53 files changed, 2286 insertions, 2052 deletions

diff --git a/src/common/CMakeLists.txt b/src/common/CMakeLists.txt
index 0acf70a0a..943ff996e 100644
--- a/src/common/CMakeLists.txt
+++ b/src/common/CMakeLists.txt
@@ -104,6 +104,7 @@ add_library(common STATIC
     detached_tasks.h
     bit_cast.h
     bit_field.h
+    bit_set.h
     bit_util.h
     cityhash.cpp
     cityhash.h
@@ -140,7 +141,6 @@ add_library(common STATIC
     microprofile.h
     microprofileui.h
     misc.cpp
-    multi_level_queue.h
     page_table.cpp
     page_table.h
     param_package.cpp
diff --git a/src/common/bit_set.h b/src/common/bit_set.h
new file mode 100644
index 000000000..9235ad412
--- /dev/null
+++ b/src/common/bit_set.h
@@ -0,0 +1,99 @@
+/*
+ * Copyright (c) 2018-2020 Atmosphère-NX
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#pragma once
+
+#include <array>
+#include <bit>
+
+#include "common/alignment.h"
+#include "common/bit_util.h"
+#include "common/common_types.h"
+
+namespace Common {
+
+namespace impl {
+
+template <typename Storage, size_t N>
+class BitSet {
+
+public:
+    constexpr BitSet() = default;
+
+    constexpr void SetBit(size_t i) {
+        this->words[i / FlagsPerWord] |= GetBitMask(i % FlagsPerWord);
+    }
+
+    constexpr void ClearBit(size_t i) {
+        this->words[i / FlagsPerWord] &= ~GetBitMask(i % FlagsPerWord);
+    }
+
+    constexpr size_t CountLeadingZero() const {
+        for (size_t i = 0; i < NumWords; i++) {
+            if (this->words[i]) {
+                return FlagsPerWord * i + CountLeadingZeroImpl(this->words[i]);
+            }
+        }
+        return FlagsPerWord * NumWords;
+    }
+
+    constexpr size_t GetNextSet(size_t n) const {
+        for (size_t i = (n + 1) / FlagsPerWord; i < NumWords; i++) {
+            Storage word = this->words[i];
+            if (!IsAligned(n + 1, FlagsPerWord)) {
+                word &= GetBitMask(n % FlagsPerWord) - 1;
+            }
+            if (word) {
+                return FlagsPerWord * i + CountLeadingZeroImpl(word);
+            }
+        }
+        return FlagsPerWord * NumWords;
+    }
+
+private:
+    static_assert(std::is_unsigned_v<Storage>);
+    static_assert(sizeof(Storage) <= sizeof(u64));
+
+    static constexpr size_t FlagsPerWord = BitSize<Storage>();
+    static constexpr size_t NumWords = AlignUp(N, FlagsPerWord) / FlagsPerWord;
+
+    static constexpr auto CountLeadingZeroImpl(Storage word) {
+        return std::countl_zero(static_cast<unsigned long long>(word)) -
+               (BitSize<unsigned long long>() - FlagsPerWord);
+    }
+
+    static constexpr Storage GetBitMask(size_t bit) {
+        return Storage(1) << (FlagsPerWord - 1 - bit);
+    }
+
+    std::array<Storage, NumWords> words{};
+};
+
+} // namespace impl
+
+template <size_t N>
+using BitSet8 = impl::BitSet<u8, N>;
+
+template <size_t N>
+using BitSet16 = impl::BitSet<u16, N>;
+
+template <size_t N>
+using BitSet32 = impl::BitSet<u32, N>;
+
+template <size_t N>
+using BitSet64 = impl::BitSet<u64, N>;
+
+} // namespace Common
diff --git a/src/common/multi_level_queue.h b/src/common/multi_level_queue.h
deleted file mode 100644
index 4b305bf40..000000000
--- a/src/common/multi_level_queue.h
+++ /dev/null
@@ -1,345 +0,0 @@
-// Copyright 2019 TuxSH
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#pragma once
-
-#include <array>
-#include <iterator>
-#include <list>
-#include <utility>
-
-#include "common/bit_util.h"
-#include "common/common_types.h"
-
-namespace Common {
-
-/**
- * A MultiLevelQueue is a type of priority queue which has the following characteristics:
- * - iteratable through each of its elements.
- * - back can be obtained.
- * - O(1) add, lookup (both front and back)
- * - discrete priorities and a max of 64 priorities (limited domain)
- * This type of priority queue is normaly used for managing threads within an scheduler
- */
-template <typename T, std::size_t Depth>
-class MultiLevelQueue {
-public:
-    using value_type = T;
-    using reference = value_type&;
-    using const_reference = const value_type&;
-    using pointer = value_type*;
-    using const_pointer = const value_type*;
-
-    using difference_type = typename std::pointer_traits<pointer>::difference_type;
-    using size_type = std::size_t;
-
-    template <bool is_constant>
-    class iterator_impl {
-    public:
-        using iterator_category = std::bidirectional_iterator_tag;
-        using value_type = T;
-        using pointer = std::conditional_t<is_constant, T*, const T*>;
-        using reference = std::conditional_t<is_constant, const T&, T&>;
-        using difference_type = typename std::pointer_traits<pointer>::difference_type;
-
-        friend bool operator==(const iterator_impl& lhs, const iterator_impl& rhs) {
-            if (lhs.IsEnd() && rhs.IsEnd())
-                return true;
-            return std::tie(lhs.current_priority, lhs.it) == std::tie(rhs.current_priority, rhs.it);
-        }
-
-        friend bool operator!=(const iterator_impl& lhs, const iterator_impl& rhs) {
-            return !operator==(lhs, rhs);
-        }
-
-        reference operator*() const {
-            return *it;
-        }
-
-        pointer operator->() const {
-            return it.operator->();
-        }
-
-        iterator_impl& operator++() {
-            if (IsEnd()) {
-                return *this;
-            }
-
-            ++it;
-
-            if (it == GetEndItForPrio()) {
-                u64 prios = mlq.used_priorities;
-                prios &= ~((1ULL << (current_priority + 1)) - 1);
-                if (prios == 0) {
-                    current_priority = static_cast<u32>(mlq.depth());
-                } else {
-                    current_priority = CountTrailingZeroes64(prios);
-                    it = GetBeginItForPrio();
-                }
-            }
-            return *this;
-        }
-
-        iterator_impl& operator--() {
-            if (IsEnd()) {
-                if (mlq.used_priorities != 0) {
-                    current_priority = 63 - CountLeadingZeroes64(mlq.used_priorities);
-                    it = GetEndItForPrio();
-                    --it;
-                }
-            } else if (it == GetBeginItForPrio()) {
-                u64 prios = mlq.used_priorities;
-                prios &= (1ULL << current_priority) - 1;
-                if (prios != 0) {
-                    current_priority = CountTrailingZeroes64(prios);
-                    it = GetEndItForPrio();
-                    --it;
-                }
-            } else {
-                --it;
-            }
-            return *this;
-        }
-
-        iterator_impl operator++(int) {
-            const iterator_impl v{*this};
-            ++(*this);
-            return v;
-        }
-
-        iterator_impl operator--(int) {
-            const iterator_impl v{*this};
-            --(*this);
-            return v;
-        }
-
-        // allow implicit const->non-const
-        iterator_impl(const iterator_impl<false>& other)
-            : mlq(other.mlq), it(other.it), current_priority(other.current_priority) {}
-
-        iterator_impl(const iterator_impl<true>& other)
-            : mlq(other.mlq), it(other.it), current_priority(other.current_priority) {}
-
-        iterator_impl& operator=(const iterator_impl<false>& other) {
-            mlq = other.mlq;
-            it = other.it;
-            current_priority = other.current_priority;
-            return *this;
-        }
-
-        friend class iterator_impl<true>;
-        iterator_impl() = default;
-
-    private:
-        friend class MultiLevelQueue;
-        using container_ref =
-            std::conditional_t<is_constant, const MultiLevelQueue&, MultiLevelQueue&>;
-        using list_iterator = std::conditional_t<is_constant, typename std::list<T>::const_iterator,
-                                                 typename std::list<T>::iterator>;
-
-        explicit iterator_impl(container_ref mlq, list_iterator it, u32 current_priority)
-            : mlq(mlq), it(it), current_priority(current_priority) {}
-        explicit iterator_impl(container_ref mlq, u32 current_priority)
-            : mlq(mlq), it(), current_priority(current_priority) {}
-
-        bool IsEnd() const {
-            return current_priority == mlq.depth();
-        }
-
-        list_iterator GetBeginItForPrio() const {
-            return mlq.levels[current_priority].begin();
-        }
-
-        list_iterator GetEndItForPrio() const {
-            return mlq.levels[current_priority].end();
-        }
-
-        container_ref mlq;
-        list_iterator it;
-        u32 current_priority;
-    };
-
-    using iterator = iterator_impl<false>;
-    using const_iterator = iterator_impl<true>;
-
-    void add(const T& element, u32 priority, bool send_back = true) {
-        if (send_back)
-            levels[priority].push_back(element);
-        else
-            levels[priority].push_front(element);
-        used_priorities |= 1ULL << priority;
-    }
-
-    void remove(const T& element, u32 priority) {
-        auto it = ListIterateTo(levels[priority], element);
-        if (it == levels[priority].end())
-            return;
-        levels[priority].erase(it);
-        if (levels[priority].empty()) {
-            used_priorities &= ~(1ULL << priority);
-        }
-    }
-
-    void adjust(const T& element, u32 old_priority, u32 new_priority, bool adjust_front = false) {
-        remove(element, old_priority);
-        add(element, new_priority, !adjust_front);
-    }
-    void adjust(const_iterator it, u32 old_priority, u32 new_priority, bool adjust_front = false) {
-        adjust(*it, old_priority, new_priority, adjust_front);
-    }
-
-    void transfer_to_front(const T& element, u32 priority, MultiLevelQueue& other) {
-        ListSplice(other.levels[priority], other.levels[priority].begin(), levels[priority],
-                   ListIterateTo(levels[priority], element));
-
-        other.used_priorities |= 1ULL << priority;
-
-        if (levels[priority].empty()) {
-            used_priorities &= ~(1ULL << priority);
-        }
-    }
-
-    void transfer_to_front(const_iterator it, u32 priority, MultiLevelQueue& other) {
-        transfer_to_front(*it, priority, other);
-    }
-
-    void transfer_to_back(const T& element, u32 priority, MultiLevelQueue& other) {
-        ListSplice(other.levels[priority], other.levels[priority].end(), levels[priority],
-                   ListIterateTo(levels[priority], element));
-
-        other.used_priorities |= 1ULL << priority;
-
-        if (levels[priority].empty()) {
-            used_priorities &= ~(1ULL << priority);
-        }
-    }
-
-    void transfer_to_back(const_iterator it, u32 priority, MultiLevelQueue& other) {
-        transfer_to_back(*it, priority, other);
-    }
-
-    void yield(u32 priority, std::size_t n = 1) {
-        ListShiftForward(levels[priority], n);
-    }
-
-    [[nodiscard]] std::size_t depth() const {
-        return Depth;
-    }
-
-    [[nodiscard]] std::size_t size(u32 priority) const {
-        return levels[priority].size();
-    }
-
-    [[nodiscard]] std::size_t size() const {
-        u64 priorities = used_priorities;
-        std::size_t size = 0;
-        while (priorities != 0) {
-            const u64 current_priority = CountTrailingZeroes64(priorities);
-            size += levels[current_priority].size();
-            priorities &= ~(1ULL << current_priority);
-        }
-        return size;
-    }
-
-    [[nodiscard]] bool empty() const {
-        return used_priorities == 0;
-    }
-
-    [[nodiscard]] bool empty(u32 priority) const {
-        return (used_priorities & (1ULL << priority)) == 0;
-    }
-
-    [[nodiscard]] u32 highest_priority_set(u32 max_priority = 0) const {
-        const u64 priorities =
-            max_priority == 0 ? used_priorities : (used_priorities & ~((1ULL << max_priority) - 1));
-        return priorities == 0 ? Depth : static_cast<u32>(CountTrailingZeroes64(priorities));
-    }
-
-    [[nodiscard]] u32 lowest_priority_set(u32 min_priority = Depth - 1) const {
-        const u64 priorities = min_priority >= Depth - 1
-                                   ? used_priorities
-                                   : (used_priorities & ((1ULL << (min_priority + 1)) - 1));
-        return priorities == 0 ? Depth : 63 - CountLeadingZeroes64(priorities);
-    }
-
-    [[nodiscard]] const_iterator cbegin(u32 max_prio = 0) const {
-        const u32 priority = highest_priority_set(max_prio);
-        return priority == Depth ? cend()
-                                 : const_iterator{*this, levels[priority].cbegin(), priority};
-    }
-    [[nodiscard]] const_iterator begin(u32 max_prio = 0) const {
-        return cbegin(max_prio);
-    }
-    [[nodiscard]] iterator begin(u32 max_prio = 0) {
-        const u32 priority = highest_priority_set(max_prio);
-        return priority == Depth ? end() : iterator{*this, levels[priority].begin(), priority};
-    }
-
-    [[nodiscard]] const_iterator cend(u32 min_prio = Depth - 1) const {
-        return min_prio == Depth - 1 ? const_iterator{*this, Depth} : cbegin(min_prio + 1);
-    }
-    [[nodiscard]] const_iterator end(u32 min_prio = Depth - 1) const {
-        return cend(min_prio);
-    }
-    [[nodiscard]] iterator end(u32 min_prio = Depth - 1) {
-        return min_prio == Depth - 1 ? iterator{*this, Depth} : begin(min_prio + 1);
-    }
-
-    [[nodiscard]] T& front(u32 max_priority = 0) {
-        const u32 priority = highest_priority_set(max_priority);
-        return levels[priority == Depth ? 0 : priority].front();
-    }
-    [[nodiscard]] const T& front(u32 max_priority = 0) const {
-        const u32 priority = highest_priority_set(max_priority);
-        return levels[priority == Depth ? 0 : priority].front();
-    }
-
-    [[nodiscard]] T& back(u32 min_priority = Depth - 1) {
-        const u32 priority = lowest_priority_set(min_priority); // intended
-        return levels[priority == Depth ? 63 : priority].back();
-    }
-    [[nodiscard]] const T& back(u32 min_priority = Depth - 1) const {
-        const u32 priority = lowest_priority_set(min_priority); // intended
-        return levels[priority == Depth ? 63 : priority].back();
-    }
-
-    void clear() {
-        used_priorities = 0;
-        for (std::size_t i = 0; i < Depth; i++) {
-            levels[i].clear();
-        }
-    }
-
-private:
-    using const_list_iterator = typename std::list<T>::const_iterator;
-
-    static void ListShiftForward(std::list<T>& list, const std::size_t shift = 1) {
-        if (shift >= list.size()) {
-            return;
-        }
-
-        const auto begin_range = list.begin();
-        const auto end_range = std::next(begin_range, shift);
-        list.splice(list.end(), list, begin_range, end_range);
-    }
-
-    static void ListSplice(std::list<T>& in_list, const_list_iterator position,
-                           std::list<T>& out_list, const_list_iterator element) {
-        in_list.splice(position, out_list, element);
-    }
-
-    [[nodiscard]] static const_list_iterator ListIterateTo(const std::list<T>& list,
-                                                           const T& element) {
-        auto it = list.cbegin();
-        while (it != list.cend() && *it != element) {
-            ++it;
-        }
-        return it;
-    }
-
-    std::array<std::list<T>, Depth> levels;
-    u64 used_priorities = 0;
-};
-
-} // namespace Common
diff --git a/src/core/CMakeLists.txt b/src/core/CMakeLists.txt
index 56c165336..59bd3d2a6 100644
--- a/src/core/CMakeLists.txt
+++ b/src/core/CMakeLists.txt
@@ -151,10 +151,19 @@ add_library(core STATIC
     hle/kernel/code_set.cpp
     hle/kernel/code_set.h
     hle/kernel/errors.h
+    hle/kernel/global_scheduler_context.cpp
+    hle/kernel/global_scheduler_context.h
     hle/kernel/handle_table.cpp
     hle/kernel/handle_table.h
     hle/kernel/hle_ipc.cpp
     hle/kernel/hle_ipc.h
+    hle/kernel/k_affinity_mask.h
+    hle/kernel/k_priority_queue.h
+    hle/kernel/k_scheduler.cpp
+    hle/kernel/k_scheduler.h
+    hle/kernel/k_scheduler_lock.h
+    hle/kernel/k_scoped_lock.h
+    hle/kernel/k_scoped_scheduler_lock_and_sleep.h
     hle/kernel/kernel.cpp
     hle/kernel/kernel.h
     hle/kernel/memory/address_space_info.cpp
@@ -189,8 +198,6 @@ add_library(core STATIC
     hle/kernel/readable_event.h
     hle/kernel/resource_limit.cpp
     hle/kernel/resource_limit.h
-    hle/kernel/scheduler.cpp
-    hle/kernel/scheduler.h
     hle/kernel/server_port.cpp
     hle/kernel/server_port.h
     hle/kernel/server_session.cpp
diff --git a/src/core/arm/dynarmic/arm_dynarmic_32.cpp b/src/core/arm/dynarmic/arm_dynarmic_32.cpp
index 193fd7d62..e9c74b1a6 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_32.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic_32.cpp
@@ -294,6 +294,9 @@ void ARM_Dynarmic_32::InvalidateCacheRange(VAddr addr, std::size_t size) {
 }
 
 void ARM_Dynarmic_32::ClearExclusiveState() {
+    if (!jit) {
+        return;
+    }
     jit->ClearExclusiveState();
 }
 
diff --git a/src/core/arm/dynarmic/arm_dynarmic_64.cpp b/src/core/arm/dynarmic/arm_dynarmic_64.cpp
index 0f0585d0f..7a4eb88a2 100644
--- a/src/core/arm/dynarmic/arm_dynarmic_64.cpp
+++ b/src/core/arm/dynarmic/arm_dynarmic_64.cpp
@@ -15,8 +15,8 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/hardware_properties.h"
+#include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/process.h"
-#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/svc.h"
 #include "core/memory.h"
 #include "core/settings.h"
@@ -330,6 +330,9 @@ void ARM_Dynarmic_64::InvalidateCacheRange(VAddr addr, std::size_t size) {
 }
 
 void ARM_Dynarmic_64::ClearExclusiveState() {
+    if (!jit) {
+        return;
+    }
     jit->ClearExclusiveState();
 }
 
diff --git a/src/core/core.cpp b/src/core/core.cpp
index 7e3c54618..0961c0819 100644
--- a/src/core/core.cpp
+++ b/src/core/core.cpp
@@ -27,10 +27,10 @@
 #include "core/file_sys/vfs_real.h"
 #include "core/hardware_interrupt_manager.h"
 #include "core/hle/kernel/client_port.h"
+#include "core/hle/kernel/k_scheduler.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/thread.h"
 #include "core/hle/service/am/applets/applets.h"
 #include "core/hle/service/apm/controller.h"
@@ -507,14 +507,6 @@ std::size_t System::CurrentCoreIndex() const {
     return core;
 }
 
-Kernel::Scheduler& System::CurrentScheduler() {
-    return impl->kernel.CurrentScheduler();
-}
-
-const Kernel::Scheduler& System::CurrentScheduler() const {
-    return impl->kernel.CurrentScheduler();
-}
-
 Kernel::PhysicalCore& System::CurrentPhysicalCore() {
     return impl->kernel.CurrentPhysicalCore();
 }
@@ -523,22 +515,14 @@ const Kernel::PhysicalCore& System::CurrentPhysicalCore() const {
     return impl->kernel.CurrentPhysicalCore();
 }
 
-Kernel::Scheduler& System::Scheduler(std::size_t core_index) {
-    return impl->kernel.Scheduler(core_index);
-}
-
-const Kernel::Scheduler& System::Scheduler(std::size_t core_index) const {
-    return impl->kernel.Scheduler(core_index);
-}
-
 /// Gets the global scheduler
-Kernel::GlobalScheduler& System::GlobalScheduler() {
-    return impl->kernel.GlobalScheduler();
+Kernel::GlobalSchedulerContext& System::GlobalSchedulerContext() {
+    return impl->kernel.GlobalSchedulerContext();
 }
 
 /// Gets the global scheduler
-const Kernel::GlobalScheduler& System::GlobalScheduler() const {
-    return impl->kernel.GlobalScheduler();
+const Kernel::GlobalSchedulerContext& System::GlobalSchedulerContext() const {
+    return impl->kernel.GlobalSchedulerContext();
 }
 
 Kernel::Process* System::CurrentProcess() {
diff --git a/src/core/core.h b/src/core/core.h
index 29b8fb92a..579a774e4 100644
--- a/src/core/core.h
+++ b/src/core/core.h
@@ -26,11 +26,11 @@ class VfsFilesystem;
 } // namespace FileSys
 
 namespace Kernel {
-class GlobalScheduler;
+class GlobalSchedulerContext;
 class KernelCore;
 class PhysicalCore;
 class Process;
-class Scheduler;
+class KScheduler;
 } // namespace Kernel
 
 namespace Loader {
@@ -213,12 +213,6 @@ public:
     /// Gets the index of the currently running CPU core
     [[nodiscard]] std::size_t CurrentCoreIndex() const;
 
-    /// Gets the scheduler for the CPU core that is currently running
-    [[nodiscard]] Kernel::Scheduler& CurrentScheduler();
-
-    /// Gets the scheduler for the CPU core that is currently running
-    [[nodiscard]] const Kernel::Scheduler& CurrentScheduler() const;
-
     /// Gets the physical core for the CPU core that is currently running
     [[nodiscard]] Kernel::PhysicalCore& CurrentPhysicalCore();
 
@@ -261,17 +255,11 @@ public:
     /// Gets an immutable reference to the renderer.
     [[nodiscard]] const VideoCore::RendererBase& Renderer() const;
 
-    /// Gets the scheduler for the CPU core with the specified index
-    [[nodiscard]] Kernel::Scheduler& Scheduler(std::size_t core_index);
-
-    /// Gets the scheduler for the CPU core with the specified index
-    [[nodiscard]] const Kernel::Scheduler& Scheduler(std::size_t core_index) const;
-
     /// Gets the global scheduler
-    [[nodiscard]] Kernel::GlobalScheduler& GlobalScheduler();
+    [[nodiscard]] Kernel::GlobalSchedulerContext& GlobalSchedulerContext();
 
     /// Gets the global scheduler
-    [[nodiscard]] const Kernel::GlobalScheduler& GlobalScheduler() const;
+    [[nodiscard]] const Kernel::GlobalSchedulerContext& GlobalSchedulerContext() const;
 
     /// Gets the manager for the guest device memory
     [[nodiscard]] Core::DeviceMemory& DeviceMemory();
diff --git a/src/core/cpu_manager.cpp b/src/core/cpu_manager.cpp
index 0cff985e9..373395047 100644
--- a/src/core/cpu_manager.cpp
+++ b/src/core/cpu_manager.cpp
@@ -10,9 +10,9 @@
 #include "core/core.h"
 #include "core/core_timing.h"
 #include "core/cpu_manager.h"
+#include "core/hle/kernel/k_scheduler.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"
 
@@ -109,11 +109,8 @@ void* CpuManager::GetStartFuncParamater() {
 
 void CpuManager::MultiCoreRunGuestThread() {
     auto& kernel = system.Kernel();
-    {
-        auto& sched = kernel.CurrentScheduler();
-        sched.OnThreadStart();
-    }
-    auto* thread = kernel.CurrentScheduler().GetCurrentThread();
+    kernel.CurrentScheduler()->OnThreadStart();
+    auto* thread = kernel.CurrentScheduler()->GetCurrentThread();
     auto& host_context = thread->GetHostContext();
     host_context->SetRewindPoint(GuestRewindFunction, this);
     MultiCoreRunGuestLoop();
@@ -130,8 +127,8 @@ void CpuManager::MultiCoreRunGuestLoop() {
             physical_core = &kernel.CurrentPhysicalCore();
         }
         system.ExitDynarmicProfile();
-        auto& scheduler = kernel.CurrentScheduler();
-        scheduler.TryDoContextSwitch();
+        physical_core->ArmInterface().ClearExclusiveState();
+        kernel.CurrentScheduler()->RescheduleCurrentCore();
     }
 }
 
@@ -140,25 +137,21 @@ void CpuManager::MultiCoreRunIdleThread() {
     while (true) {
         auto& physical_core = kernel.CurrentPhysicalCore();
         physical_core.Idle();
-        auto& scheduler = kernel.CurrentScheduler();
-        scheduler.TryDoContextSwitch();
+        kernel.CurrentScheduler()->RescheduleCurrentCore();
     }
 }
 
 void CpuManager::MultiCoreRunSuspendThread() {
     auto& kernel = system.Kernel();
-    {
-        auto& sched = kernel.CurrentScheduler();
-        sched.OnThreadStart();
-    }
+    kernel.CurrentScheduler()->OnThreadStart();
     while (true) {
         auto core = kernel.GetCurrentHostThreadID();
-        auto& scheduler = kernel.CurrentScheduler();
+        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();
+        scheduler.RescheduleCurrentCore();
     }
 }
 
@@ -206,11 +199,8 @@ void CpuManager::MultiCorePause(bool paused) {
 
 void CpuManager::SingleCoreRunGuestThread() {
     auto& kernel = system.Kernel();
-    {
-        auto& sched = kernel.CurrentScheduler();
-        sched.OnThreadStart();
-    }
-    auto* thread = kernel.CurrentScheduler().GetCurrentThread();
+    kernel.CurrentScheduler()->OnThreadStart();
+    auto* thread = kernel.CurrentScheduler()->GetCurrentThread();
     auto& host_context = thread->GetHostContext();
     host_context->SetRewindPoint(GuestRewindFunction, this);
     SingleCoreRunGuestLoop();
@@ -218,7 +208,7 @@ void CpuManager::SingleCoreRunGuestThread() {
 
 void CpuManager::SingleCoreRunGuestLoop() {
     auto& kernel = system.Kernel();
-    auto* thread = kernel.CurrentScheduler().GetCurrentThread();
+    auto* thread = kernel.CurrentScheduler()->GetCurrentThread();
     while (true) {
         auto* physical_core = &kernel.CurrentPhysicalCore();
         system.EnterDynarmicProfile();
@@ -230,9 +220,10 @@ void CpuManager::SingleCoreRunGuestLoop() {
         thread->SetPhantomMode(true);
         system.CoreTiming().Advance();
         thread->SetPhantomMode(false);
+        physical_core->ArmInterface().ClearExclusiveState();
         PreemptSingleCore();
         auto& scheduler = kernel.Scheduler(current_core);
-        scheduler.TryDoContextSwitch();
+        scheduler.RescheduleCurrentCore();
     }
 }
 
@@ -244,51 +235,53 @@ void CpuManager::SingleCoreRunIdleThread() {
         system.CoreTiming().AddTicks(1000U);
         idle_count++;
         auto& scheduler = physical_core.Scheduler();
-        scheduler.TryDoContextSwitch();
+        scheduler.RescheduleCurrentCore();
     }
 }
 
 void CpuManager::SingleCoreRunSuspendThread() {
     auto& kernel = system.Kernel();
-    {
-        auto& sched = kernel.CurrentScheduler();
-        sched.OnThreadStart();
-    }
+    kernel.CurrentScheduler()->OnThreadStart();
     while (true) {
         auto core = kernel.GetCurrentHostThreadID();
-        auto& scheduler = kernel.CurrentScheduler();
+        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();
+        scheduler.RescheduleCurrentCore();
     }
 }
 
 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;
+    {
+        auto& scheduler = system.Kernel().Scheduler(current_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_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(scheduler.GetCurrentThread());
+
+        auto& next_scheduler = system.Kernel().Scheduler(current_core);
+        Common::Fiber::YieldTo(current_thread->GetHostContext(), next_scheduler.ControlContext());
     }
-    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;
+
+    // May have changed scheduler
+    {
+        auto& scheduler = system.Kernel().Scheduler(current_core);
+        scheduler.Reload(scheduler.GetCurrentThread());
+        auto* currrent_thread2 = scheduler.GetCurrentThread();
+        if (!currrent_thread2->IsIdleThread()) {
+            idle_count = 0;
+        }
     }
 }
 
@@ -369,8 +362,7 @@ void CpuManager::RunThread(std::size_t core) {
             return;
         }
 
-        auto& scheduler = system.Kernel().CurrentScheduler();
-        Kernel::Thread* current_thread = scheduler.GetCurrentThread();
+        auto current_thread = system.Kernel().CurrentScheduler()->GetCurrentThread();
         data.is_running = true;
         Common::Fiber::YieldTo(data.host_context, current_thread->GetHostContext());
         data.is_running = false;
diff --git a/src/core/hle/kernel/address_arbiter.cpp b/src/core/hle/kernel/address_arbiter.cpp
index 048acd30e..20ffa7d47 100644
--- a/src/core/hle/kernel/address_arbiter.cpp
+++ b/src/core/hle/kernel/address_arbiter.cpp
@@ -12,8 +12,9 @@
 #include "core/hle/kernel/address_arbiter.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.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"
@@ -58,7 +59,7 @@ ResultCode AddressArbiter::SignalToAddress(VAddr address, SignalType type, s32 v
 }
 
 ResultCode AddressArbiter::SignalToAddressOnly(VAddr address, s32 num_to_wake) {
-    SchedulerLock lock(system.Kernel());
+    KScopedSchedulerLock lock(system.Kernel());
     const std::vector<std::shared_ptr<Thread>> waiting_threads =
         GetThreadsWaitingOnAddress(address);
     WakeThreads(waiting_threads, num_to_wake);
@@ -67,7 +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());
+    KScopedSchedulerLock lock(system.Kernel());
     auto& memory = system.Memory();
 
     // Ensure that we can write to the address.
@@ -92,7 +93,7 @@ ResultCode AddressArbiter::IncrementAndSignalToAddressIfEqual(VAddr address, s32
 
 ResultCode AddressArbiter::ModifyByWaitingCountAndSignalToAddressIfEqual(VAddr address, s32 value,
                                                                          s32 num_to_wake) {
-    SchedulerLock lock(system.Kernel());
+    KScopedSchedulerLock lock(system.Kernel());
     auto& memory = system.Memory();
 
     // Ensure that we can write to the address.
@@ -153,11 +154,11 @@ ResultCode AddressArbiter::WaitForAddressIfLessThan(VAddr address, s32 value, s6
                                                     bool should_decrement) {
     auto& memory = system.Memory();
     auto& kernel = system.Kernel();
-    Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
+    Thread* current_thread = kernel.CurrentScheduler()->GetCurrentThread();
 
     Handle event_handle = InvalidHandle;
     {
-        SchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout);
+        KScopedSchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout);
 
         if (current_thread->IsPendingTermination()) {
             lock.CancelSleep();
@@ -210,7 +211,7 @@ ResultCode AddressArbiter::WaitForAddressIfLessThan(VAddr address, s32 value, s6
     }
 
     {
-        SchedulerLock lock(kernel);
+        KScopedSchedulerLock lock(kernel);
         if (current_thread->IsWaitingForArbitration()) {
             RemoveThread(SharedFrom(current_thread));
             current_thread->WaitForArbitration(false);
@@ -223,11 +224,11 @@ ResultCode AddressArbiter::WaitForAddressIfLessThan(VAddr address, s32 value, s6
 ResultCode AddressArbiter::WaitForAddressIfEqual(VAddr address, s32 value, s64 timeout) {
     auto& memory = system.Memory();
     auto& kernel = system.Kernel();
-    Thread* current_thread = system.CurrentScheduler().GetCurrentThread();
+    Thread* current_thread = kernel.CurrentScheduler()->GetCurrentThread();
 
     Handle event_handle = InvalidHandle;
     {
-        SchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout);
+        KScopedSchedulerLockAndSleep lock(kernel, event_handle, current_thread, timeout);
 
         if (current_thread->IsPendingTermination()) {
             lock.CancelSleep();
@@ -265,7 +266,7 @@ ResultCode AddressArbiter::WaitForAddressIfEqual(VAddr address, s32 value, s64 t
     }
 
     {
-        SchedulerLock lock(kernel);
+        KScopedSchedulerLock lock(kernel);
         if (current_thread->IsWaitingForArbitration()) {
             RemoveThread(SharedFrom(current_thread));
             current_thread->WaitForArbitration(false);
diff --git a/src/core/hle/kernel/global_scheduler_context.cpp b/src/core/hle/kernel/global_scheduler_context.cpp
new file mode 100644
index 000000000..a133e8ed0
--- /dev/null
+++ b/src/core/hle/kernel/global_scheduler_context.cpp
@@ -0,0 +1,52 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include <mutex>
+
+#include "common/assert.h"
+#include "core/core.h"
+#include "core/hle/kernel/global_scheduler_context.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/kernel.h"
+
+namespace Kernel {
+
+GlobalSchedulerContext::GlobalSchedulerContext(KernelCore& kernel)
+    : kernel{kernel}, scheduler_lock{kernel} {}
+
+GlobalSchedulerContext::~GlobalSchedulerContext() = default;
+
+void GlobalSchedulerContext::AddThread(std::shared_ptr<Thread> thread) {
+    std::scoped_lock lock{global_list_guard};
+    thread_list.push_back(std::move(thread));
+}
+
+void GlobalSchedulerContext::RemoveThread(std::shared_ptr<Thread> thread) {
+    std::scoped_lock lock{global_list_guard};
+    thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread),
+                      thread_list.end());
+}
+
+void GlobalSchedulerContext::PreemptThreads() {
+    // The priority levels at which the global scheduler preempts threads every 10 ms. They are
+    // ordered from Core 0 to Core 3.
+    static constexpr std::array<u32, Core::Hardware::NUM_CPU_CORES> preemption_priorities{
+        59,
+        59,
+        59,
+        63,
+    };
+
+    ASSERT(IsLocked());
+    for (u32 core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) {
+        const u32 priority = preemption_priorities[core_id];
+        kernel.Scheduler(core_id).RotateScheduledQueue(core_id, priority);
+    }
+}
+
+bool GlobalSchedulerContext::IsLocked() const {
+    return scheduler_lock.IsLockedByCurrentThread();
+}
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/global_scheduler_context.h b/src/core/hle/kernel/global_scheduler_context.h
new file mode 100644
index 000000000..5c7b89290
--- /dev/null
+++ b/src/core/hle/kernel/global_scheduler_context.h
@@ -0,0 +1,81 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <atomic>
+#include <vector>
+
+#include "common/common_types.h"
+#include "common/spin_lock.h"
+#include "core/hardware_properties.h"
+#include "core/hle/kernel/k_priority_queue.h"
+#include "core/hle/kernel/k_scheduler_lock.h"
+#include "core/hle/kernel/thread.h"
+
+namespace Kernel {
+
+class KernelCore;
+class SchedulerLock;
+
+using KSchedulerPriorityQueue =
+    KPriorityQueue<Thread, Core::Hardware::NUM_CPU_CORES, THREADPRIO_LOWEST, THREADPRIO_HIGHEST>;
+constexpr s32 HighestCoreMigrationAllowedPriority = 2;
+
+class GlobalSchedulerContext final {
+    friend class KScheduler;
+
+public:
+    using LockType = KAbstractSchedulerLock<KScheduler>;
+
+    explicit GlobalSchedulerContext(KernelCore& kernel);
+    ~GlobalSchedulerContext();
+
+    /// Adds a new thread to the scheduler
+    void AddThread(std::shared_ptr<Thread> thread);
+
+    /// Removes a thread from the scheduler
+    void RemoveThread(std::shared_ptr<Thread> thread);
+
+    /// Returns a list of all threads managed by the scheduler
+    [[nodiscard]] const std::vector<std::shared_ptr<Thread>>& GetThreadList() const {
+        return thread_list;
+    }
+
+    /**
+     * Rotates the scheduling queues of threads at a preemption priority and then does
+     * some core rebalancing. Preemption priorities can be found in the array
+     * 'preemption_priorities'.
+     *
+     * @note This operation happens every 10ms.
+     */
+    void PreemptThreads();
+
+    /// Returns true if the global scheduler lock is acquired
+    bool IsLocked() const;
+
+    [[nodiscard]] LockType& SchedulerLock() {
+        return scheduler_lock;
+    }
+
+    [[nodiscard]] const LockType& SchedulerLock() const {
+        return scheduler_lock;
+    }
+
+private:
+    friend class KScopedSchedulerLock;
+    friend class KScopedSchedulerLockAndSleep;
+
+    KernelCore& kernel;
+
+    std::atomic_bool scheduler_update_needed{};
+    KSchedulerPriorityQueue priority_queue;
+    LockType scheduler_lock;
+
+    /// Lists all thread ids that aren't deleted/etc.
+    std::vector<std::shared_ptr<Thread>> thread_list;
+    Common::SpinLock global_list_guard{};
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/handle_table.cpp b/src/core/hle/kernel/handle_table.cpp
index 3e745c18b..40988b0fd 100644
--- a/src/core/hle/kernel/handle_table.cpp
+++ b/src/core/hle/kernel/handle_table.cpp
@@ -8,9 +8,9 @@
 #include "core/core.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/process.h"
-#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
 
 namespace Kernel {
@@ -105,7 +105,7 @@ bool HandleTable::IsValid(Handle handle) const {
 
 std::shared_ptr<Object> HandleTable::GetGeneric(Handle handle) const {
     if (handle == CurrentThread) {
-        return SharedFrom(kernel.CurrentScheduler().GetCurrentThread());
+        return SharedFrom(kernel.CurrentScheduler()->GetCurrentThread());
     } else if (handle == CurrentProcess) {
         return SharedFrom(kernel.CurrentProcess());
     }
diff --git a/src/core/hle/kernel/hle_ipc.cpp b/src/core/hle/kernel/hle_ipc.cpp
index 81f85643b..e75e80ad0 100644
--- a/src/core/hle/kernel/hle_ipc.cpp
+++ b/src/core/hle/kernel/hle_ipc.cpp
@@ -17,11 +17,12 @@
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/hle_ipc.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.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"
@@ -56,9 +57,9 @@ std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread(
         writable_event = pair.writable;
     }
 
+    Handle event_handle = InvalidHandle;
     {
-        Handle event_handle = InvalidHandle;
-        SchedulerLockAndSleep lock(kernel, event_handle, thread.get(), timeout);
+        KScopedSchedulerLockAndSleep 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
@@ -74,9 +75,8 @@ std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread(
         thread->SetStatus(ThreadStatus::WaitHLEEvent);
         thread->SetSynchronizationResults(nullptr, RESULT_TIMEOUT);
         readable_event->AddWaitingThread(thread);
-        lock.Release();
-        thread->SetHLETimeEvent(event_handle);
     }
+    thread->SetHLETimeEvent(event_handle);
 
     is_thread_waiting = true;
 
diff --git a/src/core/hle/kernel/k_affinity_mask.h b/src/core/hle/kernel/k_affinity_mask.h
new file mode 100644
index 000000000..dd73781cd
--- /dev/null
+++ b/src/core/hle/kernel/k_affinity_mask.h
@@ -0,0 +1,58 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include "common/assert.h"
+#include "common/common_types.h"
+#include "core/hardware_properties.h"
+
+namespace Kernel {
+
+class KAffinityMask {
+public:
+    constexpr KAffinityMask() = default;
+
+    [[nodiscard]] constexpr u64 GetAffinityMask() const {
+        return this->mask;
+    }
+
+    constexpr void SetAffinityMask(u64 new_mask) {
+        ASSERT((new_mask & ~AllowedAffinityMask) == 0);
+        this->mask = new_mask;
+    }
+
+    [[nodiscard]] constexpr bool GetAffinity(s32 core) const {
+        return this->mask & GetCoreBit(core);
+    }
+
+    constexpr void SetAffinity(s32 core, bool set) {
+        ASSERT(0 <= core && core < static_cast<s32>(Core::Hardware::NUM_CPU_CORES));
+
+        if (set) {
+            this->mask |= GetCoreBit(core);
+        } else {
+            this->mask &= ~GetCoreBit(core);
+        }
+    }
+
+    constexpr void SetAll() {
+        this->mask = AllowedAffinityMask;
+    }
+
+private:
+    [[nodiscard]] static constexpr u64 GetCoreBit(s32 core) {
+        ASSERT(0 <= core && core < static_cast<s32>(Core::Hardware::NUM_CPU_CORES));
+        return (1ULL << core);
+    }
+
+    static constexpr u64 AllowedAffinityMask = (1ULL << Core::Hardware::NUM_CPU_CORES) - 1;
+
+    u64 mask{};
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_priority_queue.h b/src/core/hle/kernel/k_priority_queue.h
new file mode 100644
index 000000000..01a577d0c
--- /dev/null
+++ b/src/core/hle/kernel/k_priority_queue.h
@@ -0,0 +1,449 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include <array>
+
+#include "common/assert.h"
+#include "common/bit_set.h"
+#include "common/bit_util.h"
+#include "common/common_types.h"
+
+namespace Kernel {
+
+class Thread;
+
+template <typename T>
+concept KPriorityQueueAffinityMask = !std::is_reference_v<T> && requires(T & t) {
+    { t.GetAffinityMask() }
+    ->std::convertible_to<u64>;
+    {t.SetAffinityMask(std::declval<u64>())};
+
+    { t.GetAffinity(std::declval<int32_t>()) }
+    ->std::same_as<bool>;
+    {t.SetAffinity(std::declval<int32_t>(), std::declval<bool>())};
+    {t.SetAll()};
+};
+
+template <typename T>
+concept KPriorityQueueMember = !std::is_reference_v<T> && requires(T & t) {
+    {typename T::QueueEntry()};
+    {(typename T::QueueEntry()).Initialize()};
+    {(typename T::QueueEntry()).SetPrev(std::addressof(t))};
+    {(typename T::QueueEntry()).SetNext(std::addressof(t))};
+    { (typename T::QueueEntry()).GetNext() }
+    ->std::same_as<T*>;
+    { (typename T::QueueEntry()).GetPrev() }
+    ->std::same_as<T*>;
+    { t.GetPriorityQueueEntry(std::declval<s32>()) }
+    ->std::same_as<typename T::QueueEntry&>;
+
+    {t.GetAffinityMask()};
+    { typename std::remove_cvref<decltype(t.GetAffinityMask())>::type() }
+    ->KPriorityQueueAffinityMask;
+
+    { t.GetActiveCore() }
+    ->std::convertible_to<s32>;
+    { t.GetPriority() }
+    ->std::convertible_to<s32>;
+};
+
+template <typename Member, size_t _NumCores, int LowestPriority, int HighestPriority>
+requires KPriorityQueueMember<Member> class KPriorityQueue {
+public:
+    using AffinityMaskType = typename std::remove_cv_t<
+        typename std::remove_reference<decltype(std::declval<Member>().GetAffinityMask())>::type>;
+
+    static_assert(LowestPriority >= 0);
+    static_assert(HighestPriority >= 0);
+    static_assert(LowestPriority >= HighestPriority);
+    static constexpr size_t NumPriority = LowestPriority - HighestPriority + 1;
+    static constexpr size_t NumCores = _NumCores;
+
+    static constexpr bool IsValidCore(s32 core) {
+        return 0 <= core && core < static_cast<s32>(NumCores);
+    }
+
+    static constexpr bool IsValidPriority(s32 priority) {
+        return HighestPriority <= priority && priority <= LowestPriority + 1;
+    }
+
+private:
+    using Entry = typename Member::QueueEntry;
+
+public:
+    class KPerCoreQueue {
+    private:
+        std::array<Entry, NumCores> root{};
+
+    public:
+        constexpr KPerCoreQueue() {
+            for (auto& per_core_root : root) {
+                per_core_root.Initialize();
+            }
+        }
+
+        constexpr bool PushBack(s32 core, Member* member) {
+            // Get the entry associated with the member.
+            Entry& member_entry = member->GetPriorityQueueEntry(core);
+
+            // Get the entry associated with the end of the queue.
+            Member* tail = this->root[core].GetPrev();
+            Entry& tail_entry =
+                (tail != nullptr) ? tail->GetPriorityQueueEntry(core) : this->root[core];
+
+            // Link the entries.
+            member_entry.SetPrev(tail);
+            member_entry.SetNext(nullptr);
+            tail_entry.SetNext(member);
+            this->root[core].SetPrev(member);
+
+            return tail == nullptr;
+        }
+
+        constexpr bool PushFront(s32 core, Member* member) {
+            // Get the entry associated with the member.
+            Entry& member_entry = member->GetPriorityQueueEntry(core);
+
+            // Get the entry associated with the front of the queue.
+            Member* head = this->root[core].GetNext();
+            Entry& head_entry =
+                (head != nullptr) ? head->GetPriorityQueueEntry(core) : this->root[core];
+
+            // Link the entries.
+            member_entry.SetPrev(nullptr);
+            member_entry.SetNext(head);
+            head_entry.SetPrev(member);
+            this->root[core].SetNext(member);
+
+            return (head == nullptr);
+        }
+
+        constexpr bool Remove(s32 core, Member* member) {
+            // Get the entry associated with the member.
+            Entry& member_entry = member->GetPriorityQueueEntry(core);
+
+            // Get the entries associated with next and prev.
+            Member* prev = member_entry.GetPrev();
+            Member* next = member_entry.GetNext();
+            Entry& prev_entry =
+                (prev != nullptr) ? prev->GetPriorityQueueEntry(core) : this->root[core];
+            Entry& next_entry =
+                (next != nullptr) ? next->GetPriorityQueueEntry(core) : this->root[core];
+
+            // Unlink.
+            prev_entry.SetNext(next);
+            next_entry.SetPrev(prev);
+
+            return (this->GetFront(core) == nullptr);
+        }
+
+        constexpr Member* GetFront(s32 core) const {
+            return this->root[core].GetNext();
+        }
+    };
+
+    class KPriorityQueueImpl {
+    public:
+        constexpr KPriorityQueueImpl() = default;
+
+        constexpr void PushBack(s32 priority, s32 core, Member* member) {
+            ASSERT(IsValidCore(core));
+            ASSERT(IsValidPriority(priority));
+
+            if (priority > LowestPriority) {
+                return;
+            }
+
+            if (this->queues[priority].PushBack(core, member)) {
+                this->available_priorities[core].SetBit(priority);
+            }
+        }
+
+        constexpr void PushFront(s32 priority, s32 core, Member* member) {
+            ASSERT(IsValidCore(core));
+            ASSERT(IsValidPriority(priority));
+
+            if (priority > LowestPriority) {
+                return;
+            }
+
+            if (this->queues[priority].PushFront(core, member)) {
+                this->available_priorities[core].SetBit(priority);
+            }
+        }
+
+        constexpr void Remove(s32 priority, s32 core, Member* member) {
+            ASSERT(IsValidCore(core));
+            ASSERT(IsValidPriority(priority));
+
+            if (priority > LowestPriority) {
+                return;
+            }
+
+            if (this->queues[priority].Remove(core, member)) {
+                this->available_priorities[core].ClearBit(priority);
+            }
+        }
+
+        constexpr Member* GetFront(s32 core) const {
+            ASSERT(IsValidCore(core));
+
+            const s32 priority =
+                static_cast<s32>(this->available_priorities[core].CountLeadingZero());
+            if (priority <= LowestPriority) {
+                return this->queues[priority].GetFront(core);
+            } else {
+                return nullptr;
+            }
+        }
+
+        constexpr Member* GetFront(s32 priority, s32 core) const {
+            ASSERT(IsValidCore(core));
+            ASSERT(IsValidPriority(priority));
+
+            if (priority <= LowestPriority) {
+                return this->queues[priority].GetFront(core);
+            } else {
+                return nullptr;
+            }
+        }
+
+        constexpr Member* GetNext(s32 core, const Member* member) const {
+            ASSERT(IsValidCore(core));
+
+            Member* next = member->GetPriorityQueueEntry(core).GetNext();
+            if (next == nullptr) {
+                const s32 priority = static_cast<s32>(
+                    this->available_priorities[core].GetNextSet(member->GetPriority()));
+                if (priority <= LowestPriority) {
+                    next = this->queues[priority].GetFront(core);
+                }
+            }
+            return next;
+        }
+
+        constexpr void MoveToFront(s32 priority, s32 core, Member* member) {
+            ASSERT(IsValidCore(core));
+            ASSERT(IsValidPriority(priority));
+
+            if (priority <= LowestPriority) {
+                this->queues[priority].Remove(core, member);
+                this->queues[priority].PushFront(core, member);
+            }
+        }
+
+        constexpr Member* MoveToBack(s32 priority, s32 core, Member* member) {
+            ASSERT(IsValidCore(core));
+            ASSERT(IsValidPriority(priority));
+
+            if (priority <= LowestPriority) {
+                this->queues[priority].Remove(core, member);
+                this->queues[priority].PushBack(core, member);
+                return this->queues[priority].GetFront(core);
+            } else {
+                return nullptr;
+            }
+        }
+
+    private:
+        std::array<KPerCoreQueue, NumPriority> queues{};
+        std::array<Common::BitSet64<NumPriority>, NumCores> available_priorities{};
+    };
+
+private:
+    KPriorityQueueImpl scheduled_queue;
+    KPriorityQueueImpl suggested_queue;
+
+private:
+    constexpr void ClearAffinityBit(u64& affinity, s32 core) {
+        affinity &= ~(u64(1) << core);
+    }
+
+    constexpr s32 GetNextCore(u64& affinity) {
+        const s32 core = Common::CountTrailingZeroes64(affinity);
+        ClearAffinityBit(affinity, core);
+        return core;
+    }
+
+    constexpr void PushBack(s32 priority, Member* member) {
+        ASSERT(IsValidPriority(priority));
+
+        // Push onto the scheduled queue for its core, if we can.
+        u64 affinity = member->GetAffinityMask().GetAffinityMask();
+        if (const s32 core = member->GetActiveCore(); core >= 0) {
+            this->scheduled_queue.PushBack(priority, core, member);
+            ClearAffinityBit(affinity, core);
+        }
+
+        // And suggest the thread for all other cores.
+        while (affinity) {
+            this->suggested_queue.PushBack(priority, GetNextCore(affinity), member);
+        }
+    }
+
+    constexpr void PushFront(s32 priority, Member* member) {
+        ASSERT(IsValidPriority(priority));
+
+        // Push onto the scheduled queue for its core, if we can.
+        u64 affinity = member->GetAffinityMask().GetAffinityMask();
+        if (const s32 core = member->GetActiveCore(); core >= 0) {
+            this->scheduled_queue.PushFront(priority, core, member);
+            ClearAffinityBit(affinity, core);
+        }
+
+        // And suggest the thread for all other cores.
+        // Note: Nintendo pushes onto the back of the suggested queue, not the front.
+        while (affinity) {
+            this->suggested_queue.PushBack(priority, GetNextCore(affinity), member);
+        }
+    }
+
+    constexpr void Remove(s32 priority, Member* member) {
+        ASSERT(IsValidPriority(priority));
+
+        // Remove from the scheduled queue for its core.
+        u64 affinity = member->GetAffinityMask().GetAffinityMask();
+        if (const s32 core = member->GetActiveCore(); core >= 0) {
+            this->scheduled_queue.Remove(priority, core, member);
+            ClearAffinityBit(affinity, core);
+        }
+
+        // Remove from the suggested queue for all other cores.
+        while (affinity) {
+            this->suggested_queue.Remove(priority, GetNextCore(affinity), member);
+        }
+    }
+
+public:
+    constexpr KPriorityQueue() = default;
+
+    // Getters.
+    constexpr Member* GetScheduledFront(s32 core) const {
+        return this->scheduled_queue.GetFront(core);
+    }
+
+    constexpr Member* GetScheduledFront(s32 core, s32 priority) const {
+        return this->scheduled_queue.GetFront(priority, core);
+    }
+
+    constexpr Member* GetSuggestedFront(s32 core) const {
+        return this->suggested_queue.GetFront(core);
+    }
+
+    constexpr Member* GetSuggestedFront(s32 core, s32 priority) const {
+        return this->suggested_queue.GetFront(priority, core);
+    }
+
+    constexpr Member* GetScheduledNext(s32 core, const Member* member) const {
+        return this->scheduled_queue.GetNext(core, member);
+    }
+
+    constexpr Member* GetSuggestedNext(s32 core, const Member* member) const {
+        return this->suggested_queue.GetNext(core, member);
+    }
+
+    constexpr Member* GetSamePriorityNext(s32 core, const Member* member) const {
+        return member->GetPriorityQueueEntry(core).GetNext();
+    }
+
+    // Mutators.
+    constexpr void PushBack(Member* member) {
+        this->PushBack(member->GetPriority(), member);
+    }
+
+    constexpr void Remove(Member* member) {
+        this->Remove(member->GetPriority(), member);
+    }
+
+    constexpr void MoveToScheduledFront(Member* member) {
+        this->scheduled_queue.MoveToFront(member->GetPriority(), member->GetActiveCore(), member);
+    }
+
+    constexpr Thread* MoveToScheduledBack(Member* member) {
+        return this->scheduled_queue.MoveToBack(member->GetPriority(), member->GetActiveCore(),
+                                                member);
+    }
+
+    // First class fancy operations.
+    constexpr void ChangePriority(s32 prev_priority, bool is_running, Member* member) {
+        ASSERT(IsValidPriority(prev_priority));
+
+        // Remove the member from the queues.
+        const s32 new_priority = member->GetPriority();
+        this->Remove(prev_priority, member);
+
+        // And enqueue. If the member is running, we want to keep it running.
+        if (is_running) {
+            this->PushFront(new_priority, member);
+        } else {
+            this->PushBack(new_priority, member);
+        }
+    }
+
+    constexpr void ChangeAffinityMask(s32 prev_core, const AffinityMaskType& prev_affinity,
+                                      Member* member) {
+        // Get the new information.
+        const s32 priority = member->GetPriority();
+        const AffinityMaskType& new_affinity = member->GetAffinityMask();
+        const s32 new_core = member->GetActiveCore();
+
+        // Remove the member from all queues it was in before.
+        for (s32 core = 0; core < static_cast<s32>(NumCores); core++) {
+            if (prev_affinity.GetAffinity(core)) {
+                if (core == prev_core) {
+                    this->scheduled_queue.Remove(priority, core, member);
+                } else {
+                    this->suggested_queue.Remove(priority, core, member);
+                }
+            }
+        }
+
+        // And add the member to all queues it should be in now.
+        for (s32 core = 0; core < static_cast<s32>(NumCores); core++) {
+            if (new_affinity.GetAffinity(core)) {
+                if (core == new_core) {
+                    this->scheduled_queue.PushBack(priority, core, member);
+                } else {
+                    this->suggested_queue.PushBack(priority, core, member);
+                }
+            }
+        }
+    }
+
+    constexpr void ChangeCore(s32 prev_core, Member* member, bool to_front = false) {
+        // Get the new information.
+        const s32 new_core = member->GetActiveCore();
+        const s32 priority = member->GetPriority();
+
+        // We don't need to do anything if the core is the same.
+        if (prev_core != new_core) {
+            // Remove from the scheduled queue for the previous core.
+            if (prev_core >= 0) {
+                this->scheduled_queue.Remove(priority, prev_core, member);
+            }
+
+            // Remove from the suggested queue and add to the scheduled queue for the new core.
+            if (new_core >= 0) {
+                this->suggested_queue.Remove(priority, new_core, member);
+                if (to_front) {
+                    this->scheduled_queue.PushFront(priority, new_core, member);
+                } else {
+                    this->scheduled_queue.PushBack(priority, new_core, member);
+                }
+            }
+
+            // Add to the suggested queue for the previous core.
+            if (prev_core >= 0) {
+                this->suggested_queue.PushBack(priority, prev_core, member);
+            }
+        }
+    }
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_scheduler.cpp b/src/core/hle/kernel/k_scheduler.cpp
new file mode 100644
index 000000000..c5fd82a6b
--- /dev/null
+++ b/src/core/hle/kernel/k_scheduler.cpp
@@ -0,0 +1,784 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#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/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/physical_core.h"
+#include "core/hle/kernel/process.h"
+#include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
+
+namespace Kernel {
+
+static void IncrementScheduledCount(Kernel::Thread* thread) {
+    if (auto process = thread->GetOwnerProcess(); process) {
+        process->IncrementScheduledCount();
+    }
+}
+
+void KScheduler::RescheduleCores(KernelCore& kernel, u64 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::CountTrailingZeroes64(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 &= ~(1ULL << core);
+    }
+    if (must_context_switch) {
+        auto core_scheduler = kernel.CurrentScheduler();
+        kernel.ExitSVCProfile();
+        core_scheduler->RescheduleCurrentCore();
+        kernel.EnterSVCProfile();
+    }
+}
+
+u64 KScheduler::UpdateHighestPriorityThread(Thread* highest_thread) {
+    std::scoped_lock lock{guard};
+    if (Thread* prev_highest_thread = this->state.highest_priority_thread;
+        prev_highest_thread != highest_thread) {
+        if (prev_highest_thread != nullptr) {
+            IncrementScheduledCount(prev_highest_thread);
+            prev_highest_thread->SetLastScheduledTick(system.CoreTiming().GetCPUTicks());
+        }
+        if (this->state.should_count_idle) {
+            if (highest_thread != nullptr) {
+                // if (Process* process = highest_thread->GetOwnerProcess(); process != nullptr) {
+                //    process->SetRunningThread(this->core_id, highest_thread,
+                //                              this->state.idle_count);
+                //}
+            } else {
+                this->state.idle_count++;
+            }
+        }
+
+        this->state.highest_priority_thread = highest_thread;
+        this->state.needs_scheduling = true;
+        return (1ULL << this->core_id);
+    } else {
+        return 0;
+    }
+}
+
+u64 KScheduler::UpdateHighestPriorityThreadsImpl(KernelCore& kernel) {
+    ASSERT(kernel.GlobalSchedulerContext().IsLocked());
+
+    // Clear that we need to update.
+    ClearSchedulerUpdateNeeded(kernel);
+
+    u64 cores_needing_scheduling = 0, idle_cores = 0;
+    Thread* top_threads[Core::Hardware::NUM_CPU_CORES];
+    auto& priority_queue = GetPriorityQueue(kernel);
+
+    /// We want to go over all cores, finding the highest priority thread and determining if
+    /// scheduling is needed for that core.
+    for (size_t core_id = 0; core_id < Core::Hardware::NUM_CPU_CORES; core_id++) {
+        Thread* top_thread = priority_queue.GetScheduledFront(static_cast<s32>(core_id));
+        if (top_thread != nullptr) {
+            // If the thread has no waiters, we need to check if the process has a thread pinned.
+            // TODO(bunnei): Implement thread pinning
+        } else {
+            idle_cores |= (1ULL << core_id);
+        }
+
+        top_threads[core_id] = top_thread;
+        cores_needing_scheduling |=
+            kernel.Scheduler(core_id).UpdateHighestPriorityThread(top_threads[core_id]);
+    }
+
+    // Idle cores are bad. We're going to try to migrate threads to each idle core in turn.
+    while (idle_cores != 0) {
+        u32 core_id = Common::CountTrailingZeroes64(idle_cores);
+        if (Thread* suggested = priority_queue.GetSuggestedFront(core_id); suggested != nullptr) {
+            s32 migration_candidates[Core::Hardware::NUM_CPU_CORES];
+            size_t num_candidates = 0;
+
+            // While we have a suggested thread, try to migrate it!
+            while (suggested != nullptr) {
+                // Check if the suggested thread is the top thread on its core.
+                const s32 suggested_core = suggested->GetActiveCore();
+                if (Thread* top_thread =
+                        (suggested_core >= 0) ? top_threads[suggested_core] : nullptr;
+                    top_thread != suggested) {
+                    // Make sure we're not dealing with threads too high priority for migration.
+                    if (top_thread != nullptr &&
+                        top_thread->GetPriority() < HighestCoreMigrationAllowedPriority) {
+                        break;
+                    }
+
+                    // The suggested thread isn't bound to its core, so we can migrate it!
+                    suggested->SetActiveCore(core_id);
+                    priority_queue.ChangeCore(suggested_core, suggested);
+
+                    top_threads[core_id] = suggested;
+                    cores_needing_scheduling |=
+                        kernel.Scheduler(core_id).UpdateHighestPriorityThread(top_threads[core_id]);
+                    break;
+                }
+
+                // Note this core as a candidate for migration.
+                ASSERT(num_candidates < Core::Hardware::NUM_CPU_CORES);
+                migration_candidates[num_candidates++] = suggested_core;
+                suggested = priority_queue.GetSuggestedNext(core_id, suggested);
+            }
+
+            // If suggested is nullptr, we failed to migrate a specific thread. So let's try all our
+            // candidate cores' top threads.
+            if (suggested == nullptr) {
+                for (size_t i = 0; i < num_candidates; i++) {
+                    // Check if there's some other thread that can run on the candidate core.
+                    const s32 candidate_core = migration_candidates[i];
+                    suggested = top_threads[candidate_core];
+                    if (Thread* next_on_candidate_core =
+                            priority_queue.GetScheduledNext(candidate_core, suggested);
+                        next_on_candidate_core != nullptr) {
+                        // The candidate core can run some other thread! We'll migrate its current
+                        // top thread to us.
+                        top_threads[candidate_core] = next_on_candidate_core;
+                        cores_needing_scheduling |=
+                            kernel.Scheduler(candidate_core)
+                                .UpdateHighestPriorityThread(top_threads[candidate_core]);
+
+                        // Perform the migration.
+                        suggested->SetActiveCore(core_id);
+                        priority_queue.ChangeCore(candidate_core, suggested);
+
+                        top_threads[core_id] = suggested;
+                        cores_needing_scheduling |=
+                            kernel.Scheduler(core_id).UpdateHighestPriorityThread(
+                                top_threads[core_id]);
+                        break;
+                    }
+                }
+            }
+        }
+
+        idle_cores &= ~(1ULL << core_id);
+    }
+
+    return cores_needing_scheduling;
+}
+
+void KScheduler::OnThreadStateChanged(KernelCore& kernel, Thread* thread, u32 old_state) {
+    ASSERT(kernel.GlobalSchedulerContext().IsLocked());
+
+    // Check if the state has changed, because if it hasn't there's nothing to do.
+    const auto cur_state = thread->scheduling_state;
+    if (cur_state == old_state) {
+        return;
+    }
+
+    // Update the priority queues.
+    if (old_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+        // If we were previously runnable, then we're not runnable now, and we should remove.
+        GetPriorityQueue(kernel).Remove(thread);
+        IncrementScheduledCount(thread);
+        SetSchedulerUpdateNeeded(kernel);
+    } else if (cur_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+        // If we're now runnable, then we weren't previously, and we should add.
+        GetPriorityQueue(kernel).PushBack(thread);
+        IncrementScheduledCount(thread);
+        SetSchedulerUpdateNeeded(kernel);
+    }
+}
+
+void KScheduler::OnThreadPriorityChanged(KernelCore& kernel, Thread* thread, Thread* current_thread,
+                                         u32 old_priority) {
+
+    ASSERT(kernel.GlobalSchedulerContext().IsLocked());
+
+    // If the thread is runnable, we want to change its priority in the queue.
+    if (thread->scheduling_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+        GetPriorityQueue(kernel).ChangePriority(
+            old_priority, thread == kernel.CurrentScheduler()->GetCurrentThread(), thread);
+        IncrementScheduledCount(thread);
+        SetSchedulerUpdateNeeded(kernel);
+    }
+}
+
+void KScheduler::OnThreadAffinityMaskChanged(KernelCore& kernel, Thread* thread,
+                                             const KAffinityMask& old_affinity, s32 old_core) {
+    ASSERT(kernel.GlobalSchedulerContext().IsLocked());
+
+    // If the thread is runnable, we want to change its affinity in the queue.
+    if (thread->scheduling_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+        GetPriorityQueue(kernel).ChangeAffinityMask(old_core, old_affinity, thread);
+        IncrementScheduledCount(thread);
+        SetSchedulerUpdateNeeded(kernel);
+    }
+}
+
+void KScheduler::RotateScheduledQueue(s32 core_id, s32 priority) {
+    ASSERT(system.GlobalSchedulerContext().IsLocked());
+
+    // Get a reference to the priority queue.
+    auto& kernel = system.Kernel();
+    auto& priority_queue = GetPriorityQueue(kernel);
+
+    // Rotate the front of the queue to the end.
+    Thread* top_thread = priority_queue.GetScheduledFront(core_id, priority);
+    Thread* next_thread = nullptr;
+    if (top_thread != nullptr) {
+        next_thread = priority_queue.MoveToScheduledBack(top_thread);
+        if (next_thread != top_thread) {
+            IncrementScheduledCount(top_thread);
+            IncrementScheduledCount(next_thread);
+        }
+    }
+
+    // While we have a suggested thread, try to migrate it!
+    {
+        Thread* suggested = priority_queue.GetSuggestedFront(core_id, priority);
+        while (suggested != nullptr) {
+            // Check if the suggested thread is the top thread on its core.
+            const s32 suggested_core = suggested->GetActiveCore();
+            if (Thread* top_on_suggested_core =
+                    (suggested_core >= 0) ? priority_queue.GetScheduledFront(suggested_core)
+                                          : nullptr;
+                top_on_suggested_core != suggested) {
+                // If the next thread is a new thread that has been waiting longer than our
+                // suggestion, we prefer it to our suggestion.
+                if (top_thread != next_thread && next_thread != nullptr &&
+                    next_thread->GetLastScheduledTick() < suggested->GetLastScheduledTick()) {
+                    suggested = nullptr;
+                    break;
+                }
+
+                // If we're allowed to do a migration, do one.
+                // NOTE: Unlike migrations in UpdateHighestPriorityThread, this moves the suggestion
+                // to the front of the queue.
+                if (top_on_suggested_core == nullptr ||
+                    top_on_suggested_core->GetPriority() >= HighestCoreMigrationAllowedPriority) {
+                    suggested->SetActiveCore(core_id);
+                    priority_queue.ChangeCore(suggested_core, suggested, true);
+                    IncrementScheduledCount(suggested);
+                    break;
+                }
+            }
+
+            // Get the next suggestion.
+            suggested = priority_queue.GetSamePriorityNext(core_id, suggested);
+        }
+    }
+
+    // Now that we might have migrated a thread with the same priority, check if we can do better.
+
+    {
+        Thread* best_thread = priority_queue.GetScheduledFront(core_id);
+        if (best_thread == GetCurrentThread()) {
+            best_thread = priority_queue.GetScheduledNext(core_id, best_thread);
+        }
+
+        // If the best thread we can choose has a priority the same or worse than ours, try to
+        // migrate a higher priority thread.
+        if (best_thread != nullptr && best_thread->GetPriority() >= static_cast<u32>(priority)) {
+            Thread* suggested = priority_queue.GetSuggestedFront(core_id);
+            while (suggested != nullptr) {
+                // If the suggestion's priority is the same as ours, don't bother.
+                if (suggested->GetPriority() >= best_thread->GetPriority()) {
+                    break;
+                }
+
+                // Check if the suggested thread is the top thread on its core.
+                const s32 suggested_core = suggested->GetActiveCore();
+                if (Thread* top_on_suggested_core =
+                        (suggested_core >= 0) ? priority_queue.GetScheduledFront(suggested_core)
+                                              : nullptr;
+                    top_on_suggested_core != suggested) {
+                    // If we're allowed to do a migration, do one.
+                    // NOTE: Unlike migrations in UpdateHighestPriorityThread, this moves the
+                    // suggestion to the front of the queue.
+                    if (top_on_suggested_core == nullptr ||
+                        top_on_suggested_core->GetPriority() >=
+                            HighestCoreMigrationAllowedPriority) {
+                        suggested->SetActiveCore(core_id);
+                        priority_queue.ChangeCore(suggested_core, suggested, true);
+                        IncrementScheduledCount(suggested);
+                        break;
+                    }
+                }
+
+                // Get the next suggestion.
+                suggested = priority_queue.GetSuggestedNext(core_id, suggested);
+            }
+        }
+    }
+
+    // After a rotation, we need a scheduler update.
+    SetSchedulerUpdateNeeded(kernel);
+}
+
+bool KScheduler::CanSchedule(KernelCore& kernel) {
+    return kernel.CurrentScheduler()->GetCurrentThread()->GetDisableDispatchCount() <= 1;
+}
+
+bool KScheduler::IsSchedulerUpdateNeeded(const KernelCore& kernel) {
+    return kernel.GlobalSchedulerContext().scheduler_update_needed.load(std::memory_order_acquire);
+}
+
+void KScheduler::SetSchedulerUpdateNeeded(KernelCore& kernel) {
+    kernel.GlobalSchedulerContext().scheduler_update_needed.store(true, std::memory_order_release);
+}
+
+void KScheduler::ClearSchedulerUpdateNeeded(KernelCore& kernel) {
+    kernel.GlobalSchedulerContext().scheduler_update_needed.store(false, std::memory_order_release);
+}
+
+void KScheduler::DisableScheduling(KernelCore& kernel) {
+    if (auto* scheduler = kernel.CurrentScheduler(); scheduler) {
+        ASSERT(scheduler->GetCurrentThread()->GetDisableDispatchCount() >= 0);
+        scheduler->GetCurrentThread()->DisableDispatch();
+    }
+}
+
+void KScheduler::EnableScheduling(KernelCore& kernel, u64 cores_needing_scheduling,
+                                  Core::EmuThreadHandle global_thread) {
+    if (auto* scheduler = kernel.CurrentScheduler(); scheduler) {
+        scheduler->GetCurrentThread()->EnableDispatch();
+    }
+    RescheduleCores(kernel, cores_needing_scheduling, global_thread);
+}
+
+u64 KScheduler::UpdateHighestPriorityThreads(KernelCore& kernel) {
+    if (IsSchedulerUpdateNeeded(kernel)) {
+        return UpdateHighestPriorityThreadsImpl(kernel);
+    } else {
+        return 0;
+    }
+}
+
+KSchedulerPriorityQueue& KScheduler::GetPriorityQueue(KernelCore& kernel) {
+    return kernel.GlobalSchedulerContext().priority_queue;
+}
+
+void KScheduler::YieldWithoutCoreMigration() {
+    auto& kernel = system.Kernel();
+
+    // Validate preconditions.
+    ASSERT(CanSchedule(kernel));
+    ASSERT(kernel.CurrentProcess() != nullptr);
+
+    // Get the current thread and process.
+    Thread& cur_thread = *GetCurrentThread();
+    Process& cur_process = *kernel.CurrentProcess();
+
+    // If the thread's yield count matches, there's nothing for us to do.
+    if (cur_thread.GetYieldScheduleCount() == cur_process.GetScheduledCount()) {
+        return;
+    }
+
+    // Get a reference to the priority queue.
+    auto& priority_queue = GetPriorityQueue(kernel);
+
+    // Perform the yield.
+    {
+        KScopedSchedulerLock lock(kernel);
+
+        const auto cur_state = cur_thread.scheduling_state;
+        if (cur_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+            // Put the current thread at the back of the queue.
+            Thread* next_thread = priority_queue.MoveToScheduledBack(std::addressof(cur_thread));
+            IncrementScheduledCount(std::addressof(cur_thread));
+
+            // If the next thread is different, we have an update to perform.
+            if (next_thread != std::addressof(cur_thread)) {
+                SetSchedulerUpdateNeeded(kernel);
+            } else {
+                // Otherwise, set the thread's yield count so that we won't waste work until the
+                // process is scheduled again.
+                cur_thread.SetYieldScheduleCount(cur_process.GetScheduledCount());
+            }
+        }
+    }
+}
+
+void KScheduler::YieldWithCoreMigration() {
+    auto& kernel = system.Kernel();
+
+    // Validate preconditions.
+    ASSERT(CanSchedule(kernel));
+    ASSERT(kernel.CurrentProcess() != nullptr);
+
+    // Get the current thread and process.
+    Thread& cur_thread = *GetCurrentThread();
+    Process& cur_process = *kernel.CurrentProcess();
+
+    // If the thread's yield count matches, there's nothing for us to do.
+    if (cur_thread.GetYieldScheduleCount() == cur_process.GetScheduledCount()) {
+        return;
+    }
+
+    // Get a reference to the priority queue.
+    auto& priority_queue = GetPriorityQueue(kernel);
+
+    // Perform the yield.
+    {
+        KScopedSchedulerLock lock(kernel);
+
+        const auto cur_state = cur_thread.scheduling_state;
+        if (cur_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+            // Get the current active core.
+            const s32 core_id = cur_thread.GetActiveCore();
+
+            // Put the current thread at the back of the queue.
+            Thread* next_thread = priority_queue.MoveToScheduledBack(std::addressof(cur_thread));
+            IncrementScheduledCount(std::addressof(cur_thread));
+
+            // While we have a suggested thread, try to migrate it!
+            bool recheck = false;
+            Thread* suggested = priority_queue.GetSuggestedFront(core_id);
+            while (suggested != nullptr) {
+                // Check if the suggested thread is the thread running on its core.
+                const s32 suggested_core = suggested->GetActiveCore();
+
+                if (Thread* running_on_suggested_core =
+                        (suggested_core >= 0)
+                            ? kernel.Scheduler(suggested_core).state.highest_priority_thread
+                            : nullptr;
+                    running_on_suggested_core != suggested) {
+                    // If the current thread's priority is higher than our suggestion's we prefer
+                    // the next thread to the suggestion. We also prefer the next thread when the
+                    // current thread's priority is equal to the suggestions, but the next thread
+                    // has been waiting longer.
+                    if ((suggested->GetPriority() > cur_thread.GetPriority()) ||
+                        (suggested->GetPriority() == cur_thread.GetPriority() &&
+                         next_thread != std::addressof(cur_thread) &&
+                         next_thread->GetLastScheduledTick() < suggested->GetLastScheduledTick())) {
+                        suggested = nullptr;
+                        break;
+                    }
+
+                    // If we're allowed to do a migration, do one.
+                    // NOTE: Unlike migrations in UpdateHighestPriorityThread, this moves the
+                    // suggestion to the front of the queue.
+                    if (running_on_suggested_core == nullptr ||
+                        running_on_suggested_core->GetPriority() >=
+                            HighestCoreMigrationAllowedPriority) {
+                        suggested->SetActiveCore(core_id);
+                        priority_queue.ChangeCore(suggested_core, suggested, true);
+                        IncrementScheduledCount(suggested);
+                        break;
+                    } else {
+                        // We couldn't perform a migration, but we should check again on a future
+                        // yield.
+                        recheck = true;
+                    }
+                }
+
+                // Get the next suggestion.
+                suggested = priority_queue.GetSuggestedNext(core_id, suggested);
+            }
+
+            // If we still have a suggestion or the next thread is different, we have an update to
+            // perform.
+            if (suggested != nullptr || next_thread != std::addressof(cur_thread)) {
+                SetSchedulerUpdateNeeded(kernel);
+            } else if (!recheck) {
+                // Otherwise if we don't need to re-check, set the thread's yield count so that we
+                // won't waste work until the process is scheduled again.
+                cur_thread.SetYieldScheduleCount(cur_process.GetScheduledCount());
+            }
+        }
+    }
+}
+
+void KScheduler::YieldToAnyThread() {
+    auto& kernel = system.Kernel();
+
+    // Validate preconditions.
+    ASSERT(CanSchedule(kernel));
+    ASSERT(kernel.CurrentProcess() != nullptr);
+
+    // Get the current thread and process.
+    Thread& cur_thread = *GetCurrentThread();
+    Process& cur_process = *kernel.CurrentProcess();
+
+    // If the thread's yield count matches, there's nothing for us to do.
+    if (cur_thread.GetYieldScheduleCount() == cur_process.GetScheduledCount()) {
+        return;
+    }
+
+    // Get a reference to the priority queue.
+    auto& priority_queue = GetPriorityQueue(kernel);
+
+    // Perform the yield.
+    {
+        KScopedSchedulerLock lock(kernel);
+
+        const auto cur_state = cur_thread.scheduling_state;
+        if (cur_state == static_cast<u32>(ThreadSchedStatus::Runnable)) {
+            // Get the current active core.
+            const s32 core_id = cur_thread.GetActiveCore();
+
+            // Migrate the current thread to core -1.
+            cur_thread.SetActiveCore(-1);
+            priority_queue.ChangeCore(core_id, std::addressof(cur_thread));
+            IncrementScheduledCount(std::addressof(cur_thread));
+
+            // If there's nothing scheduled, we can try to perform a migration.
+            if (priority_queue.GetScheduledFront(core_id) == nullptr) {
+                // While we have a suggested thread, try to migrate it!
+                Thread* suggested = priority_queue.GetSuggestedFront(core_id);
+                while (suggested != nullptr) {
+                    // Check if the suggested thread is the top thread on its core.
+                    const s32 suggested_core = suggested->GetActiveCore();
+                    if (Thread* top_on_suggested_core =
+                            (suggested_core >= 0) ? priority_queue.GetScheduledFront(suggested_core)
+                                                  : nullptr;
+                        top_on_suggested_core != suggested) {
+                        // If we're allowed to do a migration, do one.
+                        if (top_on_suggested_core == nullptr ||
+                            top_on_suggested_core->GetPriority() >=
+                                HighestCoreMigrationAllowedPriority) {
+                            suggested->SetActiveCore(core_id);
+                            priority_queue.ChangeCore(suggested_core, suggested);
+                            IncrementScheduledCount(suggested);
+                        }
+
+                        // Regardless of whether we migrated, we had a candidate, so we're done.
+                        break;
+                    }
+
+                    // Get the next suggestion.
+                    suggested = priority_queue.GetSuggestedNext(core_id, suggested);
+                }
+
+                // If the suggestion is different from the current thread, we need to perform an
+                // update.
+                if (suggested != std::addressof(cur_thread)) {
+                    SetSchedulerUpdateNeeded(kernel);
+                } else {
+                    // Otherwise, set the thread's yield count so that we won't waste work until the
+                    // process is scheduled again.
+                    cur_thread.SetYieldScheduleCount(cur_process.GetScheduledCount());
+                }
+            } else {
+                // Otherwise, we have an update to perform.
+                SetSchedulerUpdateNeeded(kernel);
+            }
+        }
+    }
+}
+
+KScheduler::KScheduler(Core::System& system, std::size_t core_id)
+    : system(system), core_id(core_id) {
+    switch_fiber = std::make_shared<Common::Fiber>(OnSwitch, this);
+    this->state.needs_scheduling = true;
+    this->state.interrupt_task_thread_runnable = false;
+    this->state.should_count_idle = false;
+    this->state.idle_count = 0;
+    this->state.idle_thread_stack = nullptr;
+    this->state.highest_priority_thread = nullptr;
+}
+
+KScheduler::~KScheduler() = default;
+
+Thread* KScheduler::GetCurrentThread() const {
+    if (current_thread) {
+        return current_thread;
+    }
+    return idle_thread;
+}
+
+u64 KScheduler::GetLastContextSwitchTicks() const {
+    return last_context_switch_time;
+}
+
+void KScheduler::RescheduleCurrentCore() {
+    ASSERT(GetCurrentThread()->GetDisableDispatchCount() == 1);
+
+    auto& phys_core = system.Kernel().PhysicalCore(core_id);
+    if (phys_core.IsInterrupted()) {
+        phys_core.ClearInterrupt();
+    }
+    guard.lock();
+    if (this->state.needs_scheduling) {
+        Schedule();
+    } else {
+        guard.unlock();
+    }
+}
+
+void KScheduler::OnThreadStart() {
+    SwitchContextStep2();
+}
+
+void KScheduler::Unload(Thread* thread) {
+    if (thread) {
+        thread->SetIsRunning(false);
+        if (thread->IsContinuousOnSVC() && !thread->IsHLEThread()) {
+            system.ArmInterface(core_id).ExceptionalExit();
+            thread->SetContinuousOnSVC(false);
+        }
+        if (!thread->IsHLEThread() && !thread->HasExited()) {
+            Core::ARM_Interface& cpu_core = system.ArmInterface(core_id);
+            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();
+    }
+}
+
+void KScheduler::Reload(Thread* thread) {
+    if (thread) {
+        ASSERT_MSG(thread->GetSchedulingStatus() == ThreadSchedStatus::Runnable,
+                   "Thread must be runnable.");
+
+        // Cancel any outstanding wakeup events for this thread
+        thread->SetIsRunning(true);
+        thread->SetWasRunning(false);
+
+        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 = system.ArmInterface(core_id);
+            cpu_core.LoadContext(thread->GetContext32());
+            cpu_core.LoadContext(thread->GetContext64());
+            cpu_core.SetTlsAddress(thread->GetTLSAddress());
+            cpu_core.SetTPIDR_EL0(thread->GetTPIDR_EL0());
+            cpu_core.ClearExclusiveState();
+        }
+    }
+}
+
+void KScheduler::SwitchContextStep2() {
+    // Load context of new thread
+    Reload(current_thread);
+
+    RescheduleCurrentCore();
+}
+
+void KScheduler::ScheduleImpl() {
+    Thread* previous_thread = current_thread;
+    current_thread = state.highest_priority_thread;
+
+    this->state.needs_scheduling = false;
+
+    if (current_thread == previous_thread) {
+        guard.unlock();
+        return;
+    }
+
+    Process* const previous_process = system.Kernel().CurrentProcess();
+
+    UpdateLastContextSwitchTime(previous_thread, previous_process);
+
+    // Save context for previous thread
+    Unload(previous_thread);
+
+    std::shared_ptr<Common::Fiber>* old_context;
+    if (previous_thread != nullptr) {
+        old_context = &previous_thread->GetHostContext();
+    } else {
+        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 KScheduler::OnSwitch(void* this_scheduler) {
+    KScheduler* sched = static_cast<KScheduler*>(this_scheduler);
+    sched->SwitchToCurrent();
+}
+
+void KScheduler::SwitchToCurrent() {
+    while (true) {
+        {
+            std::scoped_lock lock{guard};
+            current_thread = state.highest_priority_thread;
+            this->state.needs_scheduling = false;
+        }
+        const auto is_switch_pending = [this] {
+            std::scoped_lock lock{guard};
+            return state.needs_scheduling.load(std::memory_order_relaxed);
+        };
+        do {
+            if (current_thread != nullptr && !current_thread->IsHLEThread()) {
+                current_thread->context_guard.lock();
+                if (!current_thread->IsRunnable()) {
+                    current_thread->context_guard.unlock();
+                    break;
+                }
+                if (static_cast<u32>(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);
+        } while (!is_switch_pending());
+    }
+}
+
+void KScheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) {
+    const u64 prev_switch_ticks = last_context_switch_time;
+    const u64 most_recent_switch_ticks = system.CoreTiming().GetCPUTicks();
+    const u64 update_ticks = most_recent_switch_ticks - prev_switch_ticks;
+
+    if (thread != nullptr) {
+        thread->UpdateCPUTimeTicks(update_ticks);
+    }
+
+    if (process != nullptr) {
+        process->UpdateCPUTimeTicks(update_ticks);
+    }
+
+    last_context_switch_time = most_recent_switch_ticks;
+}
+
+void KScheduler::Initialize() {
+    std::string name = "Idle Thread Id:" + std::to_string(core_id);
+    std::function<void(void*)> init_func = Core::CpuManager::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 = thread_res.Unwrap().get();
+
+    {
+        KScopedSchedulerLock lock{system.Kernel()};
+        idle_thread->SetStatus(ThreadStatus::Ready);
+    }
+}
+
+KScopedSchedulerLock::KScopedSchedulerLock(KernelCore& kernel)
+    : KScopedLock(kernel.GlobalSchedulerContext().SchedulerLock()) {}
+
+KScopedSchedulerLock::~KScopedSchedulerLock() = default;
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_scheduler.h b/src/core/hle/kernel/k_scheduler.h
new file mode 100644
index 000000000..e84abc84c
--- /dev/null
+++ b/src/core/hle/kernel/k_scheduler.h
@@ -0,0 +1,201 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include <atomic>
+
+#include "common/common_types.h"
+#include "common/spin_lock.h"
+#include "core/hle/kernel/global_scheduler_context.h"
+#include "core/hle/kernel/k_priority_queue.h"
+#include "core/hle/kernel/k_scheduler_lock.h"
+#include "core/hle/kernel/k_scoped_lock.h"
+
+namespace Common {
+class Fiber;
+}
+
+namespace Core {
+class System;
+}
+
+namespace Kernel {
+
+class KernelCore;
+class Process;
+class SchedulerLock;
+class Thread;
+
+class KScheduler final {
+public:
+    explicit KScheduler(Core::System& system, std::size_t core_id);
+    ~KScheduler();
+
+    /// Reschedules to the next available thread (call after current thread is suspended)
+    void RescheduleCurrentCore();
+
+    /// Reschedules cores pending reschedule, to be called on EnableScheduling.
+    static void RescheduleCores(KernelCore& kernel, u64 cores_pending_reschedule,
+                                Core::EmuThreadHandle global_thread);
+
+    /// The next two are for SingleCore Only.
+    /// Unload current thread before preempting core.
+    void Unload(Thread* thread);
+
+    /// Reload current thread after core preemption.
+    void Reload(Thread* thread);
+
+    /// Gets the current running thread
+    [[nodiscard]] Thread* GetCurrentThread() const;
+
+    /// Gets the timestamp for the last context switch in ticks.
+    [[nodiscard]] u64 GetLastContextSwitchTicks() const;
+
+    [[nodiscard]] bool ContextSwitchPending() const {
+        return state.needs_scheduling.load(std::memory_order_relaxed);
+    }
+
+    void Initialize();
+
+    void OnThreadStart();
+
+    [[nodiscard]] std::shared_ptr<Common::Fiber>& ControlContext() {
+        return switch_fiber;
+    }
+
+    [[nodiscard]] const std::shared_ptr<Common::Fiber>& ControlContext() const {
+        return switch_fiber;
+    }
+
+    [[nodiscard]] u64 UpdateHighestPriorityThread(Thread* highest_thread);
+
+    /**
+     * Takes a thread and moves it to the back of the it's priority list.
+     *
+     * @note This operation can be redundant and no scheduling is changed if marked as so.
+     */
+    void YieldWithoutCoreMigration();
+
+    /**
+     * Takes a thread and moves it to the back of the it's priority list.
+     * Afterwards, tries to pick a suggested thread from the suggested queue that has worse time or
+     * a better priority than the next thread in the core.
+     *
+     * @note This operation can be redundant and no scheduling is changed if marked as so.
+     */
+    void YieldWithCoreMigration();
+
+    /**
+     * Takes a thread and moves it out of the scheduling queue.
+     * and into the suggested queue. If no thread can be scheduled afterwards in that core,
+     * a suggested thread is obtained instead.
+     *
+     * @note This operation can be redundant and no scheduling is changed if marked as so.
+     */
+    void YieldToAnyThread();
+
+    /// Notify the scheduler a thread's status has changed.
+    static void OnThreadStateChanged(KernelCore& kernel, Thread* thread, u32 old_state);
+
+    /// Notify the scheduler a thread's priority has changed.
+    static void OnThreadPriorityChanged(KernelCore& kernel, Thread* thread, Thread* current_thread,
+                                        u32 old_priority);
+
+    /// Notify the scheduler a thread's core and/or affinity mask has changed.
+    static void OnThreadAffinityMaskChanged(KernelCore& kernel, Thread* thread,
+                                            const KAffinityMask& old_affinity, s32 old_core);
+
+    static bool CanSchedule(KernelCore& kernel);
+    static bool IsSchedulerUpdateNeeded(const KernelCore& kernel);
+    static void SetSchedulerUpdateNeeded(KernelCore& kernel);
+    static void ClearSchedulerUpdateNeeded(KernelCore& kernel);
+    static void DisableScheduling(KernelCore& kernel);
+    static void EnableScheduling(KernelCore& kernel, u64 cores_needing_scheduling,
+                                 Core::EmuThreadHandle global_thread);
+    [[nodiscard]] static u64 UpdateHighestPriorityThreads(KernelCore& kernel);
+
+private:
+    friend class GlobalSchedulerContext;
+
+    /**
+     * 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.
+     *
+     * 2. Second we try to get a suggested thread that's not assigned to any core or
+     *    that is not the top thread in that core.
+     *
+     * 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.
+     */
+    [[nodiscard]] static u64 UpdateHighestPriorityThreadsImpl(KernelCore& kernel);
+
+    [[nodiscard]] static KSchedulerPriorityQueue& GetPriorityQueue(KernelCore& kernel);
+
+    void RotateScheduledQueue(s32 core_id, s32 priority);
+
+    void Schedule() {
+        ASSERT(GetCurrentThread()->GetDisableDispatchCount() == 1);
+        this->ScheduleImpl();
+    }
+
+    /// Switches the CPU's active thread context to that of the specified thread
+    void ScheduleImpl();
+
+    /// 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
+     * process using the following difference:
+     *
+     * ticks += most_recent_ticks - last_context_switch_ticks
+     *
+     * The internal tick timestamp for the scheduler is simply the
+     * most recent tick count retrieved. No special arithmetic is
+     * applied to it.
+     */
+    void UpdateLastContextSwitchTime(Thread* thread, Process* process);
+
+    static void OnSwitch(void* this_scheduler);
+    void SwitchToCurrent();
+
+    Thread* current_thread{};
+    Thread* idle_thread{};
+
+    std::shared_ptr<Common::Fiber> switch_fiber{};
+
+    struct SchedulingState {
+        std::atomic<bool> needs_scheduling;
+        bool interrupt_task_thread_runnable{};
+        bool should_count_idle{};
+        u64 idle_count{};
+        Thread* highest_priority_thread{};
+        void* idle_thread_stack{};
+    };
+
+    SchedulingState state;
+
+    Core::System& system;
+    u64 last_context_switch_time{};
+    const std::size_t core_id;
+
+    Common::SpinLock guard{};
+};
+
+class KScopedSchedulerLock : KScopedLock<GlobalSchedulerContext::LockType> {
+public:
+    explicit KScopedSchedulerLock(KernelCore& kernel);
+    ~KScopedSchedulerLock();
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_scheduler_lock.h b/src/core/hle/kernel/k_scheduler_lock.h
new file mode 100644
index 000000000..2d675b39e
--- /dev/null
+++ b/src/core/hle/kernel/k_scheduler_lock.h
@@ -0,0 +1,74 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include "common/assert.h"
+#include "common/spin_lock.h"
+#include "core/hardware_properties.h"
+
+namespace Kernel {
+
+class KernelCore;
+
+template <typename SchedulerType>
+class KAbstractSchedulerLock {
+public:
+    explicit KAbstractSchedulerLock(KernelCore& kernel) : kernel{kernel} {}
+
+    bool IsLockedByCurrentThread() const {
+        return this->owner_thread == kernel.GetCurrentEmuThreadID();
+    }
+
+    void Lock() {
+        if (this->IsLockedByCurrentThread()) {
+            // If we already own the lock, we can just increment the count.
+            ASSERT(this->lock_count > 0);
+            this->lock_count++;
+        } else {
+            // Otherwise, we want to disable scheduling and acquire the spinlock.
+            SchedulerType::DisableScheduling(kernel);
+            this->spin_lock.lock();
+
+            // For debug, ensure that our state is valid.
+            ASSERT(this->lock_count == 0);
+            ASSERT(this->owner_thread == Core::EmuThreadHandle::InvalidHandle());
+
+            // Increment count, take ownership.
+            this->lock_count = 1;
+            this->owner_thread = kernel.GetCurrentEmuThreadID();
+        }
+    }
+
+    void Unlock() {
+        ASSERT(this->IsLockedByCurrentThread());
+        ASSERT(this->lock_count > 0);
+
+        // Release an instance of the lock.
+        if ((--this->lock_count) == 0) {
+            // We're no longer going to hold the lock. Take note of what cores need scheduling.
+            const u64 cores_needing_scheduling =
+                SchedulerType::UpdateHighestPriorityThreads(kernel);
+            Core::EmuThreadHandle leaving_thread = owner_thread;
+
+            // Note that we no longer hold the lock, and unlock the spinlock.
+            this->owner_thread = Core::EmuThreadHandle::InvalidHandle();
+            this->spin_lock.unlock();
+
+            // Enable scheduling, and perform a rescheduling operation.
+            SchedulerType::EnableScheduling(kernel, cores_needing_scheduling, leaving_thread);
+        }
+    }
+
+private:
+    KernelCore& kernel;
+    Common::SpinLock spin_lock{};
+    s32 lock_count{};
+    Core::EmuThreadHandle owner_thread{Core::EmuThreadHandle::InvalidHandle()};
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_scoped_lock.h b/src/core/hle/kernel/k_scoped_lock.h
new file mode 100644
index 000000000..d7cc557b2
--- /dev/null
+++ b/src/core/hle/kernel/k_scoped_lock.h
@@ -0,0 +1,41 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include "common/common_types.h"
+
+namespace Kernel {
+
+template <typename T>
+concept KLockable = !std::is_reference_v<T> && requires(T & t) {
+    { t.Lock() }
+    ->std::same_as<void>;
+    { t.Unlock() }
+    ->std::same_as<void>;
+};
+
+template <typename T>
+requires KLockable<T> class KScopedLock {
+public:
+    explicit KScopedLock(T* l) : lock_ptr(l) {
+        this->lock_ptr->Lock();
+    }
+    explicit KScopedLock(T& l) : KScopedLock(std::addressof(l)) { /* ... */
+    }
+    ~KScopedLock() {
+        this->lock_ptr->Unlock();
+    }
+
+    KScopedLock(const KScopedLock&) = delete;
+    KScopedLock(KScopedLock&&) = delete;
+
+private:
+    T* lock_ptr;
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h b/src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h
new file mode 100644
index 000000000..2bb3817fa
--- /dev/null
+++ b/src/core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h
@@ -0,0 +1,50 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
+
+#pragma once
+
+#include "common/common_types.h"
+#include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/kernel.h"
+#include "core/hle/kernel/thread.h"
+#include "core/hle/kernel/time_manager.h"
+
+namespace Kernel {
+
+class KScopedSchedulerLockAndSleep {
+public:
+    explicit KScopedSchedulerLockAndSleep(KernelCore& kernel, Handle& event_handle, Thread* t,
+                                          s64 timeout)
+        : kernel(kernel), event_handle(event_handle), thread(t), timeout_tick(timeout) {
+        event_handle = InvalidHandle;
+
+        // Lock the scheduler.
+        kernel.GlobalSchedulerContext().scheduler_lock.Lock();
+    }
+
+    ~KScopedSchedulerLockAndSleep() {
+        // Register the sleep.
+        if (this->timeout_tick > 0) {
+            kernel.TimeManager().ScheduleTimeEvent(event_handle, this->thread, this->timeout_tick);
+        }
+
+        // Unlock the scheduler.
+        kernel.GlobalSchedulerContext().scheduler_lock.Unlock();
+    }
+
+    void CancelSleep() {
+        this->timeout_tick = 0;
+    }
+
+private:
+    KernelCore& kernel;
+    Handle& event_handle;
+    Thread* thread{};
+    s64 timeout_tick{};
+};
+
+} // namespace Kernel
diff --git a/src/core/hle/kernel/kernel.cpp b/src/core/hle/kernel/kernel.cpp
index 929db696d..04cae3a43 100644
--- a/src/core/hle/kernel/kernel.cpp
+++ b/src/core/hle/kernel/kernel.cpp
@@ -27,6 +27,7 @@
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/memory/memory_layout.h"
 #include "core/hle/kernel/memory/memory_manager.h"
@@ -34,7 +35,6 @@
 #include "core/hle/kernel/physical_core.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/resource_limit.h"
-#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/shared_memory.h"
 #include "core/hle/kernel/synchronization.h"
 #include "core/hle/kernel/thread.h"
@@ -49,17 +49,18 @@ namespace Kernel {
 
 struct KernelCore::Impl {
     explicit Impl(Core::System& system, KernelCore& kernel)
-        : global_scheduler{kernel}, synchronization{system}, time_manager{system},
-          global_handle_table{kernel}, system{system} {}
+        : synchronization{system}, time_manager{system}, global_handle_table{kernel}, system{
+                                                                                          system} {}
 
     void SetMulticore(bool is_multicore) {
         this->is_multicore = is_multicore;
     }
 
     void Initialize(KernelCore& kernel) {
-        Shutdown();
         RegisterHostThread();
 
+        global_scheduler_context = std::make_unique<Kernel::GlobalSchedulerContext>(kernel);
+
         InitializePhysicalCores();
         InitializeSystemResourceLimit(kernel);
         InitializeMemoryLayout();
@@ -86,29 +87,20 @@ struct KernelCore::Impl {
             }
         }
 
-        for (std::size_t i = 0; i < cores.size(); i++) {
-            cores[i].Shutdown();
-            schedulers[i].reset();
-        }
         cores.clear();
 
         process_list.clear();
+
         current_process = nullptr;
 
         system_resource_limit = nullptr;
 
         global_handle_table.Clear();
-        preemption_event = nullptr;
 
-        global_scheduler.Shutdown();
+        preemption_event = nullptr;
 
         named_ports.clear();
 
-        for (auto& core : cores) {
-            core.Shutdown();
-        }
-        cores.clear();
-
         exclusive_monitor.reset();
 
         num_host_threads = 0;
@@ -121,7 +113,7 @@ struct KernelCore::Impl {
         exclusive_monitor =
             Core::MakeExclusiveMonitor(system.Memory(), Core::Hardware::NUM_CPU_CORES);
         for (std::size_t i = 0; i < Core::Hardware::NUM_CPU_CORES; i++) {
-            schedulers[i] = std::make_unique<Kernel::Scheduler>(system, i);
+            schedulers[i] = std::make_unique<Kernel::KScheduler>(system, i);
             cores.emplace_back(i, system, *schedulers[i], interrupts);
         }
     }
@@ -154,8 +146,8 @@ struct KernelCore::Impl {
         preemption_event = Core::Timing::CreateEvent(
             "PreemptionCallback", [this, &kernel](std::uintptr_t, std::chrono::nanoseconds) {
                 {
-                    SchedulerLock lock(kernel);
-                    global_scheduler.PreemptThreads();
+                    KScopedSchedulerLock lock(kernel);
+                    global_scheduler_context->PreemptThreads();
                 }
                 const auto time_interval = std::chrono::nanoseconds{
                     Core::Timing::msToCycles(std::chrono::milliseconds(10))};
@@ -245,7 +237,7 @@ struct KernelCore::Impl {
         if (result.host_handle >= Core::Hardware::NUM_CPU_CORES) {
             return result;
         }
-        const Kernel::Scheduler& sched = cores[result.host_handle].Scheduler();
+        const Kernel::KScheduler& sched = cores[result.host_handle].Scheduler();
         const Kernel::Thread* current = sched.GetCurrentThread();
         if (current != nullptr && !current->IsPhantomMode()) {
             result.guest_handle = current->GetGlobalHandle();
@@ -314,7 +306,7 @@ struct KernelCore::Impl {
     // Lists all processes that exist in the current session.
     std::vector<std::shared_ptr<Process>> process_list;
     Process* current_process = nullptr;
-    Kernel::GlobalScheduler global_scheduler;
+    std::unique_ptr<Kernel::GlobalSchedulerContext> global_scheduler_context;
     Kernel::Synchronization synchronization;
     Kernel::TimeManager time_manager;
 
@@ -355,7 +347,7 @@ struct KernelCore::Impl {
 
     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{};
+    std::array<std::unique_ptr<Kernel::KScheduler>, Core::Hardware::NUM_CPU_CORES> schedulers{};
 
     bool is_multicore{};
     std::thread::id single_core_thread_id{};
@@ -415,19 +407,19 @@ const std::vector<std::shared_ptr<Process>>& KernelCore::GetProcessList() const
     return impl->process_list;
 }
 
-Kernel::GlobalScheduler& KernelCore::GlobalScheduler() {
-    return impl->global_scheduler;
+Kernel::GlobalSchedulerContext& KernelCore::GlobalSchedulerContext() {
+    return *impl->global_scheduler_context;
 }
 
-const Kernel::GlobalScheduler& KernelCore::GlobalScheduler() const {
-    return impl->global_scheduler;
+const Kernel::GlobalSchedulerContext& KernelCore::GlobalSchedulerContext() const {
+    return *impl->global_scheduler_context;
 }
 
-Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) {
+Kernel::KScheduler& KernelCore::Scheduler(std::size_t id) {
     return *impl->schedulers[id];
 }
 
-const Kernel::Scheduler& KernelCore::Scheduler(std::size_t id) const {
+const Kernel::KScheduler& KernelCore::Scheduler(std::size_t id) const {
     return *impl->schedulers[id];
 }
 
@@ -451,16 +443,13 @@ const Kernel::PhysicalCore& KernelCore::CurrentPhysicalCore() const {
     return impl->cores[core_id];
 }
 
-Kernel::Scheduler& KernelCore::CurrentScheduler() {
+Kernel::KScheduler* 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];
+    if (core_id >= Core::Hardware::NUM_CPU_CORES) {
+        // This is expected when called from not a guest thread
+        return {};
+    }
+    return impl->schedulers[core_id].get();
 }
 
 std::array<Core::CPUInterruptHandler, Core::Hardware::NUM_CPU_CORES>& KernelCore::Interrupts() {
@@ -623,7 +612,7 @@ const Kernel::SharedMemory& KernelCore::GetTimeSharedMem() const {
 void KernelCore::Suspend(bool in_suspention) {
     const bool should_suspend = exception_exited || in_suspention;
     {
-        SchedulerLock lock(*this);
+        KScopedSchedulerLock 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);
diff --git a/src/core/hle/kernel/kernel.h b/src/core/hle/kernel/kernel.h
index a73a93039..5846c3f39 100644
--- a/src/core/hle/kernel/kernel.h
+++ b/src/core/hle/kernel/kernel.h
@@ -35,12 +35,12 @@ class SlabHeap;
 
 class AddressArbiter;
 class ClientPort;
-class GlobalScheduler;
+class GlobalSchedulerContext;
 class HandleTable;
 class PhysicalCore;
 class Process;
 class ResourceLimit;
-class Scheduler;
+class KScheduler;
 class SharedMemory;
 class Synchronization;
 class Thread;
@@ -102,16 +102,16 @@ public:
     const std::vector<std::shared_ptr<Process>>& GetProcessList() const;
 
     /// Gets the sole instance of the global scheduler
-    Kernel::GlobalScheduler& GlobalScheduler();
+    Kernel::GlobalSchedulerContext& GlobalSchedulerContext();
 
     /// Gets the sole instance of the global scheduler
-    const Kernel::GlobalScheduler& GlobalScheduler() const;
+    const Kernel::GlobalSchedulerContext& GlobalSchedulerContext() const;
 
     /// Gets the sole instance of the Scheduler assoviated with cpu core 'id'
-    Kernel::Scheduler& Scheduler(std::size_t id);
+    Kernel::KScheduler& Scheduler(std::size_t id);
 
     /// Gets the sole instance of the Scheduler assoviated with cpu core 'id'
-    const Kernel::Scheduler& Scheduler(std::size_t id) const;
+    const Kernel::KScheduler& Scheduler(std::size_t id) const;
 
     /// Gets the an instance of the respective physical CPU core.
     Kernel::PhysicalCore& PhysicalCore(std::size_t id);
@@ -120,10 +120,7 @@ public:
     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;
+    Kernel::KScheduler* CurrentScheduler();
 
     /// Gets the an instance of the current physical CPU core.
     Kernel::PhysicalCore& CurrentPhysicalCore();
diff --git a/src/core/hle/kernel/mutex.cpp b/src/core/hle/kernel/mutex.cpp
index 8f6c944d1..4f8075e0e 100644
--- a/src/core/hle/kernel/mutex.cpp
+++ b/src/core/hle/kernel/mutex.cpp
@@ -11,11 +11,11 @@
 #include "core/core.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/mutex.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/process.h"
-#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/thread.h"
 #include "core/hle/result.h"
 #include "core/memory.h"
@@ -73,9 +73,9 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
 
     auto& kernel = system.Kernel();
     std::shared_ptr<Thread> current_thread =
-        SharedFrom(kernel.CurrentScheduler().GetCurrentThread());
+        SharedFrom(kernel.CurrentScheduler()->GetCurrentThread());
     {
-        SchedulerLock lock(kernel);
+        KScopedSchedulerLock lock(kernel);
         // The mutex address must be 4-byte aligned
         if ((address % sizeof(u32)) != 0) {
             return ERR_INVALID_ADDRESS;
@@ -114,7 +114,7 @@ ResultCode Mutex::TryAcquire(VAddr address, Handle holding_thread_handle,
     }
 
     {
-        SchedulerLock lock(kernel);
+        KScopedSchedulerLock lock(kernel);
         auto* owner = current_thread->GetLockOwner();
         if (owner != nullptr) {
             owner->RemoveMutexWaiter(current_thread);
@@ -153,10 +153,10 @@ std::pair<ResultCode, std::shared_ptr<Thread>> Mutex::Unlock(std::shared_ptr<Thr
 
 ResultCode Mutex::Release(VAddr address) {
     auto& kernel = system.Kernel();
-    SchedulerLock lock(kernel);
+    KScopedSchedulerLock lock(kernel);
 
     std::shared_ptr<Thread> current_thread =
-        SharedFrom(kernel.CurrentScheduler().GetCurrentThread());
+        SharedFrom(kernel.CurrentScheduler()->GetCurrentThread());
 
     auto [result, new_owner] = Unlock(current_thread, address);
 
diff --git a/src/core/hle/kernel/physical_core.cpp b/src/core/hle/kernel/physical_core.cpp
index 50aca5752..7fea45f96 100644
--- a/src/core/hle/kernel/physical_core.cpp
+++ b/src/core/hle/kernel/physical_core.cpp
@@ -7,14 +7,14 @@
 #include "core/arm/dynarmic/arm_dynarmic_32.h"
 #include "core/arm/dynarmic/arm_dynarmic_64.h"
 #include "core/core.h"
+#include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/physical_core.h"
-#include "core/hle/kernel/scheduler.h"
 
 namespace Kernel {
 
 PhysicalCore::PhysicalCore(std::size_t core_index, Core::System& system,
-                           Kernel::Scheduler& scheduler, Core::CPUInterrupts& interrupts)
+                           Kernel::KScheduler& scheduler, Core::CPUInterrupts& interrupts)
     : core_index{core_index}, system{system}, scheduler{scheduler},
       interrupts{interrupts}, guard{std::make_unique<Common::SpinLock>()} {}
 
@@ -43,10 +43,6 @@ void PhysicalCore::Idle() {
     interrupts[core_index].AwaitInterrupt();
 }
 
-void PhysicalCore::Shutdown() {
-    scheduler.Shutdown();
-}
-
 bool PhysicalCore::IsInterrupted() const {
     return interrupts[core_index].IsInterrupted();
 }
diff --git a/src/core/hle/kernel/physical_core.h b/src/core/hle/kernel/physical_core.h
index 801d24c28..f2b0911aa 100644
--- a/src/core/hle/kernel/physical_core.h
+++ b/src/core/hle/kernel/physical_core.h
@@ -15,7 +15,7 @@ class SpinLock;
 }
 
 namespace Kernel {
-class Scheduler;
+class KScheduler;
 } // namespace Kernel
 
 namespace Core {
@@ -28,7 +28,7 @@ namespace Kernel {
 
 class PhysicalCore {
 public:
-    PhysicalCore(std::size_t core_index, Core::System& system, Kernel::Scheduler& scheduler,
+    PhysicalCore(std::size_t core_index, Core::System& system, Kernel::KScheduler& scheduler,
                  Core::CPUInterrupts& interrupts);
     ~PhysicalCore();
 
@@ -55,9 +55,6 @@ public:
     /// Check if this core is interrupted
     bool IsInterrupted() const;
 
-    // Shutdown this physical core.
-    void Shutdown();
-
     bool IsInitialized() const {
         return arm_interface != nullptr;
     }
@@ -82,18 +79,18 @@ public:
         return core_index;
     }
 
-    Kernel::Scheduler& Scheduler() {
+    Kernel::KScheduler& Scheduler() {
         return scheduler;
     }
 
-    const Kernel::Scheduler& Scheduler() const {
+    const Kernel::KScheduler& Scheduler() const {
         return scheduler;
     }
 
 private:
     const std::size_t core_index;
     Core::System& system;
-    Kernel::Scheduler& scheduler;
+    Kernel::KScheduler& scheduler;
     Core::CPUInterrupts& interrupts;
     std::unique_ptr<Common::SpinLock> guard;
     std::unique_ptr<Core::ARM_Interface> arm_interface;
diff --git a/src/core/hle/kernel/process.cpp b/src/core/hle/kernel/process.cpp
index b17529dee..b905b486a 100644
--- a/src/core/hle/kernel/process.cpp
+++ b/src/core/hle/kernel/process.cpp
@@ -15,13 +15,13 @@
 #include "core/file_sys/program_metadata.h"
 #include "core/hle/kernel/code_set.h"
 #include "core/hle/kernel/errors.h"
+#include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/memory/memory_block_manager.h"
 #include "core/hle/kernel/memory/page_table.h"
 #include "core/hle/kernel/memory/slab_heap.h"
 #include "core/hle/kernel/process.h"
 #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"
@@ -54,7 +54,7 @@ void SetupMainThread(Core::System& system, Process& owner_process, u32 priority,
     auto& kernel = system.Kernel();
     // Threads by default are dormant, wake up the main thread so it runs when the scheduler fires
     {
-        SchedulerLock lock{kernel};
+        KScopedSchedulerLock lock{kernel};
         thread->SetStatus(ThreadStatus::Ready);
     }
 }
@@ -213,7 +213,7 @@ void Process::UnregisterThread(const Thread* thread) {
 }
 
 ResultCode Process::ClearSignalState() {
-    SchedulerLock lock(system.Kernel());
+    KScopedSchedulerLock lock(system.Kernel());
     if (status == ProcessStatus::Exited) {
         LOG_ERROR(Kernel, "called on a terminated process instance.");
         return ERR_INVALID_STATE;
@@ -314,7 +314,7 @@ void Process::PrepareForTermination() {
             if (thread->GetOwnerProcess() != this)
                 continue;
 
-            if (thread.get() == system.CurrentScheduler().GetCurrentThread())
+            if (thread.get() == kernel.CurrentScheduler()->GetCurrentThread())
                 continue;
 
             // TODO(Subv): When are the other running/ready threads terminated?
@@ -325,7 +325,7 @@ void Process::PrepareForTermination() {
         }
     };
 
-    stop_threads(system.GlobalScheduler().GetThreadList());
+    stop_threads(system.GlobalSchedulerContext().GetThreadList());
 
     FreeTLSRegion(tls_region_address);
     tls_region_address = 0;
@@ -347,7 +347,7 @@ static auto FindTLSPageWithAvailableSlots(std::vector<TLSPage>& tls_pages) {
 }
 
 VAddr Process::CreateTLSRegion() {
-    SchedulerLock lock(system.Kernel());
+    KScopedSchedulerLock lock(system.Kernel());
     if (auto tls_page_iter{FindTLSPageWithAvailableSlots(tls_pages)};
         tls_page_iter != tls_pages.cend()) {
         return *tls_page_iter->ReserveSlot();
@@ -378,7 +378,7 @@ VAddr Process::CreateTLSRegion() {
 }
 
 void Process::FreeTLSRegion(VAddr tls_address) {
-    SchedulerLock lock(system.Kernel());
+    KScopedSchedulerLock 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) {
diff --git a/src/core/hle/kernel/process.h b/src/core/hle/kernel/process.h
index f45cb5674..e412e58aa 100644
--- a/src/core/hle/kernel/process.h
+++ b/src/core/hle/kernel/process.h
@@ -216,6 +216,16 @@ public:
         total_process_running_time_ticks += ticks;
     }
 
+    /// Gets the process schedule count, used for thread yelding
+    s64 GetScheduledCount() const {
+        return schedule_count;
+    }
+
+    /// Increments the process schedule count, used for thread yielding.
+    void IncrementScheduledCount() {
+        ++schedule_count;
+    }
+
     /// Gets 8 bytes of random data for svcGetInfo RandomEntropy
     u64 GetRandomEntropy(std::size_t index) const {
         return random_entropy.at(index);
@@ -397,6 +407,9 @@ private:
     /// Name of this process
     std::string name;
 
+    /// Schedule count of this process
+    s64 schedule_count{};
+
     /// System context
     Core::System& system;
 };
diff --git a/src/core/hle/kernel/readable_event.cpp b/src/core/hle/kernel/readable_event.cpp
index 6e286419e..cea262ce0 100644
--- a/src/core/hle/kernel/readable_event.cpp
+++ b/src/core/hle/kernel/readable_event.cpp
@@ -6,10 +6,10 @@
 #include "common/assert.h"
 #include "common/logging/log.h"
 #include "core/hle/kernel/errors.h"
+#include "core/hle/kernel/k_scheduler.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 {
@@ -39,7 +39,7 @@ void ReadableEvent::Clear() {
 }
 
 ResultCode ReadableEvent::Reset() {
-    SchedulerLock lock(kernel);
+    KScopedSchedulerLock 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
deleted file mode 100644
index 5c63b0b4a..000000000
--- a/src/core/hle/kernel/scheduler.cpp
+++ /dev/null
@@ -1,819 +0,0 @@
-// Copyright 2018 yuzu emulator team
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-//
-// SelectThreads, Yield functions originally by TuxSH.
-// licensed under GPLv2 or later under exception provided by the author.
-
-#include <algorithm>
-#include <mutex>
-#include <set>
-#include <unordered_set>
-#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"
-
-namespace Kernel {
-
-GlobalScheduler::GlobalScheduler(KernelCore& kernel) : kernel{kernel} {}
-
-GlobalScheduler::~GlobalScheduler() = default;
-
-void GlobalScheduler::AddThread(std::shared_ptr<Thread> thread) {
-    std::scoped_lock lock{global_list_guard};
-    thread_list.push_back(std::move(thread));
-}
-
-void GlobalScheduler::RemoveThread(std::shared_ptr<Thread> thread) {
-    std::scoped_lock lock{global_list_guard};
-    thread_list.erase(std::remove(thread_list.begin(), thread_list.end(), thread),
-                      thread_list.end());
-}
-
-u32 GlobalScheduler::SelectThreads() {
-    ASSERT(is_locked);
-    const auto update_thread = [](Thread* thread, Scheduler& sched) {
-        std::scoped_lock lock{sched.guard};
-        if (thread != sched.selected_thread_set.get()) {
-            if (thread == nullptr) {
-                ++sched.idle_selection_count;
-            }
-            sched.selected_thread_set = SharedFrom(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);
-        return reschedule_pending;
-    };
-    if (!is_reselection_pending.load()) {
-        return 0;
-    }
-    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 |= (1U << core);
-        }
-        top_threads[core] = top_thread;
-    }
-
-    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;
-        }
-
-        idle_cores &= ~(1U << core_id);
-    }
-    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 ||
-               static_cast<u32>(top_threads[core]->GetProcessorID()) == core);
-        if (update_thread(top_threads[core], sched)) {
-            cores_needing_context_switch |= (1U << core);
-        }
-    }
-    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
-    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
-    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++) {
-        current_threads[i] = scheduled_queue[i].empty() ? nullptr : scheduled_queue[i].front();
-    }
-
-    Thread* next_thread = scheduled_queue[core_id].front(priority);
-    Thread* winner = nullptr;
-    for (auto& thread : suggested_queue[core_id]) {
-        const s32 source_core = thread->GetProcessorID();
-        if (source_core >= 0) {
-            if (current_threads[source_core] != nullptr) {
-                if (thread == current_threads[source_core] ||
-                    current_threads[source_core]->GetPriority() < min_regular_priority) {
-                    continue;
-                }
-            }
-        }
-        if (next_thread->GetLastRunningTicks() >= thread->GetLastRunningTicks() ||
-            next_thread->GetPriority() < thread->GetPriority()) {
-            if (thread->GetPriority() <= priority) {
-                winner = thread;
-                break;
-            }
-        }
-    }
-
-    if (winner != nullptr) {
-        if (winner != yielding_thread) {
-            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());
-
-    // Remove the thread from its scheduled mlq, put it on the corresponding "suggested" one instead
-    TransferToCore(yielding_thread->GetPriority(), -1, yielding_thread);
-
-    // If the core is idle, perform load balancing, excluding the threads that have just used this
-    // function...
-    if (scheduled_queue[core_id].empty()) {
-        // Here, "current_threads" is calculated after the ""yield"", unlike yield -1
-        std::array<Thread*, Core::Hardware::NUM_CPU_CORES> current_threads;
-        for (std::size_t i = 0; i < current_threads.size(); i++) {
-            current_threads[i] = scheduled_queue[i].empty() ? nullptr : scheduled_queue[i].front();
-        }
-        for (auto& thread : suggested_queue[core_id]) {
-            const s32 source_core = thread->GetProcessorID();
-            if (source_core < 0 || thread == current_threads[source_core]) {
-                continue;
-            }
-            if (current_threads[source_core] == nullptr ||
-                current_threads[source_core]->GetPriority() >= min_regular_priority) {
-                winner = thread;
-            }
-            break;
-        }
-        if (winner != nullptr) {
-            if (winner != yielding_thread) {
-                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) {
-            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();
-            }
-        }
-
-        Thread* current_thread =
-            scheduled_queue[core_id].empty() ? nullptr : scheduled_queue[core_id].front();
-        Thread* winner = nullptr;
-        for (auto& thread : suggested_queue[core_id]) {
-            const s32 source_core = thread->GetProcessorID();
-            if (thread->GetPriority() != priority) {
-                continue;
-            }
-            if (source_core >= 0) {
-                Thread* next_thread = scheduled_queue[source_core].empty()
-                                          ? nullptr
-                                          : scheduled_queue[source_core].front();
-                if (next_thread != nullptr && next_thread->GetPriority() < 2) {
-                    break;
-                }
-                if (next_thread == thread) {
-                    continue;
-                }
-            }
-            if (current_thread != nullptr &&
-                current_thread->GetLastRunningTicks() >= thread->GetLastRunningTicks()) {
-                winner = thread;
-                break;
-            }
-        }
-
-        if (winner != nullptr) {
-            TransferToCore(winner->GetPriority(), s32(core_id), winner);
-            current_thread =
-                winner->GetPriority() <= current_thread->GetPriority() ? winner : current_thread;
-        }
-
-        if (current_thread != nullptr && current_thread->GetPriority() > priority) {
-            for (auto& thread : suggested_queue[core_id]) {
-                const s32 source_core = thread->GetProcessorID();
-                if (thread->GetPriority() < priority) {
-                    continue;
-                }
-                if (source_core >= 0) {
-                    Thread* next_thread = scheduled_queue[source_core].empty()
-                                              ? nullptr
-                                              : scheduled_queue[source_core].front();
-                    if (next_thread != nullptr && next_thread->GetPriority() < 2) {
-                        break;
-                    }
-                    if (next_thread == thread) {
-                        continue;
-                    }
-                }
-                if (current_thread != nullptr &&
-                    current_thread->GetLastRunningTicks() >= thread->GetLastRunningTicks()) {
-                    winner = thread;
-                    break;
-                }
-            }
-
-            if (winner != nullptr) {
-                TransferToCore(winner->GetPriority(), s32(core_id), winner);
-                current_thread = winner;
-            }
-        }
-
-        is_reselection_pending.store(true, std::memory_order_release);
-    }
-}
-
-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 &= ~(1U << 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) {
-        return;
-    }
-    thread->SetProcessorID(destination_core);
-    if (source_core >= 0) {
-        Unschedule(priority, static_cast<u32>(source_core), thread);
-    }
-    if (destination_core >= 0) {
-        Unsuggest(priority, static_cast<u32>(destination_core), thread);
-        Schedule(priority, static_cast<u32>(destination_core), thread);
-    }
-    if (source_core >= 0) {
-        Suggest(priority, static_cast<u32>(source_core), thread);
-    }
-}
-
-bool GlobalScheduler::AskForReselectionOrMarkRedundant(Thread* current_thread,
-                                                       const Thread* winner) {
-    if (current_thread == winner) {
-        current_thread->IncrementYieldCount();
-        return true;
-    } else {
-        is_reselection_pending.store(true, std::memory_order_release);
-        return false;
-    }
-}
-
-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();
-        suggested_queue[core].clear();
-    }
-    thread_list.clear();
-}
-
-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;
-    }
-}
-
-void GlobalScheduler::Unlock() {
-    if (--scope_lock != 0) {
-        ASSERT(scope_lock > 0);
-        return;
-    }
-    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();
-    EnableInterruptAndSchedule(cores_pending_reschedule, leaving_thread);
-}
-
-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;
-
-bool Scheduler::HaveReadyThreads() const {
-    return system.GlobalScheduler().HaveReadyThreads(core_id);
-}
-
-Thread* Scheduler::GetCurrentThread() const {
-    if (current_thread) {
-        return current_thread.get();
-    }
-    return idle_thread.get();
-}
-
-Thread* Scheduler::GetSelectedThread() const {
-    return selected_thread.get();
-}
-
-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::OnThreadStart() {
-    SwitchContextStep2();
-}
-
-void Scheduler::Unload(Thread* thread) {
-    if (thread) {
-        thread->last_running_ticks = system.CoreTiming().GetCPUTicks();
-        thread->SetIsRunning(false);
-        if (thread->IsContinuousOnSVC() && !thread->IsHLEThread()) {
-            system.ArmInterface(core_id).ExceptionalExit();
-            thread->SetContinuousOnSVC(false);
-        }
-        if (!thread->IsHLEThread() && !thread->HasExited()) {
-            Core::ARM_Interface& cpu_core = system.ArmInterface(core_id);
-            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();
-    }
-}
-
-void Scheduler::Unload() {
-    Unload(current_thread.get());
-}
-
-void Scheduler::Reload(Thread* thread) {
-    if (thread) {
-        ASSERT_MSG(thread->GetSchedulingStatus() == ThreadSchedStatus::Runnable,
-                   "Thread must be runnable.");
-
-        // 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 = system.ArmInterface(core_id);
-            cpu_core.LoadContext(thread->GetContext32());
-            cpu_core.LoadContext(thread->GetContext64());
-            cpu_core.SetTlsAddress(thread->GetTLSAddress());
-            cpu_core.SetTPIDR_EL0(thread->GetTPIDR_EL0());
-            cpu_core.ClearExclusiveState();
-        }
-    }
-}
-
-void Scheduler::Reload() {
-    Reload(current_thread.get());
-}
-
-void Scheduler::SwitchContextStep2() {
-    // Load context of new thread
-    Reload(selected_thread.get());
-
-    TryDoContextSwitch();
-}
-
-void Scheduler::SwitchContext() {
-    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;
-    }
-
-    Process* const previous_process = system.Kernel().CurrentProcess();
-
-    UpdateLastContextSwitchTime(previous_thread, previous_process);
-
-    // Save context for previous thread
-    Unload(previous_thread);
-
-    std::shared_ptr<Common::Fiber>* old_context;
-    if (previous_thread != nullptr) {
-        old_context = &previous_thread->GetHostContext();
-    } else {
-        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) {
-        {
-            std::scoped_lock lock{guard};
-            selected_thread = selected_thread_set;
-            current_thread = selected_thread;
-            is_context_switch_pending = false;
-        }
-        const auto is_switch_pending = [this] {
-            std::scoped_lock lock{guard};
-            return is_context_switch_pending;
-        };
-        do {
-            if (current_thread != nullptr && !current_thread->IsHLEThread()) {
-                current_thread->context_guard.lock();
-                if (!current_thread->IsRunnable()) {
-                    current_thread->context_guard.unlock();
-                    break;
-                }
-                if (static_cast<u32>(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);
-        } while (!is_switch_pending());
-    }
-}
-
-void Scheduler::UpdateLastContextSwitchTime(Thread* thread, Process* process) {
-    const u64 prev_switch_ticks = last_context_switch_time;
-    const u64 most_recent_switch_ticks = system.CoreTiming().GetCPUTicks();
-    const u64 update_ticks = most_recent_switch_ticks - prev_switch_ticks;
-
-    if (thread != nullptr) {
-        thread->UpdateCPUTimeTicks(update_ticks);
-    }
-
-    if (process != nullptr) {
-        process->UpdateCPUTimeTicks(update_ticks);
-    }
-
-    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 = Core::CpuManager::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;
-}
-
-SchedulerLock::SchedulerLock(KernelCore& kernel) : kernel{kernel} {
-    kernel.GlobalScheduler().Lock();
-}
-
-SchedulerLock::~SchedulerLock() {
-    kernel.GlobalScheduler().Unlock();
-}
-
-SchedulerLockAndSleep::SchedulerLockAndSleep(KernelCore& kernel, Handle& event_handle,
-                                             Thread* time_task, s64 nanoseconds)
-    : SchedulerLock{kernel}, event_handle{event_handle}, time_task{time_task}, nanoseconds{
-                                                                                   nanoseconds} {
-    event_handle = InvalidHandle;
-}
-
-SchedulerLockAndSleep::~SchedulerLockAndSleep() {
-    if (sleep_cancelled) {
-        return;
-    }
-    auto& time_manager = kernel.TimeManager();
-    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
deleted file mode 100644
index 68db4a5ef..000000000
--- a/src/core/hle/kernel/scheduler.h
+++ /dev/null
@@ -1,320 +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 <memory>
-#include <mutex>
-#include <vector>
-
-#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;
-} // namespace Core
-
-namespace Kernel {
-
-class KernelCore;
-class Process;
-class SchedulerLock;
-
-class GlobalScheduler final {
-public:
-    explicit GlobalScheduler(KernelCore& kernel);
-    ~GlobalScheduler();
-
-    /// Adds a new thread to the scheduler
-    void AddThread(std::shared_ptr<Thread> thread);
-
-    /// Removes a thread from the scheduler
-    void RemoveThread(std::shared_ptr<Thread> thread);
-
-    /// Returns a list of all threads managed by the scheduler
-    const std::vector<std::shared_ptr<Thread>>& GetThreadList() const {
-        return thread_list;
-    }
-
-    /// Notify the scheduler a thread's status has changed.
-    void AdjustSchedulingOnStatus(Thread* thread, u32 old_flags);
-
-    /// Notify the scheduler a thread's priority has changed.
-    void AdjustSchedulingOnPriority(Thread* thread, u32 old_priority);
-
-    /// 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 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.
-     *
-     * 2. Second we try to get a suggested thread that's not assigned to any core or
-     *    that is not the top thread in that core.
-     *
-     * 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.
-     */
-    u32 SelectThreads();
-
-    bool HaveReadyThreads(std::size_t core_id) const {
-        return !scheduled_queue[core_id].empty();
-    }
-
-    /**
-     * Takes a thread and moves it to the back of the it's priority list.
-     *
-     * @note This operation can be redundant and no scheduling is changed if marked as so.
-     */
-    bool YieldThread(Thread* thread);
-
-    /**
-     * Takes a thread and moves it to the back of the it's priority list.
-     * Afterwards, tries to pick a suggested thread from the suggested queue that has worse time or
-     * a better priority than the next thread in the core.
-     *
-     * @note This operation can be redundant and no scheduling is changed if marked as so.
-     */
-    bool YieldThreadAndBalanceLoad(Thread* thread);
-
-    /**
-     * Takes a thread and moves it out of the scheduling queue.
-     * and into the suggested queue. If no thread can be scheduled afterwards in that core,
-     * a suggested thread is obtained instead.
-     *
-     * @note This operation can be redundant and no scheduling is changed if marked as so.
-     */
-    bool YieldThreadAndWaitForLoadBalancing(Thread* thread);
-
-    /**
-     * Rotates the scheduling queues of threads at a preemption priority and then does
-     * some core rebalancing. Preemption priorities can be found in the array
-     * 'preemption_priorities'.
-     *
-     * @note This operation happens every 10ms.
-     */
-    void PreemptThreads();
-
-    u32 CpuCoresCount() const {
-        return Core::Hardware::NUM_CPU_CORES;
-    }
-
-    void SetReselectionPending() {
-        is_reselection_pending.store(true, std::memory_order_release);
-    }
-
-    bool IsReselectionPending() const {
-        return is_reselection_pending.load(std::memory_order_acquire);
-    }
-
-    void Shutdown();
-
-private:
-    friend class SchedulerLock;
-
-    /// Lock the scheduler to the current thread.
-    void Lock();
-
-    /// 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
-     * queue.
-     */
-    void TransferToCore(u32 priority, s32 destination_core, Thread* thread);
-
-    bool AskForReselectionOrMarkRedundant(Thread* current_thread, const Thread* winner);
-
-    static constexpr u32 min_regular_priority = 2;
-    std::array<Common::MultiLevelQueue<Thread*, THREADPRIO_COUNT>, Core::Hardware::NUM_CPU_CORES>
-        scheduled_queue;
-    std::array<Common::MultiLevelQueue<Thread*, THREADPRIO_COUNT>, Core::Hardware::NUM_CPU_CORES>
-        suggested_queue;
-    std::atomic<bool> is_reselection_pending{false};
-
-    // The priority levels at which the global scheduler preempts threads every 10 ms. They are
-    // ordered from Core 0 to Core 3.
-    std::array<u32, Core::Hardware::NUM_CPU_CORES> preemption_priorities = {59, 59, 59, 62};
-
-    /// Scheduler lock mechanisms.
-    bool is_locked{};
-    std::mutex 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;
-};
-
-class Scheduler final {
-public:
-    explicit Scheduler(Core::System& system, std::size_t core_id);
-    ~Scheduler();
-
-    /// Returns whether there are any threads that are ready to run.
-    bool HaveReadyThreads() const;
-
-    /// Reschedules to the next available thread (call after current thread is suspended)
-    void TryDoContextSwitch();
-
-    /// The next two are for SingleCore Only.
-    /// Unload current thread before preempting core.
-    void Unload(Thread* thread);
-    void Unload();
-    /// Reload current thread after core preemption.
-    void Reload(Thread* thread);
-    void Reload();
-
-    /// Gets the current running thread
-    Thread* GetCurrentThread() const;
-
-    /// Gets the currently selected thread from the top of the multilevel queue
-    Thread* GetSelectedThread() const;
-
-    /// Gets the timestamp for the last context switch in ticks.
-    u64 GetLastContextSwitchTicks() const;
-
-    bool ContextSwitchPending() const {
-        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
-     * process using the following difference:
-     *
-     * ticks += most_recent_ticks - last_context_switch_ticks
-     *
-     * The internal tick timestamp for the scheduler is simply the
-     * most recent tick count retrieved. No special arithmetic is
-     * applied to it.
-     */
-    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;
-};
-
-class SchedulerLock {
-public:
-    [[nodiscard]] explicit SchedulerLock(KernelCore& kernel);
-    ~SchedulerLock();
-
-protected:
-    KernelCore& kernel;
-};
-
-class SchedulerLockAndSleep : public SchedulerLock {
-public:
-    explicit SchedulerLockAndSleep(KernelCore& kernel, Handle& event_handle, Thread* time_task,
-                                   s64 nanoseconds);
-    ~SchedulerLockAndSleep();
-
-    void CancelSleep() {
-        sleep_cancelled = true;
-    }
-
-    void Release();
-
-private:
-    Handle& event_handle;
-    Thread* time_task;
-    s64 nanoseconds;
-    bool sleep_cancelled{};
-};
-
-} // namespace Kernel
diff --git a/src/core/hle/kernel/server_session.cpp b/src/core/hle/kernel/server_session.cpp
index ae088cf41..a35c8aa4b 100644
--- a/src/core/hle/kernel/server_session.cpp
+++ b/src/core/hle/kernel/server_session.cpp
@@ -14,9 +14,9 @@
 #include "core/hle/kernel/client_session.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/hle_ipc.h"
+#include "core/hle/kernel/k_scheduler.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"
@@ -170,7 +170,7 @@ ResultCode ServerSession::CompleteSyncRequest() {
 
     // Some service requests require the thread to block
     {
-        SchedulerLock lock(kernel);
+        KScopedSchedulerLock lock(kernel);
         if (!context.IsThreadWaiting()) {
             context.GetThread().ResumeFromWait();
             context.GetThread().SetSynchronizationResults(nullptr, result);
diff --git a/src/core/hle/kernel/svc.cpp b/src/core/hle/kernel/svc.cpp
index c8060f179..2d225392f 100644
--- a/src/core/hle/kernel/svc.cpp
+++ b/src/core/hle/kernel/svc.cpp
@@ -24,6 +24,8 @@
 #include "core/hle/kernel/client_session.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/memory/memory_block.h"
 #include "core/hle/kernel/memory/page_table.h"
@@ -32,7 +34,6 @@
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/readable_event.h"
 #include "core/hle/kernel/resource_limit.h"
-#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/shared_memory.h"
 #include "core/hle/kernel/svc.h"
 #include "core/hle/kernel/svc_types.h"
@@ -329,7 +330,8 @@ static ResultCode ConnectToNamedPort32(Core::System& system, Handle* out_handle,
 
 /// Makes a blocking IPC call to an OS service.
 static ResultCode SendSyncRequest(Core::System& system, Handle handle) {
-    const auto& handle_table = system.Kernel().CurrentProcess()->GetHandleTable();
+    auto& kernel = system.Kernel();
+    const auto& handle_table = kernel.CurrentProcess()->GetHandleTable();
     std::shared_ptr<ClientSession> session = handle_table.Get<ClientSession>(handle);
     if (!session) {
         LOG_ERROR(Kernel_SVC, "called with invalid handle=0x{:08X}", handle);
@@ -338,9 +340,9 @@ static ResultCode SendSyncRequest(Core::System& system, Handle handle) {
 
     LOG_TRACE(Kernel_SVC, "called handle=0x{:08X}({})", handle, session->GetName());
 
-    auto thread = system.CurrentScheduler().GetCurrentThread();
+    auto thread = kernel.CurrentScheduler()->GetCurrentThread();
     {
-        SchedulerLock lock(system.Kernel());
+        KScopedSchedulerLock lock(kernel);
         thread->InvalidateHLECallback();
         thread->SetStatus(ThreadStatus::WaitIPC);
         session->SendSyncRequest(SharedFrom(thread), system.Memory(), system.CoreTiming());
@@ -349,12 +351,12 @@ static ResultCode SendSyncRequest(Core::System& system, Handle handle) {
     if (thread->HasHLECallback()) {
         Handle event_handle = thread->GetHLETimeEvent();
         if (event_handle != InvalidHandle) {
-            auto& time_manager = system.Kernel().TimeManager();
+            auto& time_manager = kernel.TimeManager();
             time_manager.UnscheduleTimeEvent(event_handle);
         }
 
         {
-            SchedulerLock lock(system.Kernel());
+            KScopedSchedulerLock lock(kernel);
             auto* sync_object = thread->GetHLESyncObject();
             sync_object->RemoveWaitingThread(SharedFrom(thread));
         }
@@ -654,7 +656,6 @@ 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}",
@@ -662,13 +663,9 @@ static void Break(Core::System& system, u32 reason, u64 info1, u64 info2) {
 
         handle_debug_buffer(info1, info2);
 
-        auto* const current_thread = system.CurrentScheduler().GetCurrentThread();
+        auto* const current_thread = system.Kernel().CurrentScheduler()->GetCurrentThread();
         const auto thread_processor_id = current_thread->GetProcessorID();
         system.ArmInterface(static_cast<std::size_t>(thread_processor_id)).LogBacktrace();
-
-        // Kill the current thread
-        system.Kernel().ExceptionalExit();
-        current_thread->Stop();
     }
 }
 
@@ -918,7 +915,7 @@ static ResultCode GetInfo(Core::System& system, u64* result, u64 info_id, u64 ha
         }
 
         const auto& core_timing = system.CoreTiming();
-        const auto& scheduler = system.CurrentScheduler();
+        const auto& scheduler = *system.Kernel().CurrentScheduler();
         const auto* const current_thread = scheduler.GetCurrentThread();
         const bool same_thread = current_thread == thread.get();
 
@@ -1086,7 +1083,7 @@ static ResultCode SetThreadActivity(Core::System& system, Handle handle, u32 act
         return ERR_INVALID_HANDLE;
     }
 
-    if (thread.get() == system.CurrentScheduler().GetCurrentThread()) {
+    if (thread.get() == system.Kernel().CurrentScheduler()->GetCurrentThread()) {
         LOG_ERROR(Kernel_SVC, "The thread handle specified is the current running thread");
         return ERR_BUSY;
     }
@@ -1119,7 +1116,7 @@ static ResultCode GetThreadContext(Core::System& system, VAddr thread_context, H
         return ERR_INVALID_HANDLE;
     }
 
-    if (thread.get() == system.CurrentScheduler().GetCurrentThread()) {
+    if (thread.get() == system.Kernel().CurrentScheduler()->GetCurrentThread()) {
         LOG_ERROR(Kernel_SVC, "The thread handle specified is the current running thread");
         return ERR_BUSY;
     }
@@ -1475,7 +1472,7 @@ static void ExitProcess(Core::System& system) {
     current_process->PrepareForTermination();
 
     // Kill the current thread
-    system.CurrentScheduler().GetCurrentThread()->Stop();
+    system.Kernel().CurrentScheduler()->GetCurrentThread()->Stop();
 }
 
 static void ExitProcess32(Core::System& system) {
@@ -1575,8 +1572,8 @@ static ResultCode StartThread32(Core::System& system, Handle thread_handle) {
 static void ExitThread(Core::System& system) {
     LOG_DEBUG(Kernel_SVC, "called, pc=0x{:08X}", system.CurrentArmInterface().GetPC());
 
-    auto* const current_thread = system.CurrentScheduler().GetCurrentThread();
-    system.GlobalScheduler().RemoveThread(SharedFrom(current_thread));
+    auto* const current_thread = system.Kernel().CurrentScheduler()->GetCurrentThread();
+    system.GlobalSchedulerContext().RemoveThread(SharedFrom(current_thread));
     current_thread->Stop();
 }
 
@@ -1589,44 +1586,31 @@ static void SleepThread(Core::System& system, s64 nanoseconds) {
     LOG_DEBUG(Kernel_SVC, "called nanoseconds={}", nanoseconds);
 
     enum class SleepType : s64 {
-        YieldWithoutLoadBalancing = 0,
-        YieldWithLoadBalancing = -1,
+        YieldWithoutCoreMigration = 0,
+        YieldWithCoreMigration = -1,
         YieldAndWaitForLoadBalancing = -2,
     };
 
-    auto& scheduler = system.CurrentScheduler();
-    auto* const current_thread = scheduler.GetCurrentThread();
-    bool is_redundant = false;
-
+    auto& scheduler = *system.Kernel().CurrentScheduler();
     if (nanoseconds <= 0) {
         switch (static_cast<SleepType>(nanoseconds)) {
-        case SleepType::YieldWithoutLoadBalancing: {
-            auto pair = current_thread->YieldSimple();
-            is_redundant = pair.second;
+        case SleepType::YieldWithoutCoreMigration: {
+            scheduler.YieldWithoutCoreMigration();
             break;
         }
-        case SleepType::YieldWithLoadBalancing: {
-            auto pair = current_thread->YieldAndBalanceLoad();
-            is_redundant = pair.second;
+        case SleepType::YieldWithCoreMigration: {
+            scheduler.YieldWithCoreMigration();
             break;
         }
         case SleepType::YieldAndWaitForLoadBalancing: {
-            auto pair = current_thread->YieldAndWaitForLoadBalancing();
-            is_redundant = pair.second;
+            scheduler.YieldToAnyThread();
             break;
         }
         default:
             UNREACHABLE_MSG("Unimplemented sleep yield type '{:016X}'!", nanoseconds);
         }
     } else {
-        current_thread->Sleep(nanoseconds);
-    }
-
-    if (is_redundant && !system.Kernel().IsMulticore()) {
-        system.Kernel().ExitSVCProfile();
-        system.CoreTiming().AddTicks(1000U);
-        system.GetCpuManager().PreemptSingleCore();
-        system.Kernel().EnterSVCProfile();
+        scheduler.GetCurrentThread()->Sleep(nanoseconds);
     }
 }
 
@@ -1661,10 +1645,10 @@ 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();
+    Thread* current_thread = kernel.CurrentScheduler()->GetCurrentThread();
+    auto* const current_process = kernel.CurrentProcess();
     {
-        SchedulerLockAndSleep lock(kernel, event_handle, current_thread, nano_seconds);
+        KScopedSchedulerLockAndSleep 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);
@@ -1700,7 +1684,7 @@ static ResultCode WaitProcessWideKeyAtomic(Core::System& system, VAddr mutex_add
     }
 
     {
-        SchedulerLock lock(kernel);
+        KScopedSchedulerLock lock(kernel);
 
         auto* owner = current_thread->GetLockOwner();
         if (owner != nullptr) {
@@ -1731,7 +1715,7 @@ static void SignalProcessWideKey(Core::System& system, VAddr condition_variable_
 
     // Retrieve a list of all threads that are waiting for this condition variable.
     auto& kernel = system.Kernel();
-    SchedulerLock lock(kernel);
+    KScopedSchedulerLock lock(kernel);
     auto* const current_process = kernel.CurrentProcess();
     std::vector<std::shared_ptr<Thread>> waiting_threads =
         current_process->GetConditionVariableThreads(condition_variable_addr);
@@ -1993,7 +1977,7 @@ static ResultCode GetThreadCoreMask(Core::System& system, Handle thread_handle,
     }
 
     *core = thread->GetIdealCore();
-    *mask = thread->GetAffinityMask();
+    *mask = thread->GetAffinityMask().GetAffinityMask();
 
     return RESULT_SUCCESS;
 }
@@ -2629,7 +2613,7 @@ void Call(Core::System& system, u32 immediate) {
     auto& kernel = system.Kernel();
     kernel.EnterSVCProfile();
 
-    auto* thread = system.CurrentScheduler().GetCurrentThread();
+    auto* thread = kernel.CurrentScheduler()->GetCurrentThread();
     thread->SetContinuousOnSVC(true);
 
     const FunctionDef* info = system.CurrentProcess()->Is64BitProcess() ? GetSVCInfo64(immediate)
diff --git a/src/core/hle/kernel/synchronization.cpp b/src/core/hle/kernel/synchronization.cpp
index 8b875d853..d3f520ea2 100644
--- a/src/core/hle/kernel/synchronization.cpp
+++ b/src/core/hle/kernel/synchronization.cpp
@@ -5,8 +5,9 @@
 #include "core/core.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
 #include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/synchronization.h"
 #include "core/hle/kernel/synchronization_object.h"
 #include "core/hle/kernel/thread.h"
@@ -18,7 +19,7 @@ Synchronization::Synchronization(Core::System& system) : system{system} {}
 
 void Synchronization::SignalObject(SynchronizationObject& obj) const {
     auto& kernel = system.Kernel();
-    SchedulerLock lock(kernel);
+    KScopedSchedulerLock lock(kernel);
     if (obj.IsSignaled()) {
         for (auto thread : obj.GetWaitingThreads()) {
             if (thread->GetSchedulingStatus() == ThreadSchedStatus::Paused) {
@@ -37,10 +38,10 @@ void Synchronization::SignalObject(SynchronizationObject& obj) const {
 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();
+    auto* const thread = kernel.CurrentScheduler()->GetCurrentThread();
     Handle event_handle = InvalidHandle;
     {
-        SchedulerLockAndSleep lock(kernel, event_handle, thread, nano_seconds);
+        KScopedSchedulerLockAndSleep 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) {
@@ -89,7 +90,7 @@ std::pair<ResultCode, Handle> Synchronization::WaitFor(
     }
 
     {
-        SchedulerLock lock(kernel);
+        KScopedSchedulerLock lock(kernel);
         ResultCode signaling_result = thread->GetSignalingResult();
         SynchronizationObject* signaling_object = thread->GetSignalingObject();
         thread->SetSynchronizationObjects(nullptr);
diff --git a/src/core/hle/kernel/thread.cpp b/src/core/hle/kernel/thread.cpp
index 7d1eb2c6e..a4f9e0d97 100644
--- a/src/core/hle/kernel/thread.cpp
+++ b/src/core/hle/kernel/thread.cpp
@@ -17,10 +17,11 @@
 #include "core/hardware_properties.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
+#include "core/hle/kernel/k_scoped_scheduler_lock_and_sleep.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/object.h"
 #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"
@@ -50,7 +51,7 @@ Thread::~Thread() = default;
 
 void Thread::Stop() {
     {
-        SchedulerLock lock(kernel);
+        KScopedSchedulerLock lock(kernel);
         SetStatus(ThreadStatus::Dead);
         Signal();
         kernel.GlobalHandleTable().Close(global_handle);
@@ -67,7 +68,7 @@ void Thread::Stop() {
 }
 
 void Thread::ResumeFromWait() {
-    SchedulerLock lock(kernel);
+    KScopedSchedulerLock lock(kernel);
     switch (status) {
     case ThreadStatus::Paused:
     case ThreadStatus::WaitSynch:
@@ -99,19 +100,18 @@ void Thread::ResumeFromWait() {
 }
 
 void Thread::OnWakeUp() {
-    SchedulerLock lock(kernel);
-
+    KScopedSchedulerLock lock(kernel);
     SetStatus(ThreadStatus::Ready);
 }
 
 ResultCode Thread::Start() {
-    SchedulerLock lock(kernel);
+    KScopedSchedulerLock lock(kernel);
     SetStatus(ThreadStatus::Ready);
     return RESULT_SUCCESS;
 }
 
 void Thread::CancelWait() {
-    SchedulerLock lock(kernel);
+    KScopedSchedulerLock lock(kernel);
     if (GetSchedulingStatus() != ThreadSchedStatus::Paused || !is_waiting_on_sync) {
         is_sync_cancelled = true;
         return;
@@ -186,12 +186,14 @@ ResultVal<std::shared_ptr<Thread>> Thread::Create(Core::System& system, ThreadTy
     thread->status = ThreadStatus::Dormant;
     thread->entry_point = entry_point;
     thread->stack_top = stack_top;
+    thread->disable_count = 1;
     thread->tpidr_el0 = 0;
     thread->nominal_priority = thread->current_priority = priority;
-    thread->last_running_ticks = 0;
+    thread->schedule_count = -1;
+    thread->last_scheduled_tick = 0;
     thread->processor_id = processor_id;
     thread->ideal_core = processor_id;
-    thread->affinity_mask = 1ULL << processor_id;
+    thread->affinity_mask.SetAffinity(processor_id, true);
     thread->wait_objects = nullptr;
     thread->mutex_wait_address = 0;
     thread->condvar_wait_address = 0;
@@ -201,7 +203,7 @@ ResultVal<std::shared_ptr<Thread>> Thread::Create(Core::System& system, ThreadTy
     thread->owner_process = owner_process;
     thread->type = type_flags;
     if ((type_flags & THREADTYPE_IDLE) == 0) {
-        auto& scheduler = kernel.GlobalScheduler();
+        auto& scheduler = kernel.GlobalSchedulerContext();
         scheduler.AddThread(thread);
     }
     if (owner_process) {
@@ -225,7 +227,7 @@ ResultVal<std::shared_ptr<Thread>> Thread::Create(Core::System& system, ThreadTy
 }
 
 void Thread::SetPriority(u32 priority) {
-    SchedulerLock lock(kernel);
+    KScopedSchedulerLock lock(kernel);
     ASSERT_MSG(priority <= THREADPRIO_LOWEST && priority >= THREADPRIO_HIGHEST,
                "Invalid priority value.");
     nominal_priority = priority;
@@ -362,7 +364,7 @@ bool Thread::InvokeHLECallback(std::shared_ptr<Thread> thread) {
 }
 
 ResultCode Thread::SetActivity(ThreadActivity value) {
-    SchedulerLock lock(kernel);
+    KScopedSchedulerLock lock(kernel);
 
     auto sched_status = GetSchedulingStatus();
 
@@ -391,7 +393,7 @@ ResultCode Thread::SetActivity(ThreadActivity value) {
 ResultCode Thread::Sleep(s64 nanoseconds) {
     Handle event_handle{};
     {
-        SchedulerLockAndSleep lock(kernel, event_handle, this, nanoseconds);
+        KScopedSchedulerLockAndSleep lock(kernel, event_handle, this, nanoseconds);
         SetStatus(ThreadStatus::WaitSleep);
     }
 
@@ -402,39 +404,12 @@ ResultCode Thread::Sleep(s64 nanoseconds) {
     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};
-}
-
-std::pair<ResultCode, bool> Thread::YieldAndWaitForLoadBalancing() {
-    bool is_redundant = false;
-    {
-        SchedulerLock lock(kernel);
-        is_redundant = kernel.GlobalScheduler().YieldThreadAndWaitForLoadBalancing(this);
-    }
-    return {RESULT_SUCCESS, is_redundant};
-}
-
 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);
+    KScheduler::OnThreadStateChanged(kernel, this, old_state);
 }
 
 void Thread::RemoveSchedulingFlag(ThreadSchedFlags flag) {
@@ -442,23 +417,24 @@ void Thread::RemoveSchedulingFlag(ThreadSchedFlags flag) {
     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);
+    KScheduler::OnThreadStateChanged(kernel, this, old_state);
 }
 
 void Thread::SetSchedulingStatus(ThreadSchedStatus new_status) {
     const u32 old_state = scheduling_state;
     scheduling_state = (scheduling_state & static_cast<u32>(ThreadSchedMasks::HighMask)) |
                        static_cast<u32>(new_status);
-    kernel.GlobalScheduler().AdjustSchedulingOnStatus(this, old_state);
+    KScheduler::OnThreadStateChanged(kernel, this, old_state);
 }
 
 void Thread::SetCurrentPriority(u32 new_priority) {
     const u32 old_priority = std::exchange(current_priority, new_priority);
-    kernel.GlobalScheduler().AdjustSchedulingOnPriority(this, old_priority);
+    KScheduler::OnThreadPriorityChanged(kernel, this, kernel.CurrentScheduler()->GetCurrentThread(),
+                                        old_priority);
 }
 
 ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) {
-    SchedulerLock lock(kernel);
+    KScopedSchedulerLock 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) {
@@ -479,20 +455,21 @@ ResultCode Thread::SetCoreAndAffinityMask(s32 new_core, u64 new_affinity_mask) {
     }
     if (use_override) {
         ideal_core_override = new_core;
-        affinity_mask_override = new_affinity_mask;
     } else {
-        const u64 old_affinity_mask = std::exchange(affinity_mask, new_affinity_mask);
+        const auto old_affinity_mask = affinity_mask;
+        affinity_mask.SetAffinityMask(new_affinity_mask);
         ideal_core = new_core;
-        if (old_affinity_mask != new_affinity_mask) {
+        if (old_affinity_mask.GetAffinityMask() != new_affinity_mask) {
             const s32 old_core = processor_id;
-            if (processor_id >= 0 && ((affinity_mask >> processor_id) & 1) == 0) {
+            if (processor_id >= 0 && !affinity_mask.GetAffinity(processor_id)) {
                 if (static_cast<s32>(ideal_core) < 0) {
-                    processor_id = HighestSetCore(affinity_mask, Core::Hardware::NUM_CPU_CORES);
+                    processor_id = HighestSetCore(affinity_mask.GetAffinityMask(),
+                                                  Core::Hardware::NUM_CPU_CORES);
                 } else {
                     processor_id = ideal_core;
                 }
             }
-            kernel.GlobalScheduler().AdjustSchedulingOnAffinity(this, old_affinity_mask, old_core);
+            KScheduler::OnThreadAffinityMaskChanged(kernel, this, old_affinity_mask, old_core);
         }
     }
     return RESULT_SUCCESS;
diff --git a/src/core/hle/kernel/thread.h b/src/core/hle/kernel/thread.h
index a75071e9b..11ef29888 100644
--- a/src/core/hle/kernel/thread.h
+++ b/src/core/hle/kernel/thread.h
@@ -4,6 +4,7 @@
 
 #pragma once
 
+#include <array>
 #include <functional>
 #include <string>
 #include <utility>
@@ -12,6 +13,7 @@
 #include "common/common_types.h"
 #include "common/spin_lock.h"
 #include "core/arm/arm_interface.h"
+#include "core/hle/kernel/k_affinity_mask.h"
 #include "core/hle/kernel/object.h"
 #include "core/hle/kernel/synchronization_object.h"
 #include "core/hle/result.h"
@@ -27,10 +29,10 @@ class System;
 
 namespace Kernel {
 
-class GlobalScheduler;
+class GlobalSchedulerContext;
 class KernelCore;
 class Process;
-class Scheduler;
+class KScheduler;
 
 enum ThreadPriority : u32 {
     THREADPRIO_HIGHEST = 0,            ///< Highest thread priority
@@ -345,8 +347,12 @@ public:
 
     void SetStatus(ThreadStatus new_status);
 
-    u64 GetLastRunningTicks() const {
-        return last_running_ticks;
+    s64 GetLastScheduledTick() const {
+        return this->last_scheduled_tick;
+    }
+
+    void SetLastScheduledTick(s64 tick) {
+        this->last_scheduled_tick = tick;
     }
 
     u64 GetTotalCPUTimeTicks() const {
@@ -361,10 +367,18 @@ public:
         return processor_id;
     }
 
+    s32 GetActiveCore() const {
+        return GetProcessorID();
+    }
+
     void SetProcessorID(s32 new_core) {
         processor_id = new_core;
     }
 
+    void SetActiveCore(s32 new_core) {
+        processor_id = new_core;
+    }
+
     Process* GetOwnerProcess() {
         return owner_process;
     }
@@ -469,7 +483,7 @@ public:
         return ideal_core;
     }
 
-    u64 GetAffinityMask() const {
+    const KAffinityMask& GetAffinityMask() const {
         return affinity_mask;
     }
 
@@ -478,21 +492,12 @@ public:
     /// Sleeps this thread for the given amount of nanoseconds.
     ResultCode Sleep(s64 nanoseconds);
 
-    /// Yields this thread without rebalancing loads.
-    std::pair<ResultCode, bool> YieldSimple();
-
-    /// Yields this thread and does a load rebalancing.
-    std::pair<ResultCode, bool> YieldAndBalanceLoad();
-
-    /// Yields this thread and if the core is left idle, loads are rebalanced
-    std::pair<ResultCode, bool> YieldAndWaitForLoadBalancing();
-
-    void IncrementYieldCount() {
-        yield_count++;
+    s64 GetYieldScheduleCount() const {
+        return this->schedule_count;
     }
 
-    u64 GetYieldCount() const {
-        return yield_count;
+    void SetYieldScheduleCount(s64 count) {
+        this->schedule_count = count;
     }
 
     ThreadSchedStatus GetSchedulingStatus() const {
@@ -568,9 +573,59 @@ public:
         return has_exited;
     }
 
+    class QueueEntry {
+    public:
+        constexpr QueueEntry() = default;
+
+        constexpr void Initialize() {
+            this->prev = nullptr;
+            this->next = nullptr;
+        }
+
+        constexpr Thread* GetPrev() const {
+            return this->prev;
+        }
+        constexpr Thread* GetNext() const {
+            return this->next;
+        }
+        constexpr void SetPrev(Thread* thread) {
+            this->prev = thread;
+        }
+        constexpr void SetNext(Thread* thread) {
+            this->next = thread;
+        }
+
+    private:
+        Thread* prev{};
+        Thread* next{};
+    };
+
+    QueueEntry& GetPriorityQueueEntry(s32 core) {
+        return this->per_core_priority_queue_entry[core];
+    }
+
+    const QueueEntry& GetPriorityQueueEntry(s32 core) const {
+        return this->per_core_priority_queue_entry[core];
+    }
+
+    s32 GetDisableDispatchCount() const {
+        return disable_count;
+    }
+
+    void DisableDispatch() {
+        ASSERT(GetDisableDispatchCount() >= 0);
+        disable_count++;
+    }
+
+    void EnableDispatch() {
+        ASSERT(GetDisableDispatchCount() > 0);
+        disable_count--;
+    }
+
 private:
-    friend class GlobalScheduler;
-    friend class Scheduler;
+    friend class GlobalSchedulerContext;
+    friend class KScheduler;
+    friend class Process;
 
     void SetSchedulingStatus(ThreadSchedStatus new_status);
     void AddSchedulingFlag(ThreadSchedFlags flag);
@@ -583,12 +638,14 @@ private:
     ThreadContext64 context_64{};
     std::shared_ptr<Common::Fiber> host_context{};
 
-    u64 thread_id = 0;
-
     ThreadStatus status = ThreadStatus::Dormant;
+    u32 scheduling_state = 0;
+
+    u64 thread_id = 0;
 
     VAddr entry_point = 0;
     VAddr stack_top = 0;
+    std::atomic_int disable_count = 0;
 
     ThreadType type;
 
@@ -602,9 +659,8 @@ private:
     u32 current_priority = 0;
 
     u64 total_cpu_time_ticks = 0; ///< Total CPU running ticks.
-    u64 last_running_ticks = 0;   ///< CPU tick when thread was last running
-    u64 yield_count = 0;          ///< Number of redundant yields carried by this thread.
-                                  ///< a redundant yield is one where no scheduling is changed
+    s64 schedule_count{};
+    s64 last_scheduled_tick{};
 
     s32 processor_id = 0;
 
@@ -646,16 +702,16 @@ private:
     Handle hle_time_event;
     SynchronizationObject* hle_object;
 
-    Scheduler* scheduler = nullptr;
+    KScheduler* scheduler = nullptr;
+
+    std::array<QueueEntry, Core::Hardware::NUM_CPU_CORES> per_core_priority_queue_entry{};
 
     u32 ideal_core{0xFFFFFFFF};
-    u64 affinity_mask{0x1};
+    KAffinityMask affinity_mask{};
 
     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;
diff --git a/src/core/hle/kernel/time_manager.cpp b/src/core/hle/kernel/time_manager.cpp
index caf329bfb..79628e2b4 100644
--- a/src/core/hle/kernel/time_manager.cpp
+++ b/src/core/hle/kernel/time_manager.cpp
@@ -7,8 +7,8 @@
 #include "core/core_timing.h"
 #include "core/core_timing_util.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.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"
 
@@ -18,12 +18,18 @@ TimeManager::TimeManager(Core::System& system_) : system{system_} {
     time_manager_event_type = Core::Timing::CreateEvent(
         "Kernel::TimeManagerCallback",
         [this](std::uintptr_t thread_handle, std::chrono::nanoseconds) {
-            const SchedulerLock lock(system.Kernel());
+            const KScopedSchedulerLock lock(system.Kernel());
             const auto proper_handle = static_cast<Handle>(thread_handle);
-            if (cancelled_events[proper_handle]) {
-                return;
+
+            std::shared_ptr<Thread> thread;
+            {
+                std::lock_guard lock{mutex};
+                if (cancelled_events[proper_handle]) {
+                    return;
+                }
+                thread = system.Kernel().RetrieveThreadFromGlobalHandleTable(proper_handle);
             }
-            auto thread = this->system.Kernel().RetrieveThreadFromGlobalHandleTable(proper_handle);
+
             if (thread) {
                 // Thread can be null if process has exited
                 thread->OnWakeUp();
@@ -56,6 +62,7 @@ void TimeManager::UnscheduleTimeEvent(Handle event_handle) {
 }
 
 void TimeManager::CancelTimeEvent(Thread* time_task) {
+    std::lock_guard lock{mutex};
     const Handle event_handle = time_task->GetGlobalHandle();
     UnscheduleTimeEvent(event_handle);
 }
diff --git a/src/core/hle/service/nvflinger/buffer_queue.cpp b/src/core/hle/service/nvflinger/buffer_queue.cpp
index 191286ce9..377f47e8e 100644
--- a/src/core/hle/service/nvflinger/buffer_queue.cpp
+++ b/src/core/hle/service/nvflinger/buffer_queue.cpp
@@ -22,10 +22,11 @@ BufferQueue::BufferQueue(Kernel::KernelCore& kernel, u32 id, u64 layer_id)
 BufferQueue::~BufferQueue() = default;
 
 void BufferQueue::SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer) {
+    ASSERT(slot < buffer_slots);
     LOG_WARNING(Service, "Adding graphics buffer {}", slot);
 
     free_buffers.push_back(slot);
-    queue.push_back({
+    buffers[slot] = {
         .slot = slot,
         .status = Buffer::Status::Free,
         .igbp_buffer = igbp_buffer,
@@ -33,7 +34,7 @@ void BufferQueue::SetPreallocatedBuffer(u32 slot, const IGBPBuffer& igbp_buffer)
         .crop_rect = {},
         .swap_interval = 0,
         .multi_fence = {},
-    });
+    };
 
     buffer_wait_event.writable->Signal();
 }
@@ -44,73 +45,57 @@ std::optional<std::pair<u32, Service::Nvidia::MultiFence*>> BufferQueue::Dequeue
     if (free_buffers.empty()) {
         return std::nullopt;
     }
-
     auto f_itr = free_buffers.begin();
-    auto itr = queue.end();
+    auto slot = buffers.size();
 
     while (f_itr != free_buffers.end()) {
-        auto slot = *f_itr;
-        itr = std::find_if(queue.begin(), queue.end(), [&](const Buffer& buffer) {
-            // Only consider free buffers. Buffers become free once again after they've been
-            // Acquired and Released by the compositor, see the NVFlinger::Compose method.
-            if (buffer.status != Buffer::Status::Free) {
-                return false;
-            }
-
-            if (buffer.slot != slot) {
-                return false;
-            }
-
-            // Make sure that the parameters match.
-            return buffer.igbp_buffer.width == width && buffer.igbp_buffer.height == height;
-        });
-
-        if (itr != queue.end()) {
+        const Buffer& buffer = buffers[*f_itr];
+        if (buffer.status == Buffer::Status::Free && buffer.igbp_buffer.width == width &&
+            buffer.igbp_buffer.height == height) {
+            slot = *f_itr;
             free_buffers.erase(f_itr);
             break;
         }
         ++f_itr;
     }
-
-    if (itr == queue.end()) {
+    if (slot == buffers.size()) {
         return std::nullopt;
     }
-
-    itr->status = Buffer::Status::Dequeued;
-    return {{itr->slot, &itr->multi_fence}};
+    buffers[slot].status = Buffer::Status::Dequeued;
+    return {{buffers[slot].slot, &buffers[slot].multi_fence}};
 }
 
 const IGBPBuffer& BufferQueue::RequestBuffer(u32 slot) const {
-    auto itr = std::find_if(queue.begin(), queue.end(),
-                            [&](const Buffer& buffer) { return buffer.slot == slot; });
-    ASSERT(itr != queue.end());
-    ASSERT(itr->status == Buffer::Status::Dequeued);
-    return itr->igbp_buffer;
+    ASSERT(slot < buffers.size());
+    ASSERT(buffers[slot].status == Buffer::Status::Dequeued);
+    ASSERT(buffers[slot].slot == slot);
+
+    return buffers[slot].igbp_buffer;
 }
 
 void BufferQueue::QueueBuffer(u32 slot, BufferTransformFlags transform,
                               const Common::Rectangle<int>& crop_rect, u32 swap_interval,
                               Service::Nvidia::MultiFence& multi_fence) {
-    auto itr = std::find_if(queue.begin(), queue.end(),
-                            [&](const Buffer& buffer) { return buffer.slot == slot; });
-    ASSERT(itr != queue.end());
-    ASSERT(itr->status == Buffer::Status::Dequeued);
-    itr->status = Buffer::Status::Queued;
-    itr->transform = transform;
-    itr->crop_rect = crop_rect;
-    itr->swap_interval = swap_interval;
-    itr->multi_fence = multi_fence;
+    ASSERT(slot < buffers.size());
+    ASSERT(buffers[slot].status == Buffer::Status::Dequeued);
+    ASSERT(buffers[slot].slot == slot);
+
+    buffers[slot].status = Buffer::Status::Queued;
+    buffers[slot].transform = transform;
+    buffers[slot].crop_rect = crop_rect;
+    buffers[slot].swap_interval = swap_interval;
+    buffers[slot].multi_fence = multi_fence;
     queue_sequence.push_back(slot);
 }
 
 void BufferQueue::CancelBuffer(u32 slot, const Service::Nvidia::MultiFence& multi_fence) {
-    const auto itr = std::find_if(queue.begin(), queue.end(),
-                                  [slot](const Buffer& buffer) { return buffer.slot == slot; });
-    ASSERT(itr != queue.end());
-    ASSERT(itr->status != Buffer::Status::Free);
-    itr->status = Buffer::Status::Free;
-    itr->multi_fence = multi_fence;
-    itr->swap_interval = 0;
+    ASSERT(slot < buffers.size());
+    ASSERT(buffers[slot].status != Buffer::Status::Free);
+    ASSERT(buffers[slot].slot == slot);
+
+    buffers[slot].status = Buffer::Status::Free;
+    buffers[slot].multi_fence = multi_fence;
+    buffers[slot].swap_interval = 0;
 
     free_buffers.push_back(slot);
 
@@ -118,38 +103,39 @@ void BufferQueue::CancelBuffer(u32 slot, const Service::Nvidia::MultiFence& mult
 }
 
 std::optional<std::reference_wrapper<const BufferQueue::Buffer>> BufferQueue::AcquireBuffer() {
-    auto itr = queue.end();
+    std::size_t buffer_slot = buffers.size();
     // Iterate to find a queued buffer matching the requested slot.
-    while (itr == queue.end() && !queue_sequence.empty()) {
-        const u32 slot = queue_sequence.front();
-        itr = std::find_if(queue.begin(), queue.end(), [&slot](const Buffer& buffer) {
-            return buffer.status == Buffer::Status::Queued && buffer.slot == slot;
-        });
+    while (buffer_slot == buffers.size() && !queue_sequence.empty()) {
+        const auto slot = static_cast<std::size_t>(queue_sequence.front());
+        ASSERT(slot < buffers.size());
+        if (buffers[slot].status == Buffer::Status::Queued) {
+            ASSERT(buffers[slot].slot == slot);
+            buffer_slot = slot;
+        }
         queue_sequence.pop_front();
     }
-    if (itr == queue.end()) {
+    if (buffer_slot == buffers.size()) {
         return std::nullopt;
     }
-    itr->status = Buffer::Status::Acquired;
-    return *itr;
+    buffers[buffer_slot].status = Buffer::Status::Acquired;
+    return {{buffers[buffer_slot]}};
 }
 
 void BufferQueue::ReleaseBuffer(u32 slot) {
-    auto itr = std::find_if(queue.begin(), queue.end(),
-                            [&](const Buffer& buffer) { return buffer.slot == slot; });
-    ASSERT(itr != queue.end());
-    ASSERT(itr->status == Buffer::Status::Acquired);
-    itr->status = Buffer::Status::Free;
+    ASSERT(slot < buffers.size());
+    ASSERT(buffers[slot].status == Buffer::Status::Acquired);
+    ASSERT(buffers[slot].slot == slot);
+
+    buffers[slot].status = Buffer::Status::Free;
     free_buffers.push_back(slot);
 
     buffer_wait_event.writable->Signal();
 }
 
 void BufferQueue::Disconnect() {
-    queue.clear();
+    buffers.fill({});
     queue_sequence.clear();
-    id = 1;
-    layer_id = 1;
+    buffer_wait_event.writable->Signal();
 }
 
 u32 BufferQueue::Query(QueryType type) {
diff --git a/src/core/hle/service/nvflinger/buffer_queue.h b/src/core/hle/service/nvflinger/buffer_queue.h
index e7517c7e1..e610923cb 100644
--- a/src/core/hle/service/nvflinger/buffer_queue.h
+++ b/src/core/hle/service/nvflinger/buffer_queue.h
@@ -21,6 +21,7 @@ class KernelCore;
 
 namespace Service::NVFlinger {
 
+constexpr u32 buffer_slots = 0x40;
 struct IGBPBuffer {
     u32_le magic;
     u32_le width;
@@ -114,7 +115,7 @@ private:
     u64 layer_id;
 
     std::list<u32> free_buffers;
-    std::vector<Buffer> queue;
+    std::array<Buffer, buffer_slots> buffers;
     std::list<u32> queue_sequence;
     Kernel::EventPair buffer_wait_event;
 };
diff --git a/src/core/hle/service/time/time.cpp b/src/core/hle/service/time/time.cpp
index 7b7ac282d..abc753d5d 100644
--- a/src/core/hle/service/time/time.cpp
+++ b/src/core/hle/service/time/time.cpp
@@ -10,8 +10,8 @@
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/client_port.h"
 #include "core/hle/kernel/client_session.h"
+#include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/kernel.h"
-#include "core/hle/kernel/scheduler.h"
 #include "core/hle/service/time/interface.h"
 #include "core/hle/service/time/time.h"
 #include "core/hle/service/time/time_sharedmemory.h"
diff --git a/src/core/hle/service/vi/vi.cpp b/src/core/hle/service/vi/vi.cpp
index 5d8841ae8..45cfffe06 100644
--- a/src/core/hle/service/vi/vi.cpp
+++ b/src/core/hle/service/vi/vi.cpp
@@ -282,18 +282,24 @@ public:
     void DeserializeData() override {
         [[maybe_unused]] const std::u16string token = ReadInterfaceToken();
         data = Read<Data>();
-        buffer = Read<NVFlinger::IGBPBuffer>();
+        if (data.contains_object != 0) {
+            buffer_container = Read<BufferContainer>();
+        }
     }
 
     struct Data {
         u32_le slot;
-        INSERT_PADDING_WORDS(1);
+        u32_le contains_object;
+    };
+
+    struct BufferContainer {
         u32_le graphic_buffer_length;
         INSERT_PADDING_WORDS(1);
+        NVFlinger::IGBPBuffer buffer{};
     };
 
-    Data data;
-    NVFlinger::IGBPBuffer buffer;
+    Data data{};
+    BufferContainer buffer_container{};
 };
 
 class IGBPSetPreallocatedBufferResponseParcel : public Parcel {
@@ -547,7 +553,7 @@ private:
         case TransactionId::SetPreallocatedBuffer: {
             IGBPSetPreallocatedBufferRequestParcel request{ctx.ReadBuffer()};
 
-            buffer_queue.SetPreallocatedBuffer(request.data.slot, request.buffer);
+            buffer_queue.SetPreallocatedBuffer(request.data.slot, request.buffer_container.buffer);
 
             IGBPSetPreallocatedBufferResponseParcel response{};
             ctx.WriteBuffer(response.Serialize());
diff --git a/src/core/settings.cpp b/src/core/settings.cpp
index e9997a263..47d9ecf9a 100644
--- a/src/core/settings.cpp
+++ b/src/core/settings.cpp
@@ -72,8 +72,6 @@ void LogSettings() {
     log_setting("DataStorage_UseVirtualSd", values.use_virtual_sd);
     log_setting("DataStorage_NandDir", Common::FS::GetUserPath(Common::FS::UserPath::NANDDir));
     log_setting("DataStorage_SdmcDir", Common::FS::GetUserPath(Common::FS::UserPath::SDMCDir));
-    log_setting("Debugging_UseGdbstub", values.use_gdbstub);
-    log_setting("Debugging_GdbstubPort", values.gdbstub_port);
     log_setting("Debugging_ProgramArgs", values.program_args);
     log_setting("Services_BCATBackend", values.bcat_backend);
     log_setting("Services_BCATBoxcatLocal", values.bcat_boxcat_local);
diff --git a/src/tests/CMakeLists.txt b/src/tests/CMakeLists.txt
index 47ef30aa9..d80b0b688 100644
--- a/src/tests/CMakeLists.txt
+++ b/src/tests/CMakeLists.txt
@@ -2,7 +2,6 @@ add_executable(tests
     common/bit_field.cpp
     common/bit_utils.cpp
     common/fibers.cpp
-    common/multi_level_queue.cpp
     common/param_package.cpp
     common/ring_buffer.cpp
     core/arm/arm_test_common.cpp
diff --git a/src/tests/common/multi_level_queue.cpp b/src/tests/common/multi_level_queue.cpp
deleted file mode 100644
index cca7ec7da..000000000
--- a/src/tests/common/multi_level_queue.cpp
+++ /dev/null
@@ -1,55 +0,0 @@
-// Copyright 2019 Yuzu Emulator Project
-// Licensed under GPLv2 or any later version
-// Refer to the license.txt file included.
-
-#include <catch2/catch.hpp>
-#include <math.h>
-#include "common/common_types.h"
-#include "common/multi_level_queue.h"
-
-namespace Common {
-
-TEST_CASE("MultiLevelQueue", "[common]") {
-    std::array<f32, 8> values = {0.0, 5.0, 1.0, 9.0, 8.0, 2.0, 6.0, 7.0};
-    Common::MultiLevelQueue<f32, 64> mlq;
-    REQUIRE(mlq.empty());
-    mlq.add(values[2], 2);
-    mlq.add(values[7], 7);
-    mlq.add(values[3], 3);
-    mlq.add(values[4], 4);
-    mlq.add(values[0], 0);
-    mlq.add(values[5], 5);
-    mlq.add(values[6], 6);
-    mlq.add(values[1], 1);
-    u32 index = 0;
-    bool all_set = true;
-    for (auto& f : mlq) {
-        all_set &= (f == values[index]);
-        index++;
-    }
-    REQUIRE(all_set);
-    REQUIRE(!mlq.empty());
-    f32 v = 8.0;
-    mlq.add(v, 2);
-    v = -7.0;
-    mlq.add(v, 2, false);
-    REQUIRE(mlq.front(2) == -7.0);
-    mlq.yield(2);
-    REQUIRE(mlq.front(2) == values[2]);
-    REQUIRE(mlq.back(2) == -7.0);
-    REQUIRE(mlq.empty(8));
-    v = 10.0;
-    mlq.add(v, 8);
-    mlq.adjust(v, 8, 9);
-    REQUIRE(mlq.front(9) == v);
-    REQUIRE(mlq.empty(8));
-    REQUIRE(!mlq.empty(9));
-    mlq.adjust(values[0], 0, 9);
-    REQUIRE(mlq.highest_priority_set() == 1);
-    REQUIRE(mlq.lowest_priority_set() == 9);
-    mlq.remove(values[1], 1);
-    REQUIRE(mlq.highest_priority_set() == 2);
-    REQUIRE(mlq.empty(1));
-}
-
-} // namespace Common
diff --git a/src/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp
index 0ec5b861a..9fb254986 100644
--- a/src/yuzu/configuration/config.cpp
+++ b/src/yuzu/configuration/config.cpp
@@ -637,8 +637,6 @@ void Config::ReadDebuggingValues() {
     // Intentionally not using the QT default setting as this is intended to be changed in the ini
     Settings::values.record_frame_times =
         qt_config->value(QStringLiteral("record_frame_times"), false).toBool();
-    Settings::values.use_gdbstub = ReadSetting(QStringLiteral("use_gdbstub"), false).toBool();
-    Settings::values.gdbstub_port = ReadSetting(QStringLiteral("gdbstub_port"), 24689).toInt();
     Settings::values.program_args =
         ReadSetting(QStringLiteral("program_args"), QString{}).toString().toStdString();
     Settings::values.dump_exefs = ReadSetting(QStringLiteral("dump_exefs"), false).toBool();
@@ -1236,8 +1234,6 @@ void Config::SaveDebuggingValues() {
 
     // Intentionally not using the QT default setting as this is intended to be changed in the ini
     qt_config->setValue(QStringLiteral("record_frame_times"), Settings::values.record_frame_times);
-    WriteSetting(QStringLiteral("use_gdbstub"), Settings::values.use_gdbstub, false);
-    WriteSetting(QStringLiteral("gdbstub_port"), Settings::values.gdbstub_port, 24689);
     WriteSetting(QStringLiteral("program_args"),
                  QString::fromStdString(Settings::values.program_args), QString{});
     WriteSetting(QStringLiteral("dump_exefs"), Settings::values.dump_exefs, false);
diff --git a/src/yuzu/configuration/configure_debug.cpp b/src/yuzu/configuration/configure_debug.cpp
index 027099ab7..121873f95 100644
--- a/src/yuzu/configuration/configure_debug.cpp
+++ b/src/yuzu/configuration/configure_debug.cpp
@@ -28,9 +28,6 @@ ConfigureDebug::ConfigureDebug(QWidget* parent) : QWidget(parent), ui(new Ui::Co
 ConfigureDebug::~ConfigureDebug() = default;
 
 void ConfigureDebug::SetConfiguration() {
-    ui->toggle_gdbstub->setChecked(Settings::values.use_gdbstub);
-    ui->gdbport_spinbox->setEnabled(Settings::values.use_gdbstub);
-    ui->gdbport_spinbox->setValue(Settings::values.gdbstub_port);
     ui->toggle_console->setEnabled(!Core::System::GetInstance().IsPoweredOn());
     ui->toggle_console->setChecked(UISettings::values.show_console);
     ui->log_filter_edit->setText(QString::fromStdString(Settings::values.log_filter));
@@ -45,8 +42,6 @@ void ConfigureDebug::SetConfiguration() {
 }
 
 void ConfigureDebug::ApplyConfiguration() {
-    Settings::values.use_gdbstub = ui->toggle_gdbstub->isChecked();
-    Settings::values.gdbstub_port = ui->gdbport_spinbox->value();
     UISettings::values.show_console = ui->toggle_console->isChecked();
     Settings::values.log_filter = ui->log_filter_edit->text().toStdString();
     Settings::values.program_args = ui->homebrew_args_edit->text().toStdString();
diff --git a/src/yuzu/configuration/configure_debug.ui b/src/yuzu/configuration/configure_debug.ui
index 6f94fe304..9186aa732 100644
--- a/src/yuzu/configuration/configure_debug.ui
+++ b/src/yuzu/configuration/configure_debug.ui
@@ -7,7 +7,7 @@
     <x>0</x>
     <y>0</y>
     <width>400</width>
-    <height>467</height>
+    <height>486</height>
    </rect>
   </property>
   <property name="windowTitle">
@@ -15,57 +15,6 @@
   </property>
   <layout class="QVBoxLayout" name="verticalLayout_1">
    <item>
-    <layout class="QVBoxLayout" name="verticalLayout_2">
-     <item>
-      <widget class="QGroupBox" name="groupBox">
-       <property name="title">
-        <string>GDB</string>
-       </property>
-       <layout class="QVBoxLayout" name="verticalLayout_3">
-        <item>
-         <layout class="QHBoxLayout" name="horizontalLayout_1">
-          <item>
-           <widget class="QCheckBox" name="toggle_gdbstub">
-            <property name="text">
-             <string>Enable GDB Stub</string>
-            </property>
-           </widget>
-          </item>
-          <item>
-           <spacer name="horizontalSpacer">
-            <property name="orientation">
-             <enum>Qt::Horizontal</enum>
-            </property>
-            <property name="sizeHint" stdset="0">
-             <size>
-              <width>40</width>
-              <height>20</height>
-             </size>
-            </property>
-           </spacer>
-          </item>
-          <item>
-           <widget class="QLabel" name="label_1">
-            <property name="text">
-             <string>Port:</string>
-            </property>
-           </widget>
-          </item>
-          <item>
-           <widget class="QSpinBox" name="gdbport_spinbox">
-            <property name="maximum">
-             <number>65536</number>
-            </property>
-           </widget>
-          </item>
-         </layout>
-        </item>
-       </layout>
-      </widget>
-     </item>
-    </layout>
-   </item>
-   <item>
     <widget class="QGroupBox" name="groupBox_2">
      <property name="title">
       <string>Logging</string>
@@ -258,8 +207,6 @@
   </layout>
  </widget>
  <tabstops>
-  <tabstop>toggle_gdbstub</tabstop>
-  <tabstop>gdbport_spinbox</tabstop>
   <tabstop>log_filter_edit</tabstop>
   <tabstop>toggle_console</tabstop>
   <tabstop>open_log_button</tabstop>
@@ -269,22 +216,5 @@
   <tabstop>quest_flag</tabstop>
  </tabstops>
  <resources/>
- <connections>
-  <connection>
-   <sender>toggle_gdbstub</sender>
-   <signal>toggled(bool)</signal>
-   <receiver>gdbport_spinbox</receiver>
-   <slot>setEnabled(bool)</slot>
-   <hints>
-    <hint type="sourcelabel">
-     <x>84</x>
-     <y>157</y>
-    </hint>
-    <hint type="destinationlabel">
-     <x>342</x>
-     <y>158</y>
-    </hint>
-   </hints>
-  </connection>
- </connections>
+ <connections/>
 </ui>
diff --git a/src/yuzu/debugger/wait_tree.cpp b/src/yuzu/debugger/wait_tree.cpp
index a20824719..546a2cd4d 100644
--- a/src/yuzu/debugger/wait_tree.cpp
+++ b/src/yuzu/debugger/wait_tree.cpp
@@ -13,10 +13,10 @@
 #include "core/arm/arm_interface.h"
 #include "core/core.h"
 #include "core/hle/kernel/handle_table.h"
+#include "core/hle/kernel/k_scheduler.h"
 #include "core/hle/kernel/mutex.h"
 #include "core/hle/kernel/process.h"
 #include "core/hle/kernel/readable_event.h"
-#include "core/hle/kernel/scheduler.h"
 #include "core/hle/kernel/synchronization_object.h"
 #include "core/hle/kernel/thread.h"
 #include "core/memory.h"
@@ -101,7 +101,7 @@ std::vector<std::unique_ptr<WaitTreeThread>> WaitTreeItem::MakeThreadItemList()
     };
 
     const auto& system = Core::System::GetInstance();
-    add_threads(system.GlobalScheduler().GetThreadList());
+    add_threads(system.GlobalSchedulerContext().GetThreadList());
 
     return item_list;
 }
@@ -349,14 +349,14 @@ std::vector<std::unique_ptr<WaitTreeItem>> WaitTreeThread::GetChildren() const {
     list.push_back(std::make_unique<WaitTreeText>(tr("processor = %1").arg(processor)));
     list.push_back(
         std::make_unique<WaitTreeText>(tr("ideal core = %1").arg(thread.GetIdealCore())));
-    list.push_back(
-        std::make_unique<WaitTreeText>(tr("affinity mask = %1").arg(thread.GetAffinityMask())));
+    list.push_back(std::make_unique<WaitTreeText>(
+        tr("affinity mask = %1").arg(thread.GetAffinityMask().GetAffinityMask())));
     list.push_back(std::make_unique<WaitTreeText>(tr("thread id = %1").arg(thread.GetThreadID())));
     list.push_back(std::make_unique<WaitTreeText>(tr("priority = %1(current) / %2(normal)")
                                                       .arg(thread.GetPriority())
                                                       .arg(thread.GetNominalPriority())));
     list.push_back(std::make_unique<WaitTreeText>(
-        tr("last running ticks = %1").arg(thread.GetLastRunningTicks())));
+        tr("last running ticks = %1").arg(thread.GetLastScheduledTick())));
 
     const VAddr mutex_wait_address = thread.GetMutexWaitAddress();
     if (mutex_wait_address != 0) {
diff --git a/src/yuzu_cmd/config.cpp b/src/yuzu_cmd/config.cpp
index b70f71a08..38075c345 100644
--- a/src/yuzu_cmd/config.cpp
+++ b/src/yuzu_cmd/config.cpp
@@ -429,9 +429,6 @@ void Config::ReadValues() {
     // Debugging
     Settings::values.record_frame_times =
         sdl2_config->GetBoolean("Debugging", "record_frame_times", false);
-    Settings::values.use_gdbstub = sdl2_config->GetBoolean("Debugging", "use_gdbstub", false);
-    Settings::values.gdbstub_port =
-        static_cast<u16>(sdl2_config->GetInteger("Debugging", "gdbstub_port", 24689));
     Settings::values.program_args = sdl2_config->Get("Debugging", "program_args", "");
     Settings::values.dump_exefs = sdl2_config->GetBoolean("Debugging", "dump_exefs", false);
     Settings::values.dump_nso = sdl2_config->GetBoolean("Debugging", "dump_nso", false);
diff --git a/src/yuzu_cmd/default_ini.h b/src/yuzu_cmd/default_ini.h
index bcbbcd4ca..2d4b98d9a 100644
--- a/src/yuzu_cmd/default_ini.h
+++ b/src/yuzu_cmd/default_ini.h
@@ -318,9 +318,6 @@ log_filter = *:Trace
 [Debugging]
 # Record frame time data, can be found in the log directory. Boolean value
 record_frame_times =
-# Port for listening to GDB connections.
-use_gdbstub=false
-gdbstub_port=24689
 # Determines whether or not yuzu will dump the ExeFS of all games it attempts to load while loading them
 dump_exefs=false
 # Determines whether or not yuzu will dump all NSOs it attempts to load while loading them
diff --git a/src/yuzu_cmd/yuzu.cpp b/src/yuzu_cmd/yuzu.cpp
index c2efe1ee6..1ebc04af5 100644
--- a/src/yuzu_cmd/yuzu.cpp
+++ b/src/yuzu_cmd/yuzu.cpp
@@ -64,7 +64,6 @@ __declspec(dllexport) int AmdPowerXpressRequestHighPerformance = 1;
 static void PrintHelp(const char* argv0) {
     std::cout << "Usage: " << argv0
               << " [options] <filename>\n"
-                 "-g, --gdbport=NUMBER  Enable gdb stub on port NUMBER\n"
                  "-f, --fullscreen      Start in fullscreen mode\n"
                  "-h, --help            Display this help and exit\n"
                  "-v, --version         Output version information and exit\n"
@@ -96,8 +95,6 @@ int main(int argc, char** argv) {
     Config config;
 
     int option_index = 0;
-    bool use_gdbstub = Settings::values.use_gdbstub;
-    u32 gdb_port = static_cast<u32>(Settings::values.gdbstub_port);
 
     InitializeLogging();
 
@@ -116,26 +113,17 @@ int main(int argc, char** argv) {
     bool fullscreen = false;
 
     static struct option long_options[] = {
-        {"gdbport", required_argument, 0, 'g'}, {"fullscreen", no_argument, 0, 'f'},
-        {"help", no_argument, 0, 'h'},          {"version", no_argument, 0, 'v'},
-        {"program", optional_argument, 0, 'p'}, {0, 0, 0, 0},
+        {"fullscreen", no_argument, 0, 'f'},
+        {"help", no_argument, 0, 'h'},
+        {"version", no_argument, 0, 'v'},
+        {"program", optional_argument, 0, 'p'},
+        {0, 0, 0, 0},
     };
 
     while (optind < argc) {
         int arg = getopt_long(argc, argv, "g:fhvp::", long_options, &option_index);
         if (arg != -1) {
             switch (static_cast<char>(arg)) {
-            case 'g':
-                errno = 0;
-                gdb_port = strtoul(optarg, &endarg, 0);
-                use_gdbstub = true;
-                if (endarg == optarg)
-                    errno = EINVAL;
-                if (errno != 0) {
-                    perror("--gdbport");
-                    exit(1);
-                }
-                break;
             case 'f':
                 fullscreen = true;
                 LOG_INFO(Frontend, "Starting in fullscreen mode...");
@@ -177,8 +165,6 @@ int main(int argc, char** argv) {
     InputCommon::InputSubsystem input_subsystem;
 
     // Apply the command line arguments
-    Settings::values.gdbstub_port = gdb_port;
-    Settings::values.use_gdbstub = use_gdbstub;
     Settings::Apply(system);
 
     std::unique_ptr<EmuWindow_SDL2> emu_window;
diff --git a/src/yuzu_tester/config.cpp b/src/yuzu_tester/config.cpp
index b6cdc7c1c..91684e96e 100644
--- a/src/yuzu_tester/config.cpp
+++ b/src/yuzu_tester/config.cpp
@@ -158,7 +158,6 @@ void Config::ReadValues() {
     Settings::values.use_dev_keys = sdl2_config->GetBoolean("Miscellaneous", "use_dev_keys", false);
 
     // Debugging
-    Settings::values.use_gdbstub = false;
     Settings::values.program_args = "";
     Settings::values.dump_exefs = sdl2_config->GetBoolean("Debugging", "dump_exefs", false);
     Settings::values.dump_nso = sdl2_config->GetBoolean("Debugging", "dump_nso", false);
diff --git a/src/yuzu_tester/yuzu.cpp b/src/yuzu_tester/yuzu.cpp
index 50bd7ae41..6435ffabb 100644
--- a/src/yuzu_tester/yuzu.cpp
+++ b/src/yuzu_tester/yuzu.cpp
@@ -162,7 +162,6 @@ int main(int argc, char** argv) {
 
     Core::System& system{Core::System::GetInstance()};
 
-    Settings::values.use_gdbstub = false;
     Settings::Apply(system);
 
     const auto emu_window{std::make_unique<EmuWindow_SDL2_Hide>()};