18 files changed, 876 insertions, 100 deletions

diff --git a/src/common/intrusive_list.h b/src/common/intrusive_list.h
new file mode 100644
index 000000000..d330dc1c2
--- /dev/null
+++ b/src/common/intrusive_list.h
@@ -0,0 +1,631 @@
+// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include "common/common_funcs.h"
+#include "common/parent_of_member.h"
+
+namespace Common {
+
+// Forward declare implementation class for Node.
+namespace impl {
+
+class IntrusiveListImpl;
+
+}
+
+class IntrusiveListNode {
+    YUZU_NON_COPYABLE(IntrusiveListNode);
+
+private:
+    friend class impl::IntrusiveListImpl;
+
+    IntrusiveListNode* m_prev;
+    IntrusiveListNode* m_next;
+
+public:
+    constexpr IntrusiveListNode() : m_prev(this), m_next(this) {}
+
+    constexpr bool IsLinked() const {
+        return m_next != this;
+    }
+
+private:
+    constexpr void LinkPrev(IntrusiveListNode* node) {
+        // We can't link an already linked node.
+        ASSERT(!node->IsLinked());
+        this->SplicePrev(node, node);
+    }
+
+    constexpr void SplicePrev(IntrusiveListNode* first, IntrusiveListNode* last) {
+        // Splice a range into the list.
+        auto last_prev = last->m_prev;
+        first->m_prev = m_prev;
+        last_prev->m_next = this;
+        m_prev->m_next = first;
+        m_prev = last_prev;
+    }
+
+    constexpr void LinkNext(IntrusiveListNode* node) {
+        // We can't link an already linked node.
+        ASSERT(!node->IsLinked());
+        return this->SpliceNext(node, node);
+    }
+
+    constexpr void SpliceNext(IntrusiveListNode* first, IntrusiveListNode* last) {
+        // Splice a range into the list.
+        auto last_prev = last->m_prev;
+        first->m_prev = this;
+        last_prev->m_next = m_next;
+        m_next->m_prev = last_prev;
+        m_next = first;
+    }
+
+    constexpr void Unlink() {
+        this->Unlink(m_next);
+    }
+
+    constexpr void Unlink(IntrusiveListNode* last) {
+        // Unlink a node from a next node.
+        auto last_prev = last->m_prev;
+        m_prev->m_next = last;
+        last->m_prev = m_prev;
+        last_prev->m_next = this;
+        m_prev = last_prev;
+    }
+
+    constexpr IntrusiveListNode* GetPrev() {
+        return m_prev;
+    }
+
+    constexpr const IntrusiveListNode* GetPrev() const {
+        return m_prev;
+    }
+
+    constexpr IntrusiveListNode* GetNext() {
+        return m_next;
+    }
+
+    constexpr const IntrusiveListNode* GetNext() const {
+        return m_next;
+    }
+};
+// DEPRECATED: static_assert(std::is_literal_type<IntrusiveListNode>::value);
+
+namespace impl {
+
+class IntrusiveListImpl {
+    YUZU_NON_COPYABLE(IntrusiveListImpl);
+
+private:
+    IntrusiveListNode m_root_node;
+
+public:
+    template <bool Const>
+    class Iterator;
+
+    using value_type = IntrusiveListNode;
+    using size_type = size_t;
+    using difference_type = ptrdiff_t;
+    using pointer = value_type*;
+    using const_pointer = const value_type*;
+    using reference = value_type&;
+    using const_reference = const value_type&;
+    using iterator = Iterator<false>;
+    using const_iterator = Iterator<true>;
+    using reverse_iterator = std::reverse_iterator<iterator>;
+    using const_reverse_iterator = std::reverse_iterator<const_iterator>;
+
+    template <bool Const>
+    class Iterator {
+    public:
+        using iterator_category = std::bidirectional_iterator_tag;
+        using value_type = typename IntrusiveListImpl::value_type;
+        using difference_type = typename IntrusiveListImpl::difference_type;
+        using pointer =
+            std::conditional_t<Const, IntrusiveListImpl::const_pointer, IntrusiveListImpl::pointer>;
+        using reference = std::conditional_t<Const, IntrusiveListImpl::const_reference,
+                                             IntrusiveListImpl::reference>;
+
+    private:
+        pointer m_node;
+
+    public:
+        constexpr explicit Iterator(pointer n) : m_node(n) {}
+
+        constexpr bool operator==(const Iterator& rhs) const {
+            return m_node == rhs.m_node;
+        }
+
+        constexpr pointer operator->() const {
+            return m_node;
+        }
+
+        constexpr reference operator*() const {
+            return *m_node;
+        }
+
+        constexpr Iterator& operator++() {
+            m_node = m_node->m_next;
+            return *this;
+        }
+
+        constexpr Iterator& operator--() {
+            m_node = m_node->m_prev;
+            return *this;
+        }
+
+        constexpr Iterator operator++(int) {
+            const Iterator it{*this};
+            ++(*this);
+            return it;
+        }
+
+        constexpr Iterator operator--(int) {
+            const Iterator it{*this};
+            --(*this);
+            return it;
+        }
+
+        constexpr operator Iterator<true>() const {
+            return Iterator<true>(m_node);
+        }
+
+        constexpr Iterator<false> GetNonConstIterator() const {
+            return Iterator<false>(const_cast<IntrusiveListImpl::pointer>(m_node));
+        }
+    };
+
+public:
+    constexpr IntrusiveListImpl() : m_root_node() {}
+
+    // Iterator accessors.
+    constexpr iterator begin() {
+        return iterator(m_root_node.GetNext());
+    }
+
+    constexpr const_iterator begin() const {
+        return const_iterator(m_root_node.GetNext());
+    }
+
+    constexpr iterator end() {
+        return iterator(std::addressof(m_root_node));
+    }
+
+    constexpr const_iterator end() const {
+        return const_iterator(std::addressof(m_root_node));
+    }
+
+    constexpr iterator iterator_to(reference v) {
+        // Only allow iterator_to for values in lists.
+        ASSERT(v.IsLinked());
+        return iterator(std::addressof(v));
+    }
+
+    constexpr const_iterator iterator_to(const_reference v) const {
+        // Only allow iterator_to for values in lists.
+        ASSERT(v.IsLinked());
+        return const_iterator(std::addressof(v));
+    }
+
+    // Content management.
+    constexpr bool empty() const {
+        return !m_root_node.IsLinked();
+    }
+
+    constexpr size_type size() const {
+        return static_cast<size_type>(std::distance(this->begin(), this->end()));
+    }
+
+    constexpr reference back() {
+        return *m_root_node.GetPrev();
+    }
+
+    constexpr const_reference back() const {
+        return *m_root_node.GetPrev();
+    }
+
+    constexpr reference front() {
+        return *m_root_node.GetNext();
+    }
+
+    constexpr const_reference front() const {
+        return *m_root_node.GetNext();
+    }
+
+    constexpr void push_back(reference node) {
+        m_root_node.LinkPrev(std::addressof(node));
+    }
+
+    constexpr void push_front(reference node) {
+        m_root_node.LinkNext(std::addressof(node));
+    }
+
+    constexpr void pop_back() {
+        m_root_node.GetPrev()->Unlink();
+    }
+
+    constexpr void pop_front() {
+        m_root_node.GetNext()->Unlink();
+    }
+
+    constexpr iterator insert(const_iterator pos, reference node) {
+        pos.GetNonConstIterator()->LinkPrev(std::addressof(node));
+        return iterator(std::addressof(node));
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveListImpl& o) {
+        splice_impl(pos, o.begin(), o.end());
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveListImpl& o, const_iterator first) {
+        const_iterator last(first);
+        std::advance(last, 1);
+        splice_impl(pos, first, last);
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveListImpl& o, const_iterator first,
+                          const_iterator last) {
+        splice_impl(pos, first, last);
+    }
+
+    constexpr iterator erase(const_iterator pos) {
+        if (pos == this->end()) {
+            return this->end();
+        }
+        iterator it(pos.GetNonConstIterator());
+        (it++)->Unlink();
+        return it;
+    }
+
+    constexpr void clear() {
+        while (!this->empty()) {
+            this->pop_front();
+        }
+    }
+
+private:
+    constexpr void splice_impl(const_iterator _pos, const_iterator _first, const_iterator _last) {
+        if (_first == _last) {
+            return;
+        }
+        iterator pos(_pos.GetNonConstIterator());
+        iterator first(_first.GetNonConstIterator());
+        iterator last(_last.GetNonConstIterator());
+        first->Unlink(std::addressof(*last));
+        pos->SplicePrev(std::addressof(*first), std::addressof(*first));
+    }
+};
+
+} // namespace impl
+
+template <class T, class Traits>
+class IntrusiveList {
+    YUZU_NON_COPYABLE(IntrusiveList);
+
+private:
+    impl::IntrusiveListImpl m_impl;
+
+public:
+    template <bool Const>
+    class Iterator;
+
+    using value_type = T;
+    using size_type = size_t;
+    using difference_type = ptrdiff_t;
+    using pointer = value_type*;
+    using const_pointer = const value_type*;
+    using reference = value_type&;
+    using const_reference = const value_type&;
+    using iterator = Iterator<false>;
+    using const_iterator = Iterator<true>;
+    using reverse_iterator = std::reverse_iterator<iterator>;
+    using const_reverse_iterator = std::reverse_iterator<const_iterator>;
+
+    template <bool Const>
+    class Iterator {
+    public:
+        friend class Common::IntrusiveList<T, Traits>;
+
+        using ImplIterator =
+            std::conditional_t<Const, Common::impl::IntrusiveListImpl::const_iterator,
+                               Common::impl::IntrusiveListImpl::iterator>;
+
+        using iterator_category = std::bidirectional_iterator_tag;
+        using value_type = typename IntrusiveList::value_type;
+        using difference_type = typename IntrusiveList::difference_type;
+        using pointer =
+            std::conditional_t<Const, IntrusiveList::const_pointer, IntrusiveList::pointer>;
+        using reference =
+            std::conditional_t<Const, IntrusiveList::const_reference, IntrusiveList::reference>;
+
+    private:
+        ImplIterator m_iterator;
+
+    private:
+        constexpr explicit Iterator(ImplIterator it) : m_iterator(it) {}
+
+        constexpr ImplIterator GetImplIterator() const {
+            return m_iterator;
+        }
+
+    public:
+        constexpr bool operator==(const Iterator& rhs) const {
+            return m_iterator == rhs.m_iterator;
+        }
+
+        constexpr pointer operator->() const {
+            return std::addressof(Traits::GetParent(*m_iterator));
+        }
+
+        constexpr reference operator*() const {
+            return Traits::GetParent(*m_iterator);
+        }
+
+        constexpr Iterator& operator++() {
+            ++m_iterator;
+            return *this;
+        }
+
+        constexpr Iterator& operator--() {
+            --m_iterator;
+            return *this;
+        }
+
+        constexpr Iterator operator++(int) {
+            const Iterator it{*this};
+            ++m_iterator;
+            return it;
+        }
+
+        constexpr Iterator operator--(int) {
+            const Iterator it{*this};
+            --m_iterator;
+            return it;
+        }
+
+        constexpr operator Iterator<true>() const {
+            return Iterator<true>(m_iterator);
+        }
+    };
+
+private:
+    static constexpr IntrusiveListNode& GetNode(reference ref) {
+        return Traits::GetNode(ref);
+    }
+
+    static constexpr IntrusiveListNode const& GetNode(const_reference ref) {
+        return Traits::GetNode(ref);
+    }
+
+    static constexpr reference GetParent(IntrusiveListNode& node) {
+        return Traits::GetParent(node);
+    }
+
+    static constexpr const_reference GetParent(IntrusiveListNode const& node) {
+        return Traits::GetParent(node);
+    }
+
+public:
+    constexpr IntrusiveList() : m_impl() {}
+
+    // Iterator accessors.
+    constexpr iterator begin() {
+        return iterator(m_impl.begin());
+    }
+
+    constexpr const_iterator begin() const {
+        return const_iterator(m_impl.begin());
+    }
+
+    constexpr iterator end() {
+        return iterator(m_impl.end());
+    }
+
+    constexpr const_iterator end() const {
+        return const_iterator(m_impl.end());
+    }
+
+    constexpr const_iterator cbegin() const {
+        return this->begin();
+    }
+
+    constexpr const_iterator cend() const {
+        return this->end();
+    }
+
+    constexpr reverse_iterator rbegin() {
+        return reverse_iterator(this->end());
+    }
+
+    constexpr const_reverse_iterator rbegin() const {
+        return const_reverse_iterator(this->end());
+    }
+
+    constexpr reverse_iterator rend() {
+        return reverse_iterator(this->begin());
+    }
+
+    constexpr const_reverse_iterator rend() const {
+        return const_reverse_iterator(this->begin());
+    }
+
+    constexpr const_reverse_iterator crbegin() const {
+        return this->rbegin();
+    }
+
+    constexpr const_reverse_iterator crend() const {
+        return this->rend();
+    }
+
+    constexpr iterator iterator_to(reference v) {
+        return iterator(m_impl.iterator_to(GetNode(v)));
+    }
+
+    constexpr const_iterator iterator_to(const_reference v) const {
+        return const_iterator(m_impl.iterator_to(GetNode(v)));
+    }
+
+    // Content management.
+    constexpr bool empty() const {
+        return m_impl.empty();
+    }
+
+    constexpr size_type size() const {
+        return m_impl.size();
+    }
+
+    constexpr reference back() {
+        return GetParent(m_impl.back());
+    }
+
+    constexpr const_reference back() const {
+        return GetParent(m_impl.back());
+    }
+
+    constexpr reference front() {
+        return GetParent(m_impl.front());
+    }
+
+    constexpr const_reference front() const {
+        return GetParent(m_impl.front());
+    }
+
+    constexpr void push_back(reference ref) {
+        m_impl.push_back(GetNode(ref));
+    }
+
+    constexpr void push_front(reference ref) {
+        m_impl.push_front(GetNode(ref));
+    }
+
+    constexpr void pop_back() {
+        m_impl.pop_back();
+    }
+
+    constexpr void pop_front() {
+        m_impl.pop_front();
+    }
+
+    constexpr iterator insert(const_iterator pos, reference ref) {
+        return iterator(m_impl.insert(pos.GetImplIterator(), GetNode(ref)));
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveList& o) {
+        m_impl.splice(pos.GetImplIterator(), o.m_impl);
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveList& o, const_iterator first) {
+        m_impl.splice(pos.GetImplIterator(), o.m_impl, first.GetImplIterator());
+    }
+
+    constexpr void splice(const_iterator pos, IntrusiveList& o, const_iterator first,
+                          const_iterator last) {
+        m_impl.splice(pos.GetImplIterator(), o.m_impl, first.GetImplIterator(),
+                      last.GetImplIterator());
+    }
+
+    constexpr iterator erase(const_iterator pos) {
+        return iterator(m_impl.erase(pos.GetImplIterator()));
+    }
+
+    constexpr void clear() {
+        m_impl.clear();
+    }
+};
+
+template <auto T, class Derived = Common::impl::GetParentType<T>>
+class IntrusiveListMemberTraits;
+
+template <class Parent, IntrusiveListNode Parent::*Member, class Derived>
+class IntrusiveListMemberTraits<Member, Derived> {
+public:
+    using ListType = IntrusiveList<Derived, IntrusiveListMemberTraits>;
+
+private:
+    friend class IntrusiveList<Derived, IntrusiveListMemberTraits>;
+
+    static constexpr IntrusiveListNode& GetNode(Derived& parent) {
+        return parent.*Member;
+    }
+
+    static constexpr IntrusiveListNode const& GetNode(Derived const& parent) {
+        return parent.*Member;
+    }
+
+    static Derived& GetParent(IntrusiveListNode& node) {
+        return Common::GetParentReference<Member, Derived>(std::addressof(node));
+    }
+
+    static Derived const& GetParent(IntrusiveListNode const& node) {
+        return Common::GetParentReference<Member, Derived>(std::addressof(node));
+    }
+};
+
+template <auto T, class Derived = Common::impl::GetParentType<T>>
+class IntrusiveListMemberTraitsByNonConstexprOffsetOf;
+
+template <class Parent, IntrusiveListNode Parent::*Member, class Derived>
+class IntrusiveListMemberTraitsByNonConstexprOffsetOf<Member, Derived> {
+public:
+    using ListType = IntrusiveList<Derived, IntrusiveListMemberTraitsByNonConstexprOffsetOf>;
+
+private:
+    friend class IntrusiveList<Derived, IntrusiveListMemberTraitsByNonConstexprOffsetOf>;
+
+    static constexpr IntrusiveListNode& GetNode(Derived& parent) {
+        return parent.*Member;
+    }
+
+    static constexpr IntrusiveListNode const& GetNode(Derived const& parent) {
+        return parent.*Member;
+    }
+
+    static Derived& GetParent(IntrusiveListNode& node) {
+        return *reinterpret_cast<Derived*>(reinterpret_cast<char*>(std::addressof(node)) -
+                                           GetOffset());
+    }
+
+    static Derived const& GetParent(IntrusiveListNode const& node) {
+        return *reinterpret_cast<const Derived*>(
+            reinterpret_cast<const char*>(std::addressof(node)) - GetOffset());
+    }
+
+    static uintptr_t GetOffset() {
+        return reinterpret_cast<uintptr_t>(std::addressof(reinterpret_cast<Derived*>(0)->*Member));
+    }
+};
+
+template <class Derived>
+class IntrusiveListBaseNode : public IntrusiveListNode {};
+
+template <class Derived>
+class IntrusiveListBaseTraits {
+public:
+    using ListType = IntrusiveList<Derived, IntrusiveListBaseTraits>;
+
+private:
+    friend class IntrusiveList<Derived, IntrusiveListBaseTraits>;
+
+    static constexpr IntrusiveListNode& GetNode(Derived& parent) {
+        return static_cast<IntrusiveListNode&>(
+            static_cast<IntrusiveListBaseNode<Derived>&>(parent));
+    }
+
+    static constexpr IntrusiveListNode const& GetNode(Derived const& parent) {
+        return static_cast<const IntrusiveListNode&>(
+            static_cast<const IntrusiveListBaseNode<Derived>&>(parent));
+    }
+
+    static constexpr Derived& GetParent(IntrusiveListNode& node) {
+        return static_cast<Derived&>(static_cast<IntrusiveListBaseNode<Derived>&>(node));
+    }
+
+    static constexpr Derived const& GetParent(IntrusiveListNode const& node) {
+        return static_cast<const Derived&>(
+            static_cast<const IntrusiveListBaseNode<Derived>&>(node));
+    }
+};
+
+} // namespace Common
diff --git a/src/core/hle/kernel/k_event_info.h b/src/core/hle/kernel/k_event_info.h
index 25b3ff594..eacfa5dc6 100644
--- a/src/core/hle/kernel/k_event_info.h
+++ b/src/core/hle/kernel/k_event_info.h
@@ -5,14 +5,15 @@
 
 #include <array>
 
-#include <boost/intrusive/list.hpp>
+#include "common/intrusive_list.h"
 
 #include "core/hle/kernel/slab_helpers.h"
 #include "core/hle/kernel/svc_types.h"
 
 namespace Kernel {
 
-class KEventInfo : public KSlabAllocated<KEventInfo>, public boost::intrusive::list_base_hook<> {
+class KEventInfo : public KSlabAllocated<KEventInfo>,
+                   public Common::IntrusiveListBaseNode<KEventInfo> {
 public:
     struct InfoCreateThread {
         u32 thread_id{};
diff --git a/src/core/hle/kernel/k_object_name.h b/src/core/hle/kernel/k_object_name.h
index 2d97fc777..a8876fe37 100644
--- a/src/core/hle/kernel/k_object_name.h
+++ b/src/core/hle/kernel/k_object_name.h
@@ -5,7 +5,8 @@
 
 #include <array>
 #include <memory>
-#include <boost/intrusive/list.hpp>
+
+#include "common/intrusive_list.h"
 
 #include "core/hle/kernel/k_light_lock.h"
 #include "core/hle/kernel/slab_helpers.h"
@@ -15,13 +16,14 @@ namespace Kernel {
 
 class KObjectNameGlobalData;
 
-class KObjectName : public KSlabAllocated<KObjectName>, public boost::intrusive::list_base_hook<> {
+class KObjectName : public KSlabAllocated<KObjectName>,
+                    public Common::IntrusiveListBaseNode<KObjectName> {
 public:
     explicit KObjectName(KernelCore&) {}
     virtual ~KObjectName() = default;
 
     static constexpr size_t NameLengthMax = 12;
-    using List = boost::intrusive::list<KObjectName>;
+    using List = Common::IntrusiveListBaseTraits<KObjectName>::ListType;
 
     static Result NewFromName(KernelCore& kernel, KAutoObject* obj, const char* name);
     static Result Delete(KernelCore& kernel, KAutoObject* obj, const char* name);
diff --git a/src/core/hle/kernel/k_server_port.h b/src/core/hle/kernel/k_server_port.h
index 21c040e62..625280290 100644
--- a/src/core/hle/kernel/k_server_port.h
+++ b/src/core/hle/kernel/k_server_port.h
@@ -7,7 +7,7 @@
 #include <string>
 #include <utility>
 
-#include <boost/intrusive/list.hpp>
+#include "common/intrusive_list.h"
 
 #include "core/hle/kernel/k_server_session.h"
 #include "core/hle/kernel/k_synchronization_object.h"
@@ -42,7 +42,7 @@ public:
     bool IsSignaled() const override;
 
 private:
-    using SessionList = boost::intrusive::list<KServerSession>;
+    using SessionList = Common::IntrusiveListBaseTraits<KServerSession>::ListType;
 
     void CleanupSessions();
 
diff --git a/src/core/hle/kernel/k_server_session.h b/src/core/hle/kernel/k_server_session.h
index 5ee02f556..403891919 100644
--- a/src/core/hle/kernel/k_server_session.h
+++ b/src/core/hle/kernel/k_server_session.h
@@ -8,7 +8,7 @@
 #include <string>
 #include <utility>
 
-#include <boost/intrusive/list.hpp>
+#include "common/intrusive_list.h"
 
 #include "core/hle/kernel/k_light_lock.h"
 #include "core/hle/kernel/k_session_request.h"
@@ -27,7 +27,7 @@ class KSession;
 class KThread;
 
 class KServerSession final : public KSynchronizationObject,
-                             public boost::intrusive::list_base_hook<> {
+                             public Common::IntrusiveListBaseNode<KServerSession> {
     KERNEL_AUTOOBJECT_TRAITS(KServerSession, KSynchronizationObject);
 
     friend class ServiceThread;
@@ -67,7 +67,8 @@ private:
     KSession* m_parent{};
 
     /// List of threads which are pending a reply.
-    boost::intrusive::list<KSessionRequest> m_request_list{};
+    using RequestList = Common::IntrusiveListBaseTraits<KSessionRequest>::ListType;
+    RequestList m_request_list{};
     KSessionRequest* m_current_request{};
 
     KLightLock m_lock;
diff --git a/src/core/hle/kernel/k_session_request.h b/src/core/hle/kernel/k_session_request.h
index b5f04907b..283669e0a 100644
--- a/src/core/hle/kernel/k_session_request.h
+++ b/src/core/hle/kernel/k_session_request.h
@@ -5,6 +5,8 @@
 
 #include <array>
 
+#include "common/intrusive_list.h"
+
 #include "core/hle/kernel/k_auto_object.h"
 #include "core/hle/kernel/k_event.h"
 #include "core/hle/kernel/k_memory_block.h"
@@ -16,7 +18,7 @@ namespace Kernel {
 
 class KSessionRequest final : public KSlabAllocated<KSessionRequest>,
                               public KAutoObject,
-                              public boost::intrusive::list_base_hook<> {
+                              public Common::IntrusiveListBaseNode<KSessionRequest> {
     KERNEL_AUTOOBJECT_TRAITS(KSessionRequest, KAutoObject);
 
 public:
diff --git a/src/core/hle/kernel/k_shared_memory_info.h b/src/core/hle/kernel/k_shared_memory_info.h
index 75b73ba39..2d8ff20d6 100644
--- a/src/core/hle/kernel/k_shared_memory_info.h
+++ b/src/core/hle/kernel/k_shared_memory_info.h
@@ -3,7 +3,7 @@
 
 #pragma once
 
-#include <boost/intrusive/list.hpp>
+#include "common/intrusive_list.h"
 
 #include "core/hle/kernel/slab_helpers.h"
 
@@ -12,7 +12,7 @@ namespace Kernel {
 class KSharedMemory;
 
 class KSharedMemoryInfo final : public KSlabAllocated<KSharedMemoryInfo>,
-                                public boost::intrusive::list_base_hook<> {
+                                public Common::IntrusiveListBaseNode<KSharedMemoryInfo> {
 
 public:
     explicit KSharedMemoryInfo(KernelCore&) {}
diff --git a/src/core/hle/kernel/k_thread.h b/src/core/hle/kernel/k_thread.h
index 9c1a41128..f9814ac8f 100644
--- a/src/core/hle/kernel/k_thread.h
+++ b/src/core/hle/kernel/k_thread.h
@@ -12,7 +12,7 @@
 #include <utility>
 #include <vector>
 
-#include <boost/intrusive/list.hpp>
+#include "common/intrusive_list.h"
 
 #include "common/intrusive_red_black_tree.h"
 #include "common/spin_lock.h"
@@ -119,7 +119,7 @@ s32 GetCurrentCoreId(KernelCore& kernel);
 Core::Memory::Memory& GetCurrentMemory(KernelCore& kernel);
 
 class KThread final : public KAutoObjectWithSlabHeapAndContainer<KThread, KWorkerTask>,
-                      public boost::intrusive::list_base_hook<>,
+                      public Common::IntrusiveListBaseNode<KThread>,
                       public KTimerTask {
     KERNEL_AUTOOBJECT_TRAITS(KThread, KSynchronizationObject);
 
@@ -138,7 +138,7 @@ public:
 public:
     using ThreadContext32 = Core::ARM_Interface::ThreadContext32;
     using ThreadContext64 = Core::ARM_Interface::ThreadContext64;
-    using WaiterList = boost::intrusive::list<KThread>;
+    using WaiterList = Common::IntrusiveListBaseTraits<KThread>::ListType;
 
     /**
      * Gets the thread's current priority
@@ -750,8 +750,9 @@ private:
         ConditionVariableThreadTreeTraits::TreeType<LockWithPriorityInheritanceComparator>;
 
 public:
-    class LockWithPriorityInheritanceInfo : public KSlabAllocated<LockWithPriorityInheritanceInfo>,
-                                            public boost::intrusive::list_base_hook<> {
+    class LockWithPriorityInheritanceInfo
+        : public KSlabAllocated<LockWithPriorityInheritanceInfo>,
+          public Common::IntrusiveListBaseNode<LockWithPriorityInheritanceInfo> {
     public:
         explicit LockWithPriorityInheritanceInfo(KernelCore&) {}
 
@@ -839,7 +840,7 @@ public:
 
 private:
     using LockWithPriorityInheritanceInfoList =
-        boost::intrusive::list<LockWithPriorityInheritanceInfo>;
+        Common::IntrusiveListBaseTraits<LockWithPriorityInheritanceInfo>::ListType;
 
     ConditionVariableThreadTree* m_condvar_tree{};
     u64 m_condvar_key{};
diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp
index 0932fadc2..2f986097f 100644
--- a/src/video_core/engines/maxwell_3d.cpp
+++ b/src/video_core/engines/maxwell_3d.cpp
@@ -223,6 +223,9 @@ void Maxwell3D::ProcessMacro(u32 method, const u32* base_start, u32 amount, bool
 }
 
 void Maxwell3D::RefreshParametersImpl() {
+    if (!Settings::IsGPULevelHigh()) {
+        return;
+    }
     size_t current_index = 0;
     for (auto& segment : macro_segments) {
         if (segment.first == 0) {
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 90e35e307..4993d4709 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -1287,8 +1287,7 @@ bool AccelerateDMA::DmaBufferImageCopy(const Tegra::DMA::ImageCopy& copy_info,
     }
     const u32 buffer_size = static_cast<u32>(buffer_operand.pitch * buffer_operand.height);
     static constexpr auto sync_info = VideoCommon::ObtainBufferSynchronize::FullSynchronize;
-    const auto post_op = IS_IMAGE_UPLOAD ? VideoCommon::ObtainBufferOperation::DoNothing
-                                         : VideoCommon::ObtainBufferOperation::MarkAsWritten;
+    const auto post_op = VideoCommon::ObtainBufferOperation::DoNothing;
     const auto [buffer, offset] =
         buffer_cache.ObtainBuffer(buffer_operand.address, buffer_size, sync_info, post_op);
 
@@ -1299,7 +1298,8 @@ bool AccelerateDMA::DmaBufferImageCopy(const Tegra::DMA::ImageCopy& copy_info,
     if constexpr (IS_IMAGE_UPLOAD) {
         image->UploadMemory(buffer->Handle(), offset, copy_span);
     } else {
-        image->DownloadMemory(buffer->Handle(), offset, copy_span);
+        texture_cache.DownloadImageIntoBuffer(image, buffer->Handle(), offset, copy_span,
+                                              buffer_operand.address, buffer_size);
     }
     return true;
 }
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp
index 0b9c4a904..032a8ebc5 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp
@@ -803,30 +803,40 @@ void Image::UploadMemory(const ImageBufferMap& map,
 
 void Image::DownloadMemory(GLuint buffer_handle, size_t buffer_offset,
                            std::span<const VideoCommon::BufferImageCopy> copies) {
+    std::array buffer_handles{buffer_handle};
+    std::array buffer_offsets{buffer_offset};
+    DownloadMemory(buffer_handles, buffer_offsets, copies);
+}
+
+void Image::DownloadMemory(std::span<GLuint> buffer_handles, std::span<size_t> buffer_offsets,
+                           std::span<const VideoCommon::BufferImageCopy> copies) {
     const bool is_rescaled = True(flags & ImageFlagBits::Rescaled);
     if (is_rescaled) {
         ScaleDown();
     }
     glMemoryBarrier(GL_PIXEL_BUFFER_BARRIER_BIT); // TODO: Move this to its own API
-    glBindBuffer(GL_PIXEL_PACK_BUFFER, buffer_handle);
-    glPixelStorei(GL_PACK_ALIGNMENT, 1);
+    for (size_t i = 0; i < buffer_handles.size(); i++) {
+        auto& buffer_handle = buffer_handles[i];
+        glBindBuffer(GL_PIXEL_PACK_BUFFER, buffer_handle);
+        glPixelStorei(GL_PACK_ALIGNMENT, 1);
 
-    u32 current_row_length = std::numeric_limits<u32>::max();
-    u32 current_image_height = std::numeric_limits<u32>::max();
+        u32 current_row_length = std::numeric_limits<u32>::max();
+        u32 current_image_height = std::numeric_limits<u32>::max();
 
-    for (const VideoCommon::BufferImageCopy& copy : copies) {
-        if (copy.image_subresource.base_level >= gl_num_levels) {
-            continue;
-        }
-        if (current_row_length != copy.buffer_row_length) {
-            current_row_length = copy.buffer_row_length;
-            glPixelStorei(GL_PACK_ROW_LENGTH, current_row_length);
-        }
-        if (current_image_height != copy.buffer_image_height) {
-            current_image_height = copy.buffer_image_height;
-            glPixelStorei(GL_PACK_IMAGE_HEIGHT, current_image_height);
+        for (const VideoCommon::BufferImageCopy& copy : copies) {
+            if (copy.image_subresource.base_level >= gl_num_levels) {
+                continue;
+            }
+            if (current_row_length != copy.buffer_row_length) {
+                current_row_length = copy.buffer_row_length;
+                glPixelStorei(GL_PACK_ROW_LENGTH, current_row_length);
+            }
+            if (current_image_height != copy.buffer_image_height) {
+                current_image_height = copy.buffer_image_height;
+                glPixelStorei(GL_PACK_IMAGE_HEIGHT, current_image_height);
+            }
+            CopyImageToBuffer(copy, buffer_offsets[i]);
         }
-        CopyImageToBuffer(copy, buffer_offset);
     }
     if (is_rescaled) {
         ScaleUp(true);
diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h
index 911e4607a..0dd039ed2 100644
--- a/src/video_core/renderer_opengl/gl_texture_cache.h
+++ b/src/video_core/renderer_opengl/gl_texture_cache.h
@@ -215,6 +215,9 @@ public:
     void DownloadMemory(GLuint buffer_handle, size_t buffer_offset,
                         std::span<const VideoCommon::BufferImageCopy> copies);
 
+    void DownloadMemory(std::span<GLuint> buffer_handle, std::span<size_t> buffer_offset,
+                        std::span<const VideoCommon::BufferImageCopy> copies);
+
     void DownloadMemory(ImageBufferMap& map, std::span<const VideoCommon::BufferImageCopy> copies);
 
     GLuint StorageHandle() noexcept;
@@ -376,6 +379,7 @@ struct TextureCacheParams {
     using Sampler = OpenGL::Sampler;
     using Framebuffer = OpenGL::Framebuffer;
     using AsyncBuffer = u32;
+    using BufferType = GLuint;
 };
 
 using TextureCache = VideoCommon::TextureCache<TextureCacheParams>;
diff --git a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
index 985cc3203..a318d643e 100644
--- a/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_pipeline_cache.cpp
@@ -696,6 +696,13 @@ std::unique_ptr<ComputePipeline> PipelineCache::CreateComputePipeline(
 std::unique_ptr<ComputePipeline> PipelineCache::CreateComputePipeline(
     ShaderPools& pools, const ComputePipelineCacheKey& key, Shader::Environment& env,
     PipelineStatistics* statistics, bool build_in_parallel) try {
+    // TODO: Remove this when Intel fixes their shader compiler.
+    //       https://github.com/IGCIT/Intel-GPU-Community-Issue-Tracker-IGCIT/issues/159
+    if (device.GetDriverID() == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS) {
+        LOG_ERROR(Render_Vulkan, "Skipping 0x{:016x}", key.Hash());
+        return nullptr;
+    }
+
     LOG_INFO(Render_Vulkan, "0x{:016x}", key.Hash());
 
     Shader::Maxwell::Flow::CFG cfg{env, pools.flow_block, env.StartAddress()};
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
index 673ab478e..2559a3aa7 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
@@ -781,8 +781,7 @@ bool AccelerateDMA::DmaBufferImageCopy(const Tegra::DMA::ImageCopy& copy_info,
     }
     const u32 buffer_size = static_cast<u32>(buffer_operand.pitch * buffer_operand.height);
     static constexpr auto sync_info = VideoCommon::ObtainBufferSynchronize::FullSynchronize;
-    const auto post_op = IS_IMAGE_UPLOAD ? VideoCommon::ObtainBufferOperation::DoNothing
-                                         : VideoCommon::ObtainBufferOperation::MarkAsWritten;
+    const auto post_op = VideoCommon::ObtainBufferOperation::DoNothing;
     const auto [buffer, offset] =
         buffer_cache.ObtainBuffer(buffer_operand.address, buffer_size, sync_info, post_op);
 
@@ -793,7 +792,8 @@ bool AccelerateDMA::DmaBufferImageCopy(const Tegra::DMA::ImageCopy& copy_info,
     if constexpr (IS_IMAGE_UPLOAD) {
         image->UploadMemory(buffer->Handle(), offset, copy_span);
     } else {
-        image->DownloadMemory(buffer->Handle(), offset, copy_span);
+        texture_cache.DownloadImageIntoBuffer(image, buffer->Handle(), offset, copy_span,
+                                              buffer_operand.address, buffer_size);
     }
     return true;
 }
diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.cpp b/src/video_core/renderer_vulkan/vk_texture_cache.cpp
index ae15f6976..d0a7d8f35 100644
--- a/src/video_core/renderer_vulkan/vk_texture_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_texture_cache.cpp
@@ -1,10 +1,11 @@
 // SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
+// SPDX-License-Identifier: GPL-3.0-or-later
 
 #include <algorithm>
 #include <array>
 #include <span>
 #include <vector>
+#include <boost/container/small_vector.hpp>
 
 #include "common/bit_cast.h"
 #include "common/bit_util.h"
@@ -1343,14 +1344,31 @@ void Image::UploadMemory(const StagingBufferRef& map, std::span<const BufferImag
 
 void Image::DownloadMemory(VkBuffer buffer, VkDeviceSize offset,
                            std::span<const VideoCommon::BufferImageCopy> copies) {
+    std::array buffer_handles{
+        buffer,
+    };
+    std::array buffer_offsets{
+        offset,
+    };
+    DownloadMemory(buffer_handles, buffer_offsets, copies);
+}
+
+void Image::DownloadMemory(std::span<VkBuffer> buffers_span, std::span<VkDeviceSize> offsets_span,
+                           std::span<const VideoCommon::BufferImageCopy> copies) {
     const bool is_rescaled = True(flags & ImageFlagBits::Rescaled);
     if (is_rescaled) {
         ScaleDown();
     }
-    std::vector vk_copies = TransformBufferImageCopies(copies, offset, aspect_mask);
+    boost::container::small_vector<VkBuffer, 1> buffers_vector{};
+    boost::container::small_vector<std::vector<VkBufferImageCopy>, 1> vk_copies;
+    for (size_t index = 0; index < buffers_span.size(); index++) {
+        buffers_vector.emplace_back(buffers_span[index]);
+        vk_copies.emplace_back(
+            TransformBufferImageCopies(copies, offsets_span[index], aspect_mask));
+    }
     scheduler->RequestOutsideRenderPassOperationContext();
-    scheduler->Record([buffer, image = *original_image, aspect_mask = aspect_mask,
-                       vk_copies](vk::CommandBuffer cmdbuf) {
+    scheduler->Record([buffers = std::move(buffers_vector), image = *original_image,
+                       aspect_mask = aspect_mask, vk_copies](vk::CommandBuffer cmdbuf) {
         const VkImageMemoryBarrier read_barrier{
             .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
             .pNext = nullptr,
@@ -1369,6 +1387,20 @@ void Image::DownloadMemory(VkBuffer buffer, VkDeviceSize offset,
                 .layerCount = VK_REMAINING_ARRAY_LAYERS,
             },
         };
+        cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT,
+                               0, read_barrier);
+
+        for (size_t index = 0; index < buffers.size(); index++) {
+            cmdbuf.CopyImageToBuffer(image, VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL, buffers[index],
+                                     vk_copies[index]);
+        }
+
+        const VkMemoryBarrier memory_write_barrier{
+            .sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER,
+            .pNext = nullptr,
+            .srcAccessMask = VK_ACCESS_MEMORY_WRITE_BIT,
+            .dstAccessMask = VK_ACCESS_MEMORY_READ_BIT | VK_ACCESS_MEMORY_WRITE_BIT,
+        };
         const VkImageMemoryBarrier image_write_barrier{
             .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
             .pNext = nullptr,
@@ -1387,15 +1419,6 @@ void Image::DownloadMemory(VkBuffer buffer, VkDeviceSize offset,
                 .layerCount = VK_REMAINING_ARRAY_LAYERS,
             },
         };
-        const VkMemoryBarrier memory_write_barrier{
-            .sType = VK_STRUCTURE_TYPE_MEMORY_BARRIER,
-            .pNext = nullptr,
-            .srcAccessMask = VK_ACCESS_MEMORY_WRITE_BIT,
-            .dstAccessMask = VK_ACCESS_MEMORY_READ_BIT | VK_ACCESS_MEMORY_WRITE_BIT,
-        };
-        cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT, VK_PIPELINE_STAGE_TRANSFER_BIT,
-                               0, read_barrier);
-        cmdbuf.CopyImageToBuffer(image, VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL, buffer, vk_copies);
         cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT, VK_PIPELINE_STAGE_ALL_COMMANDS_BIT,
                                0, memory_write_barrier, nullptr, image_write_barrier);
     });
@@ -1405,7 +1428,13 @@ void Image::DownloadMemory(VkBuffer buffer, VkDeviceSize offset,
 }
 
 void Image::DownloadMemory(const StagingBufferRef& map, std::span<const BufferImageCopy> copies) {
-    DownloadMemory(map.buffer, map.offset, copies);
+    std::array buffers{
+        map.buffer,
+    };
+    std::array offsets{
+        map.offset,
+    };
+    DownloadMemory(buffers, offsets, copies);
 }
 
 bool Image::IsRescaled() const noexcept {
diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.h b/src/video_core/renderer_vulkan/vk_texture_cache.h
index d5ee23f8d..c656c5386 100644
--- a/src/video_core/renderer_vulkan/vk_texture_cache.h
+++ b/src/video_core/renderer_vulkan/vk_texture_cache.h
@@ -1,5 +1,5 @@
 // SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
-// SPDX-License-Identifier: GPL-2.0-or-later
+// SPDX-License-Identifier: GPL-3.0-or-later
 
 #pragma once
 
@@ -141,6 +141,9 @@ public:
     void DownloadMemory(VkBuffer buffer, VkDeviceSize offset,
                         std::span<const VideoCommon::BufferImageCopy> copies);
 
+    void DownloadMemory(std::span<VkBuffer> buffers, std::span<VkDeviceSize> offsets,
+                        std::span<const VideoCommon::BufferImageCopy> copies);
+
     void DownloadMemory(const StagingBufferRef& map,
                         std::span<const VideoCommon::BufferImageCopy> copies);
 
@@ -371,6 +374,7 @@ struct TextureCacheParams {
     using Sampler = Vulkan::Sampler;
     using Framebuffer = Vulkan::Framebuffer;
     using AsyncBuffer = Vulkan::StagingBufferRef;
+    using BufferType = VkBuffer;
 };
 
 using TextureCache = VideoCommon::TextureCache<TextureCacheParams>;
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index ed5c768d8..e601f8446 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -1,9 +1,10 @@
-// SPDX-FileCopyrightText: 2021 yuzu Emulator Project
+// SPDX-FileCopyrightText: 2023 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 #pragma once
 
 #include <unordered_set>
+#include <boost/container/small_vector.hpp>
 
 #include "common/alignment.h"
 #include "common/settings.h"
@@ -17,15 +18,10 @@
 
 namespace VideoCommon {
 
-using Tegra::Texture::SwizzleSource;
-using Tegra::Texture::TextureType;
 using Tegra::Texture::TICEntry;
 using Tegra::Texture::TSCEntry;
 using VideoCore::Surface::GetFormatType;
-using VideoCore::Surface::IsCopyCompatible;
 using VideoCore::Surface::PixelFormat;
-using VideoCore::Surface::PixelFormatFromDepthFormat;
-using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
 using VideoCore::Surface::SurfaceType;
 using namespace Common::Literals;
 
@@ -143,6 +139,13 @@ void TextureCache<P>::TickFrame() {
     runtime.TickFrame();
     critical_gc = 0;
     ++frame_tick;
+
+    if constexpr (IMPLEMENTS_ASYNC_DOWNLOADS) {
+        for (auto& buffer : async_buffers_death_ring) {
+            runtime.FreeDeferredStagingBuffer(buffer);
+        }
+        async_buffers_death_ring.clear();
+    }
 }
 
 template <class P>
@@ -661,25 +664,39 @@ template <class P>
 void TextureCache<P>::CommitAsyncFlushes() {
     // This is intentionally passing the value by copy
     if constexpr (IMPLEMENTS_ASYNC_DOWNLOADS) {
-        const std::span<const ImageId> download_ids = uncommitted_downloads;
+        auto& download_ids = uncommitted_downloads;
         if (download_ids.empty()) {
             committed_downloads.emplace_back(std::move(uncommitted_downloads));
             uncommitted_downloads.clear();
-            async_buffers.emplace_back(std::optional<AsyncBuffer>{});
+            async_buffers.emplace_back(std::move(uncommitted_async_buffers));
+            uncommitted_async_buffers.clear();
             return;
         }
         size_t total_size_bytes = 0;
-        for (const ImageId image_id : download_ids) {
-            total_size_bytes += slot_images[image_id].unswizzled_size_bytes;
+        size_t last_async_buffer_id = uncommitted_async_buffers.size();
+        bool any_none_dma = false;
+        for (PendingDownload& download_info : download_ids) {
+            if (download_info.is_swizzle) {
+                total_size_bytes +=
+                    Common::AlignUp(slot_images[download_info.object_id].unswizzled_size_bytes, 64);
+                any_none_dma = true;
+                download_info.async_buffer_id = last_async_buffer_id;
+            }
         }
-        auto download_map = runtime.DownloadStagingBuffer(total_size_bytes, true);
-        for (const ImageId image_id : download_ids) {
-            Image& image = slot_images[image_id];
-            const auto copies = FullDownloadCopies(image.info);
-            image.DownloadMemory(download_map, copies);
-            download_map.offset += Common::AlignUp(image.unswizzled_size_bytes, 64);
+        if (any_none_dma) {
+            auto download_map = runtime.DownloadStagingBuffer(total_size_bytes, true);
+            for (const PendingDownload& download_info : download_ids) {
+                if (download_info.is_swizzle) {
+                    Image& image = slot_images[download_info.object_id];
+                    const auto copies = FullDownloadCopies(image.info);
+                    image.DownloadMemory(download_map, copies);
+                    download_map.offset += Common::AlignUp(image.unswizzled_size_bytes, 64);
+                }
+            }
+            uncommitted_async_buffers.emplace_back(download_map);
         }
-        async_buffers.emplace_back(download_map);
+        async_buffers.emplace_back(std::move(uncommitted_async_buffers));
+        uncommitted_async_buffers.clear();
     }
     committed_downloads.emplace_back(std::move(uncommitted_downloads));
     uncommitted_downloads.clear();
@@ -691,39 +708,57 @@ void TextureCache<P>::PopAsyncFlushes() {
         return;
     }
     if constexpr (IMPLEMENTS_ASYNC_DOWNLOADS) {
-        const std::span<const ImageId> download_ids = committed_downloads.front();
+        const auto& download_ids = committed_downloads.front();
         if (download_ids.empty()) {
             committed_downloads.pop_front();
             async_buffers.pop_front();
             return;
         }
-        auto download_map = *async_buffers.front();
-        std::span<u8> download_span = download_map.mapped_span;
+        auto download_map = std::move(async_buffers.front());
         for (size_t i = download_ids.size(); i > 0; i--) {
-            const ImageBase& image = slot_images[download_ids[i - 1]];
-            const auto copies = FullDownloadCopies(image.info);
-            download_map.offset -= Common::AlignUp(image.unswizzled_size_bytes, 64);
-            std::span<u8> download_span_alt = download_span.subspan(download_map.offset);
-            SwizzleImage(*gpu_memory, image.gpu_addr, image.info, copies, download_span_alt,
-                         swizzle_data_buffer);
+            auto& download_info = download_ids[i - 1];
+            auto& download_buffer = download_map[download_info.async_buffer_id];
+            if (download_info.is_swizzle) {
+                const ImageBase& image = slot_images[download_info.object_id];
+                const auto copies = FullDownloadCopies(image.info);
+                download_buffer.offset -= Common::AlignUp(image.unswizzled_size_bytes, 64);
+                std::span<u8> download_span =
+                    download_buffer.mapped_span.subspan(download_buffer.offset);
+                SwizzleImage(*gpu_memory, image.gpu_addr, image.info, copies, download_span,
+                             swizzle_data_buffer);
+            } else {
+                const BufferDownload& buffer_info = slot_buffer_downloads[download_info.object_id];
+                std::span<u8> download_span =
+                    download_buffer.mapped_span.subspan(download_buffer.offset);
+                gpu_memory->WriteBlockUnsafe(buffer_info.address, download_span.data(),
+                                             buffer_info.size);
+                slot_buffer_downloads.erase(download_info.object_id);
+            }
+        }
+        for (auto& download_buffer : download_map) {
+            async_buffers_death_ring.emplace_back(download_buffer);
         }
-        runtime.FreeDeferredStagingBuffer(download_map);
         committed_downloads.pop_front();
         async_buffers.pop_front();
     } else {
-        const std::span<const ImageId> download_ids = committed_downloads.front();
+        const auto& download_ids = committed_downloads.front();
         if (download_ids.empty()) {
             committed_downloads.pop_front();
             return;
         }
         size_t total_size_bytes = 0;
-        for (const ImageId image_id : download_ids) {
-            total_size_bytes += slot_images[image_id].unswizzled_size_bytes;
+        for (const PendingDownload& download_info : download_ids) {
+            if (download_info.is_swizzle) {
+                total_size_bytes += slot_images[download_info.object_id].unswizzled_size_bytes;
+            }
         }
         auto download_map = runtime.DownloadStagingBuffer(total_size_bytes);
         const size_t original_offset = download_map.offset;
-        for (const ImageId image_id : download_ids) {
-            Image& image = slot_images[image_id];
+        for (const PendingDownload& download_info : download_ids) {
+            if (!download_info.is_swizzle) {
+                continue;
+            }
+            Image& image = slot_images[download_info.object_id];
             const auto copies = FullDownloadCopies(image.info);
             image.DownloadMemory(download_map, copies);
             download_map.offset += image.unswizzled_size_bytes;
@@ -732,8 +767,11 @@ void TextureCache<P>::PopAsyncFlushes() {
         runtime.Finish();
         download_map.offset = original_offset;
         std::span<u8> download_span = download_map.mapped_span;
-        for (const ImageId image_id : download_ids) {
-            const ImageBase& image = slot_images[image_id];
+        for (const PendingDownload& download_info : download_ids) {
+            if (!download_info.is_swizzle) {
+                continue;
+            }
+            const ImageBase& image = slot_images[download_info.object_id];
             const auto copies = FullDownloadCopies(image.info);
             SwizzleImage(*gpu_memory, image.gpu_addr, image.info, copies, download_span,
                          swizzle_data_buffer);
@@ -834,6 +872,33 @@ std::pair<typename TextureCache<P>::Image*, BufferImageCopy> TextureCache<P>::Dm
 }
 
 template <class P>
+void TextureCache<P>::DownloadImageIntoBuffer(typename TextureCache<P>::Image* image,
+                                              typename TextureCache<P>::BufferType buffer,
+                                              size_t buffer_offset,
+                                              std::span<const VideoCommon::BufferImageCopy> copies,
+                                              GPUVAddr address, size_t size) {
+    if constexpr (IMPLEMENTS_ASYNC_DOWNLOADS) {
+        const BufferDownload new_buffer_download{address, size};
+        auto slot = slot_buffer_downloads.insert(new_buffer_download);
+        const PendingDownload new_download{false, uncommitted_async_buffers.size(), slot};
+        uncommitted_downloads.emplace_back(new_download);
+        auto download_map = runtime.DownloadStagingBuffer(size, true);
+        uncommitted_async_buffers.emplace_back(download_map);
+        std::array buffers{
+            buffer,
+            download_map.buffer,
+        };
+        std::array buffer_offsets{
+            buffer_offset,
+            download_map.offset,
+        };
+        image->DownloadMemory(buffers, buffer_offsets, copies);
+    } else {
+        image->DownloadMemory(buffer, buffer_offset, copies);
+    }
+}
+
+template <class P>
 void TextureCache<P>::RefreshContents(Image& image, ImageId image_id) {
     if (False(image.flags & ImageFlagBits::CpuModified)) {
         // Only upload modified images
@@ -2209,7 +2274,8 @@ void TextureCache<P>::BindRenderTarget(ImageViewId* old_id, ImageViewId new_id)
     if (new_id) {
         const ImageViewBase& old_view = slot_image_views[new_id];
         if (True(old_view.flags & ImageViewFlagBits::PreemtiveDownload)) {
-            uncommitted_downloads.push_back(old_view.image_id);
+            const PendingDownload new_download{true, 0, old_view.image_id};
+            uncommitted_downloads.emplace_back(new_download);
         }
     }
     *old_id = new_id;
diff --git a/src/video_core/texture_cache/texture_cache_base.h b/src/video_core/texture_cache/texture_cache_base.h
index 5a5b4179c..758b7e212 100644
--- a/src/video_core/texture_cache/texture_cache_base.h
+++ b/src/video_core/texture_cache/texture_cache_base.h
@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: 2021 yuzu Emulator Project
+// SPDX-FileCopyrightText: 2023 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 #pragma once
@@ -40,14 +40,9 @@ struct ChannelState;
 
 namespace VideoCommon {
 
-using Tegra::Texture::SwizzleSource;
 using Tegra::Texture::TICEntry;
 using Tegra::Texture::TSCEntry;
-using VideoCore::Surface::GetFormatType;
-using VideoCore::Surface::IsCopyCompatible;
 using VideoCore::Surface::PixelFormat;
-using VideoCore::Surface::PixelFormatFromDepthFormat;
-using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
 using namespace Common::Literals;
 
 struct ImageViewInOut {
@@ -119,6 +114,7 @@ class TextureCache : public VideoCommon::ChannelSetupCaches<TextureCacheChannelI
     using Sampler = typename P::Sampler;
     using Framebuffer = typename P::Framebuffer;
     using AsyncBuffer = typename P::AsyncBuffer;
+    using BufferType = typename P::BufferType;
 
     struct BlitImages {
         ImageId dst_id;
@@ -215,6 +211,10 @@ public:
         const Tegra::DMA::ImageCopy& copy_info, const Tegra::DMA::BufferOperand& buffer_operand,
         const Tegra::DMA::ImageOperand& image_operand, ImageId image_id, bool modifies_image);
 
+    void DownloadImageIntoBuffer(Image* image, BufferType buffer, size_t buffer_offset,
+                                 std::span<const VideoCommon::BufferImageCopy> copies,
+                                 GPUVAddr address = 0, size_t size = 0);
+
     /// Return true when a CPU region is modified from the GPU
     [[nodiscard]] bool IsRegionGpuModified(VAddr addr, size_t size);
 
@@ -424,17 +424,32 @@ private:
     u64 critical_memory;
     size_t critical_gc;
 
+    struct BufferDownload {
+        GPUVAddr address;
+        size_t size;
+    };
+
+    struct PendingDownload {
+        bool is_swizzle;
+        size_t async_buffer_id;
+        SlotId object_id;
+    };
+
     SlotVector<Image> slot_images;
     SlotVector<ImageMapView> slot_map_views;
     SlotVector<ImageView> slot_image_views;
     SlotVector<ImageAlloc> slot_image_allocs;
     SlotVector<Sampler> slot_samplers;
     SlotVector<Framebuffer> slot_framebuffers;
+    SlotVector<BufferDownload> slot_buffer_downloads;
 
     // TODO: This data structure is not optimal and it should be reworked
-    std::vector<ImageId> uncommitted_downloads;
-    std::deque<std::vector<ImageId>> committed_downloads;
-    std::deque<std::optional<AsyncBuffer>> async_buffers;
+
+    std::vector<PendingDownload> uncommitted_downloads;
+    std::deque<std::vector<PendingDownload>> committed_downloads;
+    std::vector<AsyncBuffer> uncommitted_async_buffers;
+    std::deque<std::vector<AsyncBuffer>> async_buffers;
+    std::deque<AsyncBuffer> async_buffers_death_ring;
 
     struct LRUItemParams {
         using ObjectType = ImageId;