3ae 100644
--- a/src/core/cpu_manager.cpp
+++ b/src/core/cpu_manager.cpp
@@ -2,80 +2,372 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include "common/fiber.h"
+#include "common/microprofile.h"
+#include "common/thread.h"
 #include "core/arm/exclusive_monitor.h"
 #include "core/core.h"
-#include "core/core_manager.h"
 #include "core/core_timing.h"
 #include "core/cpu_manager.h"
 #include "core/gdbstub/gdbstub.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/physical_core.h"
+#include "core/hle/kernel/scheduler.h"
+#include "core/hle/kernel/thread.h"
+#include "video_core/gpu.h"
 
 namespace Core {
 
 CpuManager::CpuManager(System& system) : system{system} {}
 CpuManager::~CpuManager() = default;
 
+void CpuManager::ThreadStart(CpuManager& cpu_manager, std::size_t core) {
+    cpu_manager.RunThread(core);
+}
+
 void CpuManager::Initialize() {
-    for (std::size_t index = 0; index < core_managers.size(); ++index) {
-        core_managers[index] = std::make_unique<CoreManager>(system, index);
+    running_mode = true;
+    if (is_multicore) {
+        for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+            core_data[core].host_thread =
+                std::make_unique<std::thread>(ThreadStart, std::ref(*this), core);
+        }
+    } else {
+        core_data[0].host_thread = std::make_unique<std::thread>(ThreadStart, std::ref(*this), 0);
     }
 }
 
 void CpuManager::Shutdown() {
-    for (auto& cpu_core : core_managers) {
-        cpu_core.reset();
+    running_mode = false;
+    Pause(false);
+    if (is_multicore) {
+        for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+            core_data[core].host_thread->join();
+            core_data[core].host_thread.reset();
+        }
+    } else {
+        core_data[0].host_thread->join();
+        core_data[0].host_thread.reset();
     }
 }
 
-CoreManager& CpuManager::GetCoreManager(std::size_t index) {
-    return *core_managers.at(index);
+std::function<void(void*)> CpuManager::GetGuestThreadStartFunc() {
+    return std::function<void(void*)>(GuestThreadFunction);
 }
 
-const CoreManager& CpuManager::GetCoreManager(std::size_t index) const {
-    return *core_managers.at(index);
+std::function<void(void*)> CpuManager::GetIdleThreadStartFunc() {
+    return std::function<void(void*)>(IdleThreadFunction);
 }
 
-CoreManager& CpuManager::GetCurrentCoreManager() {
-    // Otherwise, use single-threaded mode active_core variable
-    return *core_managers[active_core];
+std::function<void(void*)> CpuManager::GetSuspendThreadStartFunc() {
+    return std::function<void(void*)>(SuspendThreadFunction);
 }
 
-const CoreManager& CpuManager::GetCurrentCoreManager() const {
-    // Otherwise, use single-threaded mode active_core variable
-    return *core_managers[active_core];
+void CpuManager::GuestThreadFunction(void* cpu_manager_) {
+    CpuManager* cpu_manager = static_cast<CpuManager*>(cpu_manager_);
+    if (cpu_manager->is_multicore) {
+        cpu_manager->MultiCoreRunGuestThread();
+    } else {
+        cpu_manager->SingleCoreRunGuestThread();
+    }
 }
 
-void CpuManager::RunLoop(bool tight_loop) {
-    if (GDBStub::IsServerEnabled()) {
-        GDBStub::HandlePacket();
+void CpuManager::GuestRewindFunction(void* cpu_manager_) {
+    CpuManager* cpu_manager = static_cast<CpuManager*>(cpu_manager_);
+    if (cpu_manager->is_multicore) {
+        cpu_manager->MultiCoreRunGuestLoop();
+    } else {
+        cpu_manager->SingleCoreRunGuestLoop();
+    }
+}
 
-        // If the loop is halted and we want to step, use a tiny (1) number of instructions to
-        // execute. Otherwise, get out of the loop function.
-        if (GDBStub::GetCpuHaltFlag()) {
-            if (GDBStub::GetCpuStepFlag()) {
-                tight_loop = false;
-            } else {
-                return;
+void CpuManager::IdleThreadFunction(void* cpu_manager_) {
+    CpuManager* cpu_manager = static_cast<CpuManager*>(cpu_manager_);
+    if (cpu_manager->is_multicore) {
+        cpu_manager->MultiCoreRunIdleThread();
+    } else {
+        cpu_manager->SingleCoreRunIdleThread();
+    }
+}
+
+void CpuManager::SuspendThreadFunction(void* cpu_manager_) {
+    CpuManager* cpu_manager = static_cast<CpuManager*>(cpu_manager_);
+    if (cpu_manager->is_multicore) {
+        cpu_manager->MultiCoreRunSuspendThread();
+    } else {
+        cpu_manager->SingleCoreRunSuspendThread();
+    }
+}
+
+void* CpuManager::GetStartFuncParamater() {
+    return static_cast<void*>(this);
+}
+
+///////////////////////////////////////////////////////////////////////////////
+///                             MultiCore                                   ///
+///////////////////////////////////////////////////////////////////////////////
+
+void CpuManager::MultiCoreRunGuestThread() {
+    auto& kernel = system.Kernel();
+    {
+        auto& sched = kernel.CurrentScheduler();
+        sched.OnThreadStart();
+    }
+    MultiCoreRunGuestLoop();
+}
+
+void CpuManager::MultiCoreRunGuestLoop() {
+    auto& kernel = system.Kernel();
+    auto* thread = kernel.CurrentScheduler().GetCurrentThread();
+    while (true) {
+        auto* physical_core = &kernel.CurrentPhysicalCore();
+        auto& arm_interface = thread->ArmInterface();
+        system.EnterDynarmicProfile();
+        while (!physical_core->IsInterrupted()) {
+            arm_interface.Run();
+            physical_core = &kernel.CurrentPhysicalCore();
+        }
+        system.ExitDynarmicProfile();
+        arm_interface.ClearExclusiveState();
+        auto& scheduler = kernel.CurrentScheduler();
+        scheduler.TryDoContextSwitch();
+    }
+}
+
+void CpuManager::MultiCoreRunIdleThread() {
+    auto& kernel = system.Kernel();
+    while (true) {
+        auto& physical_core = kernel.CurrentPhysicalCore();
+        physical_core.Idle();
+        auto& scheduler = kernel.CurrentScheduler();
+        scheduler.TryDoContextSwitch();
+    }
+}
+
+void CpuManager::MultiCoreRunSuspendThread() {
+    auto& kernel = system.Kernel();
+    {
+        auto& sched = kernel.CurrentScheduler();
+        sched.OnThreadStart();
+    }
+    while (true) {
+        auto core = kernel.GetCurrentHostThreadID();
+        auto& scheduler = kernel.CurrentScheduler();
+        Kernel::Thread* current_thread = scheduler.GetCurrentThread();
+        Common::Fiber::YieldTo(current_thread->GetHostContext(), core_data[core].host_context);
+        ASSERT(scheduler.ContextSwitchPending());
+        ASSERT(core == kernel.GetCurrentHostThreadID());
+        scheduler.TryDoContextSwitch();
+    }
+}
+
+void CpuManager::MultiCorePause(bool paused) {
+    if (!paused) {
+        bool all_not_barrier = false;
+        while (!all_not_barrier) {
+            all_not_barrier = true;
+            for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+                all_not_barrier &=
+                    !core_data[core].is_running.load() && core_data[core].initialized.load();
+            }
+        }
+        for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+            core_data[core].enter_barrier->Set();
+        }
+        if (paused_state.load()) {
+            bool all_barrier = false;
+            while (!all_barrier) {
+                all_barrier = true;
+                for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+                    all_barrier &=
+                        core_data[core].is_paused.load() && core_data[core].initialized.load();
+                }
+            }
+            for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+                core_data[core].exit_barrier->Set();
+            }
+        }
+    } else {
+        /// Wait until all cores are paused.
+        bool all_barrier = false;
+        while (!all_barrier) {
+            all_barrier = true;
+            for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+                all_barrier &=
+                    core_data[core].is_paused.load() && core_data[core].initialized.load();
             }
         }
+        /// Don't release the barrier
     }
+    paused_state = paused;
+}
+
+///////////////////////////////////////////////////////////////////////////////
+///                             SingleCore                                   ///
+///////////////////////////////////////////////////////////////////////////////
 
-    auto& core_timing = system.CoreTiming();
-    core_timing.ResetRun();
-    bool keep_running{};
-    do {
-        keep_running = false;
-        for (active_core = 0; active_core < NUM_CPU_CORES; ++active_core) {
-            core_timing.SwitchContext(active_core);
-            if (core_timing.CanCurrentContextRun()) {
-                core_managers[active_core]->RunLoop(tight_loop);
+void CpuManager::SingleCoreRunGuestThread() {
+    auto& kernel = system.Kernel();
+    {
+        auto& sched = kernel.CurrentScheduler();
+        sched.OnThreadStart();
+    }
+    SingleCoreRunGuestLoop();
+}
+
+void CpuManager::SingleCoreRunGuestLoop() {
+    auto& kernel = system.Kernel();
+    auto* thread = kernel.CurrentScheduler().GetCurrentThread();
+    while (true) {
+        auto* physical_core = &kernel.CurrentPhysicalCore();
+        auto& arm_interface = thread->ArmInterface();
+        system.EnterDynarmicProfile();
+        if (!physical_core->IsInterrupted()) {
+            arm_interface.Run();
+            physical_core = &kernel.CurrentPhysicalCore();
+        }
+        system.ExitDynarmicProfile();
+        thread->SetPhantomMode(true);
+        system.CoreTiming().Advance();
+        thread->SetPhantomMode(false);
+        arm_interface.ClearExclusiveState();
+        PreemptSingleCore();
+        auto& scheduler = kernel.Scheduler(current_core);
+        scheduler.TryDoContextSwitch();
+    }
+}
+
+void CpuManager::SingleCoreRunIdleThread() {
+    auto& kernel = system.Kernel();
+    while (true) {
+        auto& physical_core = kernel.CurrentPhysicalCore();
+        PreemptSingleCore(false);
+        system.CoreTiming().AddTicks(1000U);
+        idle_count++;
+        auto& scheduler = physical_core.Scheduler();
+        scheduler.TryDoContextSwitch();
+    }
+}
+
+void CpuManager::SingleCoreRunSuspendThread() {
+    auto& kernel = system.Kernel();
+    {
+        auto& sched = kernel.CurrentScheduler();
+        sched.OnThreadStart();
+    }
+    while (true) {
+        auto core = kernel.GetCurrentHostThreadID();
+        auto& scheduler = kernel.CurrentScheduler();
+        Kernel::Thread* current_thread = scheduler.GetCurrentThread();
+        Common::Fiber::YieldTo(current_thread->GetHostContext(), core_data[0].host_context);
+        ASSERT(scheduler.ContextSwitchPending());
+        ASSERT(core == kernel.GetCurrentHostThreadID());
+        scheduler.TryDoContextSwitch();
+    }
+}
+
+void CpuManager::PreemptSingleCore(bool from_running_enviroment) {
+    std::size_t old_core = current_core;
+    auto& scheduler = system.Kernel().Scheduler(old_core);
+    Kernel::Thread* current_thread = scheduler.GetCurrentThread();
+    if (idle_count >= 4 || from_running_enviroment) {
+        if (!from_running_enviroment) {
+            system.CoreTiming().Idle();
+            idle_count = 0;
+        }
+        current_thread->SetPhantomMode(true);
+        system.CoreTiming().Advance();
+        current_thread->SetPhantomMode(false);
+    }
+    current_core.store((current_core + 1) % Core::Hardware::NUM_CPU_CORES);
+    system.CoreTiming().ResetTicks();
+    scheduler.Unload();
+    auto& next_scheduler = system.Kernel().Scheduler(current_core);
+    Common::Fiber::YieldTo(current_thread->GetHostContext(), next_scheduler.ControlContext());
+    /// May have changed scheduler
+    auto& current_scheduler = system.Kernel().Scheduler(current_core);
+    current_scheduler.Reload();
+    auto* currrent_thread2 = current_scheduler.GetCurrentThread();
+    if (!currrent_thread2->IsIdleThread()) {
+        idle_count = 0;
+    }
+}
+
+void CpuManager::SingleCorePause(bool paused) {
+    if (!paused) {
+        bool all_not_barrier = false;
+        while (!all_not_barrier) {
+            all_not_barrier = !core_data[0].is_running.load() && core_data[0].initialized.load();
+        }
+        core_data[0].enter_barrier->Set();
+        if (paused_state.load()) {
+            bool all_barrier = false;
+            while (!all_barrier) {
+                all_barrier = core_data[0].is_paused.load() && core_data[0].initialized.load();
             }
-            keep_running |= core_timing.CanCurrentContextRun();
+            core_data[0].exit_barrier->Set();
         }
-    } while (keep_running);
+    } else {
+        /// Wait until all cores are paused.
+        bool all_barrier = false;
+        while (!all_barrier) {
+            all_barrier = core_data[0].is_paused.load() && core_data[0].initialized.load();
+        }
+        /// Don't release the barrier
+    }
+    paused_state = paused;
+}
+
+void CpuManager::Pause(bool paused) {
+    if (is_multicore) {
+        MultiCorePause(paused);
+    } else {
+        SingleCorePause(paused);
+    }
+}
 
-    if (GDBStub::IsServerEnabled()) {
-        GDBStub::SetCpuStepFlag(false);
+void CpuManager::RunThread(std::size_t core) {
+    /// Initialization
+    system.RegisterCoreThread(core);
+    std::string name;
+    if (is_multicore) {
+        name = "yuzu:CoreCPUThread_" + std::to_string(core);
+    } else {
+        name = "yuzu:CPUThread";
+    }
+    MicroProfileOnThreadCreate(name.c_str());
+    Common::SetCurrentThreadName(name.c_str());
+    Common::SetCurrentThreadPriority(Common::ThreadPriority::High);
+    auto& data = core_data[core];
+    data.enter_barrier = std::make_unique<Common::Event>();
+    data.exit_barrier = std::make_unique<Common::Event>();
+    data.host_context = Common::Fiber::ThreadToFiber();
+    data.is_running = false;
+    data.initialized = true;
+    const bool sc_sync = !is_async_gpu && !is_multicore;
+    bool sc_sync_first_use = sc_sync;
+    /// Running
+    while (running_mode) {
+        data.is_running = false;
+        data.enter_barrier->Wait();
+        if (sc_sync_first_use) {
+            system.GPU().ObtainContext();
+            sc_sync_first_use = false;
+        }
+        auto& scheduler = system.Kernel().CurrentScheduler();
+        Kernel::Thread* current_thread = scheduler.GetCurrentThread();
+        data.is_running = true;
+        Common::Fiber::YieldTo(data.host_context, current_thread->GetHostContext());
+        data.is_running = false;
+        data.is_paused = true;
+        data.exit_barrier->Wait();
+        data.is_paused = false;
     }
+    /// Time to cleanup
+    data.host_context->Exit();
+    data.enter_barrier.reset();
+    data.exit_barrier.reset();
+    data.initialized = false;
 }
 
 } // namespace Core
diff --git a/src/core/cpu_manager.h b/src/core/cpu_manager.h
index 97554d1bb..35929ed94 100644
--- a/src/core/cpu_manager.h
+++ b/src/core/cpu_manager.h
@@ -5,12 +5,19 @@
 #pragma once
 
 #include <array>
+#include <atomic>
+#include <functional>
 #include <memory>
+#include <thread>
 #include "core/hardware_properties.h"
 
+namespace Common {
+class Event;
+class Fiber;
+} // namespace Common
+
 namespace Core {
 
-class CoreManager;
 class System;
 
 class CpuManager {
@@ -24,24 +31,75 @@ public:
     CpuManager& operator=(const CpuManager&) = delete;
     CpuManager& operator=(CpuManager&&) = delete;
 
+    /// Sets if emulation is multicore or single core, must be set before Initialize
+    void SetMulticore(bool is_multicore) {
+        this->is_multicore = is_multicore;
+    }
+
+    /// Sets if emulation is using an asynchronous GPU.
+    void SetAsyncGpu(bool is_async_gpu) {
+        this->is_async_gpu = is_async_gpu;
+    }
+
     void Initialize();
     void Shutdown();
 
-    CoreManager& GetCoreManager(std::size_t index);
-    const CoreManager& GetCoreManager(std::size_t index) const;
+    void Pause(bool paused);
 
-    CoreManager& GetCurrentCoreManager();
-    const CoreManager& GetCurrentCoreManager() const;
+    std::function<void(void*)> GetGuestThreadStartFunc();
+    std::function<void(void*)> GetIdleThreadStartFunc();
+    std::function<void(void*)> GetSuspendThreadStartFunc();
+    void* GetStartFuncParamater();
 
-    std::size_t GetActiveCoreIndex() const {
-        return active_core;
-    }
+    void PreemptSingleCore(bool from_running_enviroment = true);
 
-    void RunLoop(bool tight_loop);
+    std::size_t CurrentCore() const {
+        return current_core.load();
+    }
 
 private:
-    std::array<std::unique_ptr<CoreManager>, Hardware::NUM_CPU_CORES> core_managers;
-    std::size_t active_core{}; ///< Active core, only used in single thread mode
+    static void GuestThreadFunction(void* cpu_manager);
+    static void GuestRewindFunction(void* cpu_manager);
+    static void IdleThreadFunction(void* cpu_manager);
+    static void SuspendThreadFunction(void* cpu_manager);
+
+    void MultiCoreRunGuestThread();
+    void MultiCoreRunGuestLoop();
+    void MultiCoreRunIdleThread();
+    void MultiCoreRunSuspendThread();
+    void MultiCorePause(bool paused);
+
+    void SingleCoreRunGuestThread();
+    void SingleCoreRunGuestLoop();
+    void SingleCoreRunIdleThread();
+    void SingleCoreRunSuspendThread();
+    void SingleCorePause(bool paused);
+
+    static void ThreadStart(CpuManager& cpu_manager, std::size_t core);
+
+    void RunThread(std::size_t core);
+
+    struct CoreData {
+        std::shared_ptr<Common::Fiber> host_context;
+        std::unique_ptr<Common::Event> enter_barrier;
+        std::unique_ptr<Common::Event> exit_barrier;
+        std::atomic<bool> is_running;
+        std::atomic<bool> is_paused;
+        std::atomic<bool> initialized;
+        std::unique_ptr<std::thread> host_thread;
+    };
+
+    std::atomic<bool> running_mode{};
+    std::atomic<bool> paused_state{};
+
+    std::array<CoreData, Core::Hardware::NUM_CPU_CORES> core_data{};
+
+    bool is_async_gpu{};
+    bool is_multicore{};
+    std::atomic<std::size_t> current_core{};
+    std::size_t preemption_count{};
+    std::size_t idle_count{};
+    static constexpr std::size_t max_cycle_runs = 5;
 
     System& system;
 };
diff --git a/src/core/gdbstub/gdbstub.cpp b/src/core/gdbstub/gdbstub.cpp
index 70c0f8b80..79f22a403 100644
--- a/src/core/gdbstub/gdbstub.cpp
+++ b/src/core/gdbstub/gdbstub.cpp
@@ -35,7 +35,6 @@
 #include "common/swap.h"
 #include "core/arm/arm_interface.h"
 #include "core/core.h"
-#include "core/core_manager.h"
 #include "core/gdbstub/gdbstub.h"
 #include "core/hle/kernel/memory/page_table.h"
 #include "core/hle/kernel/process.h"
diff --git a/src/core/hardware_properties.h b/src/core/hardware_properties.h
index b04e046ed..456b41e1b 100644
--- a/src/core/hardware_properties.h
+++ b/src/core/hardware_properties.h
@@ -42,6 +42,10 @@ struct EmuThreadHandle {
         constexpr u32 invalid_handle = 0xFFFFFFFF;
         return {invalid_handle, invalid_handle};
     }
+
+    bool IsInvalid() const {
+        return (*this) == InvalidHandle();
+    }
 };
 
 } // namespace Core
diff --git a/src/core/hle/kernel/address_arbiter.cpp b/src/core/hle/kernel/address_arbiter.cpp
index 8475b698c..4d2a9b35d 100644
--- a/src/core/hle/kernel/address_arbiter.cpp
+++ b/src/core/hle/kernel/address_arbiter.cpp
@@ -7,11 +7,15 @@
 
 #include "common/assert.h"
 #include "common/common_types.h"
+#include "core/arm/exclusive_monitor.h"
 #include "core/core.h"
 #include "core/hle/kernel/address_arbiter.h"
 #include "core/hle/kernel/errors.h"
+#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
 #include "core/hle/result.h"
 #include "core/memory.h"
 
@@ -20,6 +24,7 @@ namespace Kernel {
 // Wake up num_to_wake (or all) threads in a vector.
 void AddressArbiter::WakeThreads(const std::vector<std::shared_ptr<Thread>>& waiting_threads,
                                  s32 num_to_wake) {
+    auto& time_manager = system.Kernel().TimeManager();
     // Only process up to 'target' threads, unless 'target' is <= 0, in which case process
     // them all.
     std::size_t last = waiting_threads.size();
@@ -29,12 +34,10 @@ void AddressArbiter::WakeThreads(const std::vector<std::shared_ptr<Thread>>& wai
 
     // Signal the waiting threads.
     for (std::size_t i = 0; i < last; i++) {
-        ASSERT(waiting_threads[i]->GetStatus() == ThreadStatus::WaitArb);
-        waiting_threads[i]->SetWaitSynchronizationResult(RESULT_SUCCESS);
+        waiting_threads[i]->SetSynchronizationResults(nullptr, RESULT_SUCCESS);
         RemoveThread(waiting_threads[i]);
-        waiting_threads[i]->SetArbiterWaitAddress(0);
+        waiting_threads[i]->WaitForArbitration(false);
         waiting_threads[i]->ResumeFromWait();
-        system.PrepareReschedule(waiting_threads[i]->GetProcessorID());
     }
 }
 
@@ -56,6 +59,7 @@ ResultCode AddressArbiter::SignalToAddress(VAddr address, SignalType type, s32 v
 }
 
 ResultCode AddressArbiter::SignalToAddressOnly(VAddr address, s32 num_to_wake) {
+    SchedulerLock lock(system.Kernel());
     const std::vector<std::shared_ptr<Thread>> waiting_threads =
         GetThreadsWaitingOnAddress(address);
     WakeThreads(waiting_threads, num_to_wake);
@@ -64,6 +68,7 @@ ResultCode AddressArbiter::SignalToAddressOnly(VAddr address, s32 num_to_wake) {
 
 ResultCode AddressArbiter::IncrementAndSignalToAddressIfEqual(VAddr address, s32 value,
                                                               s32 num_to_wake) {
+    SchedulerLock lock(system.Kernel());
     auto& memory = system.Memory();
 
     // Ensure that we can write to the address.
@@ -71,16 +76,24 @@ ResultCode AddressArbiter::IncrementAndSignalToAddressIfEqual(VAddr address, s32
         return ERR_INVALID_ADDRESS_STATE;
     }
 
-    if (static_cast<s32>(memory.Read32(address)) != value) {
-        return ERR_INVALID_STATE;
-    }
+    const std::size_t current_core = system.CurrentCoreIndex();
+    auto& monitor = system.Monitor();
+    u32 current_value;
+    do {
+        current_value = monitor.ExclusiveRead32(current_core, address);
+
+        if (current_value != value) {
+            return ERR_INVALID_STATE;
+        }
+        current_value++;
+    } while (!monitor.ExclusiveWrite32(current_core, address, current_value));
 
-    memory.Write32(address, static_cast<u32>(value + 1));
     return SignalToAddressOnly(address, num_to_wake);
 }
 
 ResultCode AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr address, s32 value,
                                                                          s32 num_to_wake) {
+    SchedulerLock lock(system.Kernel());
     auto& memory = system.Memory();
 
     // Ensure that we can write to the address.
@@ -92,29 +105,33 @@ ResultCode AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr a
     const std::vector<std::shared_ptr<Thread>> waiting_threads =
         GetThreadsWaitingOnAddress(address);
 
-    // Determine the modified value depending on the waiting count.
+    const std::size_t current_core = system.CurrentCoreIndex();
+    auto& monitor = system.Monitor();
     s32 updated_value;
-    if (num_to_wake <= 0) {
-        if (waiting_threads.empty()) {
-            updated_value = value + 1;
-        } else {
-            updated_value = value - 1;
+    do {
+        updated_value = monitor.ExclusiveRead32(current_core, address);
+
+        if (updated_value != value) {
+            return ERR_INVALID_STATE;
         }
-    } else {
-        if (waiting_threads.empty()) {
-            updated_value = value + 1;
-        } else if (waiting_threads.size() <= static_cast<u32>(num_to_wake)) {
-            updated_value = value - 1;
+        // Determine the modified value depending on the waiting count.
+        if (num_to_wake <= 0) {
+            if (waiting_threads.empty()) {
+                updated_value = value + 1;
+            } else {
+                updated_value = value - 1;
+            }
         } else {
-            updated_value = value;
+            if (waiting_threads.empty()) {
+                updated_value = value + 1;
+            } else if (waiting_threads.size() <= static_cast<u32>(num_to_wake)) {
+                updated_value = value - 1;
+            } else {
+                updated_value = value;
+            }
         }
-    }
+    } while (!monitor.ExclusiveWrite32(current_core, address, updated_value));
 
-    if (static_cast<s32>(memory.Read32(address)) != value) {
-        return ERR_INVALID_STATE;
-    }
-
-    memory.Write32(address, static_cast<u32>(updated_value));
     WakeThreads(waiting_threads, num_to_wake);
     return RESULT_SUCCESS;
 }
@@ -136,60 +153,127 @@ ResultCode AddressArbiter::WaitForAddress(VAddr address, ArbitrationType type, s
 ResultCode AddressArbiter::WaitForAddressIfLessThan(VAddr address, s32 value, s64 timeout,
                                                     bool should_decrement) {
     auto& memory = system.Memory();
+    auto& kernel = system.Kernel();
+    Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
 
-    // Ensure that we can read the address.
-    if (!memory.IsValidVirtualAddress(address)) {
-        return ERR_INVALID_ADDRESS_STATE;
-    }
+    Handle event_handle = InvalidHandle;
+    {
+        SchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout);
+
+        if (current_thread->IsPendingTermination()) {
+            lock.CancelSleep();
+            return ERR_THREAD_TERMINATING;
+        }
+
+        // Ensure that we can read the address.
+        if (!memory.IsValidVirtualAddress(address)) {
+            lock.CancelSleep();
+            return ERR_INVALID_ADDRESS_STATE;
+        }
+
+        s32 current_value = static_cast<s32>(memory.Read32(address));
+        if (current_value >= value) {
+            lock.CancelSleep();
+            return ERR_INVALID_STATE;
+        }
+
+        current_thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT);
+
+        s32 decrement_value;
+
+        const std::size_t current_core = system.CurrentCoreIndex();
+        auto& monitor = system.Monitor();
+        do {
+            current_value = static_cast<s32>(monitor.ExclusiveRead32(current_core, address));
+            if (should_decrement) {
+                decrement_value = current_value - 1;
+            } else {
+                decrement_value = current_value;
+            }
+        } while (
+            !monitor.ExclusiveWrite32(current_core, address, static_cast<u32>(decrement_value)));
+
+        // Short-circuit without rescheduling, if timeout is zero.
+        if (timeout == 0) {
+            lock.CancelSleep();
+            return RESULT_TIMEOUT;
+        }
 
-    const s32 cur_value = static_cast<s32>(memory.Read32(address));
-    if (cur_value >= value) {
-        return ERR_INVALID_STATE;
+        current_thread->SetArbiterWaitAddress(address);
+        InsertThread(SharedFrom(current_thread));
+        current_thread->SetStatus(ThreadStatus::WaitArb);
+        current_thread->WaitForArbitration(true);
     }
 
-    if (should_decrement) {
-        memory.Write32(address, static_cast<u32>(cur_value - 1));
+    if (event_handle != InvalidHandle) {
+        auto& time_manager = kernel.TimeManager();
+        time_manager.UnscheduleTimeEvent(event_handle);
     }
 
-    // Short-circuit without rescheduling, if timeout is zero.
-    if (timeout == 0) {
-        return RESULT_TIMEOUT;
+    {
+        SchedulerLock lock(kernel);
+        if (current_thread->IsWaitingForArbitration()) {
+            RemoveThread(SharedFrom(current_thread));
+            current_thread->WaitForArbitration(false);
+        }
     }
 
-    return WaitForAddressImpl(address, timeout);
+    return current_thread->GetSignalingResult();
 }
 
 ResultCode AddressArbiter::WaitForAddressIfEqual(VAddr address, s32 value, s64 timeout) {
     auto& memory = system.Memory();
+    auto& kernel = system.Kernel();
+    Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
 
-    // Ensure that we can read the address.
-    if (!memory.IsValidVirtualAddress(address)) {
-        return ERR_INVALID_ADDRESS_STATE;
-    }
+    Handle event_handle = InvalidHandle;
+    {
+        SchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout);
+
+        if (current_thread->IsPendingTermination()) {
+            lock.CancelSleep();
+            return ERR_THREAD_TERMINATING;
+        }
+
+        // Ensure that we can read the address.
+        if (!memory.IsValidVirtualAddress(address)) {
+            lock.CancelSleep();
+            return ERR_INVALID_ADDRESS_STATE;
+        }
 
-    // Only wait for the address if equal.
-    if (static_cast<s32>(memory.Read32(address)) != value) {
-        return ERR_INVALID_STATE;
+        s32 current_value = static_cast<s32>(memory.Read32(address));
+        if (current_value != value) {
+            lock.CancelSleep();
+            return ERR_INVALID_STATE;
+        }
+
+        // Short-circuit without rescheduling, if timeout is zero.
+        if (timeout == 0) {
+            lock.CancelSleep();
+            return RESULT_TIMEOUT;
+        }
+
+        current_thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT);
+        current_thread->SetArbiterWaitAddress(address);
+        InsertThread(SharedFrom(current_thread));
+        current_thread->SetStatus(ThreadStatus::WaitArb);
+        current_thread->WaitForArbitration(true);
     }
 
-    // Short-circuit without rescheduling if timeout is zero.
-    if (timeout == 0) {
-        return RESULT_TIMEOUT;
+    if (event_handle != InvalidHandle) {
+        auto& time_manager = kernel.TimeManager();
+        time_manager.UnscheduleTimeEvent(event_handle);
     }
 
-    return WaitForAddressImpl(address, timeout);
-}
+    {
+        SchedulerLock lock(kernel);
+        if (current_thread->IsWaitingForArbitration()) {
+            RemoveThread(SharedFrom(current_thread));
+            current_thread->WaitForArbitration(false);
+        }
+    }
 
-ResultCode AddressArbiter::WaitForAddressImpl(VAddr address, s64 timeout) {
-    Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
-    current_thread->SetArbiterWaitAddress(address);
-    InsertThread(SharedFrom(current_thread));
-    current_thread->SetStatus(ThreadStatus::WaitArb);
-    current_thread->InvalidateWakeupCallback();
-    current_thread->WakeAfterDelay(timeout);
-
-    system.PrepareReschedule(current_thread->GetProcessorID());
-    return RESULT_TIMEOUT;
+    return current_thread->GetSignalingResult();
 }
 
 void AddressArbiter::HandleWakeupThread(std::shared_ptr<Thread> thread) {
@@ -221,9 +305,9 @@ void AddressArbiter::RemoveThread(std::shared_ptr<Thread> thread) {
     const auto iter = std::find_if(thread_list.cbegin(), thread_list.cend(),
                                    [&thread](const auto& entry) { return thread == entry; });
 
-    ASSERT(iter != thread_list.cend());
-
-    thread_list.erase(iter);
+    if (iter != thread_list.cend()) {
+        thread_list.erase(iter);
+    }
 }
 
 std::vector<std::shared_ptr<Thread>> AddressArbiter::GetThreadsWaitingOnAddress(
diff --git a/src/core/hle/kernel/address_arbiter.h b/src/core/hle/kernel/address_arbiter.h
index f958eee5a..0b05d533c 100644
--- a/src/core/hle/kernel/address_arbiter.h
+++ b/src/core/hle/kernel/address_arbiter.h
@@ -73,9 +73,6 @@ private:
     /// Waits on an address if the value passed is equal to the argument value.
     ResultCode WaitForAddressIfEqual(VAddr address, s32 value, s64 timeout);
 
-    // Waits on the given address with a timeout in nanoseconds
-    ResultCode WaitForAddressImpl(VAddr address, s64 timeout);
-
     /// Wake up num_to_wake (or all) threads in a vector.
     void WakeThreads(const std::vector<std::shared_ptr<Thread>>& waiting_threads, s32 num_to_wake);
 
diff --git a/src/core/hle/kernel/client_port.cpp b/src/core/hle/kernel/client_port.cpp
index 5498fd313..8aff2227a 100644
--- a/src/core/hle/kernel/client_port.cpp
+++ b/src/core/hle/kernel/client_port.cpp
@@ -34,7 +34,7 @@ ResultVal<std::shared_ptr<ClientSession>> ClientPort::Connect() {
     }
 
     // Wake the threads waiting on the ServerPort
-    server_port->WakeupAllWaitingThreads();
+    server_port->Signal();
 
     return MakeResult(std::move(client));
 }
diff --git a/src/core/hle/kernel/errors.h b/src/core/hle/kernel/errors.h
index 29bfa3621..d4e5d88cf 100644
--- a/src/core/hle/kernel/errors.h
+++ b/src/core/hle/kernel/errors.h
@@ -12,6 +12,7 @@ namespace Kernel {
 
 constexpr ResultCode ERR_MAX_CONNECTIONS_REACHED{ErrorModule::Kernel, 7};
 constexpr ResultCode ERR_INVALID_CAPABILITY_DESCRIPTOR{ErrorModule::Kernel, 14};
+constexpr ResultCode ERR_THREAD_TERMINATING{ErrorModule::Kernel, 59};
 constexpr ResultCode ERR_INVALID_SIZE{ErrorModule::Kernel, 101};
 constexpr ResultCode ERR_INVALID_ADDRESS{ErrorModule::Kernel, 102};
 constexpr ResultCode ERR_OUT_OF_RESOURCES{ErrorModule::Kernel, 103};
diff --git a/src/core/hle/kernel/hle_ipc.cpp b/src/core/hle/kernel/hle_ipc.cpp
index 0d01a7047..9277b5d08 100644
--- a/src/core/hle/kernel/hle_ipc.cpp
+++ b/src/core/hle/kernel/hle_ipc.cpp
@@ -14,14 +14,17 @@
 #include "common/common_types.h"
 #include "common/logging/log.h"
 #include "core/hle/ipc_helpers.h"
+#include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/readable_event.h"
+#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/server_session.h"
 #include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
 #include "core/hle/kernel/writable_event.h"
 #include "core/memory.h"
 
@@ -46,15 +49,6 @@ std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread(
     const std::string& reason, u64 timeout, WakeupCallback&& callback,
     std::shared_ptr<WritableEvent> writable_event) {
     // Put the client thread to sleep until the wait event is signaled or the timeout expires.
-    thread->SetWakeupCallback(
-        [context = *this, callback](ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
-                                    std::shared_ptr<SynchronizationObject> object,
-                                    std::size_t index) mutable -> bool {
-            ASSERT(thread->GetStatus() == ThreadStatus::WaitHLEEvent);
-            callback(thread, context, reason);
-            context.WriteToOutgoingCommandBuffer(*thread);
-            return true;
-        });
 
     if (!writable_event) {
         // Create event if not provided
@@ -62,14 +56,26 @@ std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread(
         writable_event = pair.writable;
     }
 
-    const auto readable_event{writable_event->GetReadableEvent()};
-    writable_event->Clear();
-    thread->SetStatus(ThreadStatus::WaitHLEEvent);
-    thread->SetSynchronizationObjects({readable_event});
-    readable_event->AddWaitingThread(thread);
-
-    if (timeout > 0) {
-        thread->WakeAfterDelay(timeout);
+    {
+        Handle event_handle = InvalidHandle;
+        SchedulerLockAndSleep lock(kernel, event_handle, thread.get(), timeout);
+        thread->SetHLECallback(
+            [context = *this, callback](std::shared_ptr<Thread> thread) mutable -> bool {
+                ThreadWakeupReason reason = thread->GetSignalingResult() == RESULT_TIMEOUT
+                                                ? ThreadWakeupReason::Timeout
+                                                : ThreadWakeupReason::Signal;
+                callback(thread, context, reason);
+                context.WriteToOutgoingCommandBuffer(*thread);
+                return true;
+            });
+        const auto readable_event{writable_event->GetReadableEvent()};
+        writable_event->Clear();
+        thread->SetHLESyncObject(readable_event.get());
+        thread->SetStatus(ThreadStatus::WaitHLEEvent);
+        thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT);
+        readable_event->AddWaitingThread(thread);
+        lock.Release();
+        thread->SetHLETimeEvent(event_handle);
     }
 
     is_thread_waiting = true;
diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp
index 7655382fa..1f2af7a1b 100644
--- a/src/core/hle/kernel/kernel.cpp
+++ b/src/core/hle/kernel/kernel.cpp
@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <array>
 #include <atomic>
 #include <bitset>
 #include <functional>
@@ -13,11 +14,15 @@
 
 #include "common/assert.h"
 #include "common/logging/log.h"
+#include "common/microprofile.h"
+#include "common/thread.h"
 #include "core/arm/arm_interface.h"
+#include "core/arm/cpu_interrupt_handler.h"
 #include "core/arm/exclusive_monitor.h"
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/cpu_manager.h"
 #include "core/device_memory.h"
 #include "core/hardware_properties.h"
 #include "core/hle/kernel/client_port.h"
@@ -39,85 +44,28 @@
 #include "core/hle/result.h"
 #include "core/memory.h"
 
-namespace Kernel {
-
-/**
- * Callback that will wake up the thread it was scheduled for
- * @param thread_handle The handle of the thread that's been awoken
- * @param cycles_late The number of CPU cycles that have passed since the desired wakeup time
- */
-static void ThreadWakeupCallback(u64 thread_handle, [[maybe_unused]] s64 cycles_late) {
-    const auto proper_handle = static_cast<Handle>(thread_handle);
-    const auto& system = Core::System::GetInstance();
-
-    // Lock the global kernel mutex when we enter the kernel HLE.
-    std::lock_guard lock{HLE::g_hle_lock};
-
-    std::shared_ptr<Thread> thread =
-        system.Kernel().RetrieveThreadFromGlobalHandleTable(proper_handle);
-    if (thread == nullptr) {
-        LOG_CRITICAL(Kernel, "Callback fired for invalid thread {:08X}", proper_handle);
-        return;
-    }
-
-    bool resume = true;
-
-    if (thread->GetStatus() == ThreadStatus::WaitSynch ||
-        thread->GetStatus() == ThreadStatus::WaitHLEEvent) {
-        // Remove the thread from each of its waiting objects' waitlists
-        for (const auto& object : thread->GetSynchronizationObjects()) {
-            object->RemoveWaitingThread(thread);
-        }
-        thread->ClearSynchronizationObjects();
-
-        // Invoke the wakeup callback before clearing the wait objects
-        if (thread->HasWakeupCallback()) {
-            resume = thread->InvokeWakeupCallback(ThreadWakeupReason::Timeout, thread, nullptr, 0);
-        }
-    } else if (thread->GetStatus() == ThreadStatus::WaitMutex ||
-               thread->GetStatus() == ThreadStatus::WaitCondVar) {
-        thread->SetMutexWaitAddress(0);
-        thread->SetWaitHandle(0);
-        if (thread->GetStatus() == ThreadStatus::WaitCondVar) {
-            thread->GetOwnerProcess()->RemoveConditionVariableThread(thread);
-            thread->SetCondVarWaitAddress(0);
-        }
-
-        auto* const lock_owner = thread->GetLockOwner();
-        // Threads waking up by timeout from WaitProcessWideKey do not perform priority inheritance
-        // and don't have a lock owner unless SignalProcessWideKey was called first and the thread
-        // wasn't awakened due to the mutex already being acquired.
-        if (lock_owner != nullptr) {
-            lock_owner->RemoveMutexWaiter(thread);
-        }
-    }
+MICROPROFILE_DEFINE(Kernel_SVC, "Kernel", "SVC", MP_RGB(70, 200, 70));
 
-    if (thread->GetStatus() == ThreadStatus::WaitArb) {
-        auto& address_arbiter = thread->GetOwnerProcess()->GetAddressArbiter();
-        address_arbiter.HandleWakeupThread(thread);
-    }
-
-    if (resume) {
-        if (thread->GetStatus() == ThreadStatus::WaitCondVar ||
-            thread->GetStatus() == ThreadStatus::WaitArb) {
-            thread->SetWaitSynchronizationResult(RESULT_TIMEOUT);
-        }
-        thread->ResumeFromWait();
-    }
-}
+namespace Kernel {
 
 struct KernelCore::Impl {
     explicit Impl(Core::System& system, KernelCore& kernel)
         : global_scheduler{kernel}, synchronization{system}, time_manager{system}, system{system} {}
 
+    void SetMulticore(bool is_multicore) {
+        this->is_multicore = is_multicore;
+    }
+
     void Initialize(KernelCore& kernel) {
         Shutdown();
+        RegisterHostThread();
 
         InitializePhysicalCores();
         InitializeSystemResourceLimit(kernel);
         InitializeMemoryLayout();
-        InitializeThreads();
-        InitializePreemption();
+        InitializePreemption(kernel);
+        InitializeSchedulers();
+        InitializeSuspendThreads();
     }
 
     void Shutdown() {
@@ -126,13 +74,26 @@ struct KernelCore::Impl {
         next_user_process_id = Process::ProcessIDMin;
         next_thread_id = 1;
 
+        for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
+            if (suspend_threads[i]) {
+                suspend_threads[i].reset();
+            }
+        }
+
+        for (std::size_t i = 0; i < cores.size(); i++) {
+            cores[i].Shutdown();
+            schedulers[i].reset();
+        }
+        cores.clear();
+
+        registered_core_threads.reset();
+
         process_list.clear();
         current_process = nullptr;
 
         system_resource_limit = nullptr;
 
         global_handle_table.Clear();
-        thread_wakeup_event_type = nullptr;
         preemption_event = nullptr;
 
         global_scheduler.Shutdown();
@@ -145,13 +106,21 @@ struct KernelCore::Impl {
         cores.clear();
 
         exclusive_monitor.reset();
+        host_thread_ids.clear();
     }
 
     void InitializePhysicalCores() {
         exclusive_monitor =
             Core::MakeExclusiveMonitor(system.Memory(), Core::Hardware::NUM_CPU_CORES);
         for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
-            cores.emplace_back(system, i, *exclusive_monitor);
+            schedulers[i] = std::make_unique<Kernel::Scheduler>(system, i);
+            cores.emplace_back(system, i, *schedulers[i], interrupts[i]);
+        }
+    }
+
+    void InitializeSchedulers() {
+        for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
+            cores[i].Scheduler().Initialize();
         }
     }
 
@@ -173,15 +142,13 @@ struct KernelCore::Impl {
         }
     }
 
-    void InitializeThreads() {
-        thread_wakeup_event_type =
-            Core::Timing::CreateEvent("ThreadWakeupCallback", ThreadWakeupCallback);
-    }
-
-    void InitializePreemption() {
-        preemption_event =
-            Core::Timing::CreateEvent("PreemptionCallback", [this](u64 userdata, s64 cycles_late) {
-                global_scheduler.PreemptThreads();
+    void InitializePreemption(KernelCore& kernel) {
+        preemption_event = Core::Timing::CreateEvent(
+            "PreemptionCallback", [this, &kernel](u64 userdata, s64 cycles_late) {
+                {
+                    SchedulerLock lock(kernel);
+                    global_scheduler.PreemptThreads();
+                }
                 s64 time_interval = Core::Timing::msToCycles(std::chrono::milliseconds(10));
                 system.CoreTiming().ScheduleEvent(time_interval, preemption_event);
             });
@@ -190,6 +157,20 @@ struct KernelCore::Impl {
         system.CoreTiming().ScheduleEvent(time_interval, preemption_event);
     }
 
+    void InitializeSuspendThreads() {
+        for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
+            std::string name = "Suspend Thread Id:" + std::to_string(i);
+            std::function<void(void*)> init_func =
+                system.GetCpuManager().GetSuspendThreadStartFunc();
+            void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater();
+            ThreadType type =
+                static_cast<ThreadType>(THREADTYPE_KERNEL | THREADTYPE_HLE | THREADTYPE_SUSPEND);
+            auto thread_res = Thread::Create(system, type, name, 0, 0, 0, static_cast<u32>(i), 0,
+                                             nullptr, std::move(init_func), init_func_parameter);
+            suspend_threads[i] = std::move(thread_res).Unwrap();
+        }
+    }
+
     void MakeCurrentProcess(Process* process) {
         current_process = process;
 
@@ -197,15 +178,17 @@ struct KernelCore::Impl {
             return;
         }
 
-        for (auto& core : cores) {
-            core.SetIs64Bit(process->Is64BitProcess());
+        u32 core_id = GetCurrentHostThreadID();
+        if (core_id < Core::Hardware::NUM_CPU_CORES) {
+            system.Memory().SetCurrentPageTable(*process, core_id);
         }
-
-        system.Memory().SetCurrentPageTable(*process);
     }
 
     void RegisterCoreThread(std::size_t core_id) {
         std::unique_lock lock{register_thread_mutex};
+        if (!is_multicore) {
+            single_core_thread_id = std::this_thread::get_id();
+        }
         const std::thread::id this_id = std::this_thread::get_id();
         const auto it = host_thread_ids.find(this_id);
         ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
@@ -219,12 +202,19 @@ struct KernelCore::Impl {
         std::unique_lock lock{register_thread_mutex};
         const std::thread::id this_id = std::this_thread::get_id();
         const auto it = host_thread_ids.find(this_id);
-        ASSERT(it == host_thread_ids.end());
+        if (it != host_thread_ids.end()) {
+            return;
+        }
         host_thread_ids[this_id] = registered_thread_ids++;
     }
 
     u32 GetCurrentHostThreadID() const {
         const std::thread::id this_id = std::this_thread::get_id();
+        if (!is_multicore) {
+            if (single_core_thread_id == this_id) {
+                return static_cast<u32>(system.GetCpuManager().CurrentCore());
+            }
+        }
         const auto it = host_thread_ids.find(this_id);
         if (it == host_thread_ids.end()) {
             return Core::INVALID_HOST_THREAD_ID;
@@ -240,7 +230,7 @@ struct KernelCore::Impl {
         }
         const Kernel::Scheduler& sched = cores[result.host_handle].Scheduler();
         const Kernel::Thread* current = sched.GetCurrentThread();
-        if (current != nullptr) {
+        if (current != nullptr && !current->IsPhantomMode()) {
             result.guest_handle = current->GetGlobalHandle();
         } else {
             result.guest_handle = InvalidHandle;
@@ -313,7 +303,6 @@ struct KernelCore::Impl {
 
     std::shared_ptr<ResourceLimit> system_resource_limit;
 
-    std::shared_ptr<Core::Timing::EventType> thread_wakeup_event_type;
     std::shared_ptr<Core::Timing::EventType> preemption_event;
 
     // This is the kernel's handle table or supervisor handle table which
@@ -343,6 +332,15 @@ struct KernelCore::Impl {
     std::shared_ptr<Kernel::SharedMemory> irs_shared_mem;
     std::shared_ptr<Kernel::SharedMemory> time_shared_mem;
 
+    std::array<std::shared_ptr<Thread>, Core::Hardware::NUM_CPU_CORES> suspend_threads{};
+    std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES> interrupts{};
+    std::array<std::unique_ptr<Kernel::Scheduler>, Core::Hardware::NUM_CPU_CORES> schedulers{};
+
+    bool is_multicore{};
+    std::thread::id single_core_thread_id{};
+
+    std::array<u64, Core::Hardware::NUM_CPU_CORES> svc_ticks{};
+
     // System context
     Core::System& system;
 };
@@ -352,6 +350,10 @@ KernelCore::~KernelCore() {
     Shutdown();
 }
 
+void KernelCore::SetMulticore(bool is_multicore) {
+    impl->SetMulticore(is_multicore);
+}
+
 void KernelCore::Initialize() {
     impl->Initialize(*this);
 }
@@ -397,11 +399,11 @@ const Kernel::GlobalScheduler& KernelCore::GlobalScheduler() const {
 }
 
 Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) {
-    return impl->cores[id].Scheduler();
+    return *impl->schedulers[id];
 }
 
 const Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) const {
-    return impl->cores[id].Scheduler();
+    return *impl->schedulers[id];
 }
 
 Kernel::PhysicalCore& KernelCore::PhysicalCore(std::size_t id) {
@@ -412,6 +414,39 @@ const Kernel::PhysicalCore& KernelCore::PhysicalCore(std::size_t id) const {
     return impl->cores[id];
 }
 
+Kernel::PhysicalCore& KernelCore::CurrentPhysicalCore() {
+    u32 core_id = impl->GetCurrentHostThreadID();
+    ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
+    return impl->cores[core_id];
+}
+
+const Kernel::PhysicalCore& KernelCore::CurrentPhysicalCore() const {
+    u32 core_id = impl->GetCurrentHostThreadID();
+    ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
+    return impl->cores[core_id];
+}
+
+Kernel::Scheduler& KernelCore::CurrentScheduler() {
+    u32 core_id = impl->GetCurrentHostThreadID();
+    ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
+    return *impl->schedulers[core_id];
+}
+
+const Kernel::Scheduler& KernelCore::CurrentScheduler() const {
+    u32 core_id = impl->GetCurrentHostThreadID();
+    ASSERT(core_id < Core::Hardware::NUM_CPU_CORES);
+    return *impl->schedulers[core_id];
+}
+
+std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>& KernelCore::Interrupts() {
+    return impl->interrupts;
+}
+
+const std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>& KernelCore::Interrupts()
+    const {
+    return impl->interrupts;
+}
+
 Kernel::Synchronization& KernelCore::Synchronization() {
     return impl->synchronization;
 }
@@ -437,15 +472,17 @@ const Core::ExclusiveMonitor& KernelCore::GetExclusiveMonitor() const {
 }
 
 void KernelCore::InvalidateAllInstructionCaches() {
-    for (std::size_t i = 0; i < impl->global_scheduler.CpuCoresCount(); i++) {
-        PhysicalCore(i).ArmInterface().ClearInstructionCache();
+    auto& threads = GlobalScheduler().GetThreadList();
+    for (auto& thread : threads) {
+        if (!thread->IsHLEThread()) {
+            auto& arm_interface = thread->ArmInterface();
+            arm_interface.ClearInstructionCache();
+        }
     }
 }
 
 void KernelCore::PrepareReschedule(std::size_t id) {
-    if (id < impl->global_scheduler.CpuCoresCount()) {
-        impl->cores[id].Stop();
-    }
+    // TODO: Reimplement, this
 }
 
 void KernelCore::AddNamedPort(std::string name, std::shared_ptr<ClientPort> port) {
@@ -481,10 +518,6 @@ u64 KernelCore::CreateNewUserProcessID() {
     return impl->next_user_process_id++;
 }
 
-const std::shared_ptr<Core::Timing::EventType>& KernelCore::ThreadWakeupCallbackEventType() const {
-    return impl->thread_wakeup_event_type;
-}
-
 Kernel::HandleTable& KernelCore::GlobalHandleTable() {
     return impl->global_handle_table;
 }
@@ -557,4 +590,34 @@ const Kernel::SharedMemory& KernelCore::GetTimeSharedMem() const {
     return *impl->time_shared_mem;
 }
 
+void KernelCore::Suspend(bool in_suspention) {
+    const bool should_suspend = exception_exited || in_suspention;
+    {
+        SchedulerLock lock(*this);
+        ThreadStatus status = should_suspend ? ThreadStatus::Ready : ThreadStatus::WaitSleep;
+        for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
+            impl->suspend_threads[i]->SetStatus(status);
+        }
+    }
+}
+
+bool KernelCore::IsMulticore() const {
+    return impl->is_multicore;
+}
+
+void KernelCore::ExceptionalExit() {
+    exception_exited = true;
+    Suspend(true);
+}
+
+void KernelCore::EnterSVCProfile() {
+    std::size_t core = impl->GetCurrentHostThreadID();
+    impl->svc_ticks[core] = MicroProfileEnter(MICROPROFILE_TOKEN(Kernel_SVC));
+}
+
+void KernelCore::ExitSVCProfile() {
+    std::size_t core = impl->GetCurrentHostThreadID();
+    MicroProfileLeave(MICROPROFILE_TOKEN(Kernel_SVC), impl->svc_ticks[core]);
+}
+
 } // namespace Kernel
diff --git a/src/core/hle/kernel/kernel.h b/src/core/hle/kernel/kernel.h
index 83de1f542..49bd47e89 100644
--- a/src/core/hle/kernel/kernel.h
+++ b/src/core/hle/kernel/kernel.h
@@ -4,15 +4,17 @@
 
 #pragma once
 
+#include <array>
 #include <memory>
 #include <string>
 #include <unordered_map>
 #include <vector>
+#include "core/hardware_properties.h"
 #include "core/hle/kernel/memory/memory_types.h"
 #include "core/hle/kernel/object.h"
 
 namespace Core {
-struct EmuThreadHandle;
+class CPUInterruptHandler;
 class ExclusiveMonitor;
 class System;
 } // namespace Core
@@ -65,6 +67,9 @@ public:
     KernelCore(KernelCore&&) = delete;
     KernelCore& operator=(KernelCore&&) = delete;
 
+    /// Sets if emulation is multicore or single core, must be set before Initialize
+    void SetMulticore(bool is_multicore);
+
     /// Resets the kernel to a clean slate for use.
     void Initialize();
 
@@ -110,6 +115,18 @@ public:
     /// Gets the an instance of the respective physical CPU core.
     const Kernel::PhysicalCore& PhysicalCore(std::size_t id) const;
 
+    /// Gets the sole instance of the Scheduler at the current running core.
+    Kernel::Scheduler& CurrentScheduler();
+
+    /// Gets the sole instance of the Scheduler at the current running core.
+    const Kernel::Scheduler& CurrentScheduler() const;
+
+    /// Gets the an instance of the current physical CPU core.
+    Kernel::PhysicalCore& CurrentPhysicalCore();
+
+    /// Gets the an instance of the current physical CPU core.
+    const Kernel::PhysicalCore& CurrentPhysicalCore() const;
+
     /// Gets the an instance of the Synchronization Interface.
     Kernel::Synchronization& Synchronization();
 
@@ -129,6 +146,10 @@ public:
 
     const Core::ExclusiveMonitor& GetExclusiveMonitor() const;
 
+    std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>& Interrupts();
+
+    const std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>& Interrupts() const;
+
     void InvalidateAllInstructionCaches();
 
     /// Adds a port to the named port table
@@ -191,6 +212,18 @@ public:
     /// Gets the shared memory object for Time services.
     const Kernel::SharedMemory& GetTimeSharedMem() const;
 
+    /// Suspend/unsuspend the OS.
+    void Suspend(bool in_suspention);
+
+    /// Exceptional exit the OS.
+    void ExceptionalExit();
+
+    bool IsMulticore() const;
+
+    void EnterSVCProfile();
+
+    void ExitSVCProfile();
+
 private:
     friend class Object;
     friend class Process;
@@ -208,9 +241,6 @@ private:
     /// Creates a new thread ID, incrementing the internal thread ID counter.
     u64 CreateNewThreadID();
 
-    /// Retrieves the event type used for thread wakeup callbacks.
-    const std::shared_ptr<Core::Timing::EventType>& ThreadWakeupCallbackEventType() const;
-
     /// Provides a reference to the global handle table.
     Kernel::HandleTable& GlobalHandleTable();
 
@@ -219,6 +249,7 @@ private:
 
     struct Impl;
     std::unique_ptr<Impl> impl;
+    bool exception_exited{};
 };
 
 } // namespace Kernel
diff --git a/src/core/hle/kernel/mutex.cpp b/src/core/hle/kernel/mutex.cpp
index 7869eb32b..8f6c944d1 100644
--- a/src/core/hle/kernel/mutex.cpp
+++ b/src/core/hle/kernel/mutex.cpp
@@ -34,8 +34,6 @@ static std::pair<std::shared_ptr<Thread>, u32> GetHighestPriorityMutexWaitingThr
         if (thread->GetMutexWaitAddress() != mutex_addr)
             continue;
 
-        ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex);
-
         ++num_waiters;
         if (highest_priority_thread == nullptr ||
             thread->GetPriority() < highest_priority_thread->GetPriority()) {
@@ -49,6 +47,7 @@ static std::pair<std::shared_ptr<Thread>, u32> GetHighestPriorityMutexWaitingThr
 /// Update the mutex owner field of all threads waiting on the mutex to point to the new owner.
 static void TransferMutexOwnership(VAddr mutex_addr, std::shared_ptr<Thread> current_thread,
                                    std::shared_ptr<Thread> new_owner) {
+    current_thread->RemoveMutexWaiter(new_owner);
     const auto threads = current_thread->GetMutexWaitingThreads();
     for (const auto& thread : threads) {
         if (thread->GetMutexWaitAddress() != mutex_addr)
@@ -72,85 +71,100 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
         return ERR_INVALID_ADDRESS;
     }
 
-    const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
+    auto& kernel = system.Kernel();
     std::shared_ptr<Thread> current_thread =
-        SharedFrom(system.CurrentScheduler().GetCurrentThread());
-    std::shared_ptr<Thread> holding_thread = handle_table.Get<Thread>(holding_thread_handle);
-    std::shared_ptr<Thread> requesting_thread = handle_table.Get<Thread>(requesting_thread_handle);
+        SharedFrom(kernel.CurrentScheduler().GetCurrentThread());
+    {
+        SchedulerLock lock(kernel);
+        // The mutex address must be 4-byte aligned
+        if ((address % sizeof(u32)) != 0) {
+            return ERR_INVALID_ADDRESS;
+        }
 
-    // TODO(Subv): It is currently unknown if it is possible to lock a mutex in behalf of another
-    // thread.
-    ASSERT(requesting_thread == current_thread);
+        const auto& handle_table = kernel.CurrentProcess()->GetHandleTable();
+        std::shared_ptr<Thread> holding_thread = handle_table.Get<Thread>(holding_thread_handle);
+        std::shared_ptr<Thread> requesting_thread =
+            handle_table.Get<Thread>(requesting_thread_handle);
 
-    const u32 addr_value = system.Memory().Read32(address);
+        // TODO(Subv): It is currently unknown if it is possible to lock a mutex in behalf of
+        // another thread.
+        ASSERT(requesting_thread == current_thread);
 
-    // If the mutex isn't being held, just return success.
-    if (addr_value != (holding_thread_handle | Mutex::MutexHasWaitersFlag)) {
-        return RESULT_SUCCESS;
-    }
+        current_thread->SetSynchronizationResults(nullptr, RESULT_SUCCESS);
 
-    if (holding_thread == nullptr) {
-        LOG_ERROR(Kernel, "Holding thread does not exist! thread_handle={:08X}",
-                  holding_thread_handle);
-        return ERR_INVALID_HANDLE;
-    }
+        const u32 addr_value = system.Memory().Read32(address);
+
+        // If the mutex isn't being held, just return success.
+        if (addr_value != (holding_thread_handle | Mutex::MutexHasWaitersFlag)) {
+            return RESULT_SUCCESS;
+        }
 
-    // Wait until the mutex is released
-    current_thread->SetMutexWaitAddress(address);
-    current_thread->SetWaitHandle(requesting_thread_handle);
+        if (holding_thread == nullptr) {
+            return ERR_INVALID_HANDLE;
+        }
 
-    current_thread->SetStatus(ThreadStatus::WaitMutex);
-    current_thread->InvalidateWakeupCallback();
+        // Wait until the mutex is released
+        current_thread->SetMutexWaitAddress(address);
+        current_thread->SetWaitHandle(requesting_thread_handle);
 
-    // Update the lock holder thread's priority to prevent priority inversion.
-    holding_thread->AddMutexWaiter(current_thread);
+        current_thread->SetStatus(ThreadStatus::WaitMutex);
 
-    system.PrepareReschedule();
+        // Update the lock holder thread's priority to prevent priority inversion.
+        holding_thread->AddMutexWaiter(current_thread);
+    }
 
-    return RESULT_SUCCESS;
+    {
+        SchedulerLock lock(kernel);
+        auto* owner = current_thread->GetLockOwner();
+        if (owner != nullptr) {
+            owner->RemoveMutexWaiter(current_thread);
+        }
+    }
+    return current_thread->GetSignalingResult();
 }
 
-ResultCode Mutex::Release(VAddr address) {
+std::pair<ResultCode, std::shared_ptr<Thread>> Mutex::Unlock(std::shared_ptr<Thread> owner,
+                                                             VAddr address) {
     // The mutex address must be 4-byte aligned
     if ((address % sizeof(u32)) != 0) {
         LOG_ERROR(Kernel, "Address is not 4-byte aligned! address={:016X}", address);
-        return ERR_INVALID_ADDRESS;
+        return {ERR_INVALID_ADDRESS, nullptr};
     }
 
-    std::shared_ptr<Thread> current_thread =
-        SharedFrom(system.CurrentScheduler().GetCurrentThread());
-    auto [thread, num_waiters] = GetHighestPriorityMutexWaitingThread(current_thread, address);
-
-    // There are no more threads waiting for the mutex, release it completely.
-    if (thread == nullptr) {
+    auto [new_owner, num_waiters] = GetHighestPriorityMutexWaitingThread(owner, address);
+    if (new_owner == nullptr) {
         system.Memory().Write32(address, 0);
-        return RESULT_SUCCESS;
+        return {RESULT_SUCCESS, nullptr};
     }
-
     // Transfer the ownership of the mutex from the previous owner to the new one.
-    TransferMutexOwnership(address, current_thread, thread);
-
-    u32 mutex_value = thread->GetWaitHandle();
-
+    TransferMutexOwnership(address, owner, new_owner);
+    u32 mutex_value = new_owner->GetWaitHandle();
     if (num_waiters >= 2) {
         // Notify the guest that there are still some threads waiting for the mutex
         mutex_value |= Mutex::MutexHasWaitersFlag;
     }
+    new_owner->SetSynchronizationResults(nullptr, RESULT_SUCCESS);
+    new_owner->SetLockOwner(nullptr);
+    new_owner->ResumeFromWait();
 
-    // Grant the mutex to the next waiting thread and resume it.
     system.Memory().Write32(address, mutex_value);
+    return {RESULT_SUCCESS, new_owner};
+}
 
-    ASSERT(thread->GetStatus() == ThreadStatus::WaitMutex);
-    thread->ResumeFromWait();
+ResultCode Mutex::Release(VAddr address) {
+    auto& kernel = system.Kernel();
+    SchedulerLock lock(kernel);
 
-    thread->SetLockOwner(nullptr);
-    thread->SetCondVarWaitAddress(0);
-    thread->SetMutexWaitAddress(0);
-    thread->SetWaitHandle(0);
-    thread->SetWaitSynchronizationResult(RESULT_SUCCESS);
+    std::shared_ptr<Thread> current_thread =
+        SharedFrom(kernel.CurrentScheduler().GetCurrentThread());
 
-    system.PrepareReschedule();
+    auto [result, new_owner] = Unlock(current_thread, address);
 
-    return RESULT_SUCCESS;
+    if (result != RESULT_SUCCESS && new_owner != nullptr) {
+        new_owner->SetSynchronizationResults(nullptr, result);
+    }
+
+    return result;
 }
+
 } // namespace Kernel
diff --git a/src/core/hle/kernel/mutex.h b/src/core/hle/kernel/mutex.h
index b904de2e8..3b81dc3df 100644
--- a/src/core/hle/kernel/mutex.h
+++ b/src/core/hle/kernel/mutex.h
@@ -28,6 +28,10 @@ public:
     ResultCode TryAcquire(VAddr address, Handle holding_thread_handle,
                           Handle requesting_thread_handle);
 
+    /// Unlocks a mutex for owner at address
+    std::pair<ResultCode, std::shared_ptr<Thread>> Unlock(std::shared_ptr<Thread> owner,
+                                                          VAddr address);
+
     /// Releases the mutex at the specified address.
     ResultCode Release(VAddr address);
 
diff --git a/src/core/hle/kernel/physical_core.cpp b/src/core/hle/kernel/physical_core.cpp
index a15011076..c6bbdb080 100644
--- a/src/core/hle/kernel/physical_core.cpp
+++ b/src/core/hle/kernel/physical_core.cpp
@@ -2,12 +2,15 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include "common/assert.h"
 #include "common/logging/log.h"
+#include "common/spin_lock.h"
 #include "core/arm/arm_interface.h"
 #ifdef ARCHITECTURE_x86_64
 #include "core/arm/dynarmic/arm_dynarmic_32.h"
 #include "core/arm/dynarmic/arm_dynarmic_64.h"
 #endif
+#include "core/arm/cpu_interrupt_handler.h"
 #include "core/arm/exclusive_monitor.h"
 #include "core/arm/unicorn/arm_unicorn.h"
 #include "core/core.h"
@@ -17,50 +20,37 @@
 
 namespace Kernel {
 
-PhysicalCore::PhysicalCore(Core::System& system, std::size_t id,
-                           Core::ExclusiveMonitor& exclusive_monitor)
-    : core_index{id} {
-#ifdef ARCHITECTURE_x86_64
-    arm_interface_32 =
-        std::make_unique<Core::ARM_Dynarmic_32>(system, exclusive_monitor, core_index);
-    arm_interface_64 =
-        std::make_unique<Core::ARM_Dynarmic_64>(system, exclusive_monitor, core_index);
-
-#else
-    using Core::ARM_Unicorn;
-    arm_interface_32 = std::make_unique<ARM_Unicorn>(system, ARM_Unicorn::Arch::AArch32);
-    arm_interface_64 = std::make_unique<ARM_Unicorn>(system, ARM_Unicorn::Arch::AArch64);
-    LOG_WARNING(Core, "CPU JIT requested, but Dynarmic not available");
-#endif
+PhysicalCore::PhysicalCore(Core::System& system, std::size_t id, Kernel::Scheduler& scheduler,
+                           Core::CPUInterruptHandler& interrupt_handler)
+    : interrupt_handler{interrupt_handler}, core_index{id}, scheduler{scheduler} {
 
-    scheduler = std::make_unique<Kernel::Scheduler>(system, core_index);
+    guard = std::make_unique<Common::SpinLock>();
 }
 
 PhysicalCore::~PhysicalCore() = default;
 
-void PhysicalCore::Run() {
-    arm_interface->Run();
-    arm_interface->ClearExclusiveState();
+void PhysicalCore::Idle() {
+    interrupt_handler.AwaitInterrupt();
 }
 
-void PhysicalCore::Step() {
-    arm_interface->Step();
+void PhysicalCore::Shutdown() {
+    scheduler.Shutdown();
 }
 
-void PhysicalCore::Stop() {
-    arm_interface->PrepareReschedule();
+bool PhysicalCore::IsInterrupted() const {
+    return interrupt_handler.IsInterrupted();
 }
 
-void PhysicalCore::Shutdown() {
-    scheduler->Shutdown();
+void PhysicalCore::Interrupt() {
+    guard->lock();
+    interrupt_handler.SetInterrupt(true);
+    guard->unlock();
 }
 
-void PhysicalCore::SetIs64Bit(bool is_64_bit) {
-    if (is_64_bit) {
-        arm_interface = arm_interface_64.get();
-    } else {
-        arm_interface = arm_interface_32.get();
-    }
+void PhysicalCore::ClearInterrupt() {
+    guard->lock();
+    interrupt_handler.SetInterrupt(false);
+    guard->unlock();
 }
 
 } // namespace Kernel
diff --git a/src/core/hle/kernel/physical_core.h b/src/core/hle/kernel/physical_core.h
index 3269166be..d7a7a951c 100644
--- a/src/core/hle/kernel/physical_core.h
+++ b/src/core/hle/kernel/physical_core.h
@@ -7,12 +7,17 @@
 #include <cstddef>
 #include <memory>
 
+namespace Common {
+class SpinLock;
+}
+
 namespace Kernel {
 class Scheduler;
 } // namespace Kernel
 
 namespace Core {
 class ARM_Interface;
+class CPUInterruptHandler;
 class ExclusiveMonitor;
 class System;
 } // namespace Core
@@ -21,7 +26,8 @@ namespace Kernel {
 
 class PhysicalCore {
 public:
-    PhysicalCore(Core::System& system, std::size_t id, Core::ExclusiveMonitor& exclusive_monitor);
+    PhysicalCore(Core::System& system, std::size_t id, Kernel::Scheduler& scheduler,
+                 Core::CPUInterruptHandler& interrupt_handler);
     ~PhysicalCore();
 
     PhysicalCore(const PhysicalCore&) = delete;
@@ -30,23 +36,18 @@ public:
     PhysicalCore(PhysicalCore&&) = default;
     PhysicalCore& operator=(PhysicalCore&&) = default;
 
-    /// Execute current jit state
-    void Run();
-    /// Execute a single instruction in current jit.
-    void Step();
-    /// Stop JIT execution/exit
-    void Stop();
+    void Idle();
+    /// Interrupt this physical core.
+    void Interrupt();
 
-    // Shutdown this physical core.
-    void Shutdown();
+    /// Clear this core's interrupt
+    void ClearInterrupt();
 
-    Core::ARM_Interface& ArmInterface() {
-        return *arm_interface;
-    }
+    /// Check if this core is interrupted
+    bool IsInterrupted() const;
 
-    const Core::ARM_Interface& ArmInterface() const {
-        return *arm_interface;
-    }
+    // Shutdown this physical core.
+    void Shutdown();
 
     bool IsMainCore() const {
         return core_index == 0;
@@ -61,21 +62,18 @@ public:
     }
 
     Kernel::Scheduler& Scheduler() {
-        return *scheduler;
+        return scheduler;
     }
 
     const Kernel::Scheduler& Scheduler() const {
-        return *scheduler;
+        return scheduler;
     }
 
-    void SetIs64Bit(bool is_64_bit);
-
 private:
+    Core::CPUInterruptHandler& interrupt_handler;
     std::size_t core_index;
-    std::unique_ptr<Core::ARM_Interface> arm_interface_32;
-    std::unique_ptr<Core::ARM_Interface> arm_interface_64;
-    std::unique_ptr<Kernel::Scheduler> scheduler;
-    Core::ARM_Interface* arm_interface{};
+    Kernel::Scheduler& scheduler;
+    std::unique_ptr<Common::SpinLock> guard;
 };
 
 } // namespace Kernel
diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp
index c4c5199b1..f9d7c024d 100644
--- a/src/core/hle/kernel/process.cpp
+++ b/src/core/hle/kernel/process.cpp
@@ -22,6 +22,7 @@
 #include "core/hle/kernel/resource_limit.h"
 #include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
+#include "core/hle/lock.h"
 #include "core/memory.h"
 #include "core/settings.h"
 
@@ -30,14 +31,15 @@ namespace {
 /**
  * Sets up the primary application thread
  *
+ * @param system The system instance to create the main thread under.
  * @param owner_process The parent process for the main thread
- * @param kernel The kernel instance to create the main thread under.
  * @param priority The priority to give the main thread
  */
-void SetupMainThread(Process& owner_process, KernelCore& kernel, u32 priority, VAddr stack_top) {
+void SetupMainThread(Core::System& system, Process& owner_process, u32 priority, VAddr stack_top) {
     const VAddr entry_point = owner_process.PageTable().GetCodeRegionStart();
-    auto thread_res = Thread::Create(kernel, "main", entry_point, priority, 0,
-                                     owner_process.GetIdealCore(), stack_top, owner_process);
+    ThreadType type = THREADTYPE_USER;
+    auto thread_res = Thread::Create(system, type, "main", entry_point, priority, 0,
+                                     owner_process.GetIdealCore(), stack_top, &owner_process);
 
     std::shared_ptr<Thread> thread = std::move(thread_res).Unwrap();
 
@@ -48,8 +50,12 @@ void SetupMainThread(Process& owner_process, KernelCore& kernel, u32 priority, V
     thread->GetContext32().cpu_registers[1] = thread_handle;
     thread->GetContext64().cpu_registers[1] = thread_handle;
 
+    auto& kernel = system.Kernel();
     // Threads by default are dormant, wake up the main thread so it runs when the scheduler fires
-    thread->ResumeFromWait();
+    {
+        SchedulerLock lock{kernel};
+        thread->SetStatus(ThreadStatus::Ready);
+    }
 }
 } // Anonymous namespace
 
@@ -182,7 +188,6 @@ void Process::RemoveConditionVariableThread(std::shared_ptr<Thread> thread) {
         }
         ++it;
     }
-    UNREACHABLE();
 }
 
 std::vector<std::shared_ptr<Thread>> Process::GetConditionVariableThreads(
@@ -207,6 +212,7 @@ void Process::UnregisterThread(const Thread* thread) {
 }
 
 ResultCode Process::ClearSignalState() {
+    SchedulerLock lock(system.Kernel());
     if (status == ProcessStatus::Exited) {
         LOG_ERROR(Kernel, "called on a terminated process instance.");
         return ERR_INVALID_STATE;
@@ -294,7 +300,7 @@ void Process::Run(s32 main_thread_priority, u64 stack_size) {
 
     ChangeStatus(ProcessStatus::Running);
 
-    SetupMainThread(*this, kernel, main_thread_priority, main_thread_stack_top);
+    SetupMainThread(system, *this, main_thread_priority, main_thread_stack_top);
     resource_limit->Reserve(ResourceType::Threads, 1);
     resource_limit->Reserve(ResourceType::PhysicalMemory, main_thread_stack_size);
 }
@@ -340,6 +346,7 @@ static auto FindTLSPageWithAvailableSlots(std::vector<TLSPage>& tls_pages) {
 }
 
 VAddr Process::CreateTLSRegion() {
+    SchedulerLock lock(system.Kernel());
     if (auto tls_page_iter{FindTLSPageWithAvailableSlots(tls_pages)};
         tls_page_iter != tls_pages.cend()) {
         return *tls_page_iter->ReserveSlot();
@@ -370,6 +377,7 @@ VAddr Process::CreateTLSRegion() {
 }
 
 void Process::FreeTLSRegion(VAddr tls_address) {
+    SchedulerLock lock(system.Kernel());
     const VAddr aligned_address = Common::AlignDown(tls_address, Core::Memory::PAGE_SIZE);
     auto iter =
         std::find_if(tls_pages.begin(), tls_pages.end(), [aligned_address](const auto& page) {
@@ -384,6 +392,7 @@ void Process::FreeTLSRegion(VAddr tls_address) {
 }
 
 void Process::LoadModule(CodeSet code_set, VAddr base_addr) {
+    std::lock_guard lock{HLE::g_hle_lock};
     const auto ReprotectSegment = [&](const CodeSet::Segment& segment,
                                       Memory::MemoryPermission permission) {
         page_table->SetCodeMemoryPermission(segment.addr + base_addr, segment.size, permission);
diff --git a/src/core/hle/kernel/readable_event.cpp b/src/core/hle/kernel/readable_event.cpp
index ef5e19e63..6e286419e 100644
--- a/src/core/hle/kernel/readable_event.cpp
+++ b/src/core/hle/kernel/readable_event.cpp
@@ -6,8 +6,10 @@
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "core/hle/kernel/errors.h"
+#include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/readable_event.h"
+#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
 
 namespace Kernel {
@@ -37,6 +39,7 @@ void ReadableEvent::Clear() {
 }
 
 ResultCode ReadableEvent::Reset() {
+    SchedulerLock lock(kernel);
     if (!is_signaled) {
         LOG_TRACE(Kernel, "Handle is not signaled! object_id={}, object_type={}, object_name={}",
                   GetObjectId(), GetTypeName(), GetName());
diff --git a/src/core/hle/kernel/scheduler.cpp b/src/core/hle/kernel/scheduler.cpp
index 1140c72a3..2b12c0dbf 100644
--- a/src/core/hle/kernel/scheduler.cpp
+++ b/src/core/hle/kernel/scheduler.cpp
@@ -11,11 +11,15 @@
 #include <utility>
 
 #include "common/assert.h"
+#include "common/bit_util.h"
+#include "common/fiber.h"
 #include "common/logging/log.h"
 #include "core/arm/arm_interface.h"
 #include "core/core.h"
 #include "core/core_timing.h"
+#include "core/cpu_manager.h"
 #include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/physical_core.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/time_manager.h"
@@ -27,103 +31,151 @@ GlobalScheduler::GlobalScheduler(KernelCore& kernel) : kernel{kernel} {}
 GlobalScheduler::~GlobalScheduler() = default;
 
 void GlobalScheduler::AddThread(std::shared_ptr<Thread> thread) {
+    global_list_guard.lock();
     thread_list.push_back(std::move(thread));
+    global_list_guard.unlock();
 }
 
 void GlobalScheduler::RemoveThread(std::shared_ptr<Thread> thread) {
+    global_list_guard.lock();
     thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread),
                       thread_list.end());
+    global_list_guard.unlock();
 }
 
-void GlobalScheduler::UnloadThread(std::size_t core) {
-    Scheduler& sched = kernel.Scheduler(core);
-    sched.UnloadThread();
-}
-
-void GlobalScheduler::SelectThread(std::size_t core) {
+u32 GlobalScheduler::SelectThreads() {
+    ASSERT(is_locked);
     const auto update_thread = [](Thread* thread, Scheduler& sched) {
-        if (thread != sched.selected_thread.get()) {
+        sched.guard.lock();
+        if (thread != sched.selected_thread_set.get()) {
             if (thread == nullptr) {
                 ++sched.idle_selection_count;
             }
-            sched.selected_thread = SharedFrom(thread);
+            sched.selected_thread_set = SharedFrom(thread);
         }
-        sched.is_context_switch_pending = sched.selected_thread != sched.current_thread;
+        const bool reschedule_pending =
+            sched.is_context_switch_pending || (sched.selected_thread_set != sched.current_thread);
+        sched.is_context_switch_pending = reschedule_pending;
         std::atomic_thread_fence(std::memory_order_seq_cst);
+        sched.guard.unlock();
+        return reschedule_pending;
     };
-    Scheduler& sched = kernel.Scheduler(core);
-    Thread* current_thread = nullptr;
-    // Step 1: Get top thread in schedule queue.
-    current_thread = scheduled_queue[core].empty() ? nullptr : scheduled_queue[core].front();
-    if (current_thread) {
-        update_thread(current_thread, sched);
-        return;
+    if (!is_reselection_pending.load()) {
+        return 0;
     }
-    // Step 2: Try selecting a suggested thread.
-    Thread* winner = nullptr;
-    std::set<s32> sug_cores;
-    for (auto thread : suggested_queue[core]) {
-        s32 this_core = thread->GetProcessorID();
-        Thread* thread_on_core = nullptr;
-        if (this_core >= 0) {
-            thread_on_core = scheduled_queue[this_core].front();
-        }
-        if (this_core < 0 || thread != thread_on_core) {
-            winner = thread;
-            break;
+    std::array<Thread*, Core::Hardware::NUM_CPU_CORES> top_threads{};
+
+    u32 idle_cores{};
+
+    // Step 1: Get top thread in schedule queue.
+    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+        Thread* top_thread =
+            scheduled_queue[core].empty() ? nullptr : scheduled_queue[core].front();
+        if (top_thread != nullptr) {
+            // TODO(Blinkhawk): Implement Thread Pinning
+        } else {
+            idle_cores |= (1ul << core);
         }
-        sug_cores.insert(this_core);
+        top_threads[core] = top_thread;
     }
-    // if we got a suggested thread, select it, else do a second pass.
-    if (winner && winner->GetPriority() > 2) {
-        if (winner->IsRunning()) {
-            UnloadThread(static_cast<u32>(winner->GetProcessorID()));
+
+    while (idle_cores != 0) {
+        u32 core_id = Common::CountTrailingZeroes32(idle_cores);
+
+        if (!suggested_queue[core_id].empty()) {
+            std::array<s32, Core::Hardware::NUM_CPU_CORES> migration_candidates{};
+            std::size_t num_candidates = 0;
+            auto iter = suggested_queue[core_id].begin();
+            Thread* suggested = nullptr;
+            // Step 2: Try selecting a suggested thread.
+            while (iter != suggested_queue[core_id].end()) {
+                suggested = *iter;
+                iter++;
+                s32 suggested_core_id = suggested->GetProcessorID();
+                Thread* top_thread =
+                    suggested_core_id >= 0 ? top_threads[suggested_core_id] : nullptr;
+                if (top_thread != suggested) {
+                    if (top_thread != nullptr &&
+                        top_thread->GetPriority() < THREADPRIO_MAX_CORE_MIGRATION) {
+                        suggested = nullptr;
+                        break;
+                        // There's a too high thread to do core migration, cancel
+                    }
+                    TransferToCore(suggested->GetPriority(), static_cast<s32>(core_id), suggested);
+                    break;
+                }
+                suggested = nullptr;
+                migration_candidates[num_candidates++] = suggested_core_id;
+            }
+            // Step 3: Select a suggested thread from another core
+            if (suggested == nullptr) {
+                for (std::size_t i = 0; i < num_candidates; i++) {
+                    s32 candidate_core = migration_candidates[i];
+                    suggested = top_threads[candidate_core];
+                    auto it = scheduled_queue[candidate_core].begin();
+                    it++;
+                    Thread* next = it != scheduled_queue[candidate_core].end() ? *it : nullptr;
+                    if (next != nullptr) {
+                        TransferToCore(suggested->GetPriority(), static_cast<s32>(core_id),
+                                       suggested);
+                        top_threads[candidate_core] = next;
+                        break;
+                    } else {
+                        suggested = nullptr;
+                    }
+                }
+            }
+            top_threads[core_id] = suggested;
         }
-        TransferToCore(winner->GetPriority(), static_cast<s32>(core), winner);
-        update_thread(winner, sched);
-        return;
+
+        idle_cores &= ~(1ul << core_id);
     }
-    // Step 3: Select a suggested thread from another core
-    for (auto& src_core : sug_cores) {
-        auto it = scheduled_queue[src_core].begin();
-        it++;
-        if (it != scheduled_queue[src_core].end()) {
-            Thread* thread_on_core = scheduled_queue[src_core].front();
-            Thread* to_change = *it;
-            if (thread_on_core->IsRunning() || to_change->IsRunning()) {
-                UnloadThread(static_cast<u32>(src_core));
-            }
-            TransferToCore(thread_on_core->GetPriority(), static_cast<s32>(core), thread_on_core);
-            current_thread = thread_on_core;
-            break;
+    u32 cores_needing_context_switch{};
+    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+        Scheduler& sched = kernel.Scheduler(core);
+        ASSERT(top_threads[core] == nullptr || top_threads[core]->GetProcessorID() == core);
+        if (update_thread(top_threads[core], sched)) {
+            cores_needing_context_switch |= (1ul << core);
         }
     }
-    update_thread(current_thread, sched);
+    return cores_needing_context_switch;
 }
 
 bool GlobalScheduler::YieldThread(Thread* yielding_thread) {
+    ASSERT(is_locked);
     // Note: caller should use critical section, etc.
+    if (!yielding_thread->IsRunnable()) {
+        // Normally this case shouldn't happen except for SetThreadActivity.
+        is_reselection_pending.store(true, std::memory_order_release);
+        return false;
+    }
     const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID());
     const u32 priority = yielding_thread->GetPriority();
 
     // Yield the thread
-    const Thread* const winner = scheduled_queue[core_id].front(priority);
-    ASSERT_MSG(yielding_thread == winner, "Thread yielding without being in front");
-    scheduled_queue[core_id].yield(priority);
+    Reschedule(priority, core_id, yielding_thread);
+    const Thread* const winner = scheduled_queue[core_id].front();
+    if (kernel.GetCurrentHostThreadID() != core_id) {
+        is_reselection_pending.store(true, std::memory_order_release);
+    }
 
     return AskForReselectionOrMarkRedundant(yielding_thread, winner);
 }
 
 bool GlobalScheduler::YieldThreadAndBalanceLoad(Thread* yielding_thread) {
+    ASSERT(is_locked);
     // Note: caller should check if !thread.IsSchedulerOperationRedundant and use critical section,
     // etc.
+    if (!yielding_thread->IsRunnable()) {
+        // Normally this case shouldn't happen except for SetThreadActivity.
+        is_reselection_pending.store(true, std::memory_order_release);
+        return false;
+    }
     const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID());
     const u32 priority = yielding_thread->GetPriority();
 
     // Yield the thread
-    ASSERT_MSG(yielding_thread == scheduled_queue[core_id].front(priority),
-               "Thread yielding without being in front");
-    scheduled_queue[core_id].yield(priority);
+    Reschedule(priority, core_id, yielding_thread);
 
     std::array<Thread*, Core::Hardware::NUM_CPU_CORES> current_threads;
     for (std::size_t i = 0; i < current_threads.size(); i++) {
@@ -153,21 +205,28 @@ bool GlobalScheduler::YieldThreadAndBalanceLoad(Thread* yielding_thread) {
 
     if (winner != nullptr) {
         if (winner != yielding_thread) {
-            if (winner->IsRunning()) {
-                UnloadThread(static_cast<u32>(winner->GetProcessorID()));
-            }
             TransferToCore(winner->GetPriority(), s32(core_id), winner);
         }
     } else {
         winner = next_thread;
     }
 
+    if (kernel.GetCurrentHostThreadID() != core_id) {
+        is_reselection_pending.store(true, std::memory_order_release);
+    }
+
     return AskForReselectionOrMarkRedundant(yielding_thread, winner);
 }
 
 bool GlobalScheduler::YieldThreadAndWaitForLoadBalancing(Thread* yielding_thread) {
+    ASSERT(is_locked);
     // Note: caller should check if !thread.IsSchedulerOperationRedundant and use critical section,
     // etc.
+    if (!yielding_thread->IsRunnable()) {
+        // Normally this case shouldn't happen except for SetThreadActivity.
+        is_reselection_pending.store(true, std::memory_order_release);
+        return false;
+    }
     Thread* winner = nullptr;
     const u32 core_id = static_cast<u32>(yielding_thread->GetProcessorID());
 
@@ -195,25 +254,31 @@ bool GlobalScheduler::YieldThreadAndWaitForLoadBalancing(Thread* yielding_thread
         }
         if (winner != nullptr) {
             if (winner != yielding_thread) {
-                if (winner->IsRunning()) {
-                    UnloadThread(static_cast<u32>(winner->GetProcessorID()));
-                }
                 TransferToCore(winner->GetPriority(), static_cast<s32>(core_id), winner);
             }
         } else {
             winner = yielding_thread;
         }
+    } else {
+        winner = scheduled_queue[core_id].front();
+    }
+
+    if (kernel.GetCurrentHostThreadID() != core_id) {
+        is_reselection_pending.store(true, std::memory_order_release);
     }
 
     return AskForReselectionOrMarkRedundant(yielding_thread, winner);
 }
 
 void GlobalScheduler::PreemptThreads() {
+    ASSERT(is_locked);
     for (std::size_t core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) {
         const u32 priority = preemption_priorities[core_id];
 
         if (scheduled_queue[core_id].size(priority) > 0) {
-            scheduled_queue[core_id].front(priority)->IncrementYieldCount();
+            if (scheduled_queue[core_id].size(priority) > 1) {
+                scheduled_queue[core_id].front(priority)->IncrementYieldCount();
+            }
             scheduled_queue[core_id].yield(priority);
             if (scheduled_queue[core_id].size(priority) > 1) {
                 scheduled_queue[core_id].front(priority)->IncrementYieldCount();
@@ -247,9 +312,6 @@ void GlobalScheduler::PreemptThreads() {
         }
 
         if (winner != nullptr) {
-            if (winner->IsRunning()) {
-                UnloadThread(static_cast<u32>(winner->GetProcessorID()));
-            }
             TransferToCore(winner->GetPriority(), s32(core_id), winner);
             current_thread =
                 winner->GetPriority() <= current_thread->GetPriority() ? winner : current_thread;
@@ -280,9 +342,6 @@ void GlobalScheduler::PreemptThreads() {
             }
 
             if (winner != nullptr) {
-                if (winner->IsRunning()) {
-                    UnloadThread(static_cast<u32>(winner->GetProcessorID()));
-                }
                 TransferToCore(winner->GetPriority(), s32(core_id), winner);
                 current_thread = winner;
             }
@@ -292,34 +351,65 @@ void GlobalScheduler::PreemptThreads() {
     }
 }
 
+void GlobalScheduler::EnableInterruptAndSchedule(u32 cores_pending_reschedule,
+                                                 Core::EmuThreadHandle global_thread) {
+    u32 current_core = global_thread.host_handle;
+    bool must_context_switch = global_thread.guest_handle != InvalidHandle &&
+                               (current_core < Core::Hardware::NUM_CPU_CORES);
+    while (cores_pending_reschedule != 0) {
+        u32 core = Common::CountTrailingZeroes32(cores_pending_reschedule);
+        ASSERT(core < Core::Hardware::NUM_CPU_CORES);
+        if (!must_context_switch || core != current_core) {
+            auto& phys_core = kernel.PhysicalCore(core);
+            phys_core.Interrupt();
+        } else {
+            must_context_switch = true;
+        }
+        cores_pending_reschedule &= ~(1ul << core);
+    }
+    if (must_context_switch) {
+        auto& core_scheduler = kernel.CurrentScheduler();
+        kernel.ExitSVCProfile();
+        core_scheduler.TryDoContextSwitch();
+        kernel.EnterSVCProfile();
+    }
+}
+
 void GlobalScheduler::Suggest(u32 priority, std::size_t core, Thread* thread) {
+    ASSERT(is_locked);
     suggested_queue[core].add(thread, priority);
 }
 
 void GlobalScheduler::Unsuggest(u32 priority, std::size_t core, Thread* thread) {
+    ASSERT(is_locked);
     suggested_queue[core].remove(thread, priority);
 }
 
 void GlobalScheduler::Schedule(u32 priority, std::size_t core, Thread* thread) {
+    ASSERT(is_locked);
     ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core.");
     scheduled_queue[core].add(thread, priority);
 }
 
 void GlobalScheduler::SchedulePrepend(u32 priority, std::size_t core, Thread* thread) {
+    ASSERT(is_locked);
     ASSERT_MSG(thread->GetProcessorID() == s32(core), "Thread must be assigned to this core.");
     scheduled_queue[core].add(thread, priority, false);
 }
 
 void GlobalScheduler::Reschedule(u32 priority, std::size_t core, Thread* thread) {
+    ASSERT(is_locked);
     scheduled_queue[core].remove(thread, priority);
     scheduled_queue[core].add(thread, priority);
 }
 
 void GlobalScheduler::Unschedule(u32 priority, std::size_t core, Thread* thread) {
+    ASSERT(is_locked);
     scheduled_queue[core].remove(thread, priority);
 }
 
 void GlobalScheduler::TransferToCore(u32 priority, s32 destination_core, Thread* thread) {
+    ASSERT(is_locked);
     const bool schedulable = thread->GetPriority() < THREADPRIO_COUNT;
     const s32 source_core = thread->GetProcessorID();
     if (source_core == destination_core || !schedulable) {
@@ -349,6 +439,108 @@ bool GlobalScheduler::AskForReselectionOrMarkRedundant(Thread* current_thread,
     }
 }
 
+void GlobalScheduler::AdjustSchedulingOnStatus(Thread* thread, u32 old_flags) {
+    if (old_flags == thread->scheduling_state) {
+        return;
+    }
+    ASSERT(is_locked);
+
+    if (old_flags == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+        // In this case the thread was running, now it's pausing/exitting
+        if (thread->processor_id >= 0) {
+            Unschedule(thread->current_priority, static_cast<u32>(thread->processor_id), thread);
+        }
+
+        for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+            if (core != static_cast<u32>(thread->processor_id) &&
+                ((thread->affinity_mask >> core) & 1) != 0) {
+                Unsuggest(thread->current_priority, core, thread);
+            }
+        }
+    } else if (thread->scheduling_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+        // The thread is now set to running from being stopped
+        if (thread->processor_id >= 0) {
+            Schedule(thread->current_priority, static_cast<u32>(thread->processor_id), thread);
+        }
+
+        for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+            if (core != static_cast<u32>(thread->processor_id) &&
+                ((thread->affinity_mask >> core) & 1) != 0) {
+                Suggest(thread->current_priority, core, thread);
+            }
+        }
+    }
+
+    SetReselectionPending();
+}
+
+void GlobalScheduler::AdjustSchedulingOnPriority(Thread* thread, u32 old_priority) {
+    if (thread->scheduling_state != static_cast<u32>(ThreadSchedStatus::Runnable)) {
+        return;
+    }
+    ASSERT(is_locked);
+    if (thread->processor_id >= 0) {
+        Unschedule(old_priority, static_cast<u32>(thread->processor_id), thread);
+    }
+
+    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+        if (core != static_cast<u32>(thread->processor_id) &&
+            ((thread->affinity_mask >> core) & 1) != 0) {
+            Unsuggest(old_priority, core, thread);
+        }
+    }
+
+    if (thread->processor_id >= 0) {
+        if (thread == kernel.CurrentScheduler().GetCurrentThread()) {
+            SchedulePrepend(thread->current_priority, static_cast<u32>(thread->processor_id),
+                            thread);
+        } else {
+            Schedule(thread->current_priority, static_cast<u32>(thread->processor_id), thread);
+        }
+    }
+
+    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+        if (core != static_cast<u32>(thread->processor_id) &&
+            ((thread->affinity_mask >> core) & 1) != 0) {
+            Suggest(thread->current_priority, core, thread);
+        }
+    }
+    thread->IncrementYieldCount();
+    SetReselectionPending();
+}
+
+void GlobalScheduler::AdjustSchedulingOnAffinity(Thread* thread, u64 old_affinity_mask,
+                                                 s32 old_core) {
+    if (thread->scheduling_state != static_cast<u32>(ThreadSchedStatus::Runnable) ||
+        thread->current_priority >= THREADPRIO_COUNT) {
+        return;
+    }
+    ASSERT(is_locked);
+
+    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+        if (((old_affinity_mask >> core) & 1) != 0) {
+            if (core == static_cast<u32>(old_core)) {
+                Unschedule(thread->current_priority, core, thread);
+            } else {
+                Unsuggest(thread->current_priority, core, thread);
+            }
+        }
+    }
+
+    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
+        if (((thread->affinity_mask >> core) & 1) != 0) {
+            if (core == static_cast<u32>(thread->processor_id)) {
+                Schedule(thread->current_priority, core, thread);
+            } else {
+                Suggest(thread->current_priority, core, thread);
+            }
+        }
+    }
+
+    thread->IncrementYieldCount();
+    SetReselectionPending();
+}
+
 void GlobalScheduler::Shutdown() {
     for (std::size_t core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
         scheduled_queue[core].clear();
@@ -359,10 +551,12 @@ void GlobalScheduler::Shutdown() {
 
 void GlobalScheduler::Lock() {
     Core::EmuThreadHandle current_thread = kernel.GetCurrentEmuThreadID();
+    ASSERT(!current_thread.IsInvalid());
     if (current_thread == current_owner) {
         ++scope_lock;
     } else {
         inner_lock.lock();
+        is_locked = true;
         current_owner = current_thread;
         ASSERT(current_owner != Core::EmuThreadHandle::InvalidHandle());
         scope_lock = 1;
@@ -374,17 +568,18 @@ void GlobalScheduler::Unlock() {
         ASSERT(scope_lock > 0);
         return;
     }
-    for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
-        SelectThread(i);
-    }
+    u32 cores_pending_reschedule = SelectThreads();
+    Core::EmuThreadHandle leaving_thread = current_owner;
     current_owner = Core::EmuThreadHandle::InvalidHandle();
     scope_lock = 1;
+    is_locked = false;
     inner_lock.unlock();
-    // TODO(Blinkhawk): Setup the interrupts and change context on current core.
+    EnableInterruptAndSchedule(cores_pending_reschedule, leaving_thread);
 }
 
-Scheduler::Scheduler(Core::System& system, std::size_t core_id)
-    : system{system}, core_id{core_id} {}
+Scheduler::Scheduler(Core::System& system, std::size_t core_id) : system(system), core_id(core_id) {
+    switch_fiber = std::make_shared<Common::Fiber>(std::function<void(void*)>(OnSwitch), this);
+}
 
 Scheduler::~Scheduler() = default;
 
@@ -393,56 +588,128 @@ bool Scheduler::HaveReadyThreads() const {
 }
 
 Thread* Scheduler::GetCurrentThread() const {
-    return current_thread.get();
+    if (current_thread) {
+        return current_thread.get();
+    }
+    return idle_thread.get();
 }
 
 Thread* Scheduler::GetSelectedThread() const {
     return selected_thread.get();
 }
 
-void Scheduler::SelectThreads() {
-    system.GlobalScheduler().SelectThread(core_id);
-}
-
 u64 Scheduler::GetLastContextSwitchTicks() const {
     return last_context_switch_time;
 }
 
 void Scheduler::TryDoContextSwitch() {
+    auto& phys_core = system.Kernel().CurrentPhysicalCore();
+    if (phys_core.IsInterrupted()) {
+        phys_core.ClearInterrupt();
+    }
+    guard.lock();
     if (is_context_switch_pending) {
         SwitchContext();
+    } else {
+        guard.unlock();
     }
 }
 
-void Scheduler::UnloadThread() {
-    Thread* const previous_thread = GetCurrentThread();
-    Process* const previous_process = system.Kernel().CurrentProcess();
+void Scheduler::OnThreadStart() {
+    SwitchContextStep2();
+}
 
-    UpdateLastContextSwitchTime(previous_thread, previous_process);
+void Scheduler::Unload() {
+    Thread* thread = current_thread.get();
+    if (thread) {
+        thread->SetContinuousOnSVC(false);
+        thread->last_running_ticks = system.CoreTiming().GetCPUTicks();
+        thread->SetIsRunning(false);
+        if (!thread->IsHLEThread() && !thread->HasExited()) {
+            Core::ARM_Interface& cpu_core = thread->ArmInterface();
+            cpu_core.SaveContext(thread->GetContext32());
+            cpu_core.SaveContext(thread->GetContext64());
+            // Save the TPIDR_EL0 system register in case it was modified.
+            thread->SetTPIDR_EL0(cpu_core.GetTPIDR_EL0());
+            cpu_core.ClearExclusiveState();
+        }
+        thread->context_guard.unlock();
+    }
+}
 
-    // Save context for previous thread
-    if (previous_thread) {
-        system.ArmInterface(core_id).SaveContext(previous_thread->GetContext32());
-        system.ArmInterface(core_id).SaveContext(previous_thread->GetContext64());
-        // Save the TPIDR_EL0 system register in case it was modified.
-        previous_thread->SetTPIDR_EL0(system.ArmInterface(core_id).GetTPIDR_EL0());
+void Scheduler::Reload() {
+    Thread* thread = current_thread.get();
+    if (thread) {
+        ASSERT_MSG(thread->GetSchedulingStatus() == ThreadSchedStatus::Runnable,
+                   "Thread must be runnable.");
 
-        if (previous_thread->GetStatus() == ThreadStatus::Running) {
-            // This is only the case when a reschedule is triggered without the current thread
-            // yielding execution (i.e. an event triggered, system core time-sliced, etc)
-            previous_thread->SetStatus(ThreadStatus::Ready);
+        // Cancel any outstanding wakeup events for this thread
+        thread->SetIsRunning(true);
+        thread->SetWasRunning(false);
+        thread->last_running_ticks = system.CoreTiming().GetCPUTicks();
+
+        auto* const thread_owner_process = thread->GetOwnerProcess();
+        if (thread_owner_process != nullptr) {
+            system.Kernel().MakeCurrentProcess(thread_owner_process);
+        }
+        if (!thread->IsHLEThread()) {
+            Core::ARM_Interface& cpu_core = thread->ArmInterface();
+            cpu_core.LoadContext(thread->GetContext32());
+            cpu_core.LoadContext(thread->GetContext64());
+            cpu_core.SetTlsAddress(thread->GetTLSAddress());
+            cpu_core.SetTPIDR_EL0(thread->GetTPIDR_EL0());
+            cpu_core.ChangeProcessorID(this->core_id);
+            cpu_core.ClearExclusiveState();
         }
-        previous_thread->SetIsRunning(false);
     }
-    current_thread = nullptr;
+}
+
+void Scheduler::SwitchContextStep2() {
+    Thread* previous_thread = current_thread_prev.get();
+    Thread* new_thread = selected_thread.get();
+
+    // Load context of new thread
+    Process* const previous_process =
+        previous_thread != nullptr ? previous_thread->GetOwnerProcess() : nullptr;
+
+    if (new_thread) {
+        ASSERT_MSG(new_thread->GetSchedulingStatus() == ThreadSchedStatus::Runnable,
+                   "Thread must be runnable.");
+
+        // Cancel any outstanding wakeup events for this thread
+        new_thread->SetIsRunning(true);
+        new_thread->last_running_ticks = system.CoreTiming().GetCPUTicks();
+        new_thread->SetWasRunning(false);
+
+        auto* const thread_owner_process = current_thread->GetOwnerProcess();
+        if (thread_owner_process != nullptr) {
+            system.Kernel().MakeCurrentProcess(thread_owner_process);
+        }
+        if (!new_thread->IsHLEThread()) {
+            Core::ARM_Interface& cpu_core = new_thread->ArmInterface();
+            cpu_core.LoadContext(new_thread->GetContext32());
+            cpu_core.LoadContext(new_thread->GetContext64());
+            cpu_core.SetTlsAddress(new_thread->GetTLSAddress());
+            cpu_core.SetTPIDR_EL0(new_thread->GetTPIDR_EL0());
+            cpu_core.ChangeProcessorID(this->core_id);
+            cpu_core.ClearExclusiveState();
+        }
+    }
+
+    TryDoContextSwitch();
 }
 
 void Scheduler::SwitchContext() {
-    Thread* const previous_thread = GetCurrentThread();
-    Thread* const new_thread = GetSelectedThread();
+    current_thread_prev = current_thread;
+    selected_thread = selected_thread_set;
+    Thread* previous_thread = current_thread_prev.get();
+    Thread* new_thread = selected_thread.get();
+    current_thread = selected_thread;
 
     is_context_switch_pending = false;
+
     if (new_thread == previous_thread) {
+        guard.unlock();
         return;
     }
 
@@ -452,51 +719,75 @@ void Scheduler::SwitchContext() {
 
     // Save context for previous thread
     if (previous_thread) {
-        system.ArmInterface(core_id).SaveContext(previous_thread->GetContext32());
-        system.ArmInterface(core_id).SaveContext(previous_thread->GetContext64());
-        // Save the TPIDR_EL0 system register in case it was modified.
-        previous_thread->SetTPIDR_EL0(system.ArmInterface(core_id).GetTPIDR_EL0());
-
-        if (previous_thread->GetStatus() == ThreadStatus::Running) {
-            // This is only the case when a reschedule is triggered without the current thread
-            // yielding execution (i.e. an event triggered, system core time-sliced, etc)
-            previous_thread->SetStatus(ThreadStatus::Ready);
+        if (new_thread != nullptr && new_thread->IsSuspendThread()) {
+            previous_thread->SetWasRunning(true);
         }
+        previous_thread->SetContinuousOnSVC(false);
+        previous_thread->last_running_ticks = system.CoreTiming().GetCPUTicks();
         previous_thread->SetIsRunning(false);
-    }
-
-    // Load context of new thread
-    if (new_thread) {
-        ASSERT_MSG(new_thread->GetProcessorID() == s32(this->core_id),
-                   "Thread must be assigned to this core.");
-        ASSERT_MSG(new_thread->GetStatus() == ThreadStatus::Ready,
-                   "Thread must be ready to become running.");
-
-        // Cancel any outstanding wakeup events for this thread
-        new_thread->CancelWakeupTimer();
-        current_thread = SharedFrom(new_thread);
-        new_thread->SetStatus(ThreadStatus::Running);
-        new_thread->SetIsRunning(true);
-
-        auto* const thread_owner_process = current_thread->GetOwnerProcess();
-        if (previous_process != thread_owner_process) {
-            system.Kernel().MakeCurrentProcess(thread_owner_process);
+        if (!previous_thread->IsHLEThread() && !previous_thread->HasExited()) {
+            Core::ARM_Interface& cpu_core = previous_thread->ArmInterface();
+            cpu_core.SaveContext(previous_thread->GetContext32());
+            cpu_core.SaveContext(previous_thread->GetContext64());
+            // Save the TPIDR_EL0 system register in case it was modified.
+            previous_thread->SetTPIDR_EL0(cpu_core.GetTPIDR_EL0());
+            cpu_core.ClearExclusiveState();
         }
+        previous_thread->context_guard.unlock();
+    }
 
-        system.ArmInterface(core_id).LoadContext(new_thread->GetContext32());
-        system.ArmInterface(core_id).LoadContext(new_thread->GetContext64());
-        system.ArmInterface(core_id).SetTlsAddress(new_thread->GetTLSAddress());
-        system.ArmInterface(core_id).SetTPIDR_EL0(new_thread->GetTPIDR_EL0());
+    std::shared_ptr<Common::Fiber>* old_context;
+    if (previous_thread != nullptr) {
+        old_context = &previous_thread->GetHostContext();
     } else {
-        current_thread = nullptr;
-        // Note: We do not reset the current process and current page table when idling because
-        // technically we haven't changed processes, our threads are just paused.
+        old_context = &idle_thread->GetHostContext();
+    }
+    guard.unlock();
+
+    Common::Fiber::YieldTo(*old_context, switch_fiber);
+    /// When a thread wakes up, the scheduler may have changed to other in another core.
+    auto& next_scheduler = system.Kernel().CurrentScheduler();
+    next_scheduler.SwitchContextStep2();
+}
+
+void Scheduler::OnSwitch(void* this_scheduler) {
+    Scheduler* sched = static_cast<Scheduler*>(this_scheduler);
+    sched->SwitchToCurrent();
+}
+
+void Scheduler::SwitchToCurrent() {
+    while (true) {
+        guard.lock();
+        selected_thread = selected_thread_set;
+        current_thread = selected_thread;
+        is_context_switch_pending = false;
+        guard.unlock();
+        while (!is_context_switch_pending) {
+            if (current_thread != nullptr && !current_thread->IsHLEThread()) {
+                current_thread->context_guard.lock();
+                if (!current_thread->IsRunnable()) {
+                    current_thread->context_guard.unlock();
+                    break;
+                }
+                if (current_thread->GetProcessorID() != core_id) {
+                    current_thread->context_guard.unlock();
+                    break;
+                }
+            }
+            std::shared_ptr<Common::Fiber>* next_context;
+            if (current_thread != nullptr) {
+                next_context = &current_thread->GetHostContext();
+            } else {
+                next_context = &idle_thread->GetHostContext();
+            }
+            Common::Fiber::YieldTo(switch_fiber, *next_context);
+        }
     }
 }
 
 void Scheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) {
     const u64 prev_switch_ticks = last_context_switch_time;
-    const u64 most_recent_switch_ticks = system.CoreTiming().GetTicks();
+    const u64 most_recent_switch_ticks = system.CoreTiming().GetCPUTicks();
     const u64 update_ticks = most_recent_switch_ticks - prev_switch_ticks;
 
     if (thread != nullptr) {
@@ -510,6 +801,16 @@ void Scheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) {
     last_context_switch_time = most_recent_switch_ticks;
 }
 
+void Scheduler::Initialize() {
+    std::string name = "Idle Thread Id:" + std::to_string(core_id);
+    std::function<void(void*)> init_func = system.GetCpuManager().GetIdleThreadStartFunc();
+    void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater();
+    ThreadType type = static_cast<ThreadType>(THREADTYPE_KERNEL | THREADTYPE_HLE | THREADTYPE_IDLE);
+    auto thread_res = Thread::Create(system, type, name, 0, 64, 0, static_cast<u32>(core_id), 0,
+                                     nullptr, std::move(init_func), init_func_parameter);
+    idle_thread = std::move(thread_res).Unwrap();
+}
+
 void Scheduler::Shutdown() {
     current_thread = nullptr;
     selected_thread = nullptr;
@@ -538,4 +839,13 @@ SchedulerLockAndSleep::~SchedulerLockAndSleep() {
     time_manager.ScheduleTimeEvent(event_handle, time_task, nanoseconds);
 }
 
+void SchedulerLockAndSleep::Release() {
+    if (sleep_cancelled) {
+        return;
+    }
+    auto& time_manager = kernel.TimeManager();
+    time_manager.ScheduleTimeEvent(event_handle, time_task, nanoseconds);
+    sleep_cancelled = true;
+}
+
 } // namespace Kernel
diff --git a/src/core/hle/kernel/scheduler.h b/src/core/hle/kernel/scheduler.h
index 07df33f9c..b3b4b5169 100644
--- a/src/core/hle/kernel/scheduler.h
+++ b/src/core/hle/kernel/scheduler.h
@@ -11,9 +11,14 @@
 
 #include "common/common_types.h"
 #include "common/multi_level_queue.h"
+#include "common/spin_lock.h"
 #include "core/hardware_properties.h"
 #include "core/hle/kernel/thread.h"
 
+namespace Common {
+class Fiber;
+}
+
 namespace Core {
 class ARM_Interface;
 class System;
@@ -41,41 +46,17 @@ public:
         return thread_list;
     }
 
-    /**
-     * Add a thread to the suggested queue of a cpu core. Suggested threads may be
-     * picked if no thread is scheduled to run on the core.
-     */
-    void Suggest(u32 priority, std::size_t core, Thread* thread);
-
-    /**
-     * Remove a thread to the suggested queue of a cpu core. Suggested threads may be
-     * picked if no thread is scheduled to run on the core.
-     */
-    void Unsuggest(u32 priority, std::size_t core, Thread* thread);
-
-    /**
-     * Add a thread to the scheduling queue of a cpu core. The thread is added at the
-     * back the queue in its priority level.
-     */
-    void Schedule(u32 priority, std::size_t core, Thread* thread);
-
-    /**
-     * Add a thread to the scheduling queue of a cpu core. The thread is added at the
-     * front the queue in its priority level.
-     */
-    void SchedulePrepend(u32 priority, std::size_t core, Thread* thread);
+    /// Notify the scheduler a thread's status has changed.
+    void AdjustSchedulingOnStatus(Thread* thread, u32 old_flags);
 
-    /// Reschedule an already scheduled thread based on a new priority
-    void Reschedule(u32 priority, std::size_t core, Thread* thread);
-
-    /// Unschedules a thread.
-    void Unschedule(u32 priority, std::size_t core, Thread* thread);
+    /// Notify the scheduler a thread's priority has changed.
+    void AdjustSchedulingOnPriority(Thread* thread, u32 old_priority);
 
-    /// Selects a core and forces it to unload its current thread's context
-    void UnloadThread(std::size_t core);
+    /// Notify the scheduler a thread's core and/or affinity mask has changed.
+    void AdjustSchedulingOnAffinity(Thread* thread, u64 old_affinity_mask, s32 old_core);
 
     /**
-     * Takes care of selecting the new scheduled thread in three steps:
+     * Takes care of selecting the new scheduled threads in three steps:
      *
      * 1. First a thread is selected from the top of the priority queue. If no thread
      *    is obtained then we move to step two, else we are done.
@@ -85,8 +66,10 @@ public:
      *
      * 3. Third is no suggested thread is found, we do a second pass and pick a running
      *    thread in another core and swap it with its current thread.
+     *
+     * returns the cores needing scheduling.
      */
-    void SelectThread(std::size_t core);
+    u32 SelectThreads();
 
     bool HaveReadyThreads(std::size_t core_id) const {
         return !scheduled_queue[core_id].empty();
@@ -149,6 +132,40 @@ private:
     /// Unlocks the scheduler, reselects threads, interrupts cores for rescheduling
     /// and reschedules current core if needed.
     void Unlock();
+
+    void EnableInterruptAndSchedule(u32 cores_pending_reschedule,
+                                    Core::EmuThreadHandle global_thread);
+
+    /**
+     * Add a thread to the suggested queue of a cpu core. Suggested threads may be
+     * picked if no thread is scheduled to run on the core.
+     */
+    void Suggest(u32 priority, std::size_t core, Thread* thread);
+
+    /**
+     * Remove a thread to the suggested queue of a cpu core. Suggested threads may be
+     * picked if no thread is scheduled to run on the core.
+     */
+    void Unsuggest(u32 priority, std::size_t core, Thread* thread);
+
+    /**
+     * Add a thread to the scheduling queue of a cpu core. The thread is added at the
+     * back the queue in its priority level.
+     */
+    void Schedule(u32 priority, std::size_t core, Thread* thread);
+
+    /**
+     * Add a thread to the scheduling queue of a cpu core. The thread is added at the
+     * front the queue in its priority level.
+     */
+    void SchedulePrepend(u32 priority, std::size_t core, Thread* thread);
+
+    /// Reschedule an already scheduled thread based on a new priority
+    void Reschedule(u32 priority, std::size_t core, Thread* thread);
+
+    /// Unschedules a thread.
+    void Unschedule(u32 priority, std::size_t core, Thread* thread);
+
     /**
      * Transfers a thread into an specific core. If the destination_core is -1
      * it will be unscheduled from its source code and added into its suggested
@@ -170,10 +187,13 @@ private:
     std::array<u32, Core::Hardware::NUM_CPU_CORES> preemption_priorities = {59, 59, 59, 62};
 
     /// Scheduler lock mechanisms.
-    std::mutex inner_lock{}; // TODO(Blinkhawk): Replace for a SpinLock
+    bool is_locked{};
+    Common::SpinLock inner_lock{};
     std::atomic<s64> scope_lock{};
     Core::EmuThreadHandle current_owner{Core::EmuThreadHandle::InvalidHandle()};
 
+    Common::SpinLock global_list_guard{};
+
     /// Lists all thread ids that aren't deleted/etc.
     std::vector<std::shared_ptr<Thread>> thread_list;
     KernelCore& kernel;
@@ -190,11 +210,11 @@ public:
     /// Reschedules to the next available thread (call after current thread is suspended)
     void TryDoContextSwitch();
 
-    /// Unloads currently running thread
-    void UnloadThread();
-
-    /// Select the threads in top of the scheduling multilist.
-    void SelectThreads();
+    /// The next two are for SingleCore Only.
+    /// Unload current thread before preempting core.
+    void Unload();
+    /// Reload current thread after core preemption.
+    void Reload();
 
     /// Gets the current running thread
     Thread* GetCurrentThread() const;
@@ -209,15 +229,30 @@ public:
         return is_context_switch_pending;
     }
 
+    void Initialize();
+
     /// Shutdowns the scheduler.
     void Shutdown();
 
+    void OnThreadStart();
+
+    std::shared_ptr<Common::Fiber>& ControlContext() {
+        return switch_fiber;
+    }
+
+    const std::shared_ptr<Common::Fiber>& ControlContext() const {
+        return switch_fiber;
+    }
+
 private:
     friend class GlobalScheduler;
 
     /// Switches the CPU's active thread context to that of the specified thread
     void SwitchContext();
 
+    /// When a thread wakes up, it must run this through it's new scheduler
+    void SwitchContextStep2();
+
     /**
      * Called on every context switch to update the internal timestamp
      * This also updates the running time ticks for the given thread and
@@ -231,14 +266,24 @@ private:
      */
     void UpdateLastContextSwitchTime(Thread* thread, Process* process);
 
+    static void OnSwitch(void* this_scheduler);
+    void SwitchToCurrent();
+
     std::shared_ptr<Thread> current_thread = nullptr;
     std::shared_ptr<Thread> selected_thread = nullptr;
+    std::shared_ptr<Thread> current_thread_prev = nullptr;
+    std::shared_ptr<Thread> selected_thread_set = nullptr;
+    std::shared_ptr<Thread> idle_thread = nullptr;
+
+    std::shared_ptr<Common::Fiber> switch_fiber = nullptr;
 
     Core::System& system;
     u64 last_context_switch_time = 0;
     u64 idle_selection_count = 0;
     const std::size_t core_id;
 
+    Common::SpinLock guard{};
+
     bool is_context_switch_pending = false;
 };
 
@@ -261,6 +306,8 @@ public:
         sleep_cancelled = true;
     }
 
+    void Release();
+
 private:
     Handle& event_handle;
     Thread* time_task;
diff --git a/src/core/hle/kernel/server_session.cpp b/src/core/hle/kernel/server_session.cpp
index 25438b86b..7b23a6889 100644
--- a/src/core/hle/kernel/server_session.cpp
+++ b/src/core/hle/kernel/server_session.cpp
@@ -17,6 +17,7 @@
 #include "core/hle/kernel/hle_ipc.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
+#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/server_session.h"
 #include "core/hle/kernel/session.h"
 #include "core/hle/kernel/thread.h"
@@ -168,9 +169,12 @@ ResultCode ServerSession::CompleteSyncRequest() {
     }
 
     // Some service requests require the thread to block
-    if (!context.IsThreadWaiting()) {
-        context.GetThread().ResumeFromWait();
-        context.GetThread().SetWaitSynchronizationResult(result);
+    {
+        SchedulerLock lock(kernel);
+        if (!context.IsThreadWaiting()) {
+            context.GetThread().ResumeFromWait();
+            context.GetThread().SetSynchronizationResults(nullptr, result);
+        }
     }
 
     request_queue.Pop();
@@ -180,8 +184,10 @@ ResultCode ServerSession::CompleteSyncRequest() {
 
 ResultCode ServerSession::HandleSyncRequest(std::shared_ptr<Thread> thread,
                                             Core::Memory::Memory& memory) {
-    Core::System::GetInstance().CoreTiming().ScheduleEvent(20000, request_event, {});
-    return QueueSyncRequest(std::move(thread), memory);
+    ResultCode result = QueueSyncRequest(std::move(thread), memory);
+    const u64 delay = kernel.IsMulticore() ? 0U : 20000U;
+    Core::System::GetInstance().CoreTiming().ScheduleEvent(delay, request_event, {});
+    return result;
 }
 
 } // namespace Kernel
diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp
index 4ae4529f5..5db19dcf3 100644
--- a/src/core/hle/kernel/svc.cpp
+++ b/src/core/hle/kernel/svc.cpp
@@ -10,14 +10,15 @@
 
 #include "common/alignment.h"
 #include "common/assert.h"
+#include "common/fiber.h"
 #include "common/logging/log.h"
 #include "common/microprofile.h"
 #include "common/string_util.h"
 #include "core/arm/exclusive_monitor.h"
 #include "core/core.h"
-#include "core/core_manager.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/cpu_manager.h"
 #include "core/hle/kernel/address_arbiter.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
@@ -27,6 +28,7 @@
 #include "core/hle/kernel/memory/memory_block.h"
 #include "core/hle/kernel/memory/page_table.h"
 #include "core/hle/kernel/mutex.h"
+#include "core/hle/kernel/physical_core.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/readable_event.h"
 #include "core/hle/kernel/resource_limit.h"
@@ -37,6 +39,7 @@
 #include "core/hle/kernel/svc_wrap.h"
 #include "core/hle/kernel/synchronization.h"
 #include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
 #include "core/hle/kernel/transfer_memory.h"
 #include "core/hle/kernel/writable_event.h"
 #include "core/hle/lock.h"
@@ -133,6 +136,7 @@ enum class ResourceLimitValueType {
 
 ResultVal<s64> RetrieveResourceLimitValue(Core::System& system, Handle resource_limit,
                                           u32 resource_type, ResourceLimitValueType value_type) {
+    std::lock_guard lock{HLE::g_hle_lock};
     const auto type = static_cast<ResourceType>(resource_type);
     if (!IsValidResourceType(type)) {
         LOG_ERROR(Kernel_SVC, "Invalid resource limit type: '{}'", resource_type);
@@ -160,6 +164,7 @@ ResultVal<s64> RetrieveResourceLimitValue(Core::System& system, Handle resource_
 
 /// Set the process heap to a given Size. It can both extend and shrink the heap.
 static ResultCode SetHeapSize(Core::System& system, VAddr* heap_addr, u64 heap_size) {
+    std::lock_guard lock{HLE::g_hle_lock};
     LOG_TRACE(Kernel_SVC, "called, heap_size=0x{:X}", heap_size);
 
     // Size must be a multiple of 0x200000 (2MB) and be equal to or less than 8GB.
@@ -190,6 +195,7 @@ static ResultCode SetHeapSize32(Core::System& system, u32* heap_addr, u32 heap_s
 
 static ResultCode SetMemoryAttribute(Core::System& system, VAddr address, u64 size, u32 mask,
                                      u32 attribute) {
+    std::lock_guard lock{HLE::g_hle_lock};
     LOG_DEBUG(Kernel_SVC,
               "called, address=0x{:016X}, size=0x{:X}, mask=0x{:08X}, attribute=0x{:08X}", address,
               size, mask, attribute);
@@ -226,8 +232,15 @@ static ResultCode SetMemoryAttribute(Core::System& system, VAddr address, u64 si
                                          static_cast<Memory::MemoryAttribute>(attribute));
 }
 
+static ResultCode SetMemoryAttribute32(Core::System& system, u32 address, u32 size, u32 mask,
+                                       u32 attribute) {
+    return SetMemoryAttribute(system, static_cast<VAddr>(address), static_cast<std::size_t>(size),
+                              mask, attribute);
+}
+
 /// Maps a memory range into a different range.
 static ResultCode MapMemory(Core::System& system, VAddr dst_addr, VAddr src_addr, u64 size) {
+    std::lock_guard lock{HLE::g_hle_lock};
     LOG_TRACE(Kernel_SVC, "called, dst_addr=0x{:X}, src_addr=0x{:X}, size=0x{:X}", dst_addr,
               src_addr, size);
 
@@ -241,8 +254,14 @@ static ResultCode MapMemory(Core::System& system, VAddr dst_addr, VAddr src_addr
     return page_table.Map(dst_addr, src_addr, size);
 }
 
+static ResultCode MapMemory32(Core::System& system, u32 dst_addr, u32 src_addr, u32 size) {
+    return MapMemory(system, static_cast<VAddr>(dst_addr), static_cast<VAddr>(src_addr),
+                     static_cast<std::size_t>(size));
+}
+
 /// Unmaps a region that was previously mapped with svcMapMemory
 static ResultCode UnmapMemory(Core::System& system, VAddr dst_addr, VAddr src_addr, u64 size) {
+    std::lock_guard lock{HLE::g_hle_lock};
     LOG_TRACE(Kernel_SVC, "called, dst_addr=0x{:X}, src_addr=0x{:X}, size=0x{:X}", dst_addr,
               src_addr, size);
 
@@ -256,9 +275,15 @@ static ResultCode UnmapMemory(Core::System& system, VAddr dst_addr, VAddr src_ad
     return page_table.Unmap(dst_addr, src_addr, size);
 }
 
+static ResultCode UnmapMemory32(Core::System& system, u32 dst_addr, u32 src_addr, u32 size) {
+    return UnmapMemory(system, static_cast<VAddr>(dst_addr), static_cast<VAddr>(src_addr),
+                       static_cast<std::size_t>(size));
+}
+
 /// Connect to an OS service given the port name, returns the handle to the port to out
 static ResultCode ConnectToNamedPort(Core::System& system, Handle* out_handle,
                                      VAddr port_name_address) {
+    std::lock_guard lock{HLE::g_hle_lock};
     auto& memory = system.Memory();
 
     if (!memory.IsValidVirtualAddress(port_name_address)) {
@@ -317,11 +342,30 @@ static ResultCode SendSyncRequest(Core::System& system, Handle handle) {
     LOG_TRACE(Kernel_SVC, "called handle=0x{:08X}({})", handle, session->GetName());
 
     auto thread = system.CurrentScheduler().GetCurrentThread();
-    thread->InvalidateWakeupCallback();
-    thread->SetStatus(ThreadStatus::WaitIPC);
-    system.PrepareReschedule(thread->GetProcessorID());
+    {
+        SchedulerLock lock(system.Kernel());
+        thread->InvalidateHLECallback();
+        thread->SetStatus(ThreadStatus::WaitIPC);
+        session->SendSyncRequest(SharedFrom(thread), system.Memory());
+    }
+
+    if (thread->HasHLECallback()) {
+        Handle event_handle = thread->GetHLETimeEvent();
+        if (event_handle != InvalidHandle) {
+            auto& time_manager = system.Kernel().TimeManager();
+            time_manager.UnscheduleTimeEvent(event_handle);
+        }
+
+        {
+            SchedulerLock lock(system.Kernel());
+            auto* sync_object = thread->GetHLESyncObject();
+            sync_object->RemoveWaitingThread(SharedFrom(thread));
+        }
+
+        thread->InvokeHLECallback(SharedFrom(thread));
+    }
 
-    return session->SendSyncRequest(SharedFrom(thread), system.Memory());
+    return thread->GetSignalingResult();
 }
 
 static ResultCode SendSyncRequest32(Core::System& system, Handle handle) {
@@ -383,6 +427,15 @@ static ResultCode GetProcessId(Core::System& system, u64* process_id, Handle han
     return ERR_INVALID_HANDLE;
 }
 
+static ResultCode GetProcessId32(Core::System& system, u32* process_id_low, u32* process_id_high,
+                                 Handle handle) {
+    u64 process_id{};
+    const auto result = GetProcessId(system, &process_id, handle);
+    *process_id_low = static_cast<u32>(process_id);
+    *process_id_high = static_cast<u32>(process_id >> 32);
+    return result;
+}
+
 /// Wait for the given handles to synchronize, timeout after the specified nanoseconds
 static ResultCode WaitSynchronization(Core::System& system, Handle* index, VAddr handles_address,
                                       u64 handle_count, s64 nano_seconds) {
@@ -447,10 +500,13 @@ static ResultCode CancelSynchronization(Core::System& system, Handle thread_hand
     }
 
     thread->CancelWait();
-    system.PrepareReschedule(thread->GetProcessorID());
     return RESULT_SUCCESS;
 }
 
+static ResultCode CancelSynchronization32(Core::System& system, Handle thread_handle) {
+    return CancelSynchronization(system, thread_handle);
+}
+
 /// Attempts to locks a mutex, creating it if it does not already exist
 static ResultCode ArbitrateLock(Core::System& system, Handle holding_thread_handle,
                                 VAddr mutex_addr, Handle requesting_thread_handle) {
@@ -475,6 +531,12 @@ static ResultCode ArbitrateLock(Core::System& system, Handle holding_thread_hand
                                                   requesting_thread_handle);
 }
 
+static ResultCode ArbitrateLock32(Core::System& system, Handle holding_thread_handle,
+                                  u32 mutex_addr, Handle requesting_thread_handle) {
+    return ArbitrateLock(system, holding_thread_handle, static_cast<VAddr>(mutex_addr),
+                         requesting_thread_handle);
+}
+
 /// Unlock a mutex
 static ResultCode ArbitrateUnlock(Core::System& system, VAddr mutex_addr) {
     LOG_TRACE(Kernel_SVC, "called mutex_addr=0x{:X}", mutex_addr);
@@ -494,6 +556,10 @@ static ResultCode ArbitrateUnlock(Core::System& system, VAddr mutex_addr) {
     return current_process->GetMutex().Release(mutex_addr);
 }
 
+static ResultCode ArbitrateUnlock32(Core::System& system, u32 mutex_addr) {
+    return ArbitrateUnlock(system, static_cast<VAddr>(mutex_addr));
+}
+
 enum class BreakType : u32 {
     Panic = 0,
     AssertionFailed = 1,
@@ -594,6 +660,7 @@ static void Break(Core::System& system, u32 reason, u64 info1, u64 info2) {
         info2, has_dumped_buffer ? std::make_optional(debug_buffer) : std::nullopt);
 
     if (!break_reason.signal_debugger) {
+        SchedulerLock lock(system.Kernel());
         LOG_CRITICAL(
             Debug_Emulated,
             "Emulated program broke execution! reason=0x{:016X}, info1=0x{:016X}, info2=0x{:016X}",
@@ -605,14 +672,16 @@ static void Break(Core::System& system, u32 reason, u64 info1, u64 info2) {
         const auto thread_processor_id = current_thread->GetProcessorID();
         system.ArmInterface(static_cast<std::size_t>(thread_processor_id)).LogBacktrace();
 
-        system.Kernel().CurrentProcess()->PrepareForTermination();
-
         // Kill the current thread
+        system.Kernel().ExceptionalExit();
         current_thread->Stop();
-        system.PrepareReschedule();
     }
 }
 
+static void Break32(Core::System& system, u32 reason, u32 info1, u32 info2) {
+    Break(system, reason, static_cast<u64>(info1), static_cast<u64>(info2));
+}
+
 /// Used to output a message on a debug hardware unit - does nothing on a retail unit
 static void OutputDebugString([[maybe_unused]] Core::System& system, VAddr address, u64 len) {
     if (len == 0) {
@@ -627,6 +696,7 @@ static void OutputDebugString([[maybe_unused]] Core::System& system, VAddr addre
 /// Gets system/memory information for the current process
 static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 handle,
                           u64 info_sub_id) {
+    std::lock_guard lock{HLE::g_hle_lock};
     LOG_TRACE(Kernel_SVC, "called info_id=0x{:X}, info_sub_id=0x{:X}, handle=0x{:08X}", info_id,
               info_sub_id, handle);
 
@@ -863,9 +933,9 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
         if (same_thread && info_sub_id == 0xFFFFFFFFFFFFFFFF) {
             const u64 thread_ticks = current_thread->GetTotalCPUTimeTicks();
 
-            out_ticks = thread_ticks + (core_timing.GetTicks() - prev_ctx_ticks);
+            out_ticks = thread_ticks + (core_timing.GetCPUTicks() - prev_ctx_ticks);
         } else if (same_thread && info_sub_id == system.CurrentCoreIndex()) {
-            out_ticks = core_timing.GetTicks() - prev_ctx_ticks;
+            out_ticks = core_timing.GetCPUTicks() - prev_ctx_ticks;
         }
 
         *result = out_ticks;
@@ -892,6 +962,7 @@ static ResultCode GetInfo32(Core::System& system, u32* result_low, u32* result_h
 
 /// Maps memory at a desired address
 static ResultCode MapPhysicalMemory(Core::System& system, VAddr addr, u64 size) {
+    std::lock_guard lock{HLE::g_hle_lock};
     LOG_DEBUG(Kernel_SVC, "called, addr=0x{:016X}, size=0x{:X}", addr, size);
 
     if (!Common::Is4KBAligned(addr)) {
@@ -939,8 +1010,13 @@ static ResultCode MapPhysicalMemory(Core::System& system, VAddr addr, u64 size)
     return page_table.MapPhysicalMemory(addr, size);
 }
 
+static ResultCode MapPhysicalMemory32(Core::System& system, u32 addr, u32 size) {
+    return MapPhysicalMemory(system, static_cast<VAddr>(addr), static_cast<std::size_t>(size));
+}
+
 /// Unmaps memory previously mapped via MapPhysicalMemory
 static ResultCode UnmapPhysicalMemory(Core::System& system, VAddr addr, u64 size) {
+    std::lock_guard lock{HLE::g_hle_lock};
     LOG_DEBUG(Kernel_SVC, "called, addr=0x{:016X}, size=0x{:X}", addr, size);
 
     if (!Common::Is4KBAligned(addr)) {
@@ -988,6 +1064,10 @@ static ResultCode UnmapPhysicalMemory(Core::System& system, VAddr addr, u64 size
     return page_table.UnmapPhysicalMemory(addr, size);
 }
 
+static ResultCode UnmapPhysicalMemory32(Core::System& system, u32 addr, u32 size) {
+    return UnmapPhysicalMemory(system, static_cast<VAddr>(addr), static_cast<std::size_t>(size));
+}
+
 /// Sets the thread activity
 static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 activity) {
     LOG_DEBUG(Kernel_SVC, "called, handle=0x{:08X}, activity=0x{:08X}", handle, activity);
@@ -1017,10 +1097,11 @@ static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 act
         return ERR_BUSY;
     }
 
-    thread->SetActivity(static_cast<ThreadActivity>(activity));
+    return thread->SetActivity(static_cast<ThreadActivity>(activity));
+}
 
-    system.PrepareReschedule(thread->GetProcessorID());
-    return RESULT_SUCCESS;
+static ResultCode SetThreadActivity32(Core::System& system, Handle handle, u32 activity) {
+    return SetThreadActivity(system, handle, activity);
 }
 
 /// Gets the thread context
@@ -1064,6 +1145,10 @@ static ResultCode GetThreadContext(Core::System& system, VAddr thread_context, H
     return RESULT_SUCCESS;
 }
 
+static ResultCode GetThreadContext32(Core::System& system, u32 thread_context, Handle handle) {
+    return GetThreadContext(system, static_cast<VAddr>(thread_context), handle);
+}
+
 /// Gets the priority for the specified thread
 static ResultCode GetThreadPriority(Core::System& system, u32* priority, Handle handle) {
     LOG_TRACE(Kernel_SVC, "called");
@@ -1071,6 +1156,7 @@ static ResultCode GetThreadPriority(Core::System& system, u32* priority, Handle
     const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
     const std::shared_ptr<Thread> thread = handle_table.Get<Thread>(handle);
     if (!thread) {
+        *priority = 0;
         LOG_ERROR(Kernel_SVC, "Thread handle does not exist, handle=0x{:08X}", handle);
         return ERR_INVALID_HANDLE;
     }
@@ -1105,18 +1191,26 @@ static ResultCode SetThreadPriority(Core::System& system, Handle handle, u32 pri
 
     thread->SetPriority(priority);
 
-    system.PrepareReschedule(thread->GetProcessorID());
     return RESULT_SUCCESS;
 }
 
+static ResultCode SetThreadPriority32(Core::System& system, Handle handle, u32 priority) {
+    return SetThreadPriority(system, handle, priority);
+}
+
 /// Get which CPU core is executing the current thread
 static u32 GetCurrentProcessorNumber(Core::System& system) {
     LOG_TRACE(Kernel_SVC, "called");
-    return system.CurrentScheduler().GetCurrentThread()->GetProcessorID();
+    return static_cast<u32>(system.CurrentPhysicalCore().CoreIndex());
+}
+
+static u32 GetCurrentProcessorNumber32(Core::System& system) {
+    return GetCurrentProcessorNumber(system);
 }
 
 static ResultCode MapSharedMemory(Core::System& system, Handle shared_memory_handle, VAddr addr,
                                   u64 size, u32 permissions) {
+    std::lock_guard lock{HLE::g_hle_lock};
     LOG_TRACE(Kernel_SVC,
               "called, shared_memory_handle=0x{:X}, addr=0x{:X}, size=0x{:X}, permissions=0x{:08X}",
               shared_memory_handle, addr, size, permissions);
@@ -1187,9 +1281,16 @@ static ResultCode MapSharedMemory(Core::System& system, Handle shared_memory_han
     return shared_memory->Map(*current_process, addr, size, permission_type);
 }
 
+static ResultCode MapSharedMemory32(Core::System& system, Handle shared_memory_handle, u32 addr,
+                                    u32 size, u32 permissions) {
+    return MapSharedMemory(system, shared_memory_handle, static_cast<VAddr>(addr),
+                           static_cast<std::size_t>(size), permissions);
+}
+
 static ResultCode QueryProcessMemory(Core::System& system, VAddr memory_info_address,
                                      VAddr page_info_address, Handle process_handle,
                                      VAddr address) {
+    std::lock_guard lock{HLE::g_hle_lock};
     LOG_TRACE(Kernel_SVC, "called process=0x{:08X} address={:X}", process_handle, address);
     const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
     std::shared_ptr<Process> process = handle_table.Get<Process>(process_handle);
@@ -1372,6 +1473,7 @@ static ResultCode UnmapProcessCodeMemory(Core::System& system, Handle process_ha
 /// Exits the current process
 static void ExitProcess(Core::System& system) {
     auto* current_process = system.Kernel().CurrentProcess();
+    UNIMPLEMENTED();
 
     LOG_INFO(Kernel_SVC, "Process {} exiting", current_process->GetProcessID());
     ASSERT_MSG(current_process->GetStatus() == ProcessStatus::Running,
@@ -1381,8 +1483,10 @@ static void ExitProcess(Core::System& system) {
 
     // Kill the current thread
     system.CurrentScheduler().GetCurrentThread()->Stop();
+}
 
-    system.PrepareReschedule();
+static void ExitProcess32(Core::System& system) {
+    ExitProcess(system);
 }
 
 /// Creates a new thread
@@ -1428,9 +1532,10 @@ static ResultCode CreateThread(Core::System& system, Handle* out_handle, VAddr e
 
     ASSERT(kernel.CurrentProcess()->GetResourceLimit()->Reserve(ResourceType::Threads, 1));
 
+    ThreadType type = THREADTYPE_USER;
     CASCADE_RESULT(std::shared_ptr<Thread> thread,
-                   Thread::Create(kernel, "", entry_point, priority, arg, processor_id, stack_top,
-                                  *current_process));
+                   Thread::Create(system, type, "", entry_point, priority, arg, processor_id,
+                                  stack_top, current_process));
 
     const auto new_thread_handle = current_process->GetHandleTable().Create(thread);
     if (new_thread_handle.Failed()) {
@@ -1444,11 +1549,15 @@ static ResultCode CreateThread(Core::System& system, Handle* out_handle, VAddr e
     thread->SetName(
         fmt::format("thread[entry_point={:X}, handle={:X}]", entry_point, *new_thread_handle));
 
-    system.PrepareReschedule(thread->GetProcessorID());
-
     return RESULT_SUCCESS;
 }
 
+static ResultCode CreateThread32(Core::System& system, Handle* out_handle, u32 priority,
+                                 u32 entry_point, u32 arg, u32 stack_top, s32 processor_id) {
+    return CreateThread(system, out_handle, static_cast<VAddr>(entry_point), static_cast<u64>(arg),
+                        static_cast<VAddr>(stack_top), priority, processor_id);
+}
+
 /// Starts the thread for the provided handle
 static ResultCode StartThread(Core::System& system, Handle thread_handle) {
     LOG_DEBUG(Kernel_SVC, "called thread=0x{:08X}", thread_handle);
@@ -1463,13 +1572,11 @@ static ResultCode StartThread(Core::System& system, Handle thread_handle) {
 
     ASSERT(thread->GetStatus() == ThreadStatus::Dormant);
 
-    thread->ResumeFromWait();
-
-    if (thread->GetStatus() == ThreadStatus::Ready) {
-        system.PrepareReschedule(thread->GetProcessorID());
-    }
+    return thread->Start();
+}
 
-    return RESULT_SUCCESS;
+static ResultCode StartThread32(Core::System& system, Handle thread_handle) {
+    return StartThread(system, thread_handle);
 }
 
 /// Called when a thread exits
@@ -1477,9 +1584,12 @@ static void ExitThread(Core::System& system) {
     LOG_DEBUG(Kernel_SVC, "called, pc=0x{:08X}", system.CurrentArmInterface().GetPC());
 
     auto* const current_thread = system.CurrentScheduler().GetCurrentThread();
-    current_thread->Stop();
     system.GlobalScheduler().RemoveThread(SharedFrom(current_thread));
-    system.PrepareReschedule();
+    current_thread->Stop();
+}
+
+static void ExitThread32(Core::System& system) {
+    ExitThread(system);
 }
 
 /// Sleep the current thread
@@ -1498,15 +1608,21 @@ static void SleepThread(Core::System& system, s64 nanoseconds) {
 
     if (nanoseconds <= 0) {
         switch (static_cast<SleepType>(nanoseconds)) {
-        case SleepType::YieldWithoutLoadBalancing:
-            is_redundant = current_thread->YieldSimple();
+        case SleepType::YieldWithoutLoadBalancing: {
+            auto pair = current_thread->YieldSimple();
+            is_redundant = pair.second;
             break;
-        case SleepType::YieldWithLoadBalancing:
-            is_redundant = current_thread->YieldAndBalanceLoad();
+        }
+        case SleepType::YieldWithLoadBalancing: {
+            auto pair = current_thread->YieldAndBalanceLoad();
+            is_redundant = pair.second;
             break;
-        case SleepType::YieldAndWaitForLoadBalancing:
-            is_redundant = current_thread->YieldAndWaitForLoadBalancing();
+        }
+        case SleepType::YieldAndWaitForLoadBalancing: {
+            auto pair = current_thread->YieldAndWaitForLoadBalancing();
+            is_redundant = pair.second;
             break;
+        }
         default:
             UNREACHABLE_MSG("Unimplemented sleep yield type '{:016X}'!", nanoseconds);
         }
@@ -1514,13 +1630,18 @@ static void SleepThread(Core::System& system, s64 nanoseconds) {
         current_thread->Sleep(nanoseconds);
     }
 
-    if (is_redundant) {
-        // If it's redundant, the core is pretty much idle. Some games keep idling
-        // a core while it's doing nothing, we advance timing to avoid costly continuous
-        // calls.
-        system.CoreTiming().AddTicks(2000);
+    if (is_redundant && !system.Kernel().IsMulticore()) {
+        system.Kernel().ExitSVCProfile();
+        system.CoreTiming().AddTicks(1000U);
+        system.GetCpuManager().PreemptSingleCore();
+        system.Kernel().EnterSVCProfile();
     }
-    system.PrepareReschedule(current_thread->GetProcessorID());
+}
+
+static void SleepThread32(Core::System& system, u32 nanoseconds_low, u32 nanoseconds_high) {
+    const s64 nanoseconds = static_cast<s64>(static_cast<u64>(nanoseconds_low) |
+                                             (static_cast<u64>(nanoseconds_high) << 32));
+    SleepThread(system, nanoseconds);
 }
 
 /// Wait process wide key atomic
@@ -1547,31 +1668,69 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add
     }
 
     ASSERT(condition_variable_addr == Common::AlignDown(condition_variable_addr, 4));
-
+    auto& kernel = system.Kernel();
+    Handle event_handle;
+    Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
     auto* const current_process = system.Kernel().CurrentProcess();
-    const auto& handle_table = current_process->GetHandleTable();
-    std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle);
-    ASSERT(thread);
+    {
+        SchedulerLockAndSleep lock(kernel, event_handle, current_thread, nano_seconds);
+        const auto& handle_table = current_process->GetHandleTable();
+        std::shared_ptr<Thread> thread = handle_table.Get<Thread>(thread_handle);
+        ASSERT(thread);
+
+        current_thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT);
+
+        if (thread->IsPendingTermination()) {
+            lock.CancelSleep();
+            return ERR_THREAD_TERMINATING;
+        }
+
+        const auto release_result = current_process->GetMutex().Release(mutex_addr);
+        if (release_result.IsError()) {
+            lock.CancelSleep();
+            return release_result;
+        }
+
+        if (nano_seconds == 0) {
+            lock.CancelSleep();
+            return RESULT_TIMEOUT;
+        }
 
-    const auto release_result = current_process->GetMutex().Release(mutex_addr);
-    if (release_result.IsError()) {
-        return release_result;
+        current_thread->SetCondVarWaitAddress(condition_variable_addr);
+        current_thread->SetMutexWaitAddress(mutex_addr);
+        current_thread->SetWaitHandle(thread_handle);
+        current_thread->SetStatus(ThreadStatus::WaitCondVar);
+        current_process->InsertConditionVariableThread(SharedFrom(current_thread));
     }
 
-    Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
-    current_thread->SetCondVarWaitAddress(condition_variable_addr);
-    current_thread->SetMutexWaitAddress(mutex_addr);
-    current_thread->SetWaitHandle(thread_handle);
-    current_thread->SetStatus(ThreadStatus::WaitCondVar);
-    current_thread->InvalidateWakeupCallback();
-    current_process->InsertConditionVariableThread(SharedFrom(current_thread));
+    if (event_handle != InvalidHandle) {
+        auto& time_manager = kernel.TimeManager();
+        time_manager.UnscheduleTimeEvent(event_handle);
+    }
+
+    {
+        SchedulerLock lock(kernel);
 
-    current_thread->WakeAfterDelay(nano_seconds);
+        auto* owner = current_thread->GetLockOwner();
+        if (owner != nullptr) {
+            owner->RemoveMutexWaiter(SharedFrom(current_thread));
+        }
 
+        current_process->RemoveConditionVariableThread(SharedFrom(current_thread));
+    }
     // Note: Deliberately don't attempt to inherit the lock owner's priority.
 
-    system.PrepareReschedule(current_thread->GetProcessorID());
-    return RESULT_SUCCESS;
+    return current_thread->GetSignalingResult();
+}
+
+static ResultCode WaitProcessWideKeyAtomic32(Core::System& system, u32 mutex_addr,
+                                             u32 condition_variable_addr, Handle thread_handle,
+                                             u32 nanoseconds_low, u32 nanoseconds_high) {
+    const s64 nanoseconds =
+        static_cast<s64>(nanoseconds_low | (static_cast<u64>(nanoseconds_high) << 32));
+    return WaitProcessWideKeyAtomic(system, static_cast<VAddr>(mutex_addr),
+                                    static_cast<VAddr>(condition_variable_addr), thread_handle,
+                                    nanoseconds);
 }
 
 /// Signal process wide key
@@ -1582,7 +1741,9 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_
     ASSERT(condition_variable_addr == Common::AlignDown(condition_variable_addr, 4));
 
     // Retrieve a list of all threads that are waiting for this condition variable.
-    auto* const current_process = system.Kernel().CurrentProcess();
+    auto& kernel = system.Kernel();
+    SchedulerLock lock(kernel);
+    auto* const current_process = kernel.CurrentProcess();
     std::vector<std::shared_ptr<Thread>> waiting_threads =
         current_process->GetConditionVariableThreads(condition_variable_addr);
 
@@ -1591,7 +1752,7 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_
     std::size_t last = waiting_threads.size();
     if (target > 0)
         last = std::min(waiting_threads.size(), static_cast<std::size_t>(target));
-
+    auto& time_manager = kernel.TimeManager();
     for (std::size_t index = 0; index < last; ++index) {
         auto& thread = waiting_threads[index];
 
@@ -1599,7 +1760,6 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_
 
         // liberate Cond Var Thread.
         current_process->RemoveConditionVariableThread(thread);
-        thread->SetCondVarWaitAddress(0);
 
         const std::size_t current_core = system.CurrentCoreIndex();
         auto& monitor = system.Monitor();
@@ -1610,10 +1770,8 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_
         u32 update_val = 0;
         const VAddr mutex_address = thread->GetMutexWaitAddress();
         do {
-            monitor.SetExclusive(current_core, mutex_address);
-
             // If the mutex is not yet acquired, acquire it.
-            mutex_val = memory.Read32(mutex_address);
+            mutex_val = monitor.ExclusiveRead32(current_core, mutex_address);
 
             if (mutex_val != 0) {
                 update_val = mutex_val | Mutex::MutexHasWaitersFlag;
@@ -1621,33 +1779,28 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_
                 update_val = thread->GetWaitHandle();
             }
         } while (!monitor.ExclusiveWrite32(current_core, mutex_address, update_val));
+        monitor.ClearExclusive();
         if (mutex_val == 0) {
             // We were able to acquire the mutex, resume this thread.
-            ASSERT(thread->GetStatus() == ThreadStatus::WaitCondVar);
-            thread->ResumeFromWait();
-
             auto* const lock_owner = thread->GetLockOwner();
             if (lock_owner != nullptr) {
                 lock_owner->RemoveMutexWaiter(thread);
             }
 
             thread->SetLockOwner(nullptr);
-            thread->SetMutexWaitAddress(0);
-            thread->SetWaitHandle(0);
-            thread->SetWaitSynchronizationResult(RESULT_SUCCESS);
-            system.PrepareReschedule(thread->GetProcessorID());
+            thread->SetSynchronizationResults(nullptr, RESULT_SUCCESS);
+            thread->ResumeFromWait();
         } else {
             // The mutex is already owned by some other thread, make this thread wait on it.
             const Handle owner_handle = static_cast<Handle>(mutex_val & Mutex::MutexOwnerMask);
             const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
             auto owner = handle_table.Get<Thread>(owner_handle);
             ASSERT(owner);
-            ASSERT(thread->GetStatus() == ThreadStatus::WaitCondVar);
-            thread->InvalidateWakeupCallback();
-            thread->SetStatus(ThreadStatus::WaitMutex);
+            if (thread->GetStatus() == ThreadStatus::WaitCondVar) {
+                thread->SetStatus(ThreadStatus::WaitMutex);
+            }
 
             owner->AddMutexWaiter(thread);
-            system.PrepareReschedule(thread->GetProcessorID());
         }
     }
 }
@@ -1678,12 +1831,15 @@ static ResultCode WaitForAddress(Core::System& system, VAddr address, u32 type,
     auto& address_arbiter = system.Kernel().CurrentProcess()->GetAddressArbiter();
     const ResultCode result =
         address_arbiter.WaitForAddress(address, arbitration_type, value, timeout);
-    if (result == RESULT_SUCCESS) {
-        system.PrepareReschedule();
-    }
     return result;
 }
 
+static ResultCode WaitForAddress32(Core::System& system, u32 address, u32 type, s32 value,
+                                   u32 timeout_low, u32 timeout_high) {
+    s64 timeout = static_cast<s64>(timeout_low | (static_cast<u64>(timeout_high) << 32));
+    return WaitForAddress(system, static_cast<VAddr>(address), type, value, timeout);
+}
+
 // Signals to an address (via Address Arbiter)
 static ResultCode SignalToAddress(Core::System& system, VAddr address, u32 type, s32 value,
                                   s32 num_to_wake) {
@@ -1707,6 +1863,11 @@ static ResultCode SignalToAddress(Core::System& system, VAddr address, u32 type,
     return address_arbiter.SignalToAddress(address, signal_type, value, num_to_wake);
 }
 
+static ResultCode SignalToAddress32(Core::System& system, u32 address, u32 type, s32 value,
+                                    s32 num_to_wake) {
+    return SignalToAddress(system, static_cast<VAddr>(address), type, value, num_to_wake);
+}
+
 static void KernelDebug([[maybe_unused]] Core::System& system,
                         [[maybe_unused]] u32 kernel_debug_type, [[maybe_unused]] u64 param1,
                         [[maybe_unused]] u64 param2, [[maybe_unused]] u64 param3) {
@@ -1725,14 +1886,21 @@ static u64 GetSystemTick(Core::System& system) {
     auto& core_timing = system.CoreTiming();
 
     // Returns the value of cntpct_el0 (https://switchbrew.org/wiki/SVC#svcGetSystemTick)
-    const u64 result{Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks())};
+    const u64 result{system.CoreTiming().GetClockTicks()};
 
-    // Advance time to defeat dumb games that busy-wait for the frame to end.
-    core_timing.AddTicks(400);
+    if (!system.Kernel().IsMulticore()) {
+        core_timing.AddTicks(400U);
+    }
 
     return result;
 }
 
+static void GetSystemTick32(Core::System& system, u32* time_low, u32* time_high) {
+    u64 time = GetSystemTick(system);
+    *time_low = static_cast<u32>(time);
+    *time_high = static_cast<u32>(time >> 32);
+}
+
 /// Close a handle
 static ResultCode CloseHandle(Core::System& system, Handle handle) {
     LOG_TRACE(Kernel_SVC, "Closing handle 0x{:08X}", handle);
@@ -1765,9 +1933,14 @@ static ResultCode ResetSignal(Core::System& system, Handle handle) {
     return ERR_INVALID_HANDLE;
 }
 
+static ResultCode ResetSignal32(Core::System& system, Handle handle) {
+    return ResetSignal(system, handle);
+}
+
 /// Creates a TransferMemory object
 static ResultCode CreateTransferMemory(Core::System& system, Handle* handle, VAddr addr, u64 size,
                                        u32 permissions) {
+    std::lock_guard lock{HLE::g_hle_lock};
     LOG_DEBUG(Kernel_SVC, "called addr=0x{:X}, size=0x{:X}, perms=0x{:08X}", addr, size,
               permissions);
 
@@ -1812,6 +1985,12 @@ static ResultCode CreateTransferMemory(Core::System& system, Handle* handle, VAd
     return RESULT_SUCCESS;
 }
 
+static ResultCode CreateTransferMemory32(Core::System& system, Handle* handle, u32 addr, u32 size,
+                                         u32 permissions) {
+    return CreateTransferMemory(system, handle, static_cast<VAddr>(addr),
+                                static_cast<std::size_t>(size), permissions);
+}
+
 static ResultCode GetThreadCoreMask(Core::System& system, Handle thread_handle, u32* core,
                                     u64* mask) {
     LOG_TRACE(Kernel_SVC, "called, handle=0x{:08X}", thread_handle);
@@ -1821,6 +2000,8 @@ static ResultCode GetThreadCoreMask(Core::System& system, Handle thread_handle,
     if (!thread) {
         LOG_ERROR(Kernel_SVC, "Thread handle does not exist, thread_handle=0x{:08X}",
                   thread_handle);
+        *core = 0;
+        *mask = 0;
         return ERR_INVALID_HANDLE;
     }
 
@@ -1830,6 +2011,15 @@ static ResultCode GetThreadCoreMask(Core::System& system, Handle thread_handle,
     return RESULT_SUCCESS;
 }
 
+static ResultCode GetThreadCoreMask32(Core::System& system, Handle thread_handle, u32* core,
+                                      u32* mask_low, u32* mask_high) {
+    u64 mask{};
+    const auto result = GetThreadCoreMask(system, thread_handle, core, &mask);
+    *mask_high = static_cast<u32>(mask >> 32);
+    *mask_low = static_cast<u32>(mask);
+    return result;
+}
+
 static ResultCode SetThreadCoreMask(Core::System& system, Handle thread_handle, u32 core,
                                     u64 affinity_mask) {
     LOG_DEBUG(Kernel_SVC, "called, handle=0x{:08X}, core=0x{:X}, affinity_mask=0x{:016X}",
@@ -1861,7 +2051,7 @@ static ResultCode SetThreadCoreMask(Core::System& system, Handle thread_handle,
             return ERR_INVALID_COMBINATION;
         }
 
-        if (core < Core::NUM_CPU_CORES) {
+        if (core < Core::Hardware::NUM_CPU_CORES) {
             if ((affinity_mask & (1ULL << core)) == 0) {
                 LOG_ERROR(Kernel_SVC,
                           "Core is not enabled for the current mask, core={}, mask={:016X}", core,
@@ -1883,11 +2073,14 @@ static ResultCode SetThreadCoreMask(Core::System& system, Handle thread_handle,
         return ERR_INVALID_HANDLE;
     }
 
-    system.PrepareReschedule(thread->GetProcessorID());
-    thread->ChangeCore(core, affinity_mask);
-    system.PrepareReschedule(thread->GetProcessorID());
+    return thread->SetCoreAndAffinityMask(core, affinity_mask);
+}
 
-    return RESULT_SUCCESS;
+static ResultCode SetThreadCoreMask32(Core::System& system, Handle thread_handle, u32 core,
+                                      u32 affinity_mask_low, u32 affinity_mask_high) {
+    const u64 affinity_mask =
+        static_cast<u64>(affinity_mask_low) | (static_cast<u64>(affinity_mask_high) << 32);
+    return SetThreadCoreMask(system, thread_handle, core, affinity_mask);
 }
 
 static ResultCode CreateEvent(Core::System& system, Handle* write_handle, Handle* read_handle) {
@@ -1918,6 +2111,10 @@ static ResultCode CreateEvent(Core::System& system, Handle* write_handle, Handle
     return RESULT_SUCCESS;
 }
 
+static ResultCode CreateEvent32(Core::System& system, Handle* write_handle, Handle* read_handle) {
+    return CreateEvent(system, write_handle, read_handle);
+}
+
 static ResultCode ClearEvent(Core::System& system, Handle handle) {
     LOG_TRACE(Kernel_SVC, "called, event=0x{:08X}", handle);
 
@@ -1939,6 +2136,10 @@ static ResultCode ClearEvent(Core::System& system, Handle handle) {
     return ERR_INVALID_HANDLE;
 }
 
+static ResultCode ClearEvent32(Core::System& system, Handle handle) {
+    return ClearEvent(system, handle);
+}
+
 static ResultCode SignalEvent(Core::System& system, Handle handle) {
     LOG_DEBUG(Kernel_SVC, "called. Handle=0x{:08X}", handle);
 
@@ -1951,10 +2152,13 @@ static ResultCode SignalEvent(Core::System& system, Handle handle) {
     }
 
     writable_event->Signal();
-    system.PrepareReschedule();
     return RESULT_SUCCESS;
 }
 
+static ResultCode SignalEvent32(Core::System& system, Handle handle) {
+    return SignalEvent(system, handle);
+}
+
 static ResultCode GetProcessInfo(Core::System& system, u64* out, Handle process_handle, u32 type) {
     LOG_DEBUG(Kernel_SVC, "called, handle=0x{:08X}, type=0x{:X}", process_handle, type);
 
@@ -1982,6 +2186,7 @@ static ResultCode GetProcessInfo(Core::System& system, u64* out, Handle process_
 }
 
 static ResultCode CreateResourceLimit(Core::System& system, Handle* out_handle) {
+    std::lock_guard lock{HLE::g_hle_lock};
     LOG_DEBUG(Kernel_SVC, "called");
 
     auto& kernel = system.Kernel();
@@ -2139,6 +2344,15 @@ static ResultCode GetThreadList(Core::System& system, u32* out_num_threads, VAdd
     return RESULT_SUCCESS;
 }
 
+static ResultCode FlushProcessDataCache32(Core::System& system, Handle handle, u32 address,
+                                          u32 size) {
+    // Note(Blinkhawk): For emulation purposes of the data cache this is mostly a nope
+    // as all emulation is done in the same cache level in host architecture, thus data cache
+    // does not need flushing.
+    LOG_DEBUG(Kernel_SVC, "called");
+    return RESULT_SUCCESS;
+}
+
 namespace {
 struct FunctionDef {
     using Func = void(Core::System&);
@@ -2153,57 +2367,57 @@ static const FunctionDef SVC_Table_32[] = {
     {0x00, nullptr, "Unknown"},
     {0x01, SvcWrap32<SetHeapSize32>, "SetHeapSize32"},
     {0x02, nullptr, "Unknown"},
-    {0x03, nullptr, "SetMemoryAttribute32"},
-    {0x04, nullptr, "MapMemory32"},
-    {0x05, nullptr, "UnmapMemory32"},
+    {0x03, SvcWrap32<SetMemoryAttribute32>, "SetMemoryAttribute32"},
+    {0x04, SvcWrap32<MapMemory32>, "MapMemory32"},
+    {0x05, SvcWrap32<UnmapMemory32>, "UnmapMemory32"},
     {0x06, SvcWrap32<QueryMemory32>, "QueryMemory32"},
-    {0x07, nullptr, "ExitProcess32"},
-    {0x08, nullptr, "CreateThread32"},
-    {0x09, nullptr, "StartThread32"},
-    {0x0a, nullptr, "ExitThread32"},
-    {0x0b, nullptr, "SleepThread32"},
+    {0x07, SvcWrap32<ExitProcess32>, "ExitProcess32"},
+    {0x08, SvcWrap32<CreateThread32>, "CreateThread32"},
+    {0x09, SvcWrap32<StartThread32>, "StartThread32"},
+    {0x0a, SvcWrap32<ExitThread32>, "ExitThread32"},
+    {0x0b, SvcWrap32<SleepThread32>, "SleepThread32"},
     {0x0c, SvcWrap32<GetThreadPriority32>, "GetThreadPriority32"},
-    {0x0d, nullptr, "SetThreadPriority32"},
-    {0x0e, nullptr, "GetThreadCoreMask32"},
-    {0x0f, nullptr, "SetThreadCoreMask32"},
-    {0x10, nullptr, "GetCurrentProcessorNumber32"},
-    {0x11, nullptr, "SignalEvent32"},
-    {0x12, nullptr, "ClearEvent32"},
-    {0x13, nullptr, "MapSharedMemory32"},
+    {0x0d, SvcWrap32<SetThreadPriority32>, "SetThreadPriority32"},
+    {0x0e, SvcWrap32<GetThreadCoreMask32>, "GetThreadCoreMask32"},
+    {0x0f, SvcWrap32<SetThreadCoreMask32>, "SetThreadCoreMask32"},
+    {0x10, SvcWrap32<GetCurrentProcessorNumber32>, "GetCurrentProcessorNumber32"},
+    {0x11, SvcWrap32<SignalEvent32>, "SignalEvent32"},
+    {0x12, SvcWrap32<ClearEvent32>, "ClearEvent32"},
+    {0x13, SvcWrap32<MapSharedMemory32>, "MapSharedMemory32"},
     {0x14, nullptr, "UnmapSharedMemory32"},
-    {0x15, nullptr, "CreateTransferMemory32"},
+    {0x15, SvcWrap32<CreateTransferMemory32>, "CreateTransferMemory32"},
     {0x16, SvcWrap32<CloseHandle32>, "CloseHandle32"},
-    {0x17, nullptr, "ResetSignal32"},
+    {0x17, SvcWrap32<ResetSignal32>, "ResetSignal32"},
     {0x18, SvcWrap32<WaitSynchronization32>, "WaitSynchronization32"},
-    {0x19, nullptr, "CancelSynchronization32"},
-    {0x1a, nullptr, "ArbitrateLock32"},
-    {0x1b, nullptr, "ArbitrateUnlock32"},
-    {0x1c, nullptr, "WaitProcessWideKeyAtomic32"},
+    {0x19, SvcWrap32<CancelSynchronization32>, "CancelSynchronization32"},
+    {0x1a, SvcWrap32<ArbitrateLock32>, "ArbitrateLock32"},
+    {0x1b, SvcWrap32<ArbitrateUnlock32>, "ArbitrateUnlock32"},
+    {0x1c, SvcWrap32<WaitProcessWideKeyAtomic32>, "WaitProcessWideKeyAtomic32"},
     {0x1d, SvcWrap32<SignalProcessWideKey32>, "SignalProcessWideKey32"},
-    {0x1e, nullptr, "GetSystemTick32"},
+    {0x1e, SvcWrap32<GetSystemTick32>, "GetSystemTick32"},
     {0x1f, SvcWrap32<ConnectToNamedPort32>, "ConnectToNamedPort32"},
     {0x20, nullptr, "Unknown"},
     {0x21, SvcWrap32<SendSyncRequest32>, "SendSyncRequest32"},
     {0x22, nullptr, "SendSyncRequestWithUserBuffer32"},
     {0x23, nullptr, "Unknown"},
-    {0x24, nullptr, "GetProcessId32"},
+    {0x24, SvcWrap32<GetProcessId32>, "GetProcessId32"},
     {0x25, SvcWrap32<GetThreadId32>, "GetThreadId32"},
-    {0x26, nullptr, "Break32"},
+    {0x26, SvcWrap32<Break32>, "Break32"},
     {0x27, nullptr, "OutputDebugString32"},
     {0x28, nullptr, "Unknown"},
     {0x29, SvcWrap32<GetInfo32>, "GetInfo32"},
     {0x2a, nullptr, "Unknown"},
     {0x2b, nullptr, "Unknown"},
-    {0x2c, nullptr, "MapPhysicalMemory32"},
-    {0x2d, nullptr, "UnmapPhysicalMemory32"},
+    {0x2c, SvcWrap32<MapPhysicalMemory32>, "MapPhysicalMemory32"},
+    {0x2d, SvcWrap32<UnmapPhysicalMemory32>, "UnmapPhysicalMemory32"},
     {0x2e, nullptr, "Unknown"},
     {0x2f, nullptr, "Unknown"},
     {0x30, nullptr, "Unknown"},
     {0x31, nullptr, "Unknown"},
-    {0x32, nullptr, "SetThreadActivity32"},
-    {0x33, nullptr, "GetThreadContext32"},
-    {0x34, nullptr, "WaitForAddress32"},
-    {0x35, nullptr, "SignalToAddress32"},
+    {0x32, SvcWrap32<SetThreadActivity32>, "SetThreadActivity32"},
+    {0x33, SvcWrap32<GetThreadContext32>, "GetThreadContext32"},
+    {0x34, SvcWrap32<WaitForAddress32>, "WaitForAddress32"},
+    {0x35, SvcWrap32<SignalToAddress32>, "SignalToAddress32"},
     {0x36, nullptr, "Unknown"},
     {0x37, nullptr, "Unknown"},
     {0x38, nullptr, "Unknown"},
@@ -2219,7 +2433,7 @@ static const FunctionDef SVC_Table_32[] = {
     {0x42, nullptr, "Unknown"},
     {0x43, nullptr, "ReplyAndReceive32"},
     {0x44, nullptr, "Unknown"},
-    {0x45, nullptr, "CreateEvent32"},
+    {0x45, SvcWrap32<CreateEvent32>, "CreateEvent32"},
     {0x46, nullptr, "Unknown"},
     {0x47, nullptr, "Unknown"},
     {0x48, nullptr, "Unknown"},
@@ -2245,7 +2459,7 @@ static const FunctionDef SVC_Table_32[] = {
     {0x5c, nullptr, "Unknown"},
     {0x5d, nullptr, "Unknown"},
     {0x5e, nullptr, "Unknown"},
-    {0x5F, nullptr, "FlushProcessDataCache32"},
+    {0x5F, SvcWrap32<FlushProcessDataCache32>, "FlushProcessDataCache32"},
     {0x60, nullptr, "Unknown"},
     {0x61, nullptr, "Unknown"},
     {0x62, nullptr, "Unknown"},
@@ -2423,13 +2637,10 @@ static const FunctionDef* GetSVCInfo64(u32 func_num) {
     return &SVC_Table_64[func_num];
 }
 
-MICROPROFILE_DEFINE(Kernel_SVC, "Kernel", "SVC", MP_RGB(70, 200, 70));
-
 void Call(Core::System& system, u32 immediate) {
-    MICROPROFILE_SCOPE(Kernel_SVC);
-
-    // Lock the global kernel mutex when we enter the kernel HLE.
-    std::lock_guard lock{HLE::g_hle_lock};
+    system.ExitDynarmicProfile();
+    auto& kernel = system.Kernel();
+    kernel.EnterSVCProfile();
 
     const FunctionDef* info = system.CurrentProcess()->Is64BitProcess() ? GetSVCInfo64(immediate)
                                                                         : GetSVCInfo32(immediate);
@@ -2442,6 +2653,9 @@ void Call(Core::System& system, u32 immediate) {
     } else {
         LOG_CRITICAL(Kernel_SVC, "Unknown SVC function 0x{:X}", immediate);
     }
+
+    kernel.ExitSVCProfile();
+    system.EnterDynarmicProfile();
 }
 
 } // namespace Kernel::Svc
diff --git a/src/core/hle/kernel/svc_wrap.h b/src/core/hle/kernel/svc_wrap.h
index 7d735e3fa..0b6dd9df0 100644
--- a/src/core/hle/kernel/svc_wrap.h
+++ b/src/core/hle/kernel/svc_wrap.h
@@ -350,13 +350,50 @@ void SvcWrap64(Core::System& system) {
     func(system, static_cast<u32>(Param(system, 0)), Param(system, 1), Param(system, 2));
 }
 
-// Used by QueryMemory32
+// Used by QueryMemory32, ArbitrateLock32
 template <ResultCode func(Core::System&, u32, u32, u32)>
 void SvcWrap32(Core::System& system) {
     FuncReturn32(system,
                  func(system, Param32(system, 0), Param32(system, 1), Param32(system, 2)).raw);
 }
 
+// Used by Break32
+template <void func(Core::System&, u32, u32, u32)>
+void SvcWrap32(Core::System& system) {
+    func(system, Param32(system, 0), Param32(system, 1), Param32(system, 2));
+}
+
+// Used by ExitProcess32, ExitThread32
+template <void func(Core::System&)>
+void SvcWrap32(Core::System& system) {
+    func(system);
+}
+
+// Used by GetCurrentProcessorNumber32
+template <u32 func(Core::System&)>
+void SvcWrap32(Core::System& system) {
+    FuncReturn32(system, func(system));
+}
+
+// Used by SleepThread32
+template <void func(Core::System&, u32, u32)>
+void SvcWrap32(Core::System& system) {
+    func(system, Param32(system, 0), Param32(system, 1));
+}
+
+// Used by CreateThread32
+template <ResultCode func(Core::System&, Handle*, u32, u32, u32, u32, s32)>
+void SvcWrap32(Core::System& system) {
+    Handle param_1 = 0;
+
+    const u32 retval = func(system, &param_1, Param32(system, 0), Param32(system, 1),
+                            Param32(system, 2), Param32(system, 3), Param32(system, 4))
+                           .raw;
+
+    system.CurrentArmInterface().SetReg(1, param_1);
+    FuncReturn(system, retval);
+}
+
 // Used by GetInfo32
 template <ResultCode func(Core::System&, u32*, u32*, u32, u32, u32, u32)>
 void SvcWrap32(Core::System& system) {
@@ -393,18 +430,114 @@ void SvcWrap32(Core::System& system) {
     FuncReturn(system, retval);
 }
 
+// Used by GetSystemTick32
+template <void func(Core::System&, u32*, u32*)>
+void SvcWrap32(Core::System& system) {
+    u32 param_1 = 0;
+    u32 param_2 = 0;
+
+    func(system, &param_1, &param_2);
+    system.CurrentArmInterface().SetReg(0, param_1);
+    system.CurrentArmInterface().SetReg(1, param_2);
+}
+
+// Used by CreateEvent32
+template <ResultCode func(Core::System&, Handle*, Handle*)>
+void SvcWrap32(Core::System& system) {
+    Handle param_1 = 0;
+    Handle param_2 = 0;
+
+    const u32 retval = func(system, &param_1, &param_2).raw;
+    system.CurrentArmInterface().SetReg(1, param_1);
+    system.CurrentArmInterface().SetReg(2, param_2);
+    FuncReturn(system, retval);
+}
+
+// Used by GetThreadId32
+template <ResultCode func(Core::System&, Handle, u32*, u32*, u32*)>
+void SvcWrap32(Core::System& system) {
+    u32 param_1 = 0;
+    u32 param_2 = 0;
+    u32 param_3 = 0;
+
+    const u32 retval = func(system, Param32(system, 2), &param_1, &param_2, &param_3).raw;
+    system.CurrentArmInterface().SetReg(1, param_1);
+    system.CurrentArmInterface().SetReg(2, param_2);
+    system.CurrentArmInterface().SetReg(3, param_3);
+    FuncReturn(system, retval);
+}
+
 // Used by SignalProcessWideKey32
 template <void func(Core::System&, u32, s32)>
 void SvcWrap32(Core::System& system) {
     func(system, static_cast<u32>(Param(system, 0)), static_cast<s32>(Param(system, 1)));
 }
 
-// Used by SendSyncRequest32
+// Used by SetThreadPriority32
+template <ResultCode func(Core::System&, Handle, u32)>
+void SvcWrap32(Core::System& system) {
+    const u32 retval =
+        func(system, static_cast<Handle>(Param(system, 0)), static_cast<u32>(Param(system, 1))).raw;
+    FuncReturn(system, retval);
+}
+
+// Used by SetThreadCoreMask32
+template <ResultCode func(Core::System&, Handle, u32, u32, u32)>
+void SvcWrap32(Core::System& system) {
+    const u32 retval =
+        func(system, static_cast<Handle>(Param(system, 0)), static_cast<u32>(Param(system, 1)),
+             static_cast<u32>(Param(system, 2)), static_cast<u32>(Param(system, 3)))
+            .raw;
+    FuncReturn(system, retval);
+}
+
+// Used by WaitProcessWideKeyAtomic32
+template <ResultCode func(Core::System&, u32, u32, Handle, u32, u32)>
+void SvcWrap32(Core::System& system) {
+    const u32 retval =
+        func(system, static_cast<u32>(Param(system, 0)), static_cast<u32>(Param(system, 1)),
+             static_cast<Handle>(Param(system, 2)), static_cast<u32>(Param(system, 3)),
+             static_cast<u32>(Param(system, 4)))
+            .raw;
+    FuncReturn(system, retval);
+}
+
+// Used by WaitForAddress32
+template <ResultCode func(Core::System&, u32, u32, s32, u32, u32)>
+void SvcWrap32(Core::System& system) {
+    const u32 retval = func(system, static_cast<u32>(Param(system, 0)),
+                            static_cast<u32>(Param(system, 1)), static_cast<s32>(Param(system, 2)),
+                            static_cast<u32>(Param(system, 3)), static_cast<u32>(Param(system, 4)))
+                           .raw;
+    FuncReturn(system, retval);
+}
+
+// Used by SignalToAddress32
+template <ResultCode func(Core::System&, u32, u32, s32, s32)>
+void SvcWrap32(Core::System& system) {
+    const u32 retval =
+        func(system, static_cast<u32>(Param(system, 0)), static_cast<u32>(Param(system, 1)),
+             static_cast<s32>(Param(system, 2)), static_cast<s32>(Param(system, 3)))
+            .raw;
+    FuncReturn(system, retval);
+}
+
+// Used by SendSyncRequest32, ArbitrateUnlock32
 template <ResultCode func(Core::System&, u32)>
 void SvcWrap32(Core::System& system) {
     FuncReturn(system, func(system, static_cast<u32>(Param(system, 0))).raw);
 }
 
+// Used by CreateTransferMemory32
+template <ResultCode func(Core::System&, Handle*, u32, u32, u32)>
+void SvcWrap32(Core::System& system) {
+    Handle handle = 0;
+    const u32 retval =
+        func(system, &handle, Param32(system, 1), Param32(system, 2), Param32(system, 3)).raw;
+    system.CurrentArmInterface().SetReg(1, handle);
+    FuncReturn(system, retval);
+}
+
 // Used by WaitSynchronization32
 template <ResultCode func(Core::System&, u32, u32, s32, u32, Handle*)>
 void SvcWrap32(Core::System& system) {
diff --git a/src/core/hle/kernel/synchronization.cpp b/src/core/hle/kernel/synchronization.cpp
index dc37fad1a..851b702a5 100644
--- a/src/core/hle/kernel/synchronization.cpp
+++ b/src/core/hle/kernel/synchronization.cpp
@@ -10,78 +10,107 @@
 #include "core/hle/kernel/synchronization.h"
 #include "core/hle/kernel/synchronization_object.h"
 #include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
 
 namespace Kernel {
 
-/// Default thread wakeup callback for WaitSynchronization
-static bool DefaultThreadWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
-                                        std::shared_ptr<SynchronizationObject> object,
-                                        std::size_t index) {
-    ASSERT(thread->GetStatus() == ThreadStatus::WaitSynch);
-
-    if (reason == ThreadWakeupReason::Timeout) {
-        thread->SetWaitSynchronizationResult(RESULT_TIMEOUT);
-        return true;
-    }
-
-    ASSERT(reason == ThreadWakeupReason::Signal);
-    thread->SetWaitSynchronizationResult(RESULT_SUCCESS);
-    thread->SetWaitSynchronizationOutput(static_cast<u32>(index));
-    return true;
-}
-
 Synchronization::Synchronization(Core::System& system) : system{system} {}
 
 void Synchronization::SignalObject(SynchronizationObject& obj) const {
+    auto& kernel = system.Kernel();
+    SchedulerLock lock(kernel);
+    auto& time_manager = kernel.TimeManager();
     if (obj.IsSignaled()) {
-        obj.WakeupAllWaitingThreads();
+        for (auto thread : obj.GetWaitingThreads()) {
+            if (thread->GetSchedulingStatus() == ThreadSchedStatus::Paused) {
+                if (thread->GetStatus() != ThreadStatus::WaitHLEEvent) {
+                    ASSERT(thread->GetStatus() == ThreadStatus::WaitSynch);
+                    ASSERT(thread->IsWaitingSync());
+                }
+                thread->SetSynchronizationResults(&obj, RESULT_SUCCESS);
+                thread->ResumeFromWait();
+            }
+        }
+        obj.ClearWaitingThreads();
     }
 }
 
 std::pair<ResultCode, Handle> Synchronization::WaitFor(
     std::vector<std::shared_ptr<SynchronizationObject>>& sync_objects, s64 nano_seconds) {
+    auto& kernel = system.Kernel();
     auto* const thread = system.CurrentScheduler().GetCurrentThread();
-    // Find the first object that is acquirable in the provided list of objects
-    const auto itr = std::find_if(sync_objects.begin(), sync_objects.end(),
-                                  [thread](const std::shared_ptr<SynchronizationObject>& object) {
-                                      return object->IsSignaled();
-                                  });
-
-    if (itr != sync_objects.end()) {
-        // We found a ready object, acquire it and set the result value
-        SynchronizationObject* object = itr->get();
-        object->Acquire(thread);
-        const u32 index = static_cast<s32>(std::distance(sync_objects.begin(), itr));
-        return {RESULT_SUCCESS, index};
+    Handle event_handle = InvalidHandle;
+    {
+        SchedulerLockAndSleep lock(kernel, event_handle, thread, nano_seconds);
+        const auto itr =
+            std::find_if(sync_objects.begin(), sync_objects.end(),
+                         [thread](const std::shared_ptr<SynchronizationObject>& object) {
+                             return object->IsSignaled();
+                         });
+
+        if (itr != sync_objects.end()) {
+            // We found a ready object, acquire it and set the result value
+            SynchronizationObject* object = itr->get();
+            object->Acquire(thread);
+            const u32 index = static_cast<s32>(std::distance(sync_objects.begin(), itr));
+            lock.CancelSleep();
+            return {RESULT_SUCCESS, index};
+        }
+
+        if (nano_seconds == 0) {
+            lock.CancelSleep();
+            return {RESULT_TIMEOUT, InvalidHandle};
+        }
+
+        if (thread->IsPendingTermination()) {
+            lock.CancelSleep();
+            return {ERR_THREAD_TERMINATING, InvalidHandle};
+        }
+
+        if (thread->IsSyncCancelled()) {
+            thread->SetSyncCancelled(false);
+            lock.CancelSleep();
+            return {ERR_SYNCHRONIZATION_CANCELED, InvalidHandle};
+        }
+
+        for (auto& object : sync_objects) {
+            object->AddWaitingThread(SharedFrom(thread));
+        }
+
+        thread->SetSynchronizationObjects(&sync_objects);
+        thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT);
+        thread->SetStatus(ThreadStatus::WaitSynch);
+        thread->SetWaitingSync(true);
     }
+    thread->SetWaitingSync(false);
 
-    // No objects were ready to be acquired, prepare to suspend the thread.
-
-    // If a timeout value of 0 was provided, just return the Timeout error code instead of
-    // suspending the thread.
-    if (nano_seconds == 0) {
-        return {RESULT_TIMEOUT, InvalidHandle};
+    if (event_handle != InvalidHandle) {
+        auto& time_manager = kernel.TimeManager();
+        time_manager.UnscheduleTimeEvent(event_handle);
     }
 
-    if (thread->IsSyncCancelled()) {
-        thread->SetSyncCancelled(false);
-        return {ERR_SYNCHRONIZATION_CANCELED, InvalidHandle};
+    {
+        SchedulerLock lock(kernel);
+        ResultCode signaling_result = thread->GetSignalingResult();
+        SynchronizationObject* signaling_object = thread->GetSignalingObject();
+        thread->SetSynchronizationObjects(nullptr);
+        auto shared_thread = SharedFrom(thread);
+        for (auto& obj : sync_objects) {
+            obj->RemoveWaitingThread(shared_thread);
+        }
+        if (signaling_object != nullptr) {
+            const auto itr = std::find_if(
+                sync_objects.begin(), sync_objects.end(),
+                [signaling_object](const std::shared_ptr<SynchronizationObject>& object) {
+                    return object.get() == signaling_object;
+                });
+            ASSERT(itr != sync_objects.end());
+            signaling_object->Acquire(thread);
+            const u32 index = static_cast<s32>(std::distance(sync_objects.begin(), itr));
+            return {signaling_result, index};
+        }
+        return {signaling_result, -1};
     }
-
-    for (auto& object : sync_objects) {
-        object->AddWaitingThread(SharedFrom(thread));
-    }
-
-    thread->SetSynchronizationObjects(std::move(sync_objects));
-    thread->SetStatus(ThreadStatus::WaitSynch);
-
-    // Create an event to wake the thread up after the specified nanosecond delay has passed
-    thread->WakeAfterDelay(nano_seconds);
-    thread->SetWakeupCallback(DefaultThreadWakeupCallback);
-
-    system.PrepareReschedule(thread->GetProcessorID());
-
-    return {RESULT_TIMEOUT, InvalidHandle};
 }
 
 } // namespace Kernel
diff --git a/src/core/hle/kernel/synchronization_object.cpp b/src/core/hle/kernel/synchronization_object.cpp
index 43f3eef18..ba4d39157 100644
--- a/src/core/hle/kernel/synchronization_object.cpp
+++ b/src/core/hle/kernel/synchronization_object.cpp
@@ -38,68 +38,8 @@ void SynchronizationObject::RemoveWaitingThread(std::shared_ptr<Thread> thread)
         waiting_threads.erase(itr);
 }
 
-std::shared_ptr<Thread> SynchronizationObject::GetHighestPriorityReadyThread() const {
-    Thread* candidate = nullptr;
-    u32 candidate_priority = THREADPRIO_LOWEST + 1;
-
-    for (const auto& thread : waiting_threads) {
-        const ThreadStatus thread_status = thread->GetStatus();
-
-        // The list of waiting threads must not contain threads that are not waiting to be awakened.
-        ASSERT_MSG(thread_status == ThreadStatus::WaitSynch ||
-                       thread_status == ThreadStatus::WaitHLEEvent,
-                   "Inconsistent thread statuses in waiting_threads");
-
-        if (thread->GetPriority() >= candidate_priority)
-            continue;
-
-        if (ShouldWait(thread.get()))
-            continue;
-
-        candidate = thread.get();
-        candidate_priority = thread->GetPriority();
-    }
-
-    return SharedFrom(candidate);
-}
-
-void SynchronizationObject::WakeupWaitingThread(std::shared_ptr<Thread> thread) {
-    ASSERT(!ShouldWait(thread.get()));
-
-    if (!thread) {
-        return;
-    }
-
-    if (thread->IsSleepingOnWait()) {
-        for (const auto& object : thread->GetSynchronizationObjects()) {
-            ASSERT(!object->ShouldWait(thread.get()));
-            object->Acquire(thread.get());
-        }
-    } else {
-        Acquire(thread.get());
-    }
-
-    const std::size_t index = thread->GetSynchronizationObjectIndex(SharedFrom(this));
-
-    thread->ClearSynchronizationObjects();
-
-    thread->CancelWakeupTimer();
-
-    bool resume = true;
-    if (thread->HasWakeupCallback()) {
-        resume = thread->InvokeWakeupCallback(ThreadWakeupReason::Signal, thread, SharedFrom(this),
-                                              index);
-    }
-    if (resume) {
-        thread->ResumeFromWait();
-        kernel.PrepareReschedule(thread->GetProcessorID());
-    }
-}
-
-void SynchronizationObject::WakeupAllWaitingThreads() {
-    while (auto thread = GetHighestPriorityReadyThread()) {
-        WakeupWaitingThread(thread);
-    }
+void SynchronizationObject::ClearWaitingThreads() {
+    waiting_threads.clear();
 }
 
 const std::vector<std::shared_ptr<Thread>>& SynchronizationObject::GetWaitingThreads() const {
diff --git a/src/core/hle/kernel/synchronization_object.h b/src/core/hle/kernel/synchronization_object.h
index 741c31faf..f89b24204 100644
--- a/src/core/hle/kernel/synchronization_object.h
+++ b/src/core/hle/kernel/synchronization_object.h
@@ -12,6 +12,7 @@
 namespace Kernel {
 
 class KernelCore;
+class Synchronization;
 class Thread;
 
 /// Class that represents a Kernel object that a thread can be waiting on
@@ -49,24 +50,11 @@ public:
      */
     void RemoveWaitingThread(std::shared_ptr<Thread> thread);
 
-    /**
-     * Wake up all threads waiting on this object that can be awoken, in priority order,
-     * and set the synchronization result and output of the thread.
-     */
-    void WakeupAllWaitingThreads();
-
-    /**
-     * Wakes up a single thread waiting on this object.
-     * @param thread Thread that is waiting on this object to wakeup.
-     */
-    void WakeupWaitingThread(std::shared_ptr<Thread> thread);
-
-    /// Obtains the highest priority thread that is ready to run from this object's waiting list.
-    std::shared_ptr<Thread> GetHighestPriorityReadyThread() const;
-
     /// Get a const reference to the waiting threads list for debug use
     const std::vector<std::shared_ptr<Thread>>& GetWaitingThreads() const;
 
+    void ClearWaitingThreads();
+
 protected:
     bool is_signaled{}; // Tells if this sync object is signalled;
 
diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp
index db7f379ac..2b1092697 100644
--- a/src/core/hle/kernel/thread.cpp
+++ b/src/core/hle/kernel/thread.cpp
@@ -9,12 +9,21 @@
 
 #include "common/assert.h"
 #include "common/common_types.h"
+#include "common/fiber.h"
 #include "common/logging/log.h"
 #include "common/thread_queue_list.h"
 #include "core/arm/arm_interface.h"
+#ifdef ARCHITECTURE_x86_64
+#include "core/arm/dynarmic/arm_dynarmic_32.h"
+#include "core/arm/dynarmic/arm_dynarmic_64.h"
+#endif
+#include "core/arm/cpu_interrupt_handler.h"
+#include "core/arm/exclusive_monitor.h"
+#include "core/arm/unicorn/arm_unicorn.h"
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
+#include "core/cpu_manager.h"
 #include "core/hardware_properties.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
@@ -23,6 +32,7 @@
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
 #include "core/hle/result.h"
 #include "core/memory.h"
 
@@ -44,46 +54,26 @@ Thread::Thread(KernelCore& kernel) : SynchronizationObject{kernel} {}
 Thread::~Thread() = default;
 
 void Thread::Stop() {
-    // Cancel any outstanding wakeup events for this thread
-    Core::System::GetInstance().CoreTiming().UnscheduleEvent(kernel.ThreadWakeupCallbackEventType(),
-                                                             global_handle);
-    kernel.GlobalHandleTable().Close(global_handle);
-    global_handle = 0;
-    SetStatus(ThreadStatus::Dead);
-    Signal();
-
-    // Clean up any dangling references in objects that this thread was waiting for
-    for (auto& wait_object : wait_objects) {
-        wait_object->RemoveWaitingThread(SharedFrom(this));
-    }
-    wait_objects.clear();
-
-    owner_process->UnregisterThread(this);
-
-    // Mark the TLS slot in the thread's page as free.
-    owner_process->FreeTLSRegion(tls_address);
-}
-
-void Thread::WakeAfterDelay(s64 nanoseconds) {
-    // Don't schedule a wakeup if the thread wants to wait forever
-    if (nanoseconds == -1)
-        return;
+    {
+        SchedulerLock lock(kernel);
+        SetStatus(ThreadStatus::Dead);
+        Signal();
+        kernel.GlobalHandleTable().Close(global_handle);
 
-    // This function might be called from any thread so we have to be cautious and use the
-    // thread-safe version of ScheduleEvent.
-    const s64 cycles = Core::Timing::nsToCycles(std::chrono::nanoseconds{nanoseconds});
-    Core::System::GetInstance().CoreTiming().ScheduleEvent(
-        cycles, kernel.ThreadWakeupCallbackEventType(), global_handle);
-}
+        if (owner_process) {
+            owner_process->UnregisterThread(this);
 
-void Thread::CancelWakeupTimer() {
-    Core::System::GetInstance().CoreTiming().UnscheduleEvent(kernel.ThreadWakeupCallbackEventType(),
-                                                             global_handle);
+            // Mark the TLS slot in the thread's page as free.
+            owner_process->FreeTLSRegion(tls_address);
+        }
+        arm_interface.reset();
+        has_exited = true;
+    }
+    global_handle = 0;
 }
 
 void Thread::ResumeFromWait() {
-    ASSERT_MSG(wait_objects.empty(), "Thread is waking up while waiting for objects");
-
+    SchedulerLock lock(kernel);
     switch (status) {
     case ThreadStatus::Paused:
     case ThreadStatus::WaitSynch:
@@ -99,7 +89,7 @@ void Thread::ResumeFromWait() {
     case ThreadStatus::Ready:
         // The thread's wakeup callback must have already been cleared when the thread was first
         // awoken.
-        ASSERT(wakeup_callback == nullptr);
+        ASSERT(hle_callback == nullptr);
         // If the thread is waiting on multiple wait objects, it might be awoken more than once
         // before actually resuming. We can ignore subsequent wakeups if the thread status has
         // already been set to ThreadStatus::Ready.
@@ -115,24 +105,31 @@ void Thread::ResumeFromWait() {
         return;
     }
 
-    wakeup_callback = nullptr;
+    SetStatus(ThreadStatus::Ready);
+}
+
+void Thread::OnWakeUp() {
+    SchedulerLock lock(kernel);
 
-    if (activity == ThreadActivity::Paused) {
-        SetStatus(ThreadStatus::Paused);
-        return;
-    }
+    SetStatus(ThreadStatus::Ready);
+}
 
+ResultCode Thread::Start() {
+    SchedulerLock lock(kernel);
     SetStatus(ThreadStatus::Ready);
+    return RESULT_SUCCESS;
 }
 
 void Thread::CancelWait() {
-    if (GetSchedulingStatus() != ThreadSchedStatus::Paused) {
+    SchedulerLock lock(kernel);
+    if (GetSchedulingStatus() != ThreadSchedStatus::Paused || !is_waiting_on_sync) {
         is_sync_cancelled = true;
         return;
     }
+    // TODO(Blinkhawk): Implement cancel of server session
     is_sync_cancelled = false;
-    SetWaitSynchronizationResult(ERR_SYNCHRONIZATION_CANCELED);
-    ResumeFromWait();
+    SetSynchronizationResults(nullptr, ERR_SYNCHRONIZATION_CANCELED);
+    SetStatus(ThreadStatus::Ready);
 }
 
 static void ResetThreadContext32(Core::ARM_Interface::ThreadContext32& context, u32 stack_top,
@@ -153,12 +150,29 @@ static void ResetThreadContext64(Core::ARM_Interface::ThreadContext64& context,
     context.fpcr = 0;
 }
 
-ResultVal<std::shared_ptr<Thread>> Thread::Create(KernelCore& kernel, std::string name,
-                                                  VAddr entry_point, u32 priority, u64 arg,
-                                                  s32 processor_id, VAddr stack_top,
-                                                  Process& owner_process) {
+std::shared_ptr<Common::Fiber>& Thread::GetHostContext() {
+    return host_context;
+}
+
+ResultVal<std::shared_ptr<Thread>> Thread::Create(Core::System& system, ThreadType type_flags,
+                                                  std::string name, VAddr entry_point, u32 priority,
+                                                  u64 arg, s32 processor_id, VAddr stack_top,
+                                                  Process* owner_process) {
+    std::function<void(void*)> init_func = system.GetCpuManager().GetGuestThreadStartFunc();
+    void* init_func_parameter = system.GetCpuManager().GetStartFuncParamater();
+    return Create(system, type_flags, name, entry_point, priority, arg, processor_id, stack_top,
+                  owner_process, std::move(init_func), init_func_parameter);
+}
+
+ResultVal<std::shared_ptr<Thread>> Thread::Create(Core::System& system, ThreadType type_flags,
+                                                  std::string name, VAddr entry_point, u32 priority,
+                                                  u64 arg, s32 processor_id, VAddr stack_top,
+                                                  Process* owner_process,
+                                                  std::function<void(void*)>&& thread_start_func,
+                                                  void* thread_start_parameter) {
+    auto& kernel = system.Kernel();
     // Check if priority is in ranged. Lowest priority -> highest priority id.
-    if (priority > THREADPRIO_LOWEST) {
+    if (priority > THREADPRIO_LOWEST && ((type_flags & THREADTYPE_IDLE) == 0)) {
         LOG_ERROR(Kernel_SVC, "Invalid thread priority: {}", priority);
         return ERR_INVALID_THREAD_PRIORITY;
     }
@@ -168,11 +182,12 @@ ResultVal<std::shared_ptr<Thread>> Thread::Create(KernelCore& kernel, std::strin
         return ERR_INVALID_PROCESSOR_ID;
     }
 
-    auto& system = Core::System::GetInstance();
-    if (!system.Memory().IsValidVirtualAddress(owner_process, entry_point)) {
-        LOG_ERROR(Kernel_SVC, "(name={}): invalid entry {:016X}", name, entry_point);
-        // TODO (bunnei): Find the correct error code to use here
-        return RESULT_UNKNOWN;
+    if (owner_process) {
+        if (!system.Memory().IsValidVirtualAddress(*owner_process, entry_point)) {
+            LOG_ERROR(Kernel_SVC, "(name={}): invalid entry {:016X}", name, entry_point);
+            // TODO (bunnei): Find the correct error code to use here
+            return RESULT_UNKNOWN;
+        }
     }
 
     std::shared_ptr<Thread> thread = std::make_shared<Thread>(kernel);
@@ -183,51 +198,82 @@ ResultVal<std::shared_ptr<Thread>> Thread::Create(KernelCore& kernel, std::strin
     thread->stack_top = stack_top;
     thread->tpidr_el0 = 0;
     thread->nominal_priority = thread->current_priority = priority;
-    thread->last_running_ticks = system.CoreTiming().GetTicks();
+    thread->last_running_ticks = 0;
     thread->processor_id = processor_id;
     thread->ideal_core = processor_id;
     thread->affinity_mask = 1ULL << processor_id;
-    thread->wait_objects.clear();
+    thread->wait_objects = nullptr;
     thread->mutex_wait_address = 0;
     thread->condvar_wait_address = 0;
     thread->wait_handle = 0;
     thread->name = std::move(name);
     thread->global_handle = kernel.GlobalHandleTable().Create(thread).Unwrap();
-    thread->owner_process = &owner_process;
-    auto& scheduler = kernel.GlobalScheduler();
-    scheduler.AddThread(thread);
-    thread->tls_address = thread->owner_process->CreateTLSRegion();
-
-    thread->owner_process->RegisterThread(thread.get());
+    thread->owner_process = owner_process;
+    thread->type = type_flags;
+    if ((type_flags & THREADTYPE_IDLE) == 0) {
+        auto& scheduler = kernel.GlobalScheduler();
+        scheduler.AddThread(thread);
+    }
+    if (owner_process) {
+        thread->tls_address = thread->owner_process->CreateTLSRegion();
+        thread->owner_process->RegisterThread(thread.get());
+    } else {
+        thread->tls_address = 0;
+    }
+    // TODO(peachum): move to ScheduleThread() when scheduler is added so selected core is used
+    // to initialize the context
+    thread->arm_interface.reset();
+    if ((type_flags & THREADTYPE_HLE) == 0) {
+#ifdef ARCHITECTURE_x86_64
+        if (owner_process && !owner_process->Is64BitProcess()) {
+            thread->arm_interface = std::make_unique<Core::ARM_Dynarmic_32>(
+                system, kernel.Interrupts(), kernel.IsMulticore(), kernel.GetExclusiveMonitor(),
+                processor_id);
+        } else {
+            thread->arm_interface = std::make_unique<Core::ARM_Dynarmic_64>(
+                system, kernel.Interrupts(), kernel.IsMulticore(), kernel.GetExclusiveMonitor(),
+                processor_id);
+        }
 
-    ResetThreadContext32(thread->context_32, static_cast<u32>(stack_top),
-                         static_cast<u32>(entry_point), static_cast<u32>(arg));
-    ResetThreadContext64(thread->context_64, stack_top, entry_point, arg);
+#else
+        if (owner_process && !owner_process->Is64BitProcess()) {
+            thread->arm_interface = std::make_shared<Core::ARM_Unicorn>(
+                system, kernel.Interrupts(), kernel.IsMulticore(), ARM_Unicorn::Arch::AArch32,
+                processor_id);
+        } else {
+            thread->arm_interface = std::make_shared<Core::ARM_Unicorn>(
+                system, kernel.Interrupts(), kernel.IsMulticore(), ARM_Unicorn::Arch::AArch64,
+                processor_id);
+        }
+        LOG_WARNING(Core, "CPU JIT requested, but Dynarmic not available");
+#endif
+        ResetThreadContext32(thread->context_32, static_cast<u32>(stack_top),
+                             static_cast<u32>(entry_point), static_cast<u32>(arg));
+        ResetThreadContext64(thread->context_64, stack_top, entry_point, arg);
+    }
+    thread->host_context =
+        std::make_shared<Common::Fiber>(std::move(thread_start_func), thread_start_parameter);
 
     return MakeResult<std::shared_ptr<Thread>>(std::move(thread));
 }
 
 void Thread::SetPriority(u32 priority) {
+    SchedulerLock lock(kernel);
     ASSERT_MSG(priority <= THREADPRIO_LOWEST && priority >= THREADPRIO_HIGHEST,
                "Invalid priority value.");
     nominal_priority = priority;
     UpdatePriority();
 }
 
-void Thread::SetWaitSynchronizationResult(ResultCode result) {
-    context_32.cpu_registers[0] = result.raw;
-    context_64.cpu_registers[0] = result.raw;
-}
-
-void Thread::SetWaitSynchronizationOutput(s32 output) {
-    context_32.cpu_registers[1] = output;
-    context_64.cpu_registers[1] = output;
+void Thread::SetSynchronizationResults(SynchronizationObject* object, ResultCode result) {
+    signaling_object = object;
+    signaling_result = result;
 }
 
 s32 Thread::GetSynchronizationObjectIndex(std::shared_ptr<SynchronizationObject> object) const {
-    ASSERT_MSG(!wait_objects.empty(), "Thread is not waiting for anything");
-    const auto match = std::find(wait_objects.rbegin(), wait_objects.rend(), object);
-    return static_cast<s32>(std::distance(match, wait_objects.rend()) - 1);
+    ASSERT_MSG(!wait_objects->empty(), "Thread is not waiting for anything");
+    const auto match = std::find(wait_objects->rbegin(), wait_objects->rend(), object);
+    return static_cast<s32>(std::distance(match, wait_objects->rend()) - 1);
 }
 
 VAddr Thread::GetCommandBufferAddress() const {
@@ -236,6 +282,14 @@ VAddr Thread::GetCommandBufferAddress() const {
     return GetTLSAddress() + command_header_offset;
 }
 
+Core::ARM_Interface& Thread::ArmInterface() {
+    return *arm_interface;
+}
+
+const Core::ARM_Interface& Thread::ArmInterface() const {
+    return *arm_interface;
+}
+
 void Thread::SetStatus(ThreadStatus new_status) {
     if (new_status == status) {
         return;
@@ -257,10 +311,6 @@ void Thread::SetStatus(ThreadStatus new_status) {
         break;
     }
 
-    if (status == ThreadStatus::Running) {
-        last_running_ticks = Core::System::GetInstance().CoreTiming().GetTicks();
-    }
-
     status = new_status;
 }
 
@@ -341,75 +391,116 @@ void Thread::UpdatePriority() {
     lock_owner->UpdatePriority();
 }
 
-void Thread::ChangeCore(u32 core, u64 mask) {
-    SetCoreAndAffinityMask(core, mask);
-}
-
 bool Thread::AllSynchronizationObjectsReady() const {
-    return std::none_of(wait_objects.begin(), wait_objects.end(),
+    return std::none_of(wait_objects->begin(), wait_objects->end(),
                         [this](const std::shared_ptr<SynchronizationObject>& object) {
                             return object->ShouldWait(this);
                         });
 }
 
-bool Thread::InvokeWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
-                                  std::shared_ptr<SynchronizationObject> object,
-                                  std::size_t index) {
-    ASSERT(wakeup_callback);
-    return wakeup_callback(reason, std::move(thread), std::move(object), index);
+bool Thread::InvokeHLECallback(std::shared_ptr<Thread> thread) {
+    ASSERT(hle_callback);
+    return hle_callback(std::move(thread));
 }
 
-void Thread::SetActivity(ThreadActivity value) {
-    activity = value;
+ResultCode Thread::SetActivity(ThreadActivity value) {
+    SchedulerLock lock(kernel);
+
+    auto sched_status = GetSchedulingStatus();
+
+    if (sched_status != ThreadSchedStatus::Runnable && sched_status != ThreadSchedStatus::Paused) {
+        return ERR_INVALID_STATE;
+    }
+
+    if (IsPendingTermination()) {
+        return RESULT_SUCCESS;
+    }
 
     if (value == ThreadActivity::Paused) {
-        // Set status if not waiting
-        if (status == ThreadStatus::Ready || status == ThreadStatus::Running) {
-            SetStatus(ThreadStatus::Paused);
-            kernel.PrepareReschedule(processor_id);
+        if ((pausing_state & static_cast<u32>(ThreadSchedFlags::ThreadPauseFlag)) != 0) {
+            return ERR_INVALID_STATE;
+        }
+        AddSchedulingFlag(ThreadSchedFlags::ThreadPauseFlag);
+    } else {
+        if ((pausing_state & static_cast<u32>(ThreadSchedFlags::ThreadPauseFlag)) == 0) {
+            return ERR_INVALID_STATE;
         }
-    } else if (status == ThreadStatus::Paused) {
-        // Ready to reschedule
-        ResumeFromWait();
+        RemoveSchedulingFlag(ThreadSchedFlags::ThreadPauseFlag);
     }
+    return RESULT_SUCCESS;
 }
 
-void Thread::Sleep(s64 nanoseconds) {
-    // Sleep current thread and check for next thread to schedule
-    SetStatus(ThreadStatus::WaitSleep);
+ResultCode Thread::Sleep(s64 nanoseconds) {
+    Handle event_handle{};
+    {
+        SchedulerLockAndSleep lock(kernel, event_handle, this, nanoseconds);
+        SetStatus(ThreadStatus::WaitSleep);
+    }
 
-    // Create an event to wake the thread up after the specified nanosecond delay has passed
-    WakeAfterDelay(nanoseconds);
+    if (event_handle != InvalidHandle) {
+        auto& time_manager = kernel.TimeManager();
+        time_manager.UnscheduleTimeEvent(event_handle);
+    }
+    return RESULT_SUCCESS;
+}
+
+std::pair<ResultCode, bool> Thread::YieldSimple() {
+    bool is_redundant = false;
+    {
+        SchedulerLock lock(kernel);
+        is_redundant = kernel.GlobalScheduler().YieldThread(this);
+    }
+    return {RESULT_SUCCESS, is_redundant};
+}
+
+std::pair<ResultCode, bool> Thread::YieldAndBalanceLoad() {
+    bool is_redundant = false;
+    {
+        SchedulerLock lock(kernel);
+        is_redundant = kernel.GlobalScheduler().YieldThreadAndBalanceLoad(this);
+    }
+    return {RESULT_SUCCESS, is_redundant};
 }
 
-bool Thread::YieldSimple() {
-    auto& scheduler = kernel.GlobalScheduler();
-    return scheduler.YieldThread(this);
+std::pair<ResultCode, bool> Thread::YieldAndWaitForLoadBalancing() {
+    bool is_redundant = false;
+    {
+        SchedulerLock lock(kernel);
+        is_redundant = kernel.GlobalScheduler().YieldThreadAndWaitForLoadBalancing(this);
+    }
+    return {RESULT_SUCCESS, is_redundant};
 }
 
-bool Thread::YieldAndBalanceLoad() {
-    auto& scheduler = kernel.GlobalScheduler();
-    return scheduler.YieldThreadAndBalanceLoad(this);
+void Thread::AddSchedulingFlag(ThreadSchedFlags flag) {
+    const u32 old_state = scheduling_state;
+    pausing_state |= static_cast<u32>(flag);
+    const u32 base_scheduling = static_cast<u32>(GetSchedulingStatus());
+    scheduling_state = base_scheduling | pausing_state;
+    kernel.GlobalScheduler().AdjustSchedulingOnStatus(this, old_state);
 }
 
-bool Thread::YieldAndWaitForLoadBalancing() {
-    auto& scheduler = kernel.GlobalScheduler();
-    return scheduler.YieldThreadAndWaitForLoadBalancing(this);
+void Thread::RemoveSchedulingFlag(ThreadSchedFlags flag) {
+    const u32 old_state = scheduling_state;
+    pausing_state &= ~static_cast<u32>(flag);
+    const u32 base_scheduling = static_cast<u32>(GetSchedulingStatus());
+    scheduling_state = base_scheduling | pausing_state;
+    kernel.GlobalScheduler().AdjustSchedulingOnStatus(this, old_state);
 }
 
 void Thread::SetSchedulingStatus(ThreadSchedStatus new_status) {
-    const u32 old_flags = scheduling_state;
+    const u32 old_state = scheduling_state;
     scheduling_state = (scheduling_state & static_cast<u32>(ThreadSchedMasks::HighMask)) |
                        static_cast<u32>(new_status);
-    AdjustSchedulingOnStatus(old_flags);
+    kernel.GlobalScheduler().AdjustSchedulingOnStatus(this, old_state);
 }
 
 void Thread::SetCurrentPriority(u32 new_priority) {
     const u32 old_priority = std::exchange(current_priority, new_priority);
-    AdjustSchedulingOnPriority(old_priority);
+    kernel.GlobalScheduler().AdjustSchedulingOnPriority(this, old_priority);
 }
 
 ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) {
+    SchedulerLock lock(kernel);
     const auto HighestSetCore = [](u64 mask, u32 max_cores) {
         for (s32 core = static_cast<s32>(max_cores - 1); core >= 0; core--) {
             if (((mask >> core) & 1) != 0) {
@@ -443,111 +534,12 @@ ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) {
                     processor_id = ideal_core;
                 }
             }
-            AdjustSchedulingOnAffinity(old_affinity_mask, old_core);
+            kernel.GlobalScheduler().AdjustSchedulingOnAffinity(this, old_affinity_mask, old_core);
         }
     }
     return RESULT_SUCCESS;
 }
 
-void Thread::AdjustSchedulingOnStatus(u32 old_flags) {
-    if (old_flags == scheduling_state) {
-        return;
-    }
-
-    auto& scheduler = kernel.GlobalScheduler();
-    if (static_cast<ThreadSchedStatus>(old_flags & static_cast<u32>(ThreadSchedMasks::LowMask)) ==
-        ThreadSchedStatus::Runnable) {
-        // In this case the thread was running, now it's pausing/exitting
-        if (processor_id >= 0) {
-            scheduler.Unschedule(current_priority, static_cast<u32>(processor_id), this);
-        }
-
-        for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
-            if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
-                scheduler.Unsuggest(current_priority, core, this);
-            }
-        }
-    } else if (GetSchedulingStatus() == ThreadSchedStatus::Runnable) {
-        // The thread is now set to running from being stopped
-        if (processor_id >= 0) {
-            scheduler.Schedule(current_priority, static_cast<u32>(processor_id), this);
-        }
-
-        for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
-            if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
-                scheduler.Suggest(current_priority, core, this);
-            }
-        }
-    }
-
-    scheduler.SetReselectionPending();
-}
-
-void Thread::AdjustSchedulingOnPriority(u32 old_priority) {
-    if (GetSchedulingStatus() != ThreadSchedStatus::Runnable) {
-        return;
-    }
-    auto& scheduler = kernel.GlobalScheduler();
-    if (processor_id >= 0) {
-        scheduler.Unschedule(old_priority, static_cast<u32>(processor_id), this);
-    }
-
-    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
-        if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
-            scheduler.Unsuggest(old_priority, core, this);
-        }
-    }
-
-    // Add thread to the new priority queues.
-    Thread* current_thread = GetCurrentThread();
-
-    if (processor_id >= 0) {
-        if (current_thread == this) {
-            scheduler.SchedulePrepend(current_priority, static_cast<u32>(processor_id), this);
-        } else {
-            scheduler.Schedule(current_priority, static_cast<u32>(processor_id), this);
-        }
-    }
-
-    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
-        if (core != static_cast<u32>(processor_id) && ((affinity_mask >> core) & 1) != 0) {
-            scheduler.Suggest(current_priority, core, this);
-        }
-    }
-
-    scheduler.SetReselectionPending();
-}
-
-void Thread::AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core) {
-    auto& scheduler = kernel.GlobalScheduler();
-    if (GetSchedulingStatus() != ThreadSchedStatus::Runnable ||
-        current_priority >= THREADPRIO_COUNT) {
-        return;
-    }
-
-    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
-        if (((old_affinity_mask >> core) & 1) != 0) {
-            if (core == static_cast<u32>(old_core)) {
-                scheduler.Unschedule(current_priority, core, this);
-            } else {
-                scheduler.Unsuggest(current_priority, core, this);
-            }
-        }
-    }
-
-    for (u32 core = 0; core < Core::Hardware::NUM_CPU_CORES; core++) {
-        if (((affinity_mask >> core) & 1) != 0) {
-            if (core == static_cast<u32>(processor_id)) {
-                scheduler.Schedule(current_priority, core, this);
-            } else {
-                scheduler.Suggest(current_priority, core, this);
-            }
-        }
-    }
-
-    scheduler.SetReselectionPending();
-}
-
 ////////////////////////////////////////////////////////////////////////////////////////////////////
 
 /**
diff --git a/src/core/hle/kernel/thread.h b/src/core/hle/kernel/thread.h
index 23fdef8a4..c0342c462 100644
--- a/src/core/hle/kernel/thread.h
+++ b/src/core/hle/kernel/thread.h
@@ -6,26 +6,47 @@
 
 #include <functional>
 #include <string>
+#include <utility>
 #include <vector>
 
 #include "common/common_types.h"
+#include "common/spin_lock.h"
 #include "core/arm/arm_interface.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/synchronization_object.h"
 #include "core/hle/result.h"
 
+namespace Common {
+class Fiber;
+}
+
+namespace Core {
+class ARM_Interface;
+class System;
+} // namespace Core
+
 namespace Kernel {
 
+class GlobalScheduler;
 class KernelCore;
 class Process;
 class Scheduler;
 
 enum ThreadPriority : u32 {
-    THREADPRIO_HIGHEST = 0,       ///< Highest thread priority
-    THREADPRIO_USERLAND_MAX = 24, ///< Highest thread priority for userland apps
-    THREADPRIO_DEFAULT = 44,      ///< Default thread priority for userland apps
-    THREADPRIO_LOWEST = 63,       ///< Lowest thread priority
-    THREADPRIO_COUNT = 64,        ///< Total number of possible thread priorities.
+    THREADPRIO_HIGHEST = 0,            ///< Highest thread priority
+    THREADPRIO_MAX_CORE_MIGRATION = 2, ///< Highest priority for a core migration
+    THREADPRIO_USERLAND_MAX = 24,      ///< Highest thread priority for userland apps
+    THREADPRIO_DEFAULT = 44,           ///< Default thread priority for userland apps
+    THREADPRIO_LOWEST = 63,            ///< Lowest thread priority
+    THREADPRIO_COUNT = 64,             ///< Total number of possible thread priorities.
+};
+
+enum ThreadType : u32 {
+    THREADTYPE_USER = 0x1,
+    THREADTYPE_KERNEL = 0x2,
+    THREADTYPE_HLE = 0x4,
+    THREADTYPE_IDLE = 0x8,
+    THREADTYPE_SUSPEND = 0x10,
 };
 
 enum ThreadProcessorId : s32 {
@@ -107,26 +128,45 @@ public:
 
     using ThreadSynchronizationObjects = std::vector<std::shared_ptr<SynchronizationObject>>;
 
-    using WakeupCallback =
-        std::function<bool(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
-                           std::shared_ptr<SynchronizationObject> object, std::size_t index)>;
+    using HLECallback = std::function<bool(std::shared_ptr<Thread> thread)>;
+
+    /**
+     * Creates and returns a new thread. The new thread is immediately scheduled
+     * @param system The instance of the whole system
+     * @param name The friendly name desired for the thread
+     * @param entry_point The address at which the thread should start execution
+     * @param priority The thread's priority
+     * @param arg User data to pass to the thread
+     * @param processor_id The ID(s) of the processors on which the thread is desired to be run
+     * @param stack_top The address of the thread's stack top
+     * @param owner_process The parent process for the thread, if null, it's a kernel thread
+     * @return A shared pointer to the newly created thread
+     */
+    static ResultVal<std::shared_ptr<Thread>> Create(Core::System& system, ThreadType type_flags,
+                                                     std::string name, VAddr entry_point,
+                                                     u32 priority, u64 arg, s32 processor_id,
+                                                     VAddr stack_top, Process* owner_process);
 
     /**
      * Creates and returns a new thread. The new thread is immediately scheduled
-     * @param kernel The kernel instance this thread will be created under.
+     * @param system The instance of the whole system
      * @param name The friendly name desired for the thread
      * @param entry_point The address at which the thread should start execution
      * @param priority The thread's priority
      * @param arg User data to pass to the thread
      * @param processor_id The ID(s) of the processors on which the thread is desired to be run
      * @param stack_top The address of the thread's stack top
-     * @param owner_process The parent process for the thread
+     * @param owner_process The parent process for the thread, if null, it's a kernel thread
+     * @param thread_start_func The function where the host context will start.
+     * @param thread_start_parameter The parameter which will passed to host context on init
      * @return A shared pointer to the newly created thread
      */
-    static ResultVal<std::shared_ptr<Thread>> Create(KernelCore& kernel, std::string name,
-                                                     VAddr entry_point, u32 priority, u64 arg,
-                                                     s32 processor_id, VAddr stack_top,
-                                                     Process& owner_process);
+    static ResultVal<std::shared_ptr<Thread>> Create(Core::System& system, ThreadType type_flags,
+                                                     std::string name, VAddr entry_point,
+                                                     u32 priority, u64 arg, s32 processor_id,
+                                                     VAddr stack_top, Process* owner_process,
+                                                     std::function<void(void*)>&& thread_start_func,
+                                                     void* thread_start_parameter);
 
     std::string GetName() const override {
         return name;
@@ -181,7 +221,7 @@ public:
     void UpdatePriority();
 
     /// Changes the core that the thread is running or scheduled to run on.
-    void ChangeCore(u32 core, u64 mask);
+    ResultCode SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask);
 
     /**
      * Gets the thread's thread ID
@@ -194,6 +234,10 @@ public:
     /// Resumes a thread from waiting
     void ResumeFromWait();
 
+    void OnWakeUp();
+
+    ResultCode Start();
+
     /// Cancels a waiting operation that this thread may or may not be within.
     ///
     /// When the thread is within a waiting state, this will set the thread's
@@ -202,26 +246,19 @@ public:
     ///
     void CancelWait();
 
-    /**
-     * Schedules an event to wake up the specified thread after the specified delay
-     * @param nanoseconds The time this thread will be allowed to sleep for
-     */
-    void WakeAfterDelay(s64 nanoseconds);
+    void SetSynchronizationResults(SynchronizationObject* object, ResultCode result);
 
-    /// Cancel any outstanding wakeup events for this thread
-    void CancelWakeupTimer();
+    Core::ARM_Interface& ArmInterface();
 
-    /**
-     * Sets the result after the thread awakens (from svcWaitSynchronization)
-     * @param result Value to set to the returned result
-     */
-    void SetWaitSynchronizationResult(ResultCode result);
+    const Core::ARM_Interface& ArmInterface() const;
 
-    /**
-     * Sets the output parameter value after the thread awakens (from svcWaitSynchronization)
-     * @param output Value to set to the output parameter
-     */
-    void SetWaitSynchronizationOutput(s32 output);
+    SynchronizationObject* GetSignalingObject() const {
+        return signaling_object;
+    }
+
+    ResultCode GetSignalingResult() const {
+        return signaling_result;
+    }
 
     /**
      * Retrieves the index that this particular object occupies in the list of objects
@@ -269,11 +306,6 @@ public:
      */
     VAddr GetCommandBufferAddress() const;
 
-    /// Returns whether this thread is waiting on objects from a WaitSynchronization call.
-    bool IsSleepingOnWait() const {
-        return status == ThreadStatus::WaitSynch;
-    }
-
     ThreadContext32& GetContext32() {
         return context_32;
     }
@@ -290,6 +322,28 @@ public:
         return context_64;
     }
 
+    bool IsHLEThread() const {
+        return (type & THREADTYPE_HLE) != 0;
+    }
+
+    bool IsSuspendThread() const {
+        return (type & THREADTYPE_SUSPEND) != 0;
+    }
+
+    bool IsIdleThread() const {
+        return (type & THREADTYPE_IDLE) != 0;
+    }
+
+    bool WasRunning() const {
+        return was_running;
+    }
+
+    void SetWasRunning(bool value) {
+        was_running = value;
+    }
+
+    std::shared_ptr<Common::Fiber>& GetHostContext();
+
     ThreadStatus GetStatus() const {
         return status;
     }
@@ -325,18 +379,18 @@ public:
     }
 
     const ThreadSynchronizationObjects& GetSynchronizationObjects() const {
-        return wait_objects;
+        return *wait_objects;
     }
 
-    void SetSynchronizationObjects(ThreadSynchronizationObjects objects) {
-        wait_objects = std::move(objects);
+    void SetSynchronizationObjects(ThreadSynchronizationObjects* objects) {
+        wait_objects = objects;
     }
 
     void ClearSynchronizationObjects() {
-        for (const auto& waiting_object : wait_objects) {
+        for (const auto& waiting_object : *wait_objects) {
             waiting_object->RemoveWaitingThread(SharedFrom(this));
         }
-        wait_objects.clear();
+        wait_objects->clear();
     }
 
     /// Determines whether all the objects this thread is waiting on are ready.
@@ -386,26 +440,35 @@ public:
         arb_wait_address = address;
     }
 
-    bool HasWakeupCallback() const {
-        return wakeup_callback != nullptr;
+    bool HasHLECallback() const {
+        return hle_callback != nullptr;
     }
 
-    void SetWakeupCallback(WakeupCallback callback) {
-        wakeup_callback = std::move(callback);
+    void SetHLECallback(HLECallback callback) {
+        hle_callback = std::move(callback);
     }
 
-    void InvalidateWakeupCallback() {
-        SetWakeupCallback(nullptr);
+    void SetHLETimeEvent(Handle time_event) {
+        hle_time_event = time_event;
     }
 
-    /**
-     * Invokes the thread's wakeup callback.
-     *
-     * @pre A valid wakeup callback has been set. Violating this precondition
-     *      will cause an assertion to trigger.
-     */
-    bool InvokeWakeupCallback(ThreadWakeupReason reason, std::shared_ptr<Thread> thread,
-                              std::shared_ptr<SynchronizationObject> object, std::size_t index);
+    void SetHLESyncObject(SynchronizationObject* object) {
+        hle_object = object;
+    }
+
+    Handle GetHLETimeEvent() const {
+        return hle_time_event;
+    }
+
+    SynchronizationObject* GetHLESyncObject() const {
+        return hle_object;
+    }
+
+    void InvalidateHLECallback() {
+        SetHLECallback(nullptr);
+    }
+
+    bool InvokeHLECallback(std::shared_ptr<Thread> thread);
 
     u32 GetIdealCore() const {
         return ideal_core;
@@ -415,23 +478,19 @@ public:
         return affinity_mask;
     }
 
-    ThreadActivity GetActivity() const {
-        return activity;
-    }
-
-    void SetActivity(ThreadActivity value);
+    ResultCode SetActivity(ThreadActivity value);
 
     /// Sleeps this thread for the given amount of nanoseconds.
-    void Sleep(s64 nanoseconds);
+    ResultCode Sleep(s64 nanoseconds);
 
     /// Yields this thread without rebalancing loads.
-    bool YieldSimple();
+    std::pair<ResultCode, bool> YieldSimple();
 
     /// Yields this thread and does a load rebalancing.
-    bool YieldAndBalanceLoad();
+    std::pair<ResultCode, bool> YieldAndBalanceLoad();
 
     /// Yields this thread and if the core is left idle, loads are rebalanced
-    bool YieldAndWaitForLoadBalancing();
+    std::pair<ResultCode, bool> YieldAndWaitForLoadBalancing();
 
     void IncrementYieldCount() {
         yield_count++;
@@ -446,6 +505,10 @@ public:
                                               static_cast<u32>(ThreadSchedMasks::LowMask));
     }
 
+    bool IsRunnable() const {
+        return scheduling_state == static_cast<u32>(ThreadSchedStatus::Runnable);
+    }
+
     bool IsRunning() const {
         return is_running;
     }
@@ -466,17 +529,67 @@ public:
         return global_handle;
     }
 
+    bool IsWaitingForArbitration() const {
+        return waiting_for_arbitration;
+    }
+
+    void WaitForArbitration(bool set) {
+        waiting_for_arbitration = set;
+    }
+
+    bool IsWaitingSync() const {
+        return is_waiting_on_sync;
+    }
+
+    void SetWaitingSync(bool is_waiting) {
+        is_waiting_on_sync = is_waiting;
+    }
+
+    bool IsPendingTermination() const {
+        return will_be_terminated || GetSchedulingStatus() == ThreadSchedStatus::Exited;
+    }
+
+    bool IsPaused() const {
+        return pausing_state != 0;
+    }
+
+    bool IsContinuousOnSVC() const {
+        return is_continuous_on_svc;
+    }
+
+    void SetContinuousOnSVC(bool is_continuous) {
+        is_continuous_on_svc = is_continuous;
+    }
+
+    bool IsPhantomMode() const {
+        return is_phantom_mode;
+    }
+
+    void SetPhantomMode(bool phantom) {
+        is_phantom_mode = phantom;
+    }
+
+    bool HasExited() const {
+        return has_exited;
+    }
+
 private:
+    friend class GlobalScheduler;
+    friend class Scheduler;
+
     void SetSchedulingStatus(ThreadSchedStatus new_status);
+    void AddSchedulingFlag(ThreadSchedFlags flag);
+    void RemoveSchedulingFlag(ThreadSchedFlags flag);
+
     void SetCurrentPriority(u32 new_priority);
-    ResultCode SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask);
 
-    void AdjustSchedulingOnStatus(u32 old_flags);
-    void AdjustSchedulingOnPriority(u32 old_priority);
     void AdjustSchedulingOnAffinity(u64 old_affinity_mask, s32 old_core);
 
+    Common::SpinLock context_guard{};
     ThreadContext32 context_32{};
     ThreadContext64 context_64{};
+    std::unique_ptr<Core::ARM_Interface> arm_interface{};
+    std::shared_ptr<Common::Fiber> host_context{};
 
     u64 thread_id = 0;
 
@@ -485,6 +598,8 @@ private:
     VAddr entry_point = 0;
     VAddr stack_top = 0;
 
+    ThreadType type;
+
     /// Nominal thread priority, as set by the emulated application.
     /// The nominal priority is the thread priority without priority
     /// inheritance taken into account.
@@ -509,7 +624,10 @@ private:
 
     /// Objects that the thread is waiting on, in the same order as they were
     /// passed to WaitSynchronization.
-    ThreadSynchronizationObjects wait_objects;
+    ThreadSynchronizationObjects* wait_objects;
+
+    SynchronizationObject* signaling_object;
+    ResultCode signaling_result{RESULT_SUCCESS};
 
     /// List of threads that are waiting for a mutex that is held by this thread.
     MutexWaitingThreads wait_mutex_threads;
@@ -526,30 +644,39 @@ private:
 
     /// If waiting for an AddressArbiter, this is the address being waited on.
     VAddr arb_wait_address{0};
+    bool waiting_for_arbitration{};
 
     /// Handle used as userdata to reference this object when inserting into the CoreTiming queue.
     Handle global_handle = 0;
 
-    /// Callback that will be invoked when the thread is resumed from a waiting state. If the thread
-    /// was waiting via WaitSynchronization then the object will be the last object that became
-    /// available. In case of a timeout, the object will be nullptr.
-    WakeupCallback wakeup_callback;
+    /// Callback for HLE Events
+    HLECallback hle_callback;
+    Handle hle_time_event;
+    SynchronizationObject* hle_object;
 
     Scheduler* scheduler = nullptr;
 
     u32 ideal_core{0xFFFFFFFF};
     u64 affinity_mask{0x1};
 
-    ThreadActivity activity = ThreadActivity::Normal;
-
     s32 ideal_core_override = -1;
     u64 affinity_mask_override = 0x1;
     u32 affinity_override_count = 0;
 
     u32 scheduling_state = 0;
+    u32 pausing_state = 0;
     bool is_running = false;
+    bool is_waiting_on_sync = false;
     bool is_sync_cancelled = false;
 
+    bool is_continuous_on_svc = false;
+
+    bool will_be_terminated = false;
+    bool is_phantom_mode = false;
+    bool has_exited = false;
+
+    bool was_running = false;
+
     std::string name;
 };
 
diff --git a/src/core/hle/kernel/time_manager.cpp b/src/core/hle/kernel/time_manager.cpp
index 21b290468..941305e8e 100644
--- a/src/core/hle/kernel/time_manager.cpp
+++ b/src/core/hle/kernel/time_manager.cpp
@@ -8,30 +8,37 @@
 #include "core/core_timing_util.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/kernel/time_manager.h"
 
 namespace Kernel {
 
-TimeManager::TimeManager(Core::System& system) : system{system} {
+TimeManager::TimeManager(Core::System& system_) : system{system_} {
     time_manager_event_type = Core::Timing::CreateEvent(
         "Kernel::TimeManagerCallback", [this](u64 thread_handle, [[maybe_unused]] s64 cycles_late) {
+            SchedulerLock lock(system.Kernel());
             Handle proper_handle = static_cast<Handle>(thread_handle);
+            if (cancelled_events[proper_handle]) {
+                return;
+            }
             std::shared_ptr<Thread> thread =
                 this->system.Kernel().RetrieveThreadFromGlobalHandleTable(proper_handle);
-            thread->ResumeFromWait();
+            thread->OnWakeUp();
         });
 }
 
 void TimeManager::ScheduleTimeEvent(Handle& event_handle, Thread* timetask, s64 nanoseconds) {
+    event_handle = timetask->GetGlobalHandle();
     if (nanoseconds > 0) {
         ASSERT(timetask);
-        event_handle = timetask->GetGlobalHandle();
-        const s64 cycles = Core::Timing::nsToCycles(std::chrono::nanoseconds{nanoseconds});
-        system.CoreTiming().ScheduleEvent(cycles, time_manager_event_type, event_handle);
+        ASSERT(timetask->GetStatus() != ThreadStatus::Ready);
+        ASSERT(timetask->GetStatus() != ThreadStatus::WaitMutex);
+        system.CoreTiming().ScheduleEvent(nanoseconds, time_manager_event_type, event_handle);
     } else {
         event_handle = InvalidHandle;
     }
+    cancelled_events[event_handle] = false;
 }
 
 void TimeManager::UnscheduleTimeEvent(Handle event_handle) {
@@ -39,6 +46,12 @@ void TimeManager::UnscheduleTimeEvent(Handle event_handle) {
         return;
     }
     system.CoreTiming().UnscheduleEvent(time_manager_event_type, event_handle);
+    cancelled_events[event_handle] = true;
+}
+
+void TimeManager::CancelTimeEvent(Thread* time_task) {
+    Handle event_handle = time_task->GetGlobalHandle();
+    UnscheduleTimeEvent(event_handle);
 }
 
 } // namespace Kernel
diff --git a/src/core/hle/kernel/time_manager.h b/src/core/hle/kernel/time_manager.h
index eaec486d1..307a18765 100644
--- a/src/core/hle/kernel/time_manager.h
+++ b/src/core/hle/kernel/time_manager.h
@@ -5,6 +5,7 @@
 #pragma once
 
 #include <memory>
+#include <unordered_map>
 
 #include "core/hle/kernel/object.h"
 
@@ -35,9 +36,12 @@ public:
     /// Unschedule an existing time event
     void UnscheduleTimeEvent(Handle event_handle);
 
+    void CancelTimeEvent(Thread* time_task);
+
 private:
     Core::System& system;
     std::shared_ptr<Core::Timing::EventType> time_manager_event_type;
+    std::unordered_map<Handle, bool> cancelled_events;
 };
 
 } // namespace Kernel
diff --git a/src/core/hle/service/hid/controllers/debug_pad.cpp b/src/core/hle/service/hid/controllers/debug_pad.cpp
index 1f2131ec8..cb35919e9 100644
--- a/src/core/hle/service/hid/controllers/debug_pad.cpp
+++ b/src/core/hle/service/hid/controllers/debug_pad.cpp
@@ -23,7 +23,7 @@ void Controller_DebugPad::OnRelease() {}
 
 void Controller_DebugPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data,
                                    std::size_t size) {
-    shared_memory.header.timestamp = core_timing.GetTicks();
+    shared_memory.header.timestamp = core_timing.GetCPUTicks();
     shared_memory.header.total_entry_count = 17;
 
     if (!IsControllerActivated()) {
diff --git a/src/core/hle/service/hid/controllers/gesture.cpp b/src/core/hle/service/hid/controllers/gesture.cpp
index 6e990dd00..b7b7bfeae 100644
--- a/src/core/hle/service/hid/controllers/gesture.cpp
+++ b/src/core/hle/service/hid/controllers/gesture.cpp
@@ -19,7 +19,7 @@ void Controller_Gesture::OnRelease() {}
 
 void Controller_Gesture::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data,
                                   std::size_t size) {
-    shared_memory.header.timestamp = core_timing.GetTicks();
+    shared_memory.header.timestamp = core_timing.GetCPUTicks();
     shared_memory.header.total_entry_count = 17;
 
     if (!IsControllerActivated()) {
diff --git a/src/core/hle/service/hid/controllers/keyboard.cpp b/src/core/hle/service/hid/controllers/keyboard.cpp
index 9a8d354ba..feae89525 100644
--- a/src/core/hle/service/hid/controllers/keyboard.cpp
+++ b/src/core/hle/service/hid/controllers/keyboard.cpp
@@ -21,7 +21,7 @@ void Controller_Keyboard::OnRelease() {}
 
 void Controller_Keyboard::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data,
                                    std::size_t size) {
-    shared_memory.header.timestamp = core_timing.GetTicks();
+    shared_memory.header.timestamp = core_timing.GetCPUTicks();
     shared_memory.header.total_entry_count = 17;
 
     if (!IsControllerActivated()) {
diff --git a/src/core/hle/service/hid/controllers/mouse.cpp b/src/core/hle/service/hid/controllers/mouse.cpp
index 93d88ea50..ac40989c5 100644
--- a/src/core/hle/service/hid/controllers/mouse.cpp
+++ b/src/core/hle/service/hid/controllers/mouse.cpp
@@ -19,7 +19,7 @@ void Controller_Mouse::OnRelease() {}
 
 void Controller_Mouse::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data,
                                 std::size_t size) {
-    shared_memory.header.timestamp = core_timing.GetTicks();
+    shared_memory.header.timestamp = core_timing.GetCPUTicks();
     shared_memory.header.total_entry_count = 17;
 
     if (!IsControllerActivated()) {
diff --git a/src/core/hle/service/hid/controllers/npad.cpp b/src/core/hle/service/hid/controllers/npad.cpp
index 6fbee7efa..ef67ad690 100644
--- a/src/core/hle/service/hid/controllers/npad.cpp
+++ b/src/core/hle/service/hid/controllers/npad.cpp
@@ -328,7 +328,7 @@ void Controller_NPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8*
             const auto& last_entry =
                 main_controller->npad[main_controller->common.last_entry_index];
 
-            main_controller->common.timestamp = core_timing.GetTicks();
+            main_controller->common.timestamp = core_timing.GetCPUTicks();
             main_controller->common.last_entry_index =
                 (main_controller->common.last_entry_index + 1) % 17;
 
diff --git a/src/core/hle/service/hid/controllers/stubbed.cpp b/src/core/hle/service/hid/controllers/stubbed.cpp
index 9e527d176..e7483bfa2 100644
--- a/src/core/hle/service/hid/controllers/stubbed.cpp
+++ b/src/core/hle/service/hid/controllers/stubbed.cpp
@@ -23,7 +23,7 @@ void Controller_Stubbed::OnUpdate(const Core::Timing::CoreTiming& core_timing, u
     }
 
     CommonHeader header{};
-    header.timestamp = core_timing.GetTicks();
+    header.timestamp = core_timing.GetCPUTicks();
     header.total_entry_count = 17;
     header.entry_count = 0;
     header.last_entry_index = 0;
diff --git a/src/core/hle/service/hid/controllers/touchscreen.cpp b/src/core/hle/service/hid/controllers/touchscreen.cpp
index 1c6e55566..e326f8f5c 100644
--- a/src/core/hle/service/hid/controllers/touchscreen.cpp
+++ b/src/core/hle/service/hid/controllers/touchscreen.cpp
@@ -22,7 +22,7 @@ void Controller_Touchscreen::OnRelease() {}
 
 void Controller_Touchscreen::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data,
                                       std::size_t size) {
-    shared_memory.header.timestamp = core_timing.GetTicks();
+    shared_memory.header.timestamp = core_timing.GetCPUTicks();
     shared_memory.header.total_entry_count = 17;
 
     if (!IsControllerActivated()) {
@@ -49,7 +49,7 @@ void Controller_Touchscreen::OnUpdate(const Core::Timing::CoreTiming& core_timin
         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 = core_timing.GetTicks();
+        const u64 tick = core_timing.GetCPUTicks();
         touch_entry.delta_time = tick - last_touch;
         last_touch = tick;
         touch_entry.finger = Settings::values.touchscreen.finger;
diff --git a/src/core/hle/service/hid/controllers/xpad.cpp b/src/core/hle/service/hid/controllers/xpad.cpp
index 27511b27b..2503ef241 100644
--- a/src/core/hle/service/hid/controllers/xpad.cpp
+++ b/src/core/hle/service/hid/controllers/xpad.cpp
@@ -20,7 +20,7 @@ void Controller_XPad::OnRelease() {}
 void Controller_XPad::OnUpdate(const Core::Timing::CoreTiming& core_timing, u8* data,
                                std::size_t size) {
     for (auto& xpad_entry : shared_memory.shared_memory_entries) {
-        xpad_entry.header.timestamp = core_timing.GetTicks();
+        xpad_entry.header.timestamp = core_timing.GetCPUTicks();
         xpad_entry.header.total_entry_count = 17;
 
         if (!IsControllerActivated()) {
diff --git a/src/core/hle/service/hid/hid.cpp b/src/core/hle/service/hid/hid.cpp
index 57d5edea7..e9020e0dc 100644
--- a/src/core/hle/service/hid/hid.cpp
+++ b/src/core/hle/service/hid/hid.cpp
@@ -39,11 +39,9 @@ namespace Service::HID {
 
 // Updating period for each HID device.
 // TODO(ogniK): Find actual polling rate of hid
-constexpr s64 pad_update_ticks = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 66);
-[[maybe_unused]] constexpr s64 accelerometer_update_ticks =
-    static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 100);
-[[maybe_unused]] constexpr s64 gyroscope_update_ticks =
-    static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 100);
+constexpr s64 pad_update_ticks = static_cast<s64>(1000000000 / 66);
+[[maybe_unused]] constexpr s64 accelerometer_update_ticks = static_cast<s64>(1000000000 / 100);
+[[maybe_unused]] constexpr s64 gyroscope_update_ticks = static_cast<s64>(1000000000 / 100);
 constexpr std::size_t SHARED_MEMORY_SIZE = 0x40000;
 
 IAppletResource::IAppletResource(Core::System& system)
@@ -78,8 +76,8 @@ IAppletResource::IAppletResource(Core::System& system)
 
     // Register update callbacks
     pad_update_event =
-        Core::Timing::CreateEvent("HID::UpdatePadCallback", [this](u64 userdata, s64 cycles_late) {
-            UpdateControllers(userdata, cycles_late);
+        Core::Timing::CreateEvent("HID::UpdatePadCallback", [this](u64 userdata, s64 ns_late) {
+            UpdateControllers(userdata, ns_late);
         });
 
     // TODO(shinyquagsire23): Other update callbacks? (accel, gyro?)
@@ -109,7 +107,7 @@ void IAppletResource::GetSharedMemoryHandle(Kernel::HLERequestContext& ctx) {
     rb.PushCopyObjects(shared_mem);
 }
 
-void IAppletResource::UpdateControllers(u64 userdata, s64 cycles_late) {
+void IAppletResource::UpdateControllers(u64 userdata, s64 ns_late) {
     auto& core_timing = system.CoreTiming();
 
     const bool should_reload = Settings::values.is_device_reload_pending.exchange(false);
@@ -120,7 +118,7 @@ void IAppletResource::UpdateControllers(u64 userdata, s64 cycles_late) {
         controller->OnUpdate(core_timing, shared_mem->GetPointer(), SHARED_MEMORY_SIZE);
     }
 
-    core_timing.ScheduleEvent(pad_update_ticks - cycles_late, pad_update_event);
+    core_timing.ScheduleEvent(pad_update_ticks - ns_late, pad_update_event);
 }
 
 class IActiveVibrationDeviceList final : public ServiceFramework<IActiveVibrationDeviceList> {
diff --git a/src/core/hle/service/hid/irs.cpp b/src/core/hle/service/hid/irs.cpp
index 36ed6f7da..e82fd031b 100644
--- a/src/core/hle/service/hid/irs.cpp
+++ b/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>(system.CoreTiming().GetTicks());
+    rb.PushRaw<u64>(system.CoreTiming().GetCPUTicks());
     rb.PushRaw<u32>(0);
 }
 
diff --git a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp
index 0d913334e..fba89e7a6 100644
--- a/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp
+++ b/src/core/hle/service/nvdrv/devices/nvhost_ctrl_gpu.cpp
@@ -200,8 +200,7 @@ u32 nvhost_ctrl_gpu::GetGpuTime(const std::vector<u8>& input, std::vector<u8>& o
 
     IoctlGetGpuTime params{};
     std::memcpy(&params, input.data(), input.size());
-    const auto ns = Core::Timing::CyclesToNs(system.CoreTiming().GetTicks());
-    params.gpu_time = static_cast<u64_le>(ns.count());
+    params.gpu_time = static_cast<u64_le>(system.CoreTiming().GetGlobalTimeNs().count());
     std::memcpy(output.data(), &params, output.size());
     return 0;
 }
diff --git a/src/core/hle/service/nvflinger/nvflinger.cpp b/src/core/hle/service/nvflinger/nvflinger.cpp
index 437bc5dee..2f44d3779 100644
--- a/src/core/hle/service/nvflinger/nvflinger.cpp
+++ b/src/core/hle/service/nvflinger/nvflinger.cpp
@@ -9,6 +9,7 @@
 #include "common/logging/log.h"
 #include "common/microprofile.h"
 #include "common/scope_exit.h"
+#include "common/thread.h"
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
@@ -27,8 +28,35 @@
 
 namespace Service::NVFlinger {
 
-constexpr s64 frame_ticks = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 60);
-constexpr s64 frame_ticks_30fps = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 30);
+constexpr s64 frame_ticks = static_cast<s64>(1000000000 / 60);
+constexpr s64 frame_ticks_30fps = static_cast<s64>(1000000000 / 30);
+
+void NVFlinger::VSyncThread(NVFlinger& nv_flinger) {
+    nv_flinger.SplitVSync();
+}
+
+void NVFlinger::SplitVSync() {
+    system.RegisterHostThread();
+    std::string name = "yuzu:VSyncThread";
+    MicroProfileOnThreadCreate(name.c_str());
+    Common::SetCurrentThreadName(name.c_str());
+    Common::SetCurrentThreadPriority(Common::ThreadPriority::High);
+    s64 delay = 0;
+    while (is_running) {
+        guard->lock();
+        const s64 time_start = system.CoreTiming().GetGlobalTimeNs().count();
+        Compose();
+        const auto ticks = GetNextTicks();
+        const s64 time_end = system.CoreTiming().GetGlobalTimeNs().count();
+        const s64 time_passed = time_end - time_start;
+        const s64 next_time = std::max<s64>(0, ticks - time_passed - delay);
+        guard->unlock();
+        if (next_time > 0) {
+            wait_event->WaitFor(std::chrono::nanoseconds{next_time});
+        }
+        delay = (system.CoreTiming().GetGlobalTimeNs().count() - time_end) - next_time;
+    }
+}
 
 NVFlinger::NVFlinger(Core::System& system) : system(system) {
     displays.emplace_back(0, "Default", system);
@@ -36,22 +64,36 @@ NVFlinger::NVFlinger(Core::System& system) : system(system) {
     displays.emplace_back(2, "Edid", system);
     displays.emplace_back(3, "Internal", system);
     displays.emplace_back(4, "Null", system);
+    guard = std::make_shared<std::mutex>();
 
     // Schedule the screen composition events
     composition_event =
-        Core::Timing::CreateEvent("ScreenComposition", [this](u64 userdata, s64 cycles_late) {
+        Core::Timing::CreateEvent("ScreenComposition", [this](u64 userdata, s64 ns_late) {
+            Lock();
             Compose();
-            const auto ticks =
-                Settings::values.force_30fps_mode ? frame_ticks_30fps : GetNextTicks();
-            this->system.CoreTiming().ScheduleEvent(std::max<s64>(0LL, ticks - cycles_late),
+            const auto ticks = GetNextTicks();
+            this->system.CoreTiming().ScheduleEvent(std::max<s64>(0LL, ticks - ns_late),
                                                     composition_event);
         });
-
-    system.CoreTiming().ScheduleEvent(frame_ticks, composition_event);
+    if (system.IsMulticore()) {
+        is_running = true;
+        wait_event = std::make_unique<Common::Event>();
+        vsync_thread = std::make_unique<std::thread>(VSyncThread, std::ref(*this));
+    } else {
+        system.CoreTiming().ScheduleEvent(frame_ticks, composition_event);
+    }
 }
 
 NVFlinger::~NVFlinger() {
-    system.CoreTiming().UnscheduleEvent(composition_event, 0);
+    if (system.IsMulticore()) {
+        is_running = false;
+        wait_event->Set();
+        vsync_thread->join();
+        vsync_thread.reset();
+        wait_event.reset();
+    } else {
+        system.CoreTiming().UnscheduleEvent(composition_event, 0);
+    }
 }
 
 void NVFlinger::SetNVDrvInstance(std::shared_ptr<Nvidia::Module> instance) {
@@ -199,10 +241,12 @@ void NVFlinger::Compose() {
 
         auto& gpu = system.GPU();
         const auto& multi_fence = buffer->get().multi_fence;
+        guard->unlock();
         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);
         }
+        guard->lock();
 
         MicroProfileFlip();
 
@@ -223,7 +267,7 @@ void NVFlinger::Compose() {
 
 s64 NVFlinger::GetNextTicks() const {
     constexpr s64 max_hertz = 120LL;
-    return (Core::Hardware::BASE_CLOCK_RATE * (1LL << swap_interval)) / max_hertz;
+    return (1000000000 * (1LL << swap_interval)) / max_hertz;
 }
 
 } // namespace Service::NVFlinger
diff --git a/src/core/hle/service/nvflinger/nvflinger.h b/src/core/hle/service/nvflinger/nvflinger.h
index 57a21f33b..e4959a9af 100644
--- a/src/core/hle/service/nvflinger/nvflinger.h
+++ b/src/core/hle/service/nvflinger/nvflinger.h
@@ -4,15 +4,22 @@
 
 #pragma once
 
+#include <atomic>
 #include <memory>
+#include <mutex>
 #include <optional>
 #include <string>
 #include <string_view>
+#include <thread>
 #include <vector>
 
 #include "common/common_types.h"
 #include "core/hle/kernel/object.h"
 
+namespace Common {
+class Event;
+} // namespace Common
+
 namespace Core::Timing {
 class CoreTiming;
 struct EventType;
@@ -79,6 +86,10 @@ public:
 
     s64 GetNextTicks() const;
 
+    std::unique_lock<std::mutex> Lock() {
+        return std::unique_lock{*guard};
+    }
+
 private:
     /// Finds the display identified by the specified ID.
     VI::Display* FindDisplay(u64 display_id);
@@ -92,6 +103,10 @@ private:
     /// Finds the layer identified by the specified ID in the desired display.
     const VI::Layer* FindLayer(u64 display_id, u64 layer_id) const;
 
+    static void VSyncThread(NVFlinger& nv_flinger);
+
+    void SplitVSync();
+
     std::shared_ptr<Nvidia::Module> nvdrv;
 
     std::vector<VI::Display> displays;
@@ -108,7 +123,13 @@ private:
     /// Event that handles screen composition.
     std::shared_ptr<Core::Timing::EventType> composition_event;
 
+    std::shared_ptr<std::mutex> guard;
+
     Core::System& system;
+
+    std::unique_ptr<std::thread> vsync_thread;
+    std::unique_ptr<Common::Event> wait_event;
+    std::atomic<bool> is_running{};
 };
 
 } // namespace Service::NVFlinger
diff --git a/src/core/hle/service/sm/sm.cpp b/src/core/hle/service/sm/sm.cpp
index 6ada13be4..d872de16c 100644
--- a/src/core/hle/service/sm/sm.cpp
+++ b/src/core/hle/service/sm/sm.cpp
@@ -142,7 +142,7 @@ void SM::GetService(Kernel::HLERequestContext& ctx) {
     }
 
     // Wake the threads waiting on the ServerPort
-    server_port->WakeupAllWaitingThreads();
+    server_port->Signal();
 
     LOG_DEBUG(Service_SM, "called service={} -> session={}", name, client->GetObjectId());
     IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
diff --git a/src/core/hle/service/time/standard_steady_clock_core.cpp b/src/core/hle/service/time/standard_steady_clock_core.cpp
index 1575f0b49..59a272f4a 100644
--- a/src/core/hle/service/time/standard_steady_clock_core.cpp
+++ b/src/core/hle/service/time/standard_steady_clock_core.cpp
@@ -11,9 +11,8 @@
 namespace Service::Time::Clock {
 
 TimeSpanType StandardSteadyClockCore::GetCurrentRawTimePoint(Core::System& system) {
-    const TimeSpanType ticks_time_span{TimeSpanType::FromTicks(
-        Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()),
-        Core::Hardware::CNTFREQ)};
+    const TimeSpanType ticks_time_span{
+        TimeSpanType::FromTicks(system.CoreTiming().GetClockTicks(), Core::Hardware::CNTFREQ)};
     TimeSpanType raw_time_point{setup_value.nanoseconds + ticks_time_span.nanoseconds};
 
     if (raw_time_point.nanoseconds < cached_raw_time_point.nanoseconds) {
diff --git a/src/core/hle/service/time/tick_based_steady_clock_core.cpp b/src/core/hle/service/time/tick_based_steady_clock_core.cpp
index 44d5bc651..8baaa2a6a 100644
--- a/src/core/hle/service/time/tick_based_steady_clock_core.cpp
+++ b/src/core/hle/service/time/tick_based_steady_clock_core.cpp
@@ -11,9 +11,8 @@
 namespace Service::Time::Clock {
 
 SteadyClockTimePoint TickBasedSteadyClockCore::GetTimePoint(Core::System& system) {
-    const TimeSpanType ticks_time_span{TimeSpanType::FromTicks(
-        Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()),
-        Core::Hardware::CNTFREQ)};
+    const TimeSpanType ticks_time_span{
+        TimeSpanType::FromTicks(system.CoreTiming().GetClockTicks(), Core::Hardware::CNTFREQ)};
 
     return {ticks_time_span.ToSeconds(), GetClockSourceId()};
 }
diff --git a/src/core/hle/service/time/time.cpp b/src/core/hle/service/time/time.cpp
index 67f1bbcf3..4cf58a61a 100644
--- a/src/core/hle/service/time/time.cpp
+++ b/src/core/hle/service/time/time.cpp
@@ -234,9 +234,8 @@ void Module::Interface::CalculateMonotonicSystemClockBaseTimePoint(Kernel::HLERe
     const auto current_time_point{steady_clock_core.GetCurrentTimePoint(system)};
 
     if (current_time_point.clock_source_id == context.steady_time_point.clock_source_id) {
-        const auto ticks{Clock::TimeSpanType::FromTicks(
-            Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()),
-            Core::Hardware::CNTFREQ)};
+        const auto ticks{Clock::TimeSpanType::FromTicks(system.CoreTiming().GetClockTicks(),
+                                                        Core::Hardware::CNTFREQ)};
         const s64 base_time_point{context.offset + current_time_point.time_point -
                                   ticks.ToSeconds()};
         IPC::ResponseBuilder rb{ctx, (sizeof(s64) / 4) + 2};
diff --git a/src/core/hle/service/time/time_sharedmemory.cpp b/src/core/hle/service/time/time_sharedmemory.cpp
index 999ec1e51..e0ae9f874 100644
--- a/src/core/hle/service/time/time_sharedmemory.cpp
+++ b/src/core/hle/service/time/time_sharedmemory.cpp
@@ -30,8 +30,7 @@ void SharedMemory::SetupStandardSteadyClock(Core::System& system,
                                             const Common::UUID& clock_source_id,
                                             Clock::TimeSpanType current_time_point) {
     const Clock::TimeSpanType ticks_time_span{Clock::TimeSpanType::FromTicks(
-        Core::Timing::CpuCyclesToClockCycles(system.CoreTiming().GetTicks()),
-        Core::Hardware::CNTFREQ)};
+        system.CoreTiming().GetClockTicks(), Core::Hardware::CNTFREQ)};
     const Clock::SteadyClockContext context{
         static_cast<u64>(current_time_point.nanoseconds - ticks_time_span.nanoseconds),
         clock_source_id};
diff --git a/src/core/hle/service/vi/vi.cpp b/src/core/hle/service/vi/vi.cpp
index 46e14c2a3..157092074 100644
--- a/src/core/hle/service/vi/vi.cpp
+++ b/src/core/hle/service/vi/vi.cpp
@@ -511,6 +511,7 @@ private:
         LOG_DEBUG(Service_VI, "called. id=0x{:08X} transaction={:X}, flags=0x{:08X}", id,
                   static_cast<u32>(transaction), flags);
 
+        nv_flinger->Lock();
         auto& buffer_queue = nv_flinger->FindBufferQueue(id);
 
         switch (transaction) {
@@ -550,6 +551,7 @@ private:
                     [=](std::shared_ptr<Kernel::Thread> thread, Kernel::HLERequestContext& ctx,
                         Kernel::ThreadWakeupReason reason) {
                         // Repeat TransactParcel DequeueBuffer when a buffer is available
+                        nv_flinger->Lock();
                         auto& buffer_queue = nv_flinger->FindBufferQueue(id);
                         auto result = buffer_queue.DequeueBuffer(width, height);
                         ASSERT_MSG(result != std::nullopt, "Could not dequeue buffer.");
diff --git a/src/core/host_timing.cpp b/src/core/host_timing.cpp
deleted file mode 100644
index 2f40de1a1..000000000
--- a/src/core/host_timing.cpp
+++ /dev/null
@@ -1,206 +0,0 @@
-// Copyright 2020 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include "core/host_timing.h"
-
-#include <algorithm>
-#include <mutex>
-#include <string>
-#include <tuple>
-
-#include "common/assert.h"
-#include "core/core_timing_util.h"
-
-namespace Core::HostTiming {
-
-std::shared_ptr<EventType> CreateEvent(std::string name, TimedCallback&& callback) {
-    return std::make_shared<EventType>(std::move(callback), std::move(name));
-}
-
-struct CoreTiming::Event {
-    u64 time;
-    u64 fifo_order;
-    u64 userdata;
-    std::weak_ptr<EventType> type;
-
-    // Sort by time, unless the times are the same, in which case sort by
-    // the order added to the queue
-    friend bool operator>(const Event& left, const Event& right) {
-        return std::tie(left.time, left.fifo_order) > std::tie(right.time, right.fifo_order);
-    }
-
-    friend bool operator<(const Event& left, const Event& right) {
-        return std::tie(left.time, left.fifo_order) < std::tie(right.time, right.fifo_order);
-    }
-};
-
-CoreTiming::CoreTiming() {
-    clock =
-        Common::CreateBestMatchingClock(Core::Hardware::BASE_CLOCK_RATE, Core::Hardware::CNTFREQ);
-}
-
-CoreTiming::~CoreTiming() = default;
-
-void CoreTiming::ThreadEntry(CoreTiming& instance) {
-    instance.ThreadLoop();
-}
-
-void CoreTiming::Initialize() {
-    event_fifo_id = 0;
-    const auto empty_timed_callback = [](u64, s64) {};
-    ev_lost = CreateEvent("_lost_event", empty_timed_callback);
-    timer_thread = std::make_unique<std::thread>(ThreadEntry, std::ref(*this));
-}
-
-void CoreTiming::Shutdown() {
-    paused = true;
-    shutting_down = true;
-    event.Set();
-    timer_thread->join();
-    ClearPendingEvents();
-    timer_thread.reset();
-    has_started = false;
-}
-
-void CoreTiming::Pause(bool is_paused) {
-    paused = is_paused;
-}
-
-void CoreTiming::SyncPause(bool is_paused) {
-    if (is_paused == paused && paused_set == paused) {
-        return;
-    }
-    Pause(is_paused);
-    event.Set();
-    while (paused_set != is_paused)
-        ;
-}
-
-bool CoreTiming::IsRunning() const {
-    return !paused_set;
-}
-
-bool CoreTiming::HasPendingEvents() const {
-    return !(wait_set && event_queue.empty());
-}
-
-void CoreTiming::ScheduleEvent(s64 ns_into_future, const std::shared_ptr<EventType>& event_type,
-                               u64 userdata) {
-    basic_lock.lock();
-    const u64 timeout = static_cast<u64>(GetGlobalTimeNs().count() + ns_into_future);
-
-    event_queue.emplace_back(Event{timeout, event_fifo_id++, userdata, event_type});
-
-    std::push_heap(event_queue.begin(), event_queue.end(), std::greater<>());
-    basic_lock.unlock();
-    event.Set();
-}
-
-void CoreTiming::UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u64 userdata) {
-    basic_lock.lock();
-    const auto itr = std::remove_if(event_queue.begin(), event_queue.end(), [&](const Event& e) {
-        return e.type.lock().get() == event_type.get() && e.userdata == userdata;
-    });
-
-    // Removing random items breaks the invariant so we have to re-establish it.
-    if (itr != event_queue.end()) {
-        event_queue.erase(itr, event_queue.end());
-        std::make_heap(event_queue.begin(), event_queue.end(), std::greater<>());
-    }
-    basic_lock.unlock();
-}
-
-void CoreTiming::AddTicks(std::size_t core_index, u64 ticks) {
-    ticks_count[core_index] += ticks;
-}
-
-void CoreTiming::ResetTicks(std::size_t core_index) {
-    ticks_count[core_index] = 0;
-}
-
-u64 CoreTiming::GetCPUTicks() const {
-    return clock->GetCPUCycles();
-}
-
-u64 CoreTiming::GetClockTicks() const {
-    return clock->GetClockCycles();
-}
-
-void CoreTiming::ClearPendingEvents() {
-    event_queue.clear();
-}
-
-void CoreTiming::RemoveEvent(const std::shared_ptr<EventType>& event_type) {
-    basic_lock.lock();
-
-    const auto itr = std::remove_if(event_queue.begin(), event_queue.end(), [&](const Event& e) {
-        return e.type.lock().get() == event_type.get();
-    });
-
-    // Removing random items breaks the invariant so we have to re-establish it.
-    if (itr != event_queue.end()) {
-        event_queue.erase(itr, event_queue.end());
-        std::make_heap(event_queue.begin(), event_queue.end(), std::greater<>());
-    }
-    basic_lock.unlock();
-}
-
-std::optional<u64> CoreTiming::Advance() {
-    advance_lock.lock();
-    basic_lock.lock();
-    global_timer = GetGlobalTimeNs().count();
-
-    while (!event_queue.empty() && event_queue.front().time <= global_timer) {
-        Event evt = std::move(event_queue.front());
-        std::pop_heap(event_queue.begin(), event_queue.end(), std::greater<>());
-        event_queue.pop_back();
-        basic_lock.unlock();
-
-        if (auto event_type{evt.type.lock()}) {
-            event_type->callback(evt.userdata, global_timer - evt.time);
-        }
-
-        basic_lock.lock();
-    }
-
-    if (!event_queue.empty()) {
-        const u64 next_time = event_queue.front().time - global_timer;
-        basic_lock.unlock();
-        advance_lock.unlock();
-        return next_time;
-    } else {
-        basic_lock.unlock();
-        advance_lock.unlock();
-        return std::nullopt;
-    }
-}
-
-void CoreTiming::ThreadLoop() {
-    has_started = true;
-    while (!shutting_down) {
-        while (!paused) {
-            paused_set = false;
-            const auto next_time = Advance();
-            if (next_time) {
-                std::chrono::nanoseconds next_time_ns = std::chrono::nanoseconds(*next_time);
-                event.WaitFor(next_time_ns);
-            } else {
-                wait_set = true;
-                event.Wait();
-            }
-            wait_set = false;
-        }
-        paused_set = true;
-    }
-}
-
-std::chrono::nanoseconds CoreTiming::GetGlobalTimeNs() const {
-    return clock->GetTimeNS();
-}
-
-std::chrono::microseconds CoreTiming::GetGlobalTimeUs() const {
-    return clock->GetTimeUS();
-}
-
-} // namespace Core::HostTiming
diff --git a/src/core/host_timing.h b/src/core/host_timing.h
deleted file mode 100644
index be6b68d7c..000000000
--- a/src/core/host_timing.h
+++ /dev/null
@@ -1,160 +0,0 @@
-// Copyright 2020 yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <atomic>
-#include <chrono>
-#include <functional>
-#include <memory>
-#include <mutex>
-#include <optional>
-#include <string>
-#include <thread>
-#include <vector>
-
-#include "common/common_types.h"
-#include "common/spin_lock.h"
-#include "common/thread.h"
-#include "common/threadsafe_queue.h"
-#include "common/wall_clock.h"
-#include "core/hardware_properties.h"
-
-namespace Core::HostTiming {
-
-/// A callback that may be scheduled for a particular core timing event.
-using TimedCallback = std::function<void(u64 userdata, s64 cycles_late)>;
-
-/// Contains the characteristics of a particular event.
-struct EventType {
-    EventType(TimedCallback&& callback, std::string&& name)
-        : callback{std::move(callback)}, name{std::move(name)} {}
-
-    /// The event's callback function.
-    TimedCallback callback;
-    /// A pointer to the name of the event.
-    const std::string name;
-};
-
-/**
- * This is a system to schedule events into the emulated machine's future. Time is measured
- * in main CPU clock cycles.
- *
- * To schedule an event, you first have to register its type. This is where you pass in the
- * callback. You then schedule events using the type id you get back.
- *
- * The int cyclesLate that the callbacks get is how many cycles late it was.
- * So to schedule a new event on a regular basis:
- * inside callback:
- *   ScheduleEvent(periodInCycles - cyclesLate, callback, "whatever")
- */
-class CoreTiming {
-public:
-    CoreTiming();
-    ~CoreTiming();
-
-    CoreTiming(const CoreTiming&) = delete;
-    CoreTiming(CoreTiming&&) = delete;
-
-    CoreTiming& operator=(const CoreTiming&) = delete;
-    CoreTiming& operator=(CoreTiming&&) = delete;
-
-    /// CoreTiming begins at the boundary of timing slice -1. An initial call to Advance() is
-    /// required to end slice - 1 and start slice 0 before the first cycle of code is executed.
-    void Initialize();
-
-    /// Tears down all timing related functionality.
-    void Shutdown();
-
-    /// Pauses/Unpauses the execution of the timer thread.
-    void Pause(bool is_paused);
-
-    /// Pauses/Unpauses the execution of the timer thread and waits until paused.
-    void SyncPause(bool is_paused);
-
-    /// Checks if core timing is running.
-    bool IsRunning() const;
-
-    /// Checks if the timer thread has started.
-    bool HasStarted() const {
-        return has_started;
-    }
-
-    /// Checks if there are any pending time events.
-    bool HasPendingEvents() const;
-
-    /// Schedules an event in core timing
-    void ScheduleEvent(s64 ns_into_future, const std::shared_ptr<EventType>& event_type,
-                       u64 userdata = 0);
-
-    void UnscheduleEvent(const std::shared_ptr<EventType>& event_type, u64 userdata);
-
-    /// We only permit one event of each type in the queue at a time.
-    void RemoveEvent(const std::shared_ptr<EventType>& event_type);
-
-    void AddTicks(std::size_t core_index, u64 ticks);
-
-    void ResetTicks(std::size_t core_index);
-
-    /// Returns current time in emulated CPU cycles
-    u64 GetCPUTicks() const;
-
-    /// Returns current time in emulated in Clock cycles
-    u64 GetClockTicks() const;
-
-    /// Returns current time in microseconds.
-    std::chrono::microseconds GetGlobalTimeUs() const;
-
-    /// Returns current time in nanoseconds.
-    std::chrono::nanoseconds GetGlobalTimeNs() const;
-
-    /// Checks for events manually and returns time in nanoseconds for next event, threadsafe.
-    std::optional<u64> Advance();
-
-private:
-    struct Event;
-
-    /// Clear all pending events. This should ONLY be done on exit.
-    void ClearPendingEvents();
-
-    static void ThreadEntry(CoreTiming& instance);
-    void ThreadLoop();
-
-    std::unique_ptr<Common::WallClock> clock;
-
-    u64 global_timer = 0;
-
-    std::chrono::nanoseconds start_point;
-
-    // The queue is a min-heap using std::make_heap/push_heap/pop_heap.
-    // We don't use std::priority_queue because we need to be able to serialize, unserialize and
-    // erase arbitrary events (RemoveEvent()) regardless of the queue order. These aren't
-    // accomodated by the standard adaptor class.
-    std::vector<Event> event_queue;
-    u64 event_fifo_id = 0;
-
-    std::shared_ptr<EventType> ev_lost;
-    Common::Event event{};
-    Common::SpinLock basic_lock{};
-    Common::SpinLock advance_lock{};
-    std::unique_ptr<std::thread> timer_thread;
-    std::atomic<bool> paused{};
-    std::atomic<bool> paused_set{};
-    std::atomic<bool> wait_set{};
-    std::atomic<bool> shutting_down{};
-    std::atomic<bool> has_started{};
-
-    std::array<std::atomic<u64>, Core::Hardware::NUM_CPU_CORES> ticks_count{};
-};
-
-/// Creates a core timing event with the given name and callback.
-///
-/// @param name     The name of the core timing event to create.
-/// @param callback The callback to execute for the event.
-///
-/// @returns An EventType instance representing the created event.
-///
-std::shared_ptr<EventType> CreateEvent(std::string name, TimedCallback&& callback);
-
-} // namespace Core::HostTiming
diff --git a/src/core/memory.cpp b/src/core/memory.cpp
index 9d87045a0..7def00768 100644
--- a/src/core/memory.cpp
+++ b/src/core/memory.cpp
@@ -8,6 +8,7 @@
 #include <utility>
 
 #include "common/assert.h"
+#include "common/atomic_ops.h"
 #include "common/common_types.h"
 #include "common/logging/log.h"
 #include "common/page_table.h"
@@ -29,15 +30,12 @@ namespace Core::Memory {
 struct Memory::Impl {
     explicit Impl(Core::System& system_) : system{system_} {}
 
-    void SetCurrentPageTable(Kernel::Process& process) {
+    void SetCurrentPageTable(Kernel::Process& process, u32 core_id) {
         current_page_table = &process.PageTable().PageTableImpl();
 
         const std::size_t address_space_width = process.PageTable().GetAddressSpaceWidth();
 
-        system.ArmInterface(0).PageTableChanged(*current_page_table, address_space_width);
-        system.ArmInterface(1).PageTableChanged(*current_page_table, address_space_width);
-        system.ArmInterface(2).PageTableChanged(*current_page_table, address_space_width);
-        system.ArmInterface(3).PageTableChanged(*current_page_table, address_space_width);
+        system.ArmInterface(core_id).PageTableChanged(*current_page_table, address_space_width);
     }
 
     void MapMemoryRegion(Common::PageTable& page_table, VAddr base, u64 size, PAddr target) {
@@ -179,6 +177,22 @@ struct Memory::Impl {
         }
     }
 
+    bool WriteExclusive8(const VAddr addr, const u8 data, const u8 expected) {
+        return WriteExclusive<u8>(addr, data, expected);
+    }
+
+    bool WriteExclusive16(const VAddr addr, const u16 data, const u16 expected) {
+        return WriteExclusive<u16_le>(addr, data, expected);
+    }
+
+    bool WriteExclusive32(const VAddr addr, const u32 data, const u32 expected) {
+        return WriteExclusive<u32_le>(addr, data, expected);
+    }
+
+    bool WriteExclusive64(const VAddr addr, const u64 data, const u64 expected) {
+        return WriteExclusive<u64_le>(addr, data, expected);
+    }
+
     std::string ReadCString(VAddr vaddr, std::size_t max_length) {
         std::string string;
         string.reserve(max_length);
@@ -682,6 +696,67 @@ struct Memory::Impl {
         }
     }
 
+    template <typename T>
+    bool WriteExclusive(const VAddr vaddr, const T data, const T expected) {
+        u8* page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
+        if (page_pointer != nullptr) {
+            // NOTE: Avoid adding any extra logic to this fast-path block
+            T volatile* pointer = reinterpret_cast<T volatile*>(&page_pointer[vaddr]);
+            return Common::AtomicCompareAndSwap(pointer, data, expected);
+        }
+
+        const Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
+        switch (type) {
+        case Common::PageType::Unmapped:
+            LOG_ERROR(HW_Memory, "Unmapped Write{} 0x{:08X} @ 0x{:016X}", sizeof(data) * 8,
+                      static_cast<u32>(data), vaddr);
+            return true;
+        case Common::PageType::Memory:
+            ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr);
+            break;
+        case Common::PageType::RasterizerCachedMemory: {
+            u8* host_ptr{GetPointerFromRasterizerCachedMemory(vaddr)};
+            system.GPU().InvalidateRegion(vaddr, sizeof(T));
+            T volatile* pointer = reinterpret_cast<T volatile*>(&host_ptr);
+            return Common::AtomicCompareAndSwap(pointer, data, expected);
+            break;
+        }
+        default:
+            UNREACHABLE();
+        }
+        return true;
+    }
+
+    bool WriteExclusive128(const VAddr vaddr, const u128 data, const u128 expected) {
+        u8* const page_pointer = current_page_table->pointers[vaddr >> PAGE_BITS];
+        if (page_pointer != nullptr) {
+            // NOTE: Avoid adding any extra logic to this fast-path block
+            u64 volatile* pointer = reinterpret_cast<u64 volatile*>(&page_pointer[vaddr]);
+            return Common::AtomicCompareAndSwap(pointer, data, expected);
+        }
+
+        const Common::PageType type = current_page_table->attributes[vaddr >> PAGE_BITS];
+        switch (type) {
+        case Common::PageType::Unmapped:
+            LOG_ERROR(HW_Memory, "Unmapped Write{} 0x{:08X} @ 0x{:016X}{:016X}", sizeof(data) * 8,
+                      static_cast<u64>(data[1]), static_cast<u64>(data[0]), vaddr);
+            return true;
+        case Common::PageType::Memory:
+            ASSERT_MSG(false, "Mapped memory page without a pointer @ {:016X}", vaddr);
+            break;
+        case Common::PageType::RasterizerCachedMemory: {
+            u8* host_ptr{GetPointerFromRasterizerCachedMemory(vaddr)};
+            system.GPU().InvalidateRegion(vaddr, sizeof(u128));
+            u64 volatile* pointer = reinterpret_cast<u64 volatile*>(&host_ptr);
+            return Common::AtomicCompareAndSwap(pointer, data, expected);
+            break;
+        }
+        default:
+            UNREACHABLE();
+        }
+        return true;
+    }
+
     Common::PageTable* current_page_table = nullptr;
     Core::System& system;
 };
@@ -689,8 +764,8 @@ struct Memory::Impl {
 Memory::Memory(Core::System& system) : impl{std::make_unique<Impl>(system)} {}
 Memory::~Memory() = default;
 
-void Memory::SetCurrentPageTable(Kernel::Process& process) {
-    impl->SetCurrentPageTable(process);
+void Memory::SetCurrentPageTable(Kernel::Process& process, u32 core_id) {
+    impl->SetCurrentPageTable(process, core_id);
 }
 
 void Memory::MapMemoryRegion(Common::PageTable& page_table, VAddr base, u64 size, PAddr target) {
@@ -764,6 +839,26 @@ void Memory::Write64(VAddr addr, u64 data) {
     impl->Write64(addr, data);
 }
 
+bool Memory::WriteExclusive8(VAddr addr, u8 data, u8 expected) {
+    return impl->WriteExclusive8(addr, data, expected);
+}
+
+bool Memory::WriteExclusive16(VAddr addr, u16 data, u16 expected) {
+    return impl->WriteExclusive16(addr, data, expected);
+}
+
+bool Memory::WriteExclusive32(VAddr addr, u32 data, u32 expected) {
+    return impl->WriteExclusive32(addr, data, expected);
+}
+
+bool Memory::WriteExclusive64(VAddr addr, u64 data, u64 expected) {
+    return impl->WriteExclusive64(addr, data, expected);
+}
+
+bool Memory::WriteExclusive128(VAddr addr, u128 data, u128 expected) {
+    return impl->WriteExclusive128(addr, data, expected);
+}
+
 std::string Memory::ReadCString(VAddr vaddr, std::size_t max_length) {
     return impl->ReadCString(vaddr, max_length);
 }
diff --git a/src/core/memory.h b/src/core/memory.h
index 9292f3b0a..4a1cc63f4 100644
--- a/src/core/memory.h
+++ b/src/core/memory.h
@@ -64,7 +64,7 @@ public:
      *
      * @param process The process to use the page table of.
      */
-    void SetCurrentPageTable(Kernel::Process& process);
+    void SetCurrentPageTable(Kernel::Process& process, u32 core_id);
 
     /**
      * Maps an allocated buffer onto a region of the emulated process address space.
@@ -245,6 +245,71 @@ public:
     void Write64(VAddr addr, u64 data);
 
     /**
+     * Writes a 8-bit unsigned integer to the given virtual address in
+     * the current process' address space if and only if the address contains
+     * the expected value. This operation is atomic.
+     *
+     * @param addr The virtual address to write the 8-bit unsigned integer to.
+     * @param data The 8-bit unsigned integer to write to the given virtual address.
+     * @param expected The 8-bit unsigned integer to check against the given virtual address.
+     *
+     * @post The memory range [addr, sizeof(data)) contains the given data value.
+     */
+    bool WriteExclusive8(VAddr addr, u8 data, u8 expected);
+
+    /**
+     * Writes a 16-bit unsigned integer to the given virtual address in
+     * the current process' address space if and only if the address contains
+     * the expected value. This operation is atomic.
+     *
+     * @param addr The virtual address to write the 16-bit unsigned integer to.
+     * @param data The 16-bit unsigned integer to write to the given virtual address.
+     * @param expected The 16-bit unsigned integer to check against the given virtual address.
+     *
+     * @post The memory range [addr, sizeof(data)) contains the given data value.
+     */
+    bool WriteExclusive16(VAddr addr, u16 data, u16 expected);
+
+    /**
+     * Writes a 32-bit unsigned integer to the given virtual address in
+     * the current process' address space if and only if the address contains
+     * the expected value. This operation is atomic.
+     *
+     * @param addr The virtual address to write the 32-bit unsigned integer to.
+     * @param data The 32-bit unsigned integer to write to the given virtual address.
+     * @param expected The 32-bit unsigned integer to check against the given virtual address.
+     *
+     * @post The memory range [addr, sizeof(data)) contains the given data value.
+     */
+    bool WriteExclusive32(VAddr addr, u32 data, u32 expected);
+
+    /**
+     * Writes a 64-bit unsigned integer to the given virtual address in
+     * the current process' address space if and only if the address contains
+     * the expected value. This operation is atomic.
+     *
+     * @param addr The virtual address to write the 64-bit unsigned integer to.
+     * @param data The 64-bit unsigned integer to write to the given virtual address.
+     * @param expected The 64-bit unsigned integer to check against the given virtual address.
+     *
+     * @post The memory range [addr, sizeof(data)) contains the given data value.
+     */
+    bool WriteExclusive64(VAddr addr, u64 data, u64 expected);
+
+    /**
+     * Writes a 128-bit unsigned integer to the given virtual address in
+     * the current process' address space if and only if the address contains
+     * the expected value. This operation is atomic.
+     *
+     * @param addr The virtual address to write the 128-bit unsigned integer to.
+     * @param data The 128-bit unsigned integer to write to the given virtual address.
+     * @param expected The 128-bit unsigned integer to check against the given virtual address.
+     *
+     * @post The memory range [addr, sizeof(data)) contains the given data value.
+     */
+    bool WriteExclusive128(VAddr addr, u128 data, u128 expected);
+
+    /**
      * Reads a null-terminated string from the given virtual address.
      * This function will continually read characters until either:
      *
diff --git a/src/core/memory/cheat_engine.cpp b/src/core/memory/cheat_engine.cpp
index b139e8465..53d27859b 100644
--- a/src/core/memory/cheat_engine.cpp
+++ b/src/core/memory/cheat_engine.cpp
@@ -20,7 +20,7 @@
 
 namespace Core::Memory {
 
-constexpr s64 CHEAT_ENGINE_TICKS = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 12);
+constexpr s64 CHEAT_ENGINE_TICKS = static_cast<s64>(1000000000 / 12);
 constexpr u32 KEYPAD_BITMASK = 0x3FFFFFF;
 
 StandardVmCallbacks::StandardVmCallbacks(Core::System& system, const CheatProcessMetadata& metadata)
@@ -190,7 +190,7 @@ CheatEngine::~CheatEngine() {
 void CheatEngine::Initialize() {
     event = Core::Timing::CreateEvent(
         "CheatEngine::FrameCallback::" + Common::HexToString(metadata.main_nso_build_id),
-        [this](u64 userdata, s64 cycles_late) { FrameCallback(userdata, cycles_late); });
+        [this](u64 userdata, s64 ns_late) { FrameCallback(userdata, ns_late); });
     core_timing.ScheduleEvent(CHEAT_ENGINE_TICKS, event);
 
     metadata.process_id = system.CurrentProcess()->GetProcessID();
@@ -217,7 +217,7 @@ void CheatEngine::Reload(std::vector<CheatEntry> cheats) {
 
 MICROPROFILE_DEFINE(Cheat_Engine, "Add-Ons", "Cheat Engine", MP_RGB(70, 200, 70));
 
-void CheatEngine::FrameCallback(u64 userdata, s64 cycles_late) {
+void CheatEngine::FrameCallback(u64 userdata, s64 ns_late) {
     if (is_pending_reload.exchange(false)) {
         vm.LoadProgram(cheats);
     }
@@ -230,7 +230,7 @@ void CheatEngine::FrameCallback(u64 userdata, s64 cycles_late) {
 
     vm.Execute(metadata);
 
-    core_timing.ScheduleEvent(CHEAT_ENGINE_TICKS - cycles_late, event);
+    core_timing.ScheduleEvent(CHEAT_ENGINE_TICKS - ns_late, event);
 }
 
 } // namespace Core::Memory
diff --git a/src/core/perf_stats.cpp b/src/core/perf_stats.cpp
index f1ae9d4df..9f3a6b811 100644
--- a/src/core/perf_stats.cpp
+++ b/src/core/perf_stats.cpp
@@ -119,7 +119,7 @@ double PerfStats::GetLastFrameTimeScale() {
 }
 
 void FrameLimiter::DoFrameLimiting(microseconds current_system_time_us) {
-    if (!Settings::values.use_frame_limit) {
+    if (!Settings::values.use_frame_limit || Settings::values.use_multi_core) {
         return;
     }
 
diff --git a/src/core/tools/freezer.cpp b/src/core/tools/freezer.cpp
index b2c6c537e..8b0c50d11 100644
--- a/src/core/tools/freezer.cpp
+++ b/src/core/tools/freezer.cpp
@@ -14,7 +14,7 @@
 namespace Tools {
 namespace {
 
-constexpr s64 MEMORY_FREEZER_TICKS = static_cast<s64>(Core::Hardware::BASE_CLOCK_RATE / 60);
+constexpr s64 MEMORY_FREEZER_TICKS = static_cast<s64>(1000000000 / 60);
 
 u64 MemoryReadWidth(Core::Memory::Memory& memory, u32 width, VAddr addr) {
     switch (width) {
@@ -57,7 +57,7 @@ Freezer::Freezer(Core::Timing::CoreTiming& core_timing_, Core::Memory::Memory& m
     : core_timing{core_timing_}, memory{memory_} {
     event = Core::Timing::CreateEvent(
         "MemoryFreezer::FrameCallback",
-        [this](u64 userdata, s64 cycles_late) { FrameCallback(userdata, cycles_late); });
+        [this](u64 userdata, s64 ns_late) { FrameCallback(userdata, ns_late); });
     core_timing.ScheduleEvent(MEMORY_FREEZER_TICKS, event);
 }
 
@@ -158,7 +158,7 @@ std::vector<Freezer::Entry> Freezer::GetEntries() const {
     return entries;
 }
 
-void Freezer::FrameCallback(u64 userdata, s64 cycles_late) {
+void Freezer::FrameCallback(u64 userdata, s64 ns_late) {
     if (!IsActive()) {
         LOG_DEBUG(Common_Memory, "Memory freezer has been deactivated, ending callback events.");
         return;
@@ -173,7 +173,7 @@ void Freezer::FrameCallback(u64 userdata, s64 cycles_late) {
         MemoryWriteWidth(memory, entry.width, entry.address, entry.value);
     }
 
-    core_timing.ScheduleEvent(MEMORY_FREEZER_TICKS - cycles_late, event);
+    core_timing.ScheduleEvent(MEMORY_FREEZER_TICKS - ns_late, event);
 }
 
 void Freezer::FillEntryReads() {
diff --git a/src/tests/CMakeLists.txt b/src/tests/CMakeLists.txt
index 3f750b51c..47ef30aa9 100644
--- a/src/tests/CMakeLists.txt
+++ b/src/tests/CMakeLists.txt
@@ -8,7 +8,6 @@ add_executable(tests
     core/arm/arm_test_common.cpp
     core/arm/arm_test_common.h
     core/core_timing.cpp
-    core/host_timing.cpp
     tests.cpp
 )
 
diff --git a/src/tests/common/fibers.cpp b/src/tests/common/fibers.cpp
index 12536b6d8..4fd92428f 100644
--- a/src/tests/common/fibers.cpp
+++ b/src/tests/common/fibers.cpp
@@ -68,7 +68,7 @@ static void ThreadStart1(u32 id, TestControl1& test_control) {
  *  doing all the work required.
  */
 TEST_CASE("Fibers::Setup", "[common]") {
-    constexpr u32 num_threads = 7;
+    constexpr std::size_t num_threads = 7;
     TestControl1 test_control{};
     test_control.thread_fibers.resize(num_threads);
     test_control.work_fibers.resize(num_threads);
diff --git a/src/tests/core/core_timing.cpp b/src/tests/core/core_timing.cpp
index ff2d11cc8..e66db1940 100644
--- a/src/tests/core/core_timing.cpp
+++ b/src/tests/core/core_timing.cpp
@@ -18,29 +18,26 @@ namespace {
 // Numbers are chosen randomly to make sure the correct one is given.
 constexpr std::array<u64, 5> CB_IDS{{42, 144, 93, 1026, UINT64_C(0xFFFF7FFFF7FFFF)}};
 constexpr int MAX_SLICE_LENGTH = 10000; // Copied from CoreTiming internals
+constexpr std::array<u64, 5> calls_order{{2, 0, 1, 4, 3}};
+std::array<s64, 5> delays{};
 
 std::bitset<CB_IDS.size()> callbacks_ran_flags;
 u64 expected_callback = 0;
-s64 lateness = 0;
 
 template <unsigned int IDX>
-void CallbackTemplate(u64 userdata, s64 cycles_late) {
+void HostCallbackTemplate(u64 userdata, s64 nanoseconds_late) {
     static_assert(IDX < CB_IDS.size(), "IDX out of range");
     callbacks_ran_flags.set(IDX);
     REQUIRE(CB_IDS[IDX] == userdata);
-    REQUIRE(CB_IDS[IDX] == expected_callback);
-    REQUIRE(lateness == cycles_late);
-}
-
-u64 callbacks_done = 0;
-
-void EmptyCallback(u64 userdata, s64 cycles_late) {
-    ++callbacks_done;
+    REQUIRE(CB_IDS[IDX] == CB_IDS[calls_order[expected_callback]]);
+    delays[IDX] = nanoseconds_late;
+    ++expected_callback;
 }
 
 struct ScopeInit final {
     ScopeInit() {
-        core_timing.Initialize();
+        core_timing.SetMulticore(true);
+        core_timing.Initialize([]() {});
     }
     ~ScopeInit() {
         core_timing.Shutdown();
@@ -49,110 +46,101 @@ struct ScopeInit final {
     Core::Timing::CoreTiming core_timing;
 };
 
-void AdvanceAndCheck(Core::Timing::CoreTiming& core_timing, u32 idx, u32 context = 0,
-                     int expected_lateness = 0, int cpu_downcount = 0) {
-    callbacks_ran_flags = 0;
-    expected_callback = CB_IDS[idx];
-    lateness = expected_lateness;
-
-    // Pretend we executed X cycles of instructions.
-    core_timing.SwitchContext(context);
-    core_timing.AddTicks(core_timing.GetDowncount() - cpu_downcount);
-    core_timing.Advance();
-    core_timing.SwitchContext((context + 1) % 4);
+#pragma optimize("", off)
 
-    REQUIRE(decltype(callbacks_ran_flags)().set(idx) == callbacks_ran_flags);
+u64 TestTimerSpeed(Core::Timing::CoreTiming& core_timing) {
+    u64 start = core_timing.GetGlobalTimeNs().count();
+    u64 placebo = 0;
+    for (std::size_t i = 0; i < 1000; i++) {
+        placebo += core_timing.GetGlobalTimeNs().count();
+    }
+    u64 end = core_timing.GetGlobalTimeNs().count();
+    return (end - start);
 }
+
+#pragma optimize("", on)
+
 } // Anonymous namespace
 
 TEST_CASE("CoreTiming[BasicOrder]", "[core]") {
     ScopeInit guard;
     auto& core_timing = guard.core_timing;
+    std::vector<std::shared_ptr<Core::Timing::EventType>> events{
+        Core::Timing::CreateEvent("callbackA", HostCallbackTemplate<0>),
+        Core::Timing::CreateEvent("callbackB", HostCallbackTemplate<1>),
+        Core::Timing::CreateEvent("callbackC", HostCallbackTemplate<2>),
+        Core::Timing::CreateEvent("callbackD", HostCallbackTemplate<3>),
+        Core::Timing::CreateEvent("callbackE", HostCallbackTemplate<4>),
+    };
+
+    expected_callback = 0;
+
+    core_timing.SyncPause(true);
+
+    u64 one_micro = 1000U;
+    for (std::size_t i = 0; i < events.size(); i++) {
+        u64 order = calls_order[i];
+        core_timing.ScheduleEvent(i * one_micro + 100U, events[order], CB_IDS[order]);
+    }
+    /// test pause
+    REQUIRE(callbacks_ran_flags.none());
 
-    std::shared_ptr<Core::Timing::EventType> cb_a =
-        Core::Timing::CreateEvent("callbackA", CallbackTemplate<0>);
-    std::shared_ptr<Core::Timing::EventType> cb_b =
-        Core::Timing::CreateEvent("callbackB", CallbackTemplate<1>);
-    std::shared_ptr<Core::Timing::EventType> cb_c =
-        Core::Timing::CreateEvent("callbackC", CallbackTemplate<2>);
-    std::shared_ptr<Core::Timing::EventType> cb_d =
-        Core::Timing::CreateEvent("callbackD", CallbackTemplate<3>);
-    std::shared_ptr<Core::Timing::EventType> cb_e =
-        Core::Timing::CreateEvent("callbackE", CallbackTemplate<4>);
-
-    // Enter slice 0
-    core_timing.ResetRun();
-
-    // D -> B -> C -> A -> E
-    core_timing.SwitchContext(0);
-    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(core_timing, 3, 0);
-    AdvanceAndCheck(core_timing, 1, 1);
-    AdvanceAndCheck(core_timing, 2, 2);
-    AdvanceAndCheck(core_timing, 0, 3);
-    AdvanceAndCheck(core_timing, 4, 0);
-}
-
-TEST_CASE("CoreTiming[FairSharing]", "[core]") {
+    core_timing.Pause(false); // No need to sync
 
-    ScopeInit guard;
-    auto& core_timing = guard.core_timing;
+    while (core_timing.HasPendingEvents())
+        ;
 
-    std::shared_ptr<Core::Timing::EventType> empty_callback =
-        Core::Timing::CreateEvent("empty_callback", EmptyCallback);
+    REQUIRE(callbacks_ran_flags.all());
 
-    callbacks_done = 0;
-    u64 MAX_CALLBACKS = 10;
-    for (std::size_t i = 0; i < 10; i++) {
-        core_timing.ScheduleEvent(i * 3333U, empty_callback, 0);
+    for (std::size_t i = 0; i < delays.size(); i++) {
+        const double delay = static_cast<double>(delays[i]);
+        const double micro = delay / 1000.0f;
+        const double mili = micro / 1000.0f;
+        printf("HostTimer Pausing Delay[%zu]: %.3f %.6f\n", i, micro, mili);
     }
-
-    const s64 advances = MAX_SLICE_LENGTH / 10;
-    core_timing.ResetRun();
-    u64 current_time = core_timing.GetTicks();
-    bool keep_running{};
-    do {
-        keep_running = false;
-        for (u32 active_core = 0; active_core < 4; ++active_core) {
-            core_timing.SwitchContext(active_core);
-            if (core_timing.CanCurrentContextRun()) {
-                core_timing.AddTicks(std::min<s64>(advances, core_timing.GetDowncount()));
-                core_timing.Advance();
-            }
-            keep_running |= core_timing.CanCurrentContextRun();
-        }
-    } while (keep_running);
-    u64 current_time_2 = core_timing.GetTicks();
-
-    REQUIRE(MAX_CALLBACKS == callbacks_done);
-    REQUIRE(current_time_2 == current_time + MAX_SLICE_LENGTH * 4);
 }
 
-TEST_CASE("Core::Timing[PredictableLateness]", "[core]") {
+TEST_CASE("CoreTiming[BasicOrderNoPausing]", "[core]") {
     ScopeInit guard;
     auto& core_timing = guard.core_timing;
+    std::vector<std::shared_ptr<Core::Timing::EventType>> events{
+        Core::Timing::CreateEvent("callbackA", HostCallbackTemplate<0>),
+        Core::Timing::CreateEvent("callbackB", HostCallbackTemplate<1>),
+        Core::Timing::CreateEvent("callbackC", HostCallbackTemplate<2>),
+        Core::Timing::CreateEvent("callbackD", HostCallbackTemplate<3>),
+        Core::Timing::CreateEvent("callbackE", HostCallbackTemplate<4>),
+    };
+
+    core_timing.SyncPause(true);
+    core_timing.SyncPause(false);
+
+    expected_callback = 0;
+
+    u64 start = core_timing.GetGlobalTimeNs().count();
+    u64 one_micro = 1000U;
+    for (std::size_t i = 0; i < events.size(); i++) {
+        u64 order = calls_order[i];
+        core_timing.ScheduleEvent(i * one_micro + 100U, events[order], CB_IDS[order]);
+    }
+    u64 end = core_timing.GetGlobalTimeNs().count();
+    const double scheduling_time = static_cast<double>(end - start);
+    const double timer_time = static_cast<double>(TestTimerSpeed(core_timing));
 
-    std::shared_ptr<Core::Timing::EventType> cb_a =
-        Core::Timing::CreateEvent("callbackA", CallbackTemplate<0>);
-    std::shared_ptr<Core::Timing::EventType> cb_b =
-        Core::Timing::CreateEvent("callbackB", CallbackTemplate<1>);
+    while (core_timing.HasPendingEvents())
+        ;
 
-    // Enter slice 0
-    core_timing.ResetRun();
+    REQUIRE(callbacks_ran_flags.all());
 
-    core_timing.ScheduleEvent(100, cb_a, CB_IDS[0]);
-    core_timing.ScheduleEvent(200, cb_b, CB_IDS[1]);
+    for (std::size_t i = 0; i < delays.size(); i++) {
+        const double delay = static_cast<double>(delays[i]);
+        const double micro = delay / 1000.0f;
+        const double mili = micro / 1000.0f;
+        printf("HostTimer No Pausing Delay[%zu]: %.3f %.6f\n", i, micro, mili);
+    }
 
-    AdvanceAndCheck(core_timing, 0, 0, 10, -10); // (100 - 10)
-    AdvanceAndCheck(core_timing, 1, 1, 50, -50);
+    const double micro = scheduling_time / 1000.0f;
+    const double mili = micro / 1000.0f;
+    printf("HostTimer No Pausing Scheduling Time: %.3f %.6f\n", micro, mili);
+    printf("HostTimer No Pausing Timer Time: %.3f %.6f\n", timer_time / 1000.f,
+           timer_time / 1000000.f);
 }
diff --git a/src/tests/core/host_timing.cpp b/src/tests/core/host_timing.cpp
deleted file mode 100644
index 556254098..000000000
--- a/src/tests/core/host_timing.cpp
+++ /dev/null
@@ -1,142 +0,0 @@
-// Copyright 2016 Dolphin Emulator Project / 2017 Dolphin Emulator Project
-// Licensed under GPLv2+
-// Refer to the license.txt file included.
-
-#include <catch2/catch.hpp>
-
-#include <array>
-#include <bitset>
-#include <cstdlib>
-#include <memory>
-#include <string>
-
-#include "common/file_util.h"
-#include "core/core.h"
-#include "core/host_timing.h"
-
-// Numbers are chosen randomly to make sure the correct one is given.
-static constexpr std::array<u64, 5> CB_IDS{{42, 144, 93, 1026, UINT64_C(0xFFFF7FFFF7FFFF)}};
-static constexpr int MAX_SLICE_LENGTH = 10000; // Copied from CoreTiming internals
-static constexpr std::array<u64, 5> calls_order{{2, 0, 1, 4, 3}};
-static std::array<s64, 5> delays{};
-
-static std::bitset<CB_IDS.size()> callbacks_ran_flags;
-static u64 expected_callback = 0;
-
-template <unsigned int IDX>
-void HostCallbackTemplate(u64 userdata, s64 nanoseconds_late) {
-    static_assert(IDX < CB_IDS.size(), "IDX out of range");
-    callbacks_ran_flags.set(IDX);
-    REQUIRE(CB_IDS[IDX] == userdata);
-    REQUIRE(CB_IDS[IDX] == CB_IDS[calls_order[expected_callback]]);
-    delays[IDX] = nanoseconds_late;
-    ++expected_callback;
-}
-
-struct ScopeInit final {
-    ScopeInit() {
-        core_timing.Initialize();
-    }
-    ~ScopeInit() {
-        core_timing.Shutdown();
-    }
-
-    Core::HostTiming::CoreTiming core_timing;
-};
-
-#pragma optimize("", off)
-
-static u64 TestTimerSpeed(Core::HostTiming::CoreTiming& core_timing) {
-    u64 start = core_timing.GetGlobalTimeNs().count();
-    u64 placebo = 0;
-    for (std::size_t i = 0; i < 1000; i++) {
-        placebo += core_timing.GetGlobalTimeNs().count();
-    }
-    u64 end = core_timing.GetGlobalTimeNs().count();
-    return (end - start);
-}
-
-#pragma optimize("", on)
-
-TEST_CASE("HostTiming[BasicOrder]", "[core]") {
-    ScopeInit guard;
-    auto& core_timing = guard.core_timing;
-    std::vector<std::shared_ptr<Core::HostTiming::EventType>> events{
-        Core::HostTiming::CreateEvent("callbackA", HostCallbackTemplate<0>),
-        Core::HostTiming::CreateEvent("callbackB", HostCallbackTemplate<1>),
-        Core::HostTiming::CreateEvent("callbackC", HostCallbackTemplate<2>),
-        Core::HostTiming::CreateEvent("callbackD", HostCallbackTemplate<3>),
-        Core::HostTiming::CreateEvent("callbackE", HostCallbackTemplate<4>),
-    };
-
-    expected_callback = 0;
-
-    core_timing.SyncPause(true);
-
-    u64 one_micro = 1000U;
-    for (std::size_t i = 0; i < events.size(); i++) {
-        u64 order = calls_order[i];
-        core_timing.ScheduleEvent(i * one_micro + 100U, events[order], CB_IDS[order]);
-    }
-    /// test pause
-    REQUIRE(callbacks_ran_flags.none());
-
-    core_timing.Pause(false); // No need to sync
-
-    while (core_timing.HasPendingEvents())
-        ;
-
-    REQUIRE(callbacks_ran_flags.all());
-
-    for (std::size_t i = 0; i < delays.size(); i++) {
-        const double delay = static_cast<double>(delays[i]);
-        const double micro = delay / 1000.0f;
-        const double mili = micro / 1000.0f;
-        printf("HostTimer Pausing Delay[%zu]: %.3f %.6f\n", i, micro, mili);
-    }
-}
-
-TEST_CASE("HostTiming[BasicOrderNoPausing]", "[core]") {
-    ScopeInit guard;
-    auto& core_timing = guard.core_timing;
-    std::vector<std::shared_ptr<Core::HostTiming::EventType>> events{
-        Core::HostTiming::CreateEvent("callbackA", HostCallbackTemplate<0>),
-        Core::HostTiming::CreateEvent("callbackB", HostCallbackTemplate<1>),
-        Core::HostTiming::CreateEvent("callbackC", HostCallbackTemplate<2>),
-        Core::HostTiming::CreateEvent("callbackD", HostCallbackTemplate<3>),
-        Core::HostTiming::CreateEvent("callbackE", HostCallbackTemplate<4>),
-    };
-
-    core_timing.SyncPause(true);
-    core_timing.SyncPause(false);
-
-    expected_callback = 0;
-
-    u64 start = core_timing.GetGlobalTimeNs().count();
-    u64 one_micro = 1000U;
-    for (std::size_t i = 0; i < events.size(); i++) {
-        u64 order = calls_order[i];
-        core_timing.ScheduleEvent(i * one_micro + 100U, events[order], CB_IDS[order]);
-    }
-    u64 end = core_timing.GetGlobalTimeNs().count();
-    const double scheduling_time = static_cast<double>(end - start);
-    const double timer_time = static_cast<double>(TestTimerSpeed(core_timing));
-
-    while (core_timing.HasPendingEvents())
-        ;
-
-    REQUIRE(callbacks_ran_flags.all());
-
-    for (std::size_t i = 0; i < delays.size(); i++) {
-        const double delay = static_cast<double>(delays[i]);
-        const double micro = delay / 1000.0f;
-        const double mili = micro / 1000.0f;
-        printf("HostTimer No Pausing Delay[%zu]: %.3f %.6f\n", i, micro, mili);
-    }
-
-    const double micro = scheduling_time / 1000.0f;
-    const double mili = micro / 1000.0f;
-    printf("HostTimer No Pausing Scheduling Time: %.3f %.6f\n", micro, mili);
-    printf("HostTimer No Pausing Timer Time: %.3f %.6f\n", timer_time / 1000.f,
-           timer_time / 1000000.f);
-}
diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp
index 8eb017f65..482e49711 100644
--- a/src/video_core/gpu.cpp
+++ b/src/video_core/gpu.cpp
@@ -2,6 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <chrono>
+
 #include "common/assert.h"
 #include "common/microprofile.h"
 #include "core/core.h"
@@ -154,8 +156,7 @@ u64 GPU::GetTicks() const {
     constexpr u64 gpu_ticks_num = 384;
     constexpr u64 gpu_ticks_den = 625;
 
-    const u64 cpu_ticks = system.CoreTiming().GetTicks();
-    u64 nanoseconds = Core::Timing::CyclesToNs(cpu_ticks).count();
+    u64 nanoseconds = system.CoreTiming().GetGlobalTimeNs().count();
     if (Settings::values.use_fast_gpu_time) {
         nanoseconds /= 256;
     }
diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h
index a1b4c305c..2c42483bd 100644
--- a/src/video_core/gpu.h
+++ b/src/video_core/gpu.h
@@ -284,6 +284,12 @@ public:
     /// core timing events.
     virtual void Start() = 0;
 
+    /// Obtain the CPU Context
+    virtual void ObtainContext() = 0;
+
+    /// Release the CPU Context
+    virtual void ReleaseContext() = 0;
+
     /// Push GPU command entries to be processed
     virtual void PushGPUEntries(Tegra::CommandList&& entries) = 0;
 
diff --git a/src/video_core/gpu_asynch.cpp b/src/video_core/gpu_asynch.cpp
index 53305ab43..7b855f63e 100644
--- a/src/video_core/gpu_asynch.cpp
+++ b/src/video_core/gpu_asynch.cpp
@@ -19,10 +19,17 @@ GPUAsynch::GPUAsynch(Core::System& system, std::unique_ptr<VideoCore::RendererBa
 GPUAsynch::~GPUAsynch() = default;
 
 void GPUAsynch::Start() {
-    cpu_context->MakeCurrent();
     gpu_thread.StartThread(*renderer, *gpu_context, *dma_pusher);
 }
 
+void GPUAsynch::ObtainContext() {
+    cpu_context->MakeCurrent();
+}
+
+void GPUAsynch::ReleaseContext() {
+    cpu_context->DoneCurrent();
+}
+
 void GPUAsynch::PushGPUEntries(Tegra::CommandList&& entries) {
     gpu_thread.SubmitList(std::move(entries));
 }
diff --git a/src/video_core/gpu_asynch.h b/src/video_core/gpu_asynch.h
index 517658612..15e9f1d38 100644
--- a/src/video_core/gpu_asynch.h
+++ b/src/video_core/gpu_asynch.h
@@ -25,6 +25,8 @@ public:
     ~GPUAsynch() override;
 
     void Start() override;
+    void ObtainContext() override;
+    void ReleaseContext() override;
     void PushGPUEntries(Tegra::CommandList&& entries) override;
     void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override;
     void FlushRegion(VAddr addr, u64 size) override;
diff --git a/src/video_core/gpu_synch.cpp b/src/video_core/gpu_synch.cpp
index 6f38a672a..aaeb9811d 100644
--- a/src/video_core/gpu_synch.cpp
+++ b/src/video_core/gpu_synch.cpp
@@ -13,10 +13,16 @@ GPUSynch::GPUSynch(Core::System& system, std::unique_ptr<VideoCore::RendererBase
 
 GPUSynch::~GPUSynch() = default;
 
-void GPUSynch::Start() {
+void GPUSynch::Start() {}
+
+void GPUSynch::ObtainContext() {
     context->MakeCurrent();
 }
 
+void GPUSynch::ReleaseContext() {
+    context->DoneCurrent();
+}
+
 void GPUSynch::PushGPUEntries(Tegra::CommandList&& entries) {
     dma_pusher->Push(std::move(entries));
     dma_pusher->DispatchCalls();
diff --git a/src/video_core/gpu_synch.h b/src/video_core/gpu_synch.h
index 4a6e9a01d..762c20aa5 100644
--- a/src/video_core/gpu_synch.h
+++ b/src/video_core/gpu_synch.h
@@ -24,6 +24,8 @@ public:
     ~GPUSynch() override;
 
     void Start() override;
+    void ObtainContext() override;
+    void ReleaseContext() override;
     void PushGPUEntries(Tegra::CommandList&& entries) override;
     void SwapBuffers(const Tegra::FramebufferConfig* framebuffer) override;
     void FlushRegion(VAddr addr, u64 size) override;
diff --git a/src/video_core/gpu_thread.cpp b/src/video_core/gpu_thread.cpp
index c3bb4fe06..738c6f0c1 100644
--- a/src/video_core/gpu_thread.cpp
+++ b/src/video_core/gpu_thread.cpp
@@ -4,6 +4,7 @@
 
 #include "common/assert.h"
 #include "common/microprofile.h"
+#include "common/thread.h"
 #include "core/core.h"
 #include "core/frontend/emu_window.h"
 #include "core/settings.h"
@@ -18,7 +19,11 @@ namespace VideoCommon::GPUThread {
 static void RunThread(Core::System& system, VideoCore::RendererBase& renderer,
                       Core::Frontend::GraphicsContext& context, Tegra::DmaPusher& dma_pusher,
                       SynchState& state) {
-    MicroProfileOnThreadCreate("GpuThread");
+    std::string name = "yuzu:GPU";
+    MicroProfileOnThreadCreate(name.c_str());
+    Common::SetCurrentThreadName(name.c_str());
+    Common::SetCurrentThreadPriority(Common::ThreadPriority::High);
+    system.RegisterHostThread();
 
     // Wait for first GPU command before acquiring the window context
     while (state.queue.Empty())
diff --git a/src/video_core/renderer_vulkan/vk_scheduler.cpp b/src/video_core/renderer_vulkan/vk_scheduler.cpp
index 82ec9180e..56524e6f3 100644
--- a/src/video_core/renderer_vulkan/vk_scheduler.cpp
+++ b/src/video_core/renderer_vulkan/vk_scheduler.cpp
@@ -9,6 +9,7 @@
 #include <utility>
 
 #include "common/microprofile.h"
+#include "common/thread.h"
 #include "video_core/renderer_vulkan/vk_device.h"
 #include "video_core/renderer_vulkan/vk_query_cache.h"
 #include "video_core/renderer_vulkan/vk_resource_manager.h"
@@ -133,6 +134,7 @@ void VKScheduler::BindGraphicsPipeline(VkPipeline pipeline) {
 }
 
 void VKScheduler::WorkerThread() {
+    Common::SetCurrentThreadPriority(Common::ThreadPriority::High);
     std::unique_lock lock{mutex};
     do {
         cv.wait(lock, [this] { return !chunk_queue.Empty() || quit; });
diff --git a/src/yuzu/bootmanager.cpp b/src/yuzu/bootmanager.cpp
index bfeb16458..4bfce48a4 100644
--- a/src/yuzu/bootmanager.cpp
+++ b/src/yuzu/bootmanager.cpp
@@ -44,49 +44,65 @@ EmuThread::EmuThread() = default;
 EmuThread::~EmuThread() = default;
 
 void EmuThread::run() {
-    MicroProfileOnThreadCreate("EmuThread");
+    std::string name = "yuzu:EmuControlThread";
+    MicroProfileOnThreadCreate(name.c_str());
+    Common::SetCurrentThreadName(name.c_str());
+
+    auto& system = Core::System::GetInstance();
+
+    system.RegisterHostThread();
+
+    auto& gpu = system.GPU();
 
     // Main process has been loaded. Make the context current to this thread and begin GPU and CPU
     // execution.
-    Core::System::GetInstance().GPU().Start();
+    gpu.Start();
+
+    gpu.ObtainContext();
 
     emit LoadProgress(VideoCore::LoadCallbackStage::Prepare, 0, 0);
 
-    Core::System::GetInstance().Renderer().Rasterizer().LoadDiskResources(
+    system.Renderer().Rasterizer().LoadDiskResources(
         stop_run, [this](VideoCore::LoadCallbackStage stage, std::size_t value, std::size_t total) {
             emit LoadProgress(stage, value, total);
         });
 
     emit LoadProgress(VideoCore::LoadCallbackStage::Complete, 0, 0);
 
+    gpu.ReleaseContext();
+
     // Holds whether the cpu was running during the last iteration,
     // so that the DebugModeLeft signal can be emitted before the
     // next execution step
     bool was_active = false;
     while (!stop_run) {
         if (running) {
-            if (!was_active)
+            if (was_active) {
                 emit DebugModeLeft();
+            }
 
-            Core::System::ResultStatus result = Core::System::GetInstance().RunLoop();
+            running_guard = true;
+            Core::System::ResultStatus result = system.Run();
             if (result != Core::System::ResultStatus::Success) {
+                running_guard = false;
                 this->SetRunning(false);
-                emit ErrorThrown(result, Core::System::GetInstance().GetStatusDetails());
+                emit ErrorThrown(result, system.GetStatusDetails());
             }
+            running_wait.Wait();
+            result = system.Pause();
+            if (result != Core::System::ResultStatus::Success) {
+                running_guard = false;
+                this->SetRunning(false);
+                emit ErrorThrown(result, system.GetStatusDetails());
+            }
+            running_guard = false;
 
-            was_active = running || exec_step;
-            if (!was_active && !stop_run)
+            if (!stop_run) {
+                was_active = true;
                 emit DebugModeEntered();
+            }
         } else if (exec_step) {
-            if (!was_active)
-                emit DebugModeLeft();
-
-            exec_step = false;
-            Core::System::GetInstance().SingleStep();
-            emit DebugModeEntered();
-            yieldCurrentThread();
-
-            was_active = false;
+            UNIMPLEMENTED();
         } else {
             std::unique_lock lock{running_mutex};
             running_cv.wait(lock, [this] { return IsRunning() || exec_step || stop_run; });
@@ -94,7 +110,7 @@ void EmuThread::run() {
     }
 
     // Shutdown the core emulation
-    Core::System::GetInstance().Shutdown();
+    system.Shutdown();
 
 #if MICROPROFILE_ENABLED
     MicroProfileOnThreadExit();
diff --git a/src/yuzu/bootmanager.h b/src/yuzu/bootmanager.h
index 3626604ca..6c59b4d5c 100644
--- a/src/yuzu/bootmanager.h
+++ b/src/yuzu/bootmanager.h
@@ -59,6 +59,12 @@ public:
         this->running = running;
         lock.unlock();
         running_cv.notify_all();
+        if (!running) {
+            running_wait.Set();
+            /// Wait until effectively paused
+            while (running_guard)
+                ;
+        }
     }
 
     /**
@@ -84,6 +90,8 @@ private:
     std::atomic_bool stop_run{false};
     std::mutex running_mutex;
     std::condition_variable running_cv;
+    Common::Event running_wait{};
+    std::atomic_bool running_guard{false};
 
 signals:
     /**
diff --git a/src/yuzu/configuration/configure_general.cpp b/src/yuzu/configuration/configure_general.cpp
index cb95423e0..74b2ad537 100644
--- a/src/yuzu/configuration/configure_general.cpp
+++ b/src/yuzu/configuration/configure_general.cpp
@@ -23,6 +23,11 @@ ConfigureGeneral::ConfigureGeneral(QWidget* parent)
 ConfigureGeneral::~ConfigureGeneral() = default;
 
 void ConfigureGeneral::SetConfiguration() {
+    const bool runtime_lock = !Core::System::GetInstance().IsPoweredOn();
+
+    ui->use_multi_core->setEnabled(runtime_lock);
+    ui->use_multi_core->setChecked(Settings::values.use_multi_core);
+
     ui->toggle_check_exit->setChecked(UISettings::values.confirm_before_closing);
     ui->toggle_user_on_boot->setChecked(UISettings::values.select_user_on_boot);
     ui->toggle_background_pause->setChecked(UISettings::values.pause_when_in_background);
@@ -41,6 +46,7 @@ void ConfigureGeneral::ApplyConfiguration() {
 
     Settings::values.use_frame_limit = ui->toggle_frame_limit->isChecked();
     Settings::values.frame_limit = ui->frame_limit->value();
+    Settings::values.use_multi_core = ui->use_multi_core->isChecked();
 }
 
 void ConfigureGeneral::changeEvent(QEvent* event) {
diff --git a/src/yuzu/configuration/configure_general.ui b/src/yuzu/configuration/configure_general.ui
index fc3b7e65a..2711116a2 100644
--- a/src/yuzu/configuration/configure_general.ui
+++ b/src/yuzu/configuration/configure_general.ui
@@ -52,6 +52,13 @@
            </layout>
           </item>
           <item>
+           <widget class="QCheckBox" name="use_multi_core">
+            <property name="text">
+             <string>Multicore CPU Emulation</string>
+            </property>
+           </widget>
+          </item>
+          <item>
            <widget class="QCheckBox" name="toggle_check_exit">
             <property name="text">
              <string>Confirm exit while emulation is running</string>
diff --git a/src/yuzu/debugger/wait_tree.cpp b/src/yuzu/debugger/wait_tree.cpp
index c1ea25fb8..9bb0a0109 100644
--- a/src/yuzu/debugger/wait_tree.cpp
+++ b/src/yuzu/debugger/wait_tree.cpp
@@ -2,10 +2,13 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <fmt/format.h>
+
 #include "yuzu/debugger/wait_tree.h"
 #include "yuzu/util/util.h"
 
 #include "common/assert.h"
+#include "core/arm/arm_interface.h"
 #include "core/core.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/mutex.h"
@@ -59,8 +62,10 @@ std::vector<std::unique_ptr<WaitTreeThread>> WaitTreeItem::MakeThreadItemList()
     std::size_t row = 0;
     auto add_threads = [&](const std::vector<std::shared_ptr<Kernel::Thread>>& threads) {
         for (std::size_t i = 0; i < threads.size(); ++i) {
-            item_list.push_back(std::make_unique<WaitTreeThread>(*threads[i]));
-            item_list.back()->row = row;
+            if (!threads[i]->IsHLEThread()) {
+                item_list.push_back(std::make_unique<WaitTreeThread>(*threads[i]));
+                item_list.back()->row = row;
+            }
             ++row;
         }
     };
@@ -114,20 +119,21 @@ QString WaitTreeCallstack::GetText() const {
 std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeCallstack::GetChildren() const {
     std::vector<std::unique_ptr<WaitTreeItem>> list;
 
-    constexpr std::size_t BaseRegister = 29;
-    auto& memory = Core::System::GetInstance().Memory();
-    u64 base_pointer = thread.GetContext64().cpu_registers[BaseRegister];
+    if (thread.IsHLEThread()) {
+        return list;
+    }
 
-    while (base_pointer != 0) {
-        const u64 lr = memory.Read64(base_pointer + sizeof(u64));
-        if (lr == 0) {
-            break;
-        }
+    if (thread.GetOwnerProcess() == nullptr || !thread.GetOwnerProcess()->Is64BitProcess()) {
+        return list;
+    }
 
-        list.push_back(std::make_unique<WaitTreeText>(
-            tr("0x%1").arg(lr - sizeof(u32), 16, 16, QLatin1Char{'0'})));
+    auto backtrace = Core::ARM_Interface::GetBacktraceFromContext(Core::System::GetInstance(),
+                                                                  thread.GetContext64());
 
-        base_pointer = memory.Read64(base_pointer);
+    for (auto& entry : backtrace) {
+        std::string s = fmt::format("{:20}{:016X} {:016X} {:016X} {}", entry.module, entry.address,
+                                    entry.original_address, entry.offset, entry.name);
+        list.push_back(std::make_unique<WaitTreeText>(QString::fromStdString(s)));
     }
 
     return list;
@@ -206,7 +212,15 @@ QString WaitTreeThread::GetText() const {
         status = tr("running");
         break;
     case Kernel::ThreadStatus::Ready:
-        status = tr("ready");
+        if (!thread.IsPaused()) {
+            if (thread.WasRunning()) {
+                status = tr("running");
+            } else {
+                status = tr("ready");
+            }
+        } else {
+            status = tr("paused");
+        }
         break;
     case Kernel::ThreadStatus::Paused:
         status = tr("paused");
@@ -254,7 +268,15 @@ QColor WaitTreeThread::GetColor() const {
     case Kernel::ThreadStatus::Running:
         return QColor(Qt::GlobalColor::darkGreen);
     case Kernel::ThreadStatus::Ready:
-        return QColor(Qt::GlobalColor::darkBlue);
+        if (!thread.IsPaused()) {
+            if (thread.WasRunning()) {
+                return QColor(Qt::GlobalColor::darkGreen);
+            } else {
+                return QColor(Qt::GlobalColor::darkBlue);
+            }
+        } else {
+            return QColor(Qt::GlobalColor::lightGray);
+        }
     case Kernel::ThreadStatus::Paused:
         return QColor(Qt::GlobalColor::lightGray);
     case Kernel::ThreadStatus::WaitHLEEvent:
@@ -319,7 +341,7 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const {
 
     if (thread.GetStatus() == Kernel::ThreadStatus::WaitSynch) {
         list.push_back(std::make_unique<WaitTreeObjectList>(thread.GetSynchronizationObjects(),
-                                                            thread.IsSleepingOnWait()));
+                                                            thread.IsWaitingSync()));
     }
 
     list.push_back(std::make_unique<WaitTreeCallstack>(thread));
diff --git a/src/yuzu/main.cpp b/src/yuzu/main.cpp
index ba69139e5..82625e67f 100644
--- a/src/yuzu/main.cpp
+++ b/src/yuzu/main.cpp
@@ -534,14 +534,36 @@ void GMainWindow::InitializeWidgets() {
         if (emulation_running) {
             return;
         }
-        Settings::values.use_asynchronous_gpu_emulation =
-            !Settings::values.use_asynchronous_gpu_emulation;
+        bool is_async =
+            !Settings::values.use_asynchronous_gpu_emulation || Settings::values.use_multi_core;
+        Settings::values.use_asynchronous_gpu_emulation = is_async;
         async_status_button->setChecked(Settings::values.use_asynchronous_gpu_emulation);
         Settings::Apply();
     });
     async_status_button->setText(tr("ASYNC"));
     async_status_button->setCheckable(true);
     async_status_button->setChecked(Settings::values.use_asynchronous_gpu_emulation);
+
+    // Setup Multicore button
+    multicore_status_button = new QPushButton();
+    multicore_status_button->setObjectName(QStringLiteral("TogglableStatusBarButton"));
+    multicore_status_button->setFocusPolicy(Qt::NoFocus);
+    connect(multicore_status_button, &QPushButton::clicked, [&] {
+        if (emulation_running) {
+            return;
+        }
+        Settings::values.use_multi_core = !Settings::values.use_multi_core;
+        bool is_async =
+            Settings::values.use_asynchronous_gpu_emulation || Settings::values.use_multi_core;
+        Settings::values.use_asynchronous_gpu_emulation = is_async;
+        async_status_button->setChecked(Settings::values.use_asynchronous_gpu_emulation);
+        multicore_status_button->setChecked(Settings::values.use_multi_core);
+        Settings::Apply();
+    });
+    multicore_status_button->setText(tr("MULTICORE"));
+    multicore_status_button->setCheckable(true);
+    multicore_status_button->setChecked(Settings::values.use_multi_core);
+    statusBar()->insertPermanentWidget(0, multicore_status_button);
     statusBar()->insertPermanentWidget(0, async_status_button);
 
     // Setup Renderer API button
@@ -925,6 +947,8 @@ bool GMainWindow::LoadROM(const QString& filename) {
         nullptr,                                     // E-Commerce
     });
 
+    system.RegisterHostThread();
+
     const Core::System::ResultStatus result{system.Load(*render_window, filename.toStdString())};
 
     const auto drd_callout =
@@ -1041,6 +1065,7 @@ void GMainWindow::BootGame(const QString& filename) {
     }
     status_bar_update_timer.start(2000);
     async_status_button->setDisabled(true);
+    multicore_status_button->setDisabled(true);
     renderer_status_button->setDisabled(true);
 
     if (UISettings::values.hide_mouse) {
@@ -1128,6 +1153,7 @@ void GMainWindow::ShutdownGame() {
     game_fps_label->setVisible(false);
     emu_frametime_label->setVisible(false);
     async_status_button->setEnabled(true);
+    multicore_status_button->setEnabled(true);
 #ifdef HAS_VULKAN
     renderer_status_button->setEnabled(true);
 #endif
@@ -1933,7 +1959,11 @@ void GMainWindow::OnConfigure() {
     }
 
     dock_status_button->setChecked(Settings::values.use_docked_mode);
+    multicore_status_button->setChecked(Settings::values.use_multi_core);
+    Settings::values.use_asynchronous_gpu_emulation =
+        Settings::values.use_asynchronous_gpu_emulation || Settings::values.use_multi_core;
     async_status_button->setChecked(Settings::values.use_asynchronous_gpu_emulation);
+
 #ifdef HAS_VULKAN
     renderer_status_button->setChecked(Settings::values.renderer_backend ==
                                        Settings::RendererBackend::Vulkan);
@@ -2060,7 +2090,7 @@ void GMainWindow::UpdateStatusBar() {
     game_fps_label->setText(tr("Game: %1 FPS").arg(results.game_fps, 0, 'f', 0));
     emu_frametime_label->setText(tr("Frame: %1 ms").arg(results.frametime * 1000.0, 0, 'f', 2));
 
-    emu_speed_label->setVisible(true);
+    emu_speed_label->setVisible(!Settings::values.use_multi_core);
     game_fps_label->setVisible(true);
     emu_frametime_label->setVisible(true);
 }
diff --git a/src/yuzu/main.h b/src/yuzu/main.h
index d55e55cc6..5581874ed 100644
--- a/src/yuzu/main.h
+++ b/src/yuzu/main.h
@@ -235,6 +235,7 @@ private:
     QLabel* game_fps_label = nullptr;
     QLabel* emu_frametime_label = nullptr;
     QPushButton* async_status_button = nullptr;
+    QPushButton* multicore_status_button = nullptr;
     QPushButton* renderer_status_button = nullptr;
     QPushButton* dock_status_button = nullptr;
     QTimer status_bar_update_timer;
diff --git a/src/yuzu_cmd/yuzu.cpp b/src/yuzu_cmd/yuzu.cpp
index 4d2ea7e9e..e6c6a839d 100644
--- a/src/yuzu_cmd/yuzu.cpp
+++ b/src/yuzu_cmd/yuzu.cpp
@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <chrono>
 #include <iostream>
 #include <memory>
 #include <string>
@@ -236,9 +237,11 @@ int main(int argc, char** argv) {
     system.Renderer().Rasterizer().LoadDiskResources();
 
     std::thread render_thread([&emu_window] { emu_window->Present(); });
+    system.Run();
     while (emu_window->IsOpen()) {
-        system.RunLoop();
+        std::this_thread::sleep_for(std::chrono::milliseconds(1));
     }
+    system.Pause();
     render_thread.join();
 
     system.Shutdown();
diff --git a/src/yuzu_tester/yuzu.cpp b/src/yuzu_tester/yuzu.cpp
index 676e70ebd..083667baf 100644
--- a/src/yuzu_tester/yuzu.cpp
+++ b/src/yuzu_tester/yuzu.cpp
@@ -2,6 +2,7 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
+#include <chrono>
 #include <iostream>
 #include <memory>
 #include <string>
@@ -255,9 +256,11 @@ int main(int argc, char** argv) {
     system.GPU().Start();
     system.Renderer().Rasterizer().LoadDiskResources();
 
+    system.Run();
     while (!finished) {
-        system.RunLoop();
+        std::this_thread::sleep_for(std::chrono::milliseconds(1));
     }
+    system.Pause();
 
     detached_tasks.WaitForAllTasks();
     return return_value;