[vulkan] Correction on unconditional/ harcoded makeview 2D array vs compute shader 3D views

This reverts commit a9c4c8aefd.

Reviewed-on: https://git.eden-emu.dev/eden-emu/eden/pulls/4157
Reviewed-by: CamilleLaVey <camillelavey99@gmail.com>
Reviewed-by: MaranBr <maranbr@eden-emu.dev>

src/common/lru_cache.h

@@ -0,0 +1,142 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+// SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-2.0-or-later
+
+#pragma once
+
+#include <deque>
+#include <memory>
+#include <type_traits>
+
+#include "common/common_types.h"
+
+namespace Common {
+
+template <class Traits>
+class LeastRecentlyUsedCache {
+    using ObjectType = typename Traits::ObjectType;
+    using TickType = typename Traits::TickType;
+
+    struct Item {
+        ObjectType obj;
+        TickType tick;
+        Item* next{};
+        Item* prev{};
+    };
+
+public:
+    LeastRecentlyUsedCache() : first_item{}, last_item{} {}
+    ~LeastRecentlyUsedCache() = default;
+
+    size_t Insert(ObjectType obj, TickType tick) {
+        const auto new_id = Build();
+        auto& item = item_pool[new_id];
+        item.obj = obj;
+        item.tick = tick;
+        Attach(item);
+        return new_id;
+    }
+
+    void Touch(size_t id, TickType tick) {
+        auto& item = item_pool[id];
+        if (item.tick >= tick) {
+            return;
+        }
+        item.tick = tick;
+        if (&item == last_item) {
+            return;
+        }
+        Detach(item);
+        Attach(item);
+    }
+
+    void Free(size_t id) {
+        auto& item = item_pool[id];
+        Detach(item);
+        item.prev = nullptr;
+        item.next = nullptr;
+        free_items.push_back(id);
+    }
+
+    template <typename Func>
+    void ForEachItemBelow(TickType tick, Func&& func) {
+        static constexpr bool RETURNS_BOOL =
+            std::is_same_v<std::invoke_result_t<Func, ObjectType>, bool>;
+        Item* iterator = first_item;
+        while (iterator) {
+            if (static_cast<s64>(tick) - static_cast<s64>(iterator->tick) < 0) {
+                return;
+            }
+            Item* next = iterator->next;
+            if constexpr (RETURNS_BOOL) {
+                if (func(iterator->obj)) {
+                    return;
+                }
+            } else {
+                func(iterator->obj);
+            }
+            iterator = next;
+        }
+    }
+
+private:
+    size_t Build() {
+        if (free_items.empty()) {
+            const size_t item_id = item_pool.size();
+            auto& item = item_pool.emplace_back();
+            item.next = nullptr;
+            item.prev = nullptr;
+            return item_id;
+        }
+        const size_t item_id = free_items.front();
+        free_items.pop_front();
+        auto& item = item_pool[item_id];
+        item.next = nullptr;
+        item.prev = nullptr;
+        return item_id;
+    }
+
+    void Attach(Item& item) {
+        if (!first_item) {
+            first_item = &item;
+        }
+        if (!last_item) {
+            last_item = &item;
+        } else {
+            item.prev = last_item;
+            last_item->next = &item;
+            item.next = nullptr;
+            last_item = &item;
+        }
+    }
+
+    void Detach(Item& item) {
+        if (item.prev) {
+            item.prev->next = item.next;
+        }
+        if (item.next) {
+            item.next->prev = item.prev;
+        }
+        if (&item == first_item) {
+            first_item = item.next;
+            if (first_item) {
+                first_item->prev = nullptr;
+            }
+        }
+        if (&item == last_item) {
+            last_item = item.prev;
+            if (last_item) {
+                last_item->next = nullptr;
+            }
+        }
+    }
+
+    std::deque<Item> item_pool;
+    std::deque<size_t> free_items;
+    Item* first_item{};
+    Item* last_item{};
+};
+
+} // namespace Common

src/core/hle/service/acc/acc.cpp

@@ -729,7 +729,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface(ensure_token_id);
+        rb.PushIpcInterface(ctx, ensure_token_id);
     }
 
     void LoadIdTokenCacheDeprecated(HLERequestContext& ctx) {
@@ -921,7 +921,7 @@ void Module::Interface::GetProfile(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IProfile>(system, user_id, *profile_manager);
+    rb.PushIpcInterface<IProfile>(ctx, system, user_id, *profile_manager);
 }
 
 void Module::Interface::IsUserRegistrationRequestPermitted(HLERequestContext& ctx) {
@@ -993,7 +993,7 @@ void Module::Interface::GetBaasAccountManagerForApplication(HLERequestContext& c
     LOG_DEBUG(Service_ACC, "called");
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IManagerForApplication>(system, profile_manager);
+    rb.PushIpcInterface<IManagerForApplication>(ctx, system, profile_manager);
 }
 
 void Module::Interface::IsUserAccountSwitchLocked(HLERequestContext& ctx) {
@@ -1089,7 +1089,7 @@ void Module::Interface::GetProfileEditor(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IProfileEditor>(system, user_id, *profile_manager);
+    rb.PushIpcInterface<IProfileEditor>(ctx, system, user_id, *profile_manager);
 }
 
 void Module::Interface::GetBaasAccountAdministrator(HLERequestContext &ctx) {
@@ -1100,7 +1100,7 @@ void Module::Interface::GetBaasAccountAdministrator(HLERequestContext &ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IAdministrator>(system, uuid);
+    rb.PushIpcInterface<IAdministrator>(ctx, system, uuid);
 }
 
 void Module::Interface::ListQualifiedUsers(HLERequestContext& ctx) {
@@ -1143,7 +1143,7 @@ void Module::Interface::GetBaasAccountManagerForSystemService(HLERequestContext&
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IManagerForSystemService>(system, uuid);
+    rb.PushIpcInterface<IManagerForSystemService>(ctx, system, uuid);
 }
 
 void Module::Interface::StoreSaveDataThumbnailSystem(HLERequestContext& ctx) {

src/core/hle/service/acc/async_context.cpp

@@ -35,7 +35,7 @@ void IAsyncContext::GetSystemEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(completion_event->GetReadableEvent());
+    rb.PushCopyObjects(ctx, completion_event->GetReadableEvent());
 }
 
 void IAsyncContext::Cancel(HLERequestContext& ctx) {

src/core/hle/service/apm/apm_interface.cpp

@@ -82,7 +82,7 @@ void APM::OpenSession(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ISession>(system, controller);
+    rb.PushIpcInterface<ISession>(ctx, system, controller);
 }
 
 void APM::GetPerformanceMode(HLERequestContext& ctx) {
@@ -125,7 +125,7 @@ void APM_Sys::GetPerformanceEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ISession>(system, controller);
+    rb.PushIpcInterface<ISession>(ctx, system, controller);
 }
 
 void APM_Sys::SetCpuBoostMode(HLERequestContext& ctx) {

src/core/hle/service/fgm/fgm.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -45,7 +48,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IRequest>(system);
+        rb.PushIpcInterface<IRequest>(ctx, system);
     }
 };
 

src/core/hle/service/friend/friend.cpp

@@ -184,7 +184,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(readable_event.Signal(system.Kernel()));
-        rb.PushCopyObjects(readable_event);
+        rb.PushCopyObjects(ctx, readable_event);
     }
 
     void Cancel(HLERequestContext& ctx) {
@@ -400,7 +400,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(notification_event->GetReadableEvent());
+        rb.PushCopyObjects(ctx, notification_event->GetReadableEvent());
     }
 
     void Clear(HLERequestContext& ctx) {
@@ -476,7 +476,7 @@ private:
 void Module::Interface::CreateFriendService(HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IFriendService>(system);
+    rb.PushIpcInterface<IFriendService>(ctx, system);
     LOG_DEBUG(Service_Friend, "called");
 }
 
@@ -488,12 +488,12 @@ void Module::Interface::CreateNotificationService(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<INotificationService>(system, uuid);
+    rb.PushIpcInterface<INotificationService>(ctx, system, uuid);
 }
 
-Module::Interface::Interface(std::shared_ptr<Module> module_, Core::System& system_,
-                             const char* name)
-    : ServiceFramework{system_, name}, module{std::move(module_)} {}
+Module::Interface::Interface(std::shared_ptr<Module> module_, Core::System& system_, const char* name)
+    : ServiceFramework{system_, name}, module{std::move(module_)}
+{}
 
 Module::Interface::~Interface() = default;
 

src/core/hle/service/glue/arp.cpp

@@ -279,7 +279,7 @@ void ARP_W::AcquireRegistrar(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface(registrar);
+    rb.PushIpcInterface(ctx, registrar);
 }
 
 void ARP_W::UnregisterApplicationInstance(HLERequestContext& ctx) {

src/core/hle/service/glue/bgtc.cpp

@@ -28,7 +28,7 @@ void BGTC_T::OpenTaskService(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ITaskService>(system);
+    rb.PushIpcInterface<ITaskService>(ctx, system);
 }
 
 ITaskService::ITaskService(Core::System& system_) : ServiceFramework{system_, "ITaskService"} {

src/core/hle/service/glue/ectx.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -56,7 +59,7 @@ ECTX_AW::~ECTX_AW() = default;
 void ECTX_AW::CreateContextRegistrar(HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IContextRegistrar>(std::make_shared<IContextRegistrar>(system));
+    rb.PushIpcInterface<IContextRegistrar>(ctx, system);
 }
 
 } // namespace Service::Glue

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

@@ -726,7 +726,7 @@ void IHidSystemServer::AcquireConnectionTriggerTimeoutEvent(HLERequestContext& c
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(acquire_device_registered_event->GetReadableEvent());
+    rb.PushCopyObjects(ctx, acquire_device_registered_event->GetReadableEvent());
 }
 
 void IHidSystemServer::AcquireDeviceRegisteredEventForControllerSupport(HLERequestContext& ctx) {
@@ -734,7 +734,7 @@ void IHidSystemServer::AcquireDeviceRegisteredEventForControllerSupport(HLEReque
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(acquire_device_registered_event->GetReadableEvent());
+    rb.PushCopyObjects(ctx, acquire_device_registered_event->GetReadableEvent());
 }
 
 void IHidSystemServer::GetRegisteredDevices(HLERequestContext& ctx) {
@@ -759,7 +759,7 @@ void IHidSystemServer::AcquireUniquePadConnectionEventHandle(HLERequestContext&
     LOG_WARNING(Service_HID, "(STUBBED) called");
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
-    rb.PushCopyObjects(unique_pad_connection_event->GetReadableEvent());
+    rb.PushCopyObjects(ctx, unique_pad_connection_event->GetReadableEvent());
     rb.Push(ResultSuccess);
 }
 
@@ -776,7 +776,7 @@ void IHidSystemServer::AcquireJoyDetachOnBluetoothOffEventHandle(HLERequestConte
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(joy_detach_event->GetReadableEvent());
+    rb.PushCopyObjects(ctx, joy_detach_event->GetReadableEvent());
 }
 
 void IHidSystemServer::IsUsbFullKeyControllerEnabled(HLERequestContext& ctx) {

src/core/hle/service/hle_ipc.h

@@ -31,7 +31,7 @@ class Memory;
 }
 
 namespace IPC {
-class ResponseBuilder;
+struct ResponseBuilder;
 }
 
 namespace Service {
@@ -392,7 +392,7 @@ public:
     }
 
 private:
-    friend class IPC::ResponseBuilder;
+    friend struct IPC::ResponseBuilder;
 
     void ParseCommandBuffer(u32_le* src_cmdbuf, bool incoming);
 

src/core/hle/service/ipc_helpers.h

@@ -24,45 +24,7 @@ namespace IPC {
 
 constexpr Result ResultSessionClosed{ErrorModule::HIPC, 301};
 
-class RequestHelperBase {
-protected:
-    Service::HLERequestContext* context = nullptr;
-    u32* cmdbuf;
-    u32 index = 0;
-
-public:
-    explicit RequestHelperBase(u32* command_buffer) : cmdbuf(command_buffer) {}
-
-    explicit RequestHelperBase(Service::HLERequestContext& ctx)
-        : context(&ctx), cmdbuf(ctx.CommandBuffer()) {}
-
-    void Skip(u32 size_in_words, bool set_to_null) {
-        if (set_to_null) {
-            memset(cmdbuf + index, 0, size_in_words * sizeof(u32));
-        }
-        index += size_in_words;
-    }
-
-    /**
-     * Aligns the current position forward to a 16-byte boundary, padding with zeros.
-     */
-    void AlignWithPadding() {
-        if (index & 3) {
-            Skip(static_cast<u32>(4 - (index & 3)), true);
-        }
-    }
-
-    u32 GetCurrentOffset() const {
-        return index;
-    }
-
-    void SetCurrentOffset(u32 offset) {
-        index = offset;
-    }
-};
-
-class ResponseBuilder : public RequestHelperBase {
-public:
+struct ResponseBuilder {
     /// Flags used for customizing the behavior of ResponseBuilder
     enum class Flags : u32 {
         None = 0,
@@ -71,14 +33,13 @@ public:
         AlwaysMoveHandles = 1,
     };
 
-    explicit ResponseBuilder(Service::HLERequestContext& ctx, u32 normal_params_size_,
-                             u32 num_handles_to_copy_ = 0, u32 num_objects_to_move_ = 0,
-                             Flags flags = Flags::None)
-        : RequestHelperBase(ctx), normal_params_size(normal_params_size_),
-          num_handles_to_copy(num_handles_to_copy_),
-          num_objects_to_move(num_objects_to_move_), kernel{ctx.kernel} {
-
-        memset(cmdbuf, 0, sizeof(u32) * IPC::COMMAND_BUFFER_LENGTH);
+    inline explicit ResponseBuilder(Service::HLERequestContext& ctx, u32 normal_params_size_, u32 num_handles_to_copy_ = 0, u32 num_objects_to_move_ = 0, Flags flags = Flags::None)
+        : cmdbuf(ctx.CommandBuffer())
+        , normal_params_size(normal_params_size_)
+        , num_handles_to_copy(num_handles_to_copy_)
+        , num_objects_to_move(num_objects_to_move_)
+    {
+        std::memset(cmdbuf, 0, sizeof(u32) * IPC::COMMAND_BUFFER_LENGTH);
 
         IPC::CommandHeader header{};
         auto const mgr = ctx.GetManager().get();
@@ -117,9 +78,7 @@ public:
             handle_descriptor_header.num_handles_to_copy.Assign(num_handles_to_copy_);
             handle_descriptor_header.num_handles_to_move.Assign(num_handles_to_move);
             PushRaw(handle_descriptor_header);
-
             ctx.handles_offset = index;
-
             Skip(num_handles_to_copy + num_handles_to_move, true);
         }
 
@@ -131,7 +90,6 @@ public:
                 domain_header.num_objects = num_domain_objects;
                 PushRaw(domain_header);
             }
-
             IPC::DataPayloadHeader data_payload_header{};
             data_payload_header.magic = Common::MakeMagic('S', 'F', 'C', 'O');
             PushRaw(data_payload_header);
@@ -141,34 +99,39 @@ public:
 
         ctx.data_payload_offset = index;
         ctx.write_size += index;
-        ctx.domain_offset = static_cast<u32>(index + raw_data_size / sizeof(u32));
+        ctx.domain_offset = u32(index + raw_data_size / sizeof(u32));
     }
 
-    template <class T>
-    void PushIpcInterface(std::shared_ptr<T> iface) {
-        auto manager{context->GetManager()};
+    inline void Skip(u32 size_in_words, bool set_to_null) {
+        if (set_to_null) std::memset(cmdbuf + index, 0, size_in_words * sizeof(u32));
+        index += size_in_words;
+    }
+    /// @brief Aligns the current position forward to a 16-byte boundary, padding with zeros.
+    inline void AlignWithPadding() { if (index & 3) Skip(u32(4 - (index & 3)), true); }
+    inline u32 GetCurrentOffset() const { return index; }
+    inline void SetCurrentOffset(u32 offset) { index = offset; }
 
+    template <class T> inline void PushIpcInterface(Service::HLERequestContext& ctx, std::shared_ptr<T> iface) {
+        auto manager = ctx.GetManager();
         if (manager->IsDomain()) {
-            context->AddDomainObject(std::move(iface));
+            ctx.AddDomainObject(std::move(iface));
         } else {
-            ASSERT(Kernel::GetCurrentProcess(kernel).GetResourceLimit()->Reserve(kernel, Kernel::LimitableResource::SessionCountMax, 1));
+            ASSERT(Kernel::GetCurrentProcess(ctx.kernel).GetResourceLimit()->Reserve(ctx.kernel, Kernel::LimitableResource::SessionCountMax, 1));
 
-            auto* session = Kernel::KSession::Create(kernel);
-            session->Initialize(kernel, nullptr, 0);
-            Kernel::KSession::Register(kernel, session);
+            auto* session = Kernel::KSession::Create(ctx.kernel);
+            session->Initialize(ctx.kernel, nullptr, 0);
+            Kernel::KSession::Register(ctx.kernel, session);
 
-            auto next_manager = std::make_shared<Service::SessionRequestManager>(
-                kernel, manager->GetServerManager());
+            auto next_manager = std::make_shared<Service::SessionRequestManager>(ctx.kernel, manager->GetServerManager());
             next_manager->SetSessionHandler(iface);
             manager->GetServerManager().RegisterSession(&session->GetServerSession(), next_manager);
 
-            context->AddMoveObject(&session->GetClientSession());
+            ctx.AddMoveObject(&session->GetClientSession());
         }
     }
 
-    template <class T, class... Args>
-    void PushIpcInterface(Args&&... args) {
-        PushIpcInterface<T>(std::make_shared<T>(std::forward<Args>(args)...));
+    template <class T, class... Args> inline void PushIpcInterface(Service::HLERequestContext& ctx, Args&&... args) {
+        PushIpcInterface<T>(ctx, std::make_shared<T>(std::forward<Args>(args)...));
     }
 
     void PushImpl(s8 value);
@@ -184,59 +147,38 @@ public:
     void PushImpl(bool value);
     void PushImpl(Result value);
 
-    template <typename T>
-    void Push(T value) {
+    template <typename T> inline void Push(T value) {
         return PushImpl(value);
     }
 
     template <typename First, typename... Other>
     void Push(const First& first_value, const Other&... other_values);
 
-    /**
-     * Helper function for pushing strongly-typed enumeration values.
-     *
-     * @tparam Enum The enumeration type to be pushed
-     *
-     * @param value The value to push.
-     *
-     * @note The underlying size of the enumeration type is the size of the
-     *       data that gets pushed. e.g. "enum class SomeEnum : u16" will
-     *       push a u16-sized amount of data.
-     */
-    template <typename Enum>
-    void PushEnum(Enum value) {
+    /// @brief Helper function for pushing strongly-typed enumeration values.
+    /// @tparam Enum The enumeration type to be pushed
+    /// @param value The value to push.
+    /// @note The underlying size of the enumeration type is the size of the data that gets pushed.
+    /// e.g. "enum class SomeEnum : u16" will push a u16-sized amount of data.
+    template <typename Enum> inline void PushEnum(Enum value) {
         static_assert(std::is_enum_v<Enum>, "T must be an enum type within a PushEnum call.");
-        static_assert(!std::is_convertible_v<Enum, int>,
-                      "enum type in PushEnum must be a strongly typed enum.");
+        static_assert(!std::is_convertible_v<Enum, int>, "enum type in PushEnum must be a strongly typed enum.");
         Push(static_cast<std::underlying_type_t<Enum>>(value));
     }
 
-    /**
-     * @brief Copies the content of the given trivially copyable class to the buffer as a normal
-     * param
-     * @note: The input class must be correctly packed/padded to fit hardware layout.
-     */
-    template <typename T>
-    void PushRaw(const T& value);
+    /// @brief Copies the content of the given trivially copyable class to the buffer as a normal param
+    /// @note: The input class must be correctly packed/padded to fit hardware layout.
+    template <typename T> void PushRaw(const T& value);
+    template <typename... O> void PushMoveObjects(Service::HLERequestContext& ctx, O*... pointers);
+    template <typename... O> void PushMoveObjects(Service::HLERequestContext& ctx, O&... pointers);
+    template <typename... O> void PushCopyObjects(Service::HLERequestContext& ctx, O*... pointers);
+    template <typename... O> void PushCopyObjects(Service::HLERequestContext& ctx, O&... pointers);
 
-    template <typename... O>
-    void PushMoveObjects(O*... pointers);
-
-    template <typename... O>
-    void PushMoveObjects(O&... pointers);
-
-    template <typename... O>
-    void PushCopyObjects(O*... pointers);
-
-    template <typename... O>
-    void PushCopyObjects(O&... pointers);
-
-private:
+    u32* cmdbuf;
+    u32 index = 0;
     u32 normal_params_size{};
     u32 num_handles_to_copy{};
     u32 num_objects_to_move{}; ///< Domain objects or move handles, context dependent
     u32 data_payload_index{};
-    Kernel::KernelCore& kernel;
 };
 
 /// Push ///
@@ -251,8 +193,7 @@ inline void ResponseBuilder::PushImpl(u32 value) {
 
 template <typename T>
 void ResponseBuilder::PushRaw(const T& value) {
-    static_assert(std::is_trivially_copyable_v<T>,
-                  "It's undefined behavior to use memcpy with non-trivially copyable objects");
+    static_assert(std::is_trivially_copyable_v<T>, "It's undefined behavior to use memcpy with non-trivially copyable objects");
     std::memcpy(cmdbuf + index, &value, sizeof(T));
     index += (sizeof(T) + 3) / 4; // round up to word length
 }
@@ -272,8 +213,8 @@ inline void ResponseBuilder::PushImpl(s16 value) {
 }
 
 inline void ResponseBuilder::PushImpl(s64 value) {
-    PushImpl(static_cast<u32>(value));
-    PushImpl(static_cast<u32>(value >> 32));
+    PushImpl(u32(value));
+    PushImpl(u32(value >> 32));
 }
 
 inline void ResponseBuilder::PushImpl(u8 value) {
@@ -285,8 +226,8 @@ inline void ResponseBuilder::PushImpl(u16 value) {
 }
 
 inline void ResponseBuilder::PushImpl(u64 value) {
-    PushImpl(static_cast<u32>(value));
-    PushImpl(static_cast<u32>(value >> 32));
+    PushImpl(u32(value));
+    PushImpl(u32(value >> 32));
 }
 
 inline void ResponseBuilder::PushImpl(float value) {
@@ -312,90 +253,88 @@ void ResponseBuilder::Push(const First& first_value, const Other&... other_value
 }
 
 template <typename... O>
-inline void ResponseBuilder::PushCopyObjects(O*... pointers) {
+inline void ResponseBuilder::PushCopyObjects(Service::HLERequestContext& ctx, O*... pointers) {
     auto objects = {pointers...};
     for (auto& object : objects) {
-        context->AddCopyObject(object);
+        ctx.AddCopyObject(object);
     }
 }
 
 template <typename... O>
-inline void ResponseBuilder::PushCopyObjects(O&... pointers) {
+inline void ResponseBuilder::PushCopyObjects(Service::HLERequestContext& ctx, O&... pointers) {
     auto objects = {&pointers...};
     for (auto& object : objects) {
-        context->AddCopyObject(object);
+        ctx.AddCopyObject(object);
     }
 }
 
 template <typename... O>
-inline void ResponseBuilder::PushMoveObjects(O*... pointers) {
+inline void ResponseBuilder::PushMoveObjects(Service::HLERequestContext& ctx, O*... pointers) {
     auto objects = {pointers...};
     for (auto& object : objects) {
-        context->AddMoveObject(object);
+        ctx.AddMoveObject(object);
     }
 }
 
 template <typename... O>
-inline void ResponseBuilder::PushMoveObjects(O&... pointers) {
+inline void ResponseBuilder::PushMoveObjects(Service::HLERequestContext& ctx, O&... pointers) {
     auto objects = {&pointers...};
     for (auto& object : objects) {
-        context->AddMoveObject(object);
+        ctx.AddMoveObject(object);
     }
 }
 
-class RequestParser : public RequestHelperBase {
-public:
-    explicit RequestParser(u32* command_buffer) : RequestHelperBase(command_buffer) {}
-
-    explicit RequestParser(Service::HLERequestContext& ctx) : RequestHelperBase(ctx) {
+struct RequestParser {
+    inline explicit RequestParser(u32* command_buffer) : cmdbuf(command_buffer) {}
+    inline explicit RequestParser(Service::HLERequestContext& ctx)
+        : cmdbuf(ctx.CommandBuffer())
+    {
         // TIPC does not have data payload offset
         if (!ctx.IsTipc()) {
             ASSERT_MSG(ctx.GetDataPayloadOffset(), "context is incomplete");
             Skip(ctx.GetDataPayloadOffset(), false);
         }
-
         // Skip the u64 command id, it's already stored in the context
         static constexpr u32 CommandIdSize = 2;
         Skip(CommandIdSize, false);
     }
 
-    template <typename T>
-    T Pop();
+    inline void Skip(u32 size_in_words, bool set_to_null) {
+        if (set_to_null) std::memset(cmdbuf + index, 0, size_in_words * sizeof(u32));
+        index += size_in_words;
+    }
+    /// @brief Aligns the current position forward to a 16-byte boundary, padding with zeros.
+    inline void AlignWithPadding() { if (index & 3) Skip(u32(4 - (index & 3)), true); }
+    inline u32 GetCurrentOffset() const { return index; }
+    inline void SetCurrentOffset(u32 offset) { index = offset; }
 
-    template <typename T>
-    void Pop(T& value);
-
-    template <typename First, typename... Other>
-    void Pop(First& first_value, Other&... other_values);
+    template <typename T> T Pop();
+    template <typename T> void Pop(T& value);
+    template <typename First, typename... Other> void Pop(First& first_value, Other&... other_values);
 
     template <typename T>
     T PopEnum() {
         static_assert(std::is_enum_v<T>, "T must be an enum type within a PopEnum call.");
-        static_assert(!std::is_convertible_v<T, int>,
-                      "enum type in PopEnum must be a strongly typed enum.");
-        return static_cast<T>(Pop<std::underlying_type_t<T>>());
+        static_assert(!std::is_convertible_v<T, int>, "enum type in PopEnum must be a strongly typed enum.");
+        return T(Pop<std::underlying_type_t<T>>());
     }
 
-    /**
-     * @brief Reads the next normal parameters as a struct, by copying it
-     * @note: The output class must be correctly packed/padded to fit hardware layout.
-     */
-    template <typename T>
-    void PopRaw(T& value);
+    /// @brief Reads the next normal parameters as a struct, by copying it
+    /// @note: The output class must be correctly packed/padded to fit hardware layout.
+    template <typename T> void PopRaw(T& value);
 
-    /**
-     * @brief Reads the next normal parameters as a struct, by copying it into a new value
-     * @note: The output class must be correctly packed/padded to fit hardware layout.
-     */
-    template <typename T>
-    T PopRaw();
+    /// @brief Reads the next normal parameters as a struct, by copying it into a new value
+    /// @note: The output class must be correctly packed/padded to fit hardware layout.
+    template <typename T> T PopRaw();
 
-    template <class T>
-    std::weak_ptr<T> PopIpcInterface() {
-        ASSERT(context->GetManager()->IsDomain());
-        ASSERT(context->GetDomainMessageHeader().input_object_count > 0);
-        return context->GetDomainHandler<T>(Pop<u32>() - 1);
+    template <class T> [[nodiscard]] std::weak_ptr<T> PopIpcInterface(Service::HLERequestContext& ctx) {
+        ASSERT(ctx.GetManager()->IsDomain());
+        ASSERT(ctx.GetDomainMessageHeader().input_object_count > 0);
+        return ctx.GetDomainHandler<T>(Pop<u32>() - 1);
     }
+
+    u32* cmdbuf;
+    u32 index = 0;
 };
 
 /// Pop ///
@@ -407,7 +346,7 @@ inline u32 RequestParser::Pop() {
 
 template <>
 inline s32 RequestParser::Pop() {
-    return static_cast<s32>(Pop<u32>());
+    return s32(Pop<u32>());
 }
 
 // Ignore the -Wclass-memaccess warning on memcpy for non-trivially default constructible objects.
@@ -417,8 +356,7 @@ inline s32 RequestParser::Pop() {
 #endif
 template <typename T>
 void RequestParser::PopRaw(T& value) {
-    static_assert(std::is_trivially_copyable_v<T>,
-                  "It's undefined behavior to use memcpy with non-trivially copyable objects");
+    static_assert(std::is_trivially_copyable_v<T>, "It's undefined behavior to use memcpy with non-trivially copyable objects");
     std::memcpy(&value, cmdbuf + index, sizeof(T));
     index += (sizeof(T) + 3) / 4; // round up to word length
 }

src/core/hle/service/lm/lm.cpp

@@ -353,7 +353,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<ILogger>(system);
+        rb.PushIpcInterface<ILogger>(ctx, system);
     }
 };
 

src/core/hle/service/nfc/nfc.cpp

@@ -151,7 +151,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IAm>(system);
+        rb.PushIpcInterface<IAm>(ctx, system);
     }
 };
 
@@ -173,7 +173,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<MFIUser>(system);
+        rb.PushIpcInterface<MFIUser>(ctx, system);
     }
 };
 
@@ -195,7 +195,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IUser>(system);
+        rb.PushIpcInterface<IUser>(ctx, system);
     }
 };
 
@@ -217,7 +217,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<ISystem>(system);
+        rb.PushIpcInterface<ISystem>(ctx, system);
     }
 };
 

src/core/hle/service/nfc/nfc_interface.cpp

@@ -143,7 +143,7 @@ void NfcInterface::AttachAvailabilityChangeEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(GetManager()->AttachAvailabilityChangeEvent());
+    rb.PushCopyObjects(ctx, GetManager()->AttachAvailabilityChangeEvent());
 }
 
 void NfcInterface::StartDetection(HLERequestContext& ctx) {
@@ -203,7 +203,7 @@ void NfcInterface::AttachActivateEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(result);
-    rb.PushCopyObjects(out_event);
+    rb.PushCopyObjects(ctx, out_event);
 }
 
 void NfcInterface::AttachDeactivateEvent(HLERequestContext& ctx) {
@@ -217,7 +217,7 @@ void NfcInterface::AttachDeactivateEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(result);
-    rb.PushCopyObjects(out_event);
+    rb.PushCopyObjects(ctx, out_event);
 }
 
 void NfcInterface::SetNfcEnabled(HLERequestContext& ctx) {

src/core/hle/service/nfp/nfp.cpp

@@ -157,7 +157,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IUser>(system);
+        rb.PushIpcInterface<IUser>(ctx, system);
     }
 };
 
@@ -179,7 +179,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<ISystem>(system);
+        rb.PushIpcInterface<ISystem>(ctx, system);
     }
 };
 
@@ -201,7 +201,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IDebug>(system);
+        rb.PushIpcInterface<IDebug>(ctx, system);
     }
 };
 

src/core/hle/service/nifm/nifm.cpp

@@ -307,7 +307,7 @@ private:
     void GetSystemEventReadableHandle(HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 2, 2};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(evt_scan_complete->GetReadableEvent(),
+        rb.PushCopyObjects(ctx, evt_scan_complete->GetReadableEvent(),
                            evt_processing->GetReadableEvent());
     }
 
@@ -452,7 +452,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 2};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(event1->GetReadableEvent(), event2->GetReadableEvent());
+        rb.PushCopyObjects(ctx, event1->GetReadableEvent(), event2->GetReadableEvent());
     }
 
     void Cancel(HLERequestContext& ctx) {
@@ -528,7 +528,7 @@ void IGeneralService::CreateScanRequest(HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
 
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IScanRequest>(system);
+    rb.PushIpcInterface<IScanRequest>(ctx, system);
 }
 
 void IGeneralService::CreateRequest(HLERequestContext& ctx) {
@@ -537,7 +537,7 @@ void IGeneralService::CreateRequest(HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
 
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IRequest>(system);
+    rb.PushIpcInterface<IRequest>(ctx, system);
 }
 
 void IGeneralService::GetCurrentNetworkProfile(HLERequestContext& ctx) {
@@ -716,7 +716,7 @@ void IGeneralService::GetNetworkProfile(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<INetworkProfile>(system);
+    rb.PushIpcInterface<INetworkProfile>(ctx, system);
 }
 
 void IGeneralService::SetNetworkProfile(HLERequestContext& ctx) {
@@ -869,7 +869,7 @@ void IGeneralService::CreateTemporaryNetworkProfile(HLERequestContext& ctx) {
     IPC::ResponseBuilder rb{ctx, 6, 0, 1};
 
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<INetworkProfile>(system);
+    rb.PushIpcInterface<INetworkProfile>(ctx, system);
     rb.PushRaw<u128>(uuid);
 }
 
@@ -1124,7 +1124,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IGeneralService>(system);
+        rb.PushIpcInterface<IGeneralService>(ctx, system);
     }
 
     void CreateGeneralService(HLERequestContext& ctx) {
@@ -1132,7 +1132,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IGeneralService>(system);
+        rb.PushIpcInterface<IGeneralService>(ctx, system);
     }
 };
 

src/core/hle/service/nim/nim.cpp

@@ -53,7 +53,7 @@ private:
         LOG_WARNING(Service_NIM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IShopServiceAsync>(system);
+        rb.PushIpcInterface<IShopServiceAsync>(ctx, system);
     }
 };
 
@@ -75,7 +75,7 @@ private:
         LOG_WARNING(Service_NIM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IShopServiceAccessor>(system);
+        rb.PushIpcInterface<IShopServiceAccessor>(ctx, system);
     }
 };
 
@@ -336,7 +336,7 @@ private:
         LOG_DEBUG(Service_NIM, "(STUBBED) called");
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IShopServiceAccessServer>(system);
+        rb.PushIpcInterface<IShopServiceAccessServer>(ctx, system);
     }
 
     void IsLargeResourceAvailable(HLERequestContext& ctx) {
@@ -356,7 +356,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IShopServiceAccessServer>(system);
+        rb.PushIpcInterface<IShopServiceAccessServer>(ctx, system);
     }
 };
 
@@ -439,7 +439,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(finished_event->GetReadableEvent());
+        rb.PushCopyObjects(ctx, finished_event->GetReadableEvent());
     }
 
     void GetResult(HLERequestContext& ctx) {
@@ -500,7 +500,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IEnsureNetworkClockAvailabilityService>(system);
+        rb.PushIpcInterface<IEnsureNetworkClockAvailabilityService>(ctx, system);
     }
 
     // TODO(ogniK): Do we need these?

src/core/hle/service/ns/read_only_application_control_data_interface.cpp

@@ -359,8 +359,8 @@ void IReadOnlyApplicationControlDataInterface::ListApplicationTitle(HLERequestCo
 
     IPC::ResponseBuilder rb{ctx, 2, 1, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(async_value->ReadableEvent());
-    rb.PushIpcInterface(std::move(async_value));
+    rb.PushCopyObjects(ctx, async_value->ReadableEvent());
+    rb.PushIpcInterface(ctx, std::move(async_value));
 }
 
 Result IReadOnlyApplicationControlDataInterface::GetApplicationControlData3(

src/core/hle/service/nvdrv/nvdrv_interface.cpp

@@ -199,7 +199,7 @@ void NVDRV::QueryEvent(HLERequestContext& ctx) {
         IPC::ResponseBuilder rb{ctx, 3, 1};
         rb.Push(ResultSuccess);
         auto& readable_event = event->GetReadableEvent();
-        rb.PushCopyObjects(readable_event);
+        rb.PushCopyObjects(ctx, readable_event);
         rb.PushEnum(NvResult::Success);
     } else {
         LOG_ERROR(Service_NVDRV, "Invalid event request!");

src/core/hle/service/os/process.h

@@ -6,6 +6,8 @@
 
 #pragma once
 
+#include <utility>
+
 #include "common/common_types.h"
 
 namespace Core {
@@ -28,6 +30,15 @@ public:
     inline explicit Process(Core::System& system) noexcept : m_system(system) {}
     inline ~Process() { this->Finalize(); }
 
+    Process(const Process&) = delete;
+    Process& operator=(const Process&) = delete;
+    Process& operator=(Process&&) = delete;
+    inline Process(Process&& other) noexcept
+        : m_system(other.m_system), m_process(std::exchange(other.m_process, nullptr)),
+          m_main_thread_stack_size(std::exchange(other.m_main_thread_stack_size, 0)),
+          m_main_thread_priority(std::exchange(other.m_main_thread_priority, 0)),
+          m_process_started(std::exchange(other.m_process_started, false)) {}
+
     bool Initialize(Loader::AppLoader& loader, Loader::ResultStatus& out_load_result);
     void Finalize();
 

src/core/hle/service/pcv/pcv.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -124,7 +127,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IClkrstSession>(system, device_code);
+        rb.PushIpcInterface<IClkrstSession>(ctx, system, device_code);
     }
 };
 

src/core/hle/service/pm/pm.cpp

@@ -161,7 +161,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 10, 1};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(*process);
+        rb.PushCopyObjects(ctx, *process);
         rb.PushRaw(program_location);
         rb.PushRaw(override_status);
     }

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

@@ -147,7 +147,7 @@ void IAlarmService::CreateWakeupAlarm(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ISteadyClockAlarm>(system, m_alarms, AlarmType::WakeupAlarm);
+    rb.PushIpcInterface<ISteadyClockAlarm>(ctx, system, m_alarms, AlarmType::WakeupAlarm);
 }
 
 void IAlarmService::CreateBackgroundTaskAlarm(HLERequestContext& ctx) {
@@ -155,7 +155,7 @@ void IAlarmService::CreateBackgroundTaskAlarm(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ISteadyClockAlarm>(system, m_alarms, AlarmType::BackgroundTaskAlarm);
+    rb.PushIpcInterface<ISteadyClockAlarm>(ctx, system, m_alarms, AlarmType::BackgroundTaskAlarm);
 }
 
 ISteadyClockAlarm::ISteadyClockAlarm(Core::System& system_, Alarms& alarms, AlarmType type)
@@ -179,7 +179,7 @@ void ISteadyClockAlarm::GetAlarmEvent(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 1};
     rb.Push(ResultSuccess);
-    rb.PushCopyObjects(m_alarm.GetEventHandle());
+    rb.PushCopyObjects(ctx, m_alarm.GetEventHandle());
 }
 
 void ISteadyClockAlarm::Enable(HLERequestContext& ctx) {

src/core/hle/service/ptm/psm.cpp

@@ -67,7 +67,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 1};
         rb.Push(ResultSuccess);
-        rb.PushCopyObjects(state_change_event->GetReadableEvent());
+        rb.PushCopyObjects(ctx, state_change_event->GetReadableEvent());
     }
 
     void UnbindStateChangeEvent(HLERequestContext& ctx) {
@@ -186,7 +186,7 @@ void PSM::OpenSession(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<IPsmSession>(system);
+    rb.PushIpcInterface<IPsmSession>(ctx, system);
 }
 
 void PSM::GetBatteryVoltageState(HLERequestContext& ctx) {

src/core/hle/service/ptm/ts.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
@@ -82,7 +85,7 @@ void TS::OpenSession(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1};
     rb.Push(ResultSuccess);
-    rb.PushIpcInterface<ISession>(system);
+    rb.PushIpcInterface<ISession>(ctx, system);
 }
 
 } // namespace Service::PTM

src/core/hle/service/sm/sm.cpp

@@ -140,7 +140,7 @@ void SM::GetServiceCmif(HLERequestContext& ctx) {
     if (result == ResultSuccess) {
         IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
         rb.Push(result);
-        rb.PushMoveObjects(client_session);
+        rb.PushMoveObjects(ctx, client_session);
     } else {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(result);
@@ -157,7 +157,7 @@ void SM::GetServiceTipc(HLERequestContext& ctx) {
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
     rb.Push(result);
-    rb.PushMoveObjects(result == ResultSuccess ? client_session : nullptr);
+    rb.PushMoveObjects(ctx, result == ResultSuccess ? client_session : nullptr);
 }
 
 static std::string PopServiceName(IPC::RequestParser& rp) {
@@ -230,8 +230,7 @@ void SM::RegisterServiceImpl(HLERequestContext& ctx, std::string name, u32 max_s
               max_session_count, is_light);
 
     Kernel::KServerPort* server_port{};
-    if (const auto result = service_manager.RegisterService(std::addressof(server_port), name,
-                                                            max_session_count, nullptr);
+    if (const auto result = service_manager.RegisterService(std::addressof(server_port), name, max_session_count, nullptr);
         result.IsError()) {
         LOG_ERROR(Service_SM, "failed to register service with error_code={:08X}", result.raw);
         IPC::ResponseBuilder rb{ctx, 2};
@@ -241,7 +240,7 @@ void SM::RegisterServiceImpl(HLERequestContext& ctx, std::string name, u32 max_s
 
     IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
     rb.Push(ResultSuccess);
-    rb.PushMoveObjects(server_port);
+    rb.PushMoveObjects(ctx, server_port);
 }
 
 void SM::UnregisterService(HLERequestContext& ctx) {

src/core/hle/service/sm/sm_controller.cpp

@@ -60,7 +60,7 @@ void Controller::CloneCurrentObject(HLERequestContext& ctx) {
     // We succeeded.
     IPC::ResponseBuilder rb{ctx, 2, 0, 1, IPC::ResponseBuilder::Flags::AlwaysMoveHandles};
     rb.Push(ResultSuccess);
-    rb.PushMoveObjects(session->GetClientSession());
+    rb.PushMoveObjects(ctx, session->GetClientSession());
 }
 
 void Controller::CloneCurrentObjectEx(HLERequestContext& ctx) {

src/core/hle/service/ssl/ssl.cpp

@@ -543,8 +543,7 @@ private:
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(res);
         if (res == ResultSuccess) {
-            rb.PushIpcInterface<ISslConnection>(system, ssl_version, shared_data,
-                                                std::move(backend));
+            rb.PushIpcInterface<ISslConnection>(ctx, system, ssl_version, shared_data, std::move(backend));
         }
     }
 
@@ -624,12 +623,11 @@ private:
         IPC::RequestParser rp{ctx};
         const auto parameters = rp.PopRaw<Parameters>();
 
-        LOG_WARNING(Service_SSL, "(STUBBED) called, api_version={}, pid_placeholder={}",
-                    parameters.ssl_version.api_version, parameters.pid_placeholder);
+        LOG_WARNING(Service_SSL, "(STUBBED) called, api_version={}, pid_placeholder={}", parameters.ssl_version.api_version, parameters.pid_placeholder);
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<ISslContext>(system, parameters.ssl_version);
+        rb.PushIpcInterface<ISslContext>(ctx, system, parameters.ssl_version);
     }
 
     void SetInterfaceVersion(HLERequestContext& ctx) {

src/core/hle/service/usb/usb.cpp

@@ -155,7 +155,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IPdSession>(system);
+        rb.PushIpcInterface<IPdSession>(ctx, system);
     }
 };
 
@@ -199,7 +199,7 @@ private:
 
         IPC::ResponseBuilder rb{ctx, 2, 0, 1};
         rb.Push(ResultSuccess);
-        rb.PushIpcInterface<IPdCradleSession>(system);
+        rb.PushIpcInterface<IPdCradleSession>(ctx, system);
     }
 };
 

src/hid_core/resources/npad/npad.cpp

@@ -50,8 +50,8 @@ NPad::NPad(Core::HID::HIDCore& hid_core_, KernelHelpers::ServiceContext& service
             auto& controller = controller_data[aruid_index][i];
             controller.device = hid_core.GetEmulatedControllerByIndex(i);
             Core::HID::ControllerUpdateCallback engine_callback{
-                .on_change = [this, i](Core::HID::ControllerTriggerType type) {
-                    ControllerUpdate(hid_core.kernel, type, i);
+                .on_change = [this, i, kernel = &hid_core.kernel](Core::HID::ControllerTriggerType type) {
+                    ControllerUpdate(*kernel, type, i);
                 },
                 .is_npad_service = true,
             };

src/shader_recompiler/frontend/maxwell/translate_program.cpp

@@ -236,8 +236,11 @@ void LowerGeometryPassthrough(const IR::Program& program, const HostTranslateInf
 
 IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Block>& block_pool,
                              Environment& env, Flow::CFG& cfg, const HostTranslateInfo& host_info) {
+    HostTranslateInfo normalized_host_info{host_info};
+    normalized_host_info.ApplyDescriptorLimitPolicy();
+
     IR::Program program;
-    program.syntax_list = BuildASL(inst_pool, block_pool, env, cfg, host_info);
+    program.syntax_list = BuildASL(inst_pool, block_pool, env, cfg, normalized_host_info);
     program.blocks = GenerateBlocks(program.syntax_list);
     program.post_order_blocks = PostOrder(program.syntax_list.front());
     program.stage = env.ShaderStage();
@@ -260,9 +263,9 @@ IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Blo
                 program.info.passthrough.mask[i] = ((mask[i / 32] >> (i % 32)) & 1) == 0;
             }
 
-            if (!host_info.support_geometry_shader_passthrough) {
+            if (!normalized_host_info.support_geometry_shader_passthrough) {
                 program.output_vertices = GetOutputTopologyVertices(program.output_topology);
-                LowerGeometryPassthrough(program, host_info);
+                LowerGeometryPassthrough(program, normalized_host_info);
             }
         }
         break;
@@ -277,16 +280,16 @@ IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Blo
     RemoveUnreachableBlocks(program);
 
     // Replace instructions before the SSA rewrite
-    if (!host_info.support_float64) {
+    if (!normalized_host_info.support_float64) {
         Optimization::LowerFp64ToFp32(program);
     }
-    if (!host_info.support_float16) {
+    if (!normalized_host_info.support_float16) {
         Optimization::LowerFp16ToFp32(program);
     }
-    if (!host_info.support_int64) {
+    if (!normalized_host_info.support_int64) {
         Optimization::LowerInt64ToInt32(program);
     }
-    if (!host_info.support_conditional_barrier) {
+    if (!normalized_host_info.support_conditional_barrier) {
         Optimization::ConditionalBarrierPass(program);
     }
     Optimization::SsaRewritePass(program);
@@ -295,8 +298,8 @@ IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Blo
 
     Optimization::PositionPass(env, program);
 
-    Optimization::GlobalMemoryToStorageBufferPass(program, host_info);
-    Optimization::TexturePass(env, program, host_info);
+    Optimization::GlobalMemoryToStorageBufferPass(program, normalized_host_info);
+    Optimization::TexturePass(env, program, normalized_host_info);
 
     if (Settings::values.resolution_info.active || Settings::values.rescale_hack.GetValue()) {
         Optimization::RescalingPass(program);
@@ -306,7 +309,7 @@ IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Blo
         Optimization::VerificationPass(program);
     }
     Optimization::CollectShaderInfoPass(env, program);
-    Optimization::LayerPass(program, host_info);
+    Optimization::LayerPass(program, normalized_host_info);
     Optimization::VendorWorkaroundPass(program);
 
     CollectInterpolationInfo(env, program);

src/shader_recompiler/host_translate_info.h

@@ -6,6 +6,8 @@
 
 #pragma once
 
+#include "common/common_types.h"
+
 namespace Shader {
 
 // Try to keep entries here to a minimum
@@ -13,20 +15,52 @@ namespace Shader {
 
 /// Misc information about the host
 struct HostTranslateInfo {
+    static constexpr u32 DEFAULT_DESCRIPTOR_LIMIT = 1024;
+
+    u64 min_ssbo_alignment{};            ///< Minimum alignment supported by the device for SSBOs
+    u32 max_per_stage_descriptor_sampled_images{}; ///< maximum sampled descriptors per stage
+    u32 max_per_stage_resources{};                 ///< maximum resources per stage
+    u32 max_descriptor_set_samplers{};
+    u32 max_descriptor_set_uniform_buffers{};
+    u32 max_descriptor_set_uniform_buffers_dynamic{};
+    u32 max_descriptor_set_storage_buffers{};
+    u32 max_descriptor_set_storage_buffers_dynamic{};
+    u32 max_descriptor_set_sampled_images{};
+    u32 max_descriptor_set_storage_images{};
+    u32 max_descriptor_set_input_attachements{};
     bool support_float64{};      ///< True when the device supports 64-bit floats
     bool support_float16{};      ///< True when the device supports 16-bit floats
     bool support_int64{};        ///< True when the device supports 64-bit integers
     bool needs_demote_reorder{}; ///< True when the device needs DemoteToHelperInvocation reordered
     bool support_snorm_render_buffer{};  ///< True when the device supports SNORM render buffers
     bool support_viewport_index_layer{}; ///< True when the device supports gl_Layer in VS
-    u32 min_ssbo_alignment{};            ///< Minimum alignment supported by the device for SSBOs
-    u32 max_per_stage_descriptor_sampled_images{1024}; ///< maximum sampled descriptors per stage
-    u32 max_per_stage_resources{4096};                 ///< maximum resources per stage
-    u32 max_descriptor_set_sampled_images{1024};        ///< maximum sampled descriptors per set
     bool support_geometry_shader_passthrough{}; ///< True when the device supports geometry
                                                 ///< passthrough shaders
     bool support_conditional_barrier{}; ///< True when the device supports barriers in conditional
                                         ///< control flow
+
+    void ApplyDescriptorLimitPolicy() noexcept {
+        if (min_ssbo_alignment == 0) {
+            min_ssbo_alignment = 1;
+        }
+        ApplyDescriptorLimitFallback(max_per_stage_descriptor_sampled_images);
+        ApplyDescriptorLimitFallback(max_per_stage_resources);
+        ApplyDescriptorLimitFallback(max_descriptor_set_samplers);
+        ApplyDescriptorLimitFallback(max_descriptor_set_uniform_buffers);
+        ApplyDescriptorLimitFallback(max_descriptor_set_uniform_buffers_dynamic);
+        ApplyDescriptorLimitFallback(max_descriptor_set_storage_buffers);
+        ApplyDescriptorLimitFallback(max_descriptor_set_storage_buffers_dynamic);
+        ApplyDescriptorLimitFallback(max_descriptor_set_sampled_images);
+        ApplyDescriptorLimitFallback(max_descriptor_set_storage_images);
+        ApplyDescriptorLimitFallback(max_descriptor_set_input_attachements);
+    }
+
+private:
+    static void ApplyDescriptorLimitFallback(u32& limit) noexcept {
+        if (limit == 0) {
+            limit = DEFAULT_DESCRIPTOR_LIMIT;
+        }
+    }
 };
 
 } // namespace Shader

src/shader_recompiler/ir_opt/global_memory_to_storage_buffer_pass.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -545,7 +548,7 @@ void GlobalMemoryToStorageBufferPass(IR::Program& program, const HostTranslateIn
         IR::Block* const block{storage_inst.block};
         IR::Inst* const inst{storage_inst.inst};
         const IR::U32 offset{
-            StorageOffset(*block, *inst, storage_buffer, host_info.min_ssbo_alignment)};
+            StorageOffset(*block, *inst, storage_buffer, u32(host_info.min_ssbo_alignment))};
         Replace(*block, *inst, index, offset);
     }
 }

src/shader_recompiler/ir_opt/texture_pass.cpp

@@ -32,71 +32,62 @@ struct TextureInst {
 using TextureInstVector = boost::container::small_vector<TextureInst, 24>;
 
 constexpr u32 DESCRIPTOR_SIZE = 8;
-constexpr u32 DESCRIPTOR_SIZE_SHIFT = static_cast<u32>(std::countr_zero(DESCRIPTOR_SIZE));
-constexpr u32 DYNAMIC_DESCRIPTOR_CBUF_BYTES = 16 * 1024;
-constexpr u32 MAX_DYNAMIC_DESCRIPTOR_COUNT = 1024;
+constexpr u32 DESCRIPTOR_SIZE_SHIFT = u32(std::countr_zero(DESCRIPTOR_SIZE));
+constexpr u32 DESCRIPTOR_MAX_COUNT = 1024;
 
 u32 DynamicDescriptorSizeShift(const IR::U32& dynamic_offset) {
-    const IR::Inst* const inst{dynamic_offset.InstRecursive()};
-    if (!inst || inst->GetOpcode() != IR::Opcode::ShiftLeftLogical32) {
+    const IR::Inst* const inst = dynamic_offset.InstRecursive();
+    if (!inst || inst->GetOpcode() != IR::Opcode::ShiftLeftLogical32)
         return DESCRIPTOR_SIZE_SHIFT;
-    }
-    const IR::Value shift{inst->Arg(1)};
-    if (!shift.IsImmediate()) {
+    const IR::Value shift = inst->Arg(1);
+    if (!shift.IsImmediate())
         return DESCRIPTOR_SIZE_SHIFT;
-    }
-    const u32 size_shift{shift.U32()};
-    return size_shift >= DESCRIPTOR_SIZE_SHIFT && size_shift < 31 ? size_shift
-                                                                  : DESCRIPTOR_SIZE_SHIFT;
+    const u32 size_shift = shift.U32();
+    return size_shift >= DESCRIPTOR_SIZE_SHIFT && size_shift < 31 ? size_shift : DESCRIPTOR_SIZE_SHIFT;
 }
 
-u32 DynamicDescriptorCount(u32 base_offset, u32 size_shift) {
-    if (size_shift >= 31 || base_offset >= DYNAMIC_DESCRIPTOR_CBUF_BYTES) {
+u32 DynamicDescriptorCount(u32 base_offset, u32 size_shift, u32 max_descriptors) {
+    auto const descriptor_limit = (std::max)(1U, max_descriptors);
+    auto const max_cbuf_bytes = 16 * descriptor_limit;
+    if (size_shift >= 31 || base_offset >= max_cbuf_bytes)
         return 1;
-    }
-    const u32 stride{1U << size_shift};
-    const u32 available{DYNAMIC_DESCRIPTOR_CBUF_BYTES - base_offset};
-    if (available < DESCRIPTOR_SIZE) {
+    auto const stride = 1U << size_shift;
+    auto const available = max_cbuf_bytes - base_offset;
+    if (available < DESCRIPTOR_SIZE)
         return 1;
-    }
-    const u32 available_count{1U + (available - DESCRIPTOR_SIZE) / stride};
-    return std::min(MAX_DYNAMIC_DESCRIPTOR_COUNT, available_count);
+    auto const available_count = 1U + (available - DESCRIPTOR_SIZE) / stride;
+    return std::min(descriptor_limit, available_count);
 }
 
 u32 SaturatingSub(u32 lhs, u32 rhs) {
     return lhs > rhs ? lhs - rhs : 0;
 }
 
-template <typename Descriptors>
-u32 StaticDescriptorCount(const Descriptors& descriptors) {
-    u32 count{};
-    for (const auto& desc : descriptors) {
-        if (desc.count <= 1) {
-            count += desc.count;
-        }
-    }
-    return count;
+template <typename T>
+[[nodiscard]] u32 StaticDescriptorCount(T const& descriptors) noexcept {
+    return std::accumulate(descriptors.cbegin(), descriptors.cend(), 0U, [](auto const& acc, auto const& e) {
+        return acc + (e.count <= 1 ? e.count : 0);
+    });
 }
 
-u32 DynamicSampledTextureCap(const Info& info, const HostTranslateInfo& host_info,
-                             u32 dynamic_arrays) {
-    if (dynamic_arrays == 0) {
-        return MAX_DYNAMIC_DESCRIPTOR_COUNT;
+u32 DynamicSampledTextureCap(const Info& info, const HostTranslateInfo& host_info, u32 dynamic_arrays) {
+    auto const sampled_limit = (std::max)(1U, std::min(host_info.max_per_stage_descriptor_sampled_images,
+                                                       host_info.max_descriptor_set_sampled_images));
+    auto const resource_limit = (std::max)(1U, host_info.max_per_stage_resources);
+    if (dynamic_arrays > 0) {
+        auto const sampled_static_count = StaticDescriptorCount(info.texture_buffer_descriptors) + StaticDescriptorCount(info.texture_descriptors);
+        auto const resource_static_count =
+            NumDescriptors(info.constant_buffer_descriptors)
+            + NumDescriptors(info.storage_buffers_descriptors)
+            + sampled_static_count + NumDescriptors(info.image_buffer_descriptors)
+            + NumDescriptors(info.image_descriptors);
+        auto const sampled_budget = SaturatingSub(sampled_limit, sampled_static_count);
+        auto const resource_budget = SaturatingSub(resource_limit, resource_static_count);
+        auto const sampled_cap = sampled_budget / dynamic_arrays;
+        auto const resource_cap = resource_budget / dynamic_arrays;
+        return (std::max)(1U, (std::min)(sampled_cap, resource_cap));
     }
-    const u32 sampled_static_count{StaticDescriptorCount(info.texture_buffer_descriptors) +
-                                   StaticDescriptorCount(info.texture_descriptors)};
-    const u32 resource_static_count{
-        NumDescriptors(info.constant_buffer_descriptors) +
-        NumDescriptors(info.storage_buffers_descriptors) + sampled_static_count +
-        NumDescriptors(info.image_buffer_descriptors) + NumDescriptors(info.image_descriptors)};
-    const u32 sampled_limit{std::min(host_info.max_per_stage_descriptor_sampled_images,
-                                     host_info.max_descriptor_set_sampled_images)};
-    const u32 sampled_budget{SaturatingSub(sampled_limit, sampled_static_count)};
-    const u32 resource_budget{SaturatingSub(host_info.max_per_stage_resources,
-                                            resource_static_count)};
-    const u32 sampled_cap{sampled_budget / dynamic_arrays};
-    const u32 resource_cap{resource_budget / dynamic_arrays};
-    return std::max(1U, std::min({MAX_DYNAMIC_DESCRIPTOR_COUNT, sampled_cap, resource_cap}));
+    return (std::min)({DESCRIPTOR_MAX_COUNT, sampled_limit, resource_limit});
 }
 
 IR::Opcode IndexedInstruction(const IR::Inst& inst) {
@@ -304,21 +295,23 @@ static inline bool IsTexturePixelFormatIntegerCached(Environment& env,
 }
 
 
-std::optional<ConstBufferAddr> Track(const IR::Value& value, Environment& env);
-static inline std::optional<ConstBufferAddr> TrackCached(const IR::Value& v, Environment& env) {
+std::optional<ConstBufferAddr> Track(const IR::Value& value, Environment& env, const HostTranslateInfo& host_info);
+static inline std::optional<ConstBufferAddr> TrackCached(const IR::Value& v, Environment& env, const HostTranslateInfo& host_info) {
     if (const IR::Inst* key = v.InstRecursive()) {
         if (auto it = env.track_cache.find(key); it != env.track_cache.end()) return it->second;
-        auto found = Track(v, env);
+        auto found = Track(v, env, host_info);
         if (found) env.track_cache.emplace(key, *found);
         return found;
     }
-    return Track(v, env);
+    return Track(v, env, host_info);
 }
 
-std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environment& env);
+std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environment& env, const HostTranslateInfo& host_info);
 
-std::optional<ConstBufferAddr> Track(const IR::Value& value, Environment& env) {
-    return IR::BreadthFirstSearch(value, [&env](const IR::Inst* inst) { return TryGetConstBuffer(inst, env); });
+std::optional<ConstBufferAddr> Track(const IR::Value& value, Environment& env, const HostTranslateInfo& host_info) {
+    return IR::BreadthFirstSearch(value, [&env, &host_info](const IR::Inst* inst) {
+        return TryGetConstBuffer(inst, env, host_info);
+    });
 }
 
 std::optional<u32> TryGetConstant(IR::Value& value, Environment& env) {
@@ -342,13 +335,13 @@ std::optional<u32> TryGetConstant(IR::Value& value, Environment& env) {
     return ReadCbufCached(env, index_number, offset_number);
 }
 
-std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environment& env) {
+std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environment& env, const HostTranslateInfo& host_info) {
     switch (inst->GetOpcode()) {
     default:
         return std::nullopt;
     case IR::Opcode::BitwiseOr32: {
-        std::optional lhs{TrackCached(inst->Arg(0), env)};
-        std::optional rhs{TrackCached(inst->Arg(1), env)};
+        std::optional lhs{TrackCached(inst->Arg(0), env, host_info)};
+        std::optional rhs{TrackCached(inst->Arg(1), env, host_info)};
         if (!lhs || !rhs) {
             return std::nullopt;
         }
@@ -378,12 +371,11 @@ std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environme
         if (!shift.IsImmediate()) {
             return std::nullopt;
         }
-        std::optional lhs{TrackCached(inst->Arg(0), env)};
+        std::optional lhs{TrackCached(inst->Arg(0), env, host_info)};
         if (lhs) {
             lhs->shift_left = shift.U32();
         }
         return lhs;
-        break;
     }
     case IR::Opcode::BitwiseAnd32: {
         IR::Value op1{inst->Arg(0)};
@@ -407,7 +399,7 @@ std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environme
                 return std::nullopt;
             } while (false);
         }
-        std::optional lhs{TrackCached(op1, env)};
+        std::optional lhs{TrackCached(op1, env, host_info)};
         if (lhs) {
             lhs->shift_left = static_cast<u32>(std::countr_zero(op2.U32()));
         }
@@ -453,7 +445,10 @@ std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environme
     } else {
         return std::nullopt;
     }
-    const u32 size_shift{DynamicDescriptorSizeShift(dynamic_offset)};
+    auto const size_shift = DynamicDescriptorSizeShift(dynamic_offset);
+    auto const sampled_limit = (std::max)(1U, (std::min)(host_info.max_per_stage_descriptor_sampled_images,
+                                                         host_info.max_descriptor_set_sampled_images));
+    auto const resource_limit = (std::max)(1U, host_info.max_per_stage_resources);
     return ConstBufferAddr{
         .index = index.U32(),
         .offset = base_offset,
@@ -462,15 +457,15 @@ std::optional<ConstBufferAddr> TryGetConstBuffer(const IR::Inst* inst, Environme
         .secondary_offset = 0,
         .secondary_shift_left = 0,
         .dynamic_offset = dynamic_offset,
-        .count = DynamicDescriptorCount(base_offset, size_shift),
+        .count = DynamicDescriptorCount(base_offset, size_shift, (std::min)({DESCRIPTOR_MAX_COUNT, sampled_limit, resource_limit})),
         .has_secondary = false,
     };
 }
 
-TextureInst MakeInst(Environment& env, IR::Block* block, IR::Inst& inst) {
+TextureInst MakeInst(Environment& env, IR::Block* block, IR::Inst& inst, const HostTranslateInfo& host_info) {
     ConstBufferAddr addr;
     if (IsBindless(inst)) {
-        const std::optional<ConstBufferAddr> track_addr{TrackCached(inst.Arg(0), env)};
+        const std::optional<ConstBufferAddr> track_addr{TrackCached(inst.Arg(0), env, host_info)};
 
         if (!track_addr) {
             throw NotImplementedException("Failed to track bindless texture constant buffer");
@@ -506,15 +501,15 @@ u32 GetTextureHandle(Environment& env, const ConstBufferAddr& cbuf) {
     return lhs_raw | rhs_raw;
 }
 
-    [[maybe_unused]]TextureType ReadTextureType(Environment& env, const ConstBufferAddr& cbuf) {
+[[maybe_unused]] TextureType ReadTextureType(Environment& env, const ConstBufferAddr& cbuf) {
     return env.ReadTextureType(GetTextureHandle(env, cbuf));
 }
 
-    [[maybe_unused]]TexturePixelFormat ReadTexturePixelFormat(Environment& env, const ConstBufferAddr& cbuf) {
+[[maybe_unused]] TexturePixelFormat ReadTexturePixelFormat(Environment& env, const ConstBufferAddr& cbuf) {
     return env.ReadTexturePixelFormat(GetTextureHandle(env, cbuf));
 }
 
-    [[maybe_unused]]bool IsTexturePixelFormatInteger(Environment& env, const ConstBufferAddr& cbuf) {
+[[maybe_unused]] bool IsTexturePixelFormatInteger(Environment& env, const ConstBufferAddr& cbuf) {
     return env.IsTexturePixelFormatInteger(GetTextureHandle(env, cbuf));
 }
 
@@ -675,7 +670,7 @@ void TexturePass(Environment& env, IR::Program& program, const HostTranslateInfo
             if (!IsTextureInstruction(inst)) {
                 continue;
             }
-            to_replace.push_back(MakeInst(env, block, inst));
+            to_replace.push_back(MakeInst(env, block, inst, host_info));
         }
     }
     // Sort instructions to visit textures by constant buffer index, then by offset
@@ -689,8 +684,7 @@ void TexturePass(Environment& env, IR::Program& program, const HostTranslateInfo
         program.info.texture_descriptors,
         program.info.image_descriptors,
     };
-    const u32 sampled_dynamic_cap{
-        DynamicSampledTextureCap(program.info, host_info, DynamicSampledTextureArrayCount(to_replace))};
+    const u32 sampled_dynamic_cap = DynamicSampledTextureCap(program.info, host_info, DynamicSampledTextureArrayCount(to_replace));
     for (TextureInst& texture_inst : to_replace) {
         // TODO: Handle arrays
         IR::Inst* const inst{texture_inst.inst};

src/shader_recompiler/profile.h

@@ -92,7 +92,6 @@ struct Profile {
     bool has_broken_robust{};
 
     u64 min_ssbo_alignment{};
-
     u32 max_user_clip_distances{};
 };
 

src/tests/common/host_memory.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2021 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -15,12 +18,19 @@ static constexpr auto PERMS = Common::MemoryPermission::ReadWrite;
 static constexpr auto HEAP = false;
 
 TEST_CASE("HostMemory: Initialize and deinitialize", "[common]") {
-    { HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE); }
-    { HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE); }
+    {
+        HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+        REQUIRE(mem.BackingBasePointer() != nullptr);
+    }
+    {
+        HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+        REQUIRE(mem.BackingBasePointer() != nullptr);
+    }
 }
 
 TEST_CASE("HostMemory: Simple map", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x5000, 0x8000, 0x1000, PERMS, HEAP);
 
     volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
@@ -30,6 +40,7 @@ TEST_CASE("HostMemory: Simple map", "[common]") {
 
 TEST_CASE("HostMemory: Simple mirror map", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x5000, 0x3000, 0x2000, PERMS, HEAP);
     mem.Map(0x8000, 0x4000, 0x1000, PERMS, HEAP);
 
@@ -41,6 +52,7 @@ TEST_CASE("HostMemory: Simple mirror map", "[common]") {
 
 TEST_CASE("HostMemory: Simple unmap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x5000, 0x3000, 0x2000, PERMS, HEAP);
 
     volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
@@ -52,6 +64,7 @@ TEST_CASE("HostMemory: Simple unmap", "[common]") {
 
 TEST_CASE("HostMemory: Simple unmap and remap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x5000, 0x3000, 0x2000, PERMS, HEAP);
 
     volatile u8* const data = mem.VirtualBasePointer() + 0x5000;
@@ -69,6 +82,7 @@ TEST_CASE("HostMemory: Simple unmap and remap", "[common]") {
 
 TEST_CASE("HostMemory: Nieche allocation", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x0000, 0, 0x20000, PERMS, HEAP);
     mem.Unmap(0x0000, 0x4000, HEAP);
     mem.Map(0x1000, 0, 0x2000, PERMS, HEAP);
@@ -78,6 +92,7 @@ TEST_CASE("HostMemory: Nieche allocation", "[common]") {
 
 TEST_CASE("HostMemory: Full unmap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x8000, 0x4000, HEAP);
     mem.Map(0x6000, 0, 0x16000, PERMS, HEAP);
@@ -85,6 +100,7 @@ TEST_CASE("HostMemory: Full unmap", "[common]") {
 
 TEST_CASE("HostMemory: Right out of bounds unmap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x0000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x2000, 0x4000, HEAP);
     mem.Map(0x2000, 0x80000, 0x4000, PERMS, HEAP);
@@ -92,6 +108,7 @@ TEST_CASE("HostMemory: Right out of bounds unmap", "[common]") {
 
 TEST_CASE("HostMemory: Left out of bounds unmap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+
     mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x6000, 0x4000, HEAP);
     mem.Map(0x8000, 0, 0x2000, PERMS, HEAP);
@@ -99,6 +116,7 @@ TEST_CASE("HostMemory: Left out of bounds unmap", "[common]") {
 
 TEST_CASE("HostMemory: Multiple placeholder unmap", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x0000, 0, 0x4000, PERMS, HEAP);
     mem.Map(0x4000, 0, 0x1b000, PERMS, HEAP);
     mem.Unmap(0x3000, 0x1c000, HEAP);
@@ -107,6 +125,7 @@ TEST_CASE("HostMemory: Multiple placeholder unmap", "[common]") {
 
 TEST_CASE("HostMemory: Unmap between placeholders", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x0000, 0, 0x4000, PERMS, HEAP);
     mem.Map(0x4000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x2000, 0x4000, HEAP);
@@ -115,6 +134,7 @@ TEST_CASE("HostMemory: Unmap between placeholders", "[common]") {
 
 TEST_CASE("HostMemory: Unmap to origin", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0, 0x4000, PERMS, HEAP);
     mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x4000, 0x4000, HEAP);
@@ -124,6 +144,7 @@ TEST_CASE("HostMemory: Unmap to origin", "[common]") {
 
 TEST_CASE("HostMemory: Unmap to right", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0, 0x4000, PERMS, HEAP);
     mem.Map(0x8000, 0, 0x4000, PERMS, HEAP);
     mem.Unmap(0x8000, 0x4000, HEAP);
@@ -132,6 +153,7 @@ TEST_CASE("HostMemory: Unmap to right", "[common]") {
 
 TEST_CASE("HostMemory: Partial right unmap check bindings", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0x10000, 0x4000, PERMS, HEAP);
 
     volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
@@ -144,6 +166,7 @@ TEST_CASE("HostMemory: Partial right unmap check bindings", "[common]") {
 
 TEST_CASE("HostMemory: Partial left unmap check bindings", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0x10000, 0x4000, PERMS, HEAP);
 
     volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
@@ -158,6 +181,7 @@ TEST_CASE("HostMemory: Partial left unmap check bindings", "[common]") {
 
 TEST_CASE("HostMemory: Partial middle unmap check bindings", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0x10000, 0x4000, PERMS, HEAP);
 
     volatile u8* const ptr = mem.VirtualBasePointer() + 0x4000;
@@ -172,6 +196,7 @@ TEST_CASE("HostMemory: Partial middle unmap check bindings", "[common]") {
 
 TEST_CASE("HostMemory: Partial sparse middle unmap and check bindings", "[common]") {
     HostMemory mem(BACKING_SIZE, VIRTUAL_SIZE);
+    REQUIRE(mem.BackingBasePointer() != nullptr);
     mem.Map(0x4000, 0x10000, 0x2000, PERMS, HEAP);
     mem.Map(0x6000, 0x20000, 0x2000, PERMS, HEAP);
 

src/video_core/buffer_cache/buffer_base.h

@@ -109,12 +109,12 @@ public:
         return static_cast<u32>(other_cpu_addr - cpu_addr);
     }
 
-    u64 GetFrameTick() const noexcept {
-        return frame_tick;
+    size_t getLRUID() const noexcept {
+        return lru_id;
     }
 
-    void SetFrameTick(u64 tick) noexcept {
-        frame_tick = tick;
+    void setLRUID(size_t lru_id_) {
+        lru_id = lru_id_;
     }
 
     size_t SizeBytes() const {
@@ -125,7 +125,7 @@ private:
     VAddr cpu_addr = 0;
     BufferFlagBits flags{};
     int stream_score = 0;
-    u64 frame_tick = 0;
+    size_t lru_id = SIZE_MAX;
     size_t size_bytes = 0;
 };
 

src/video_core/buffer_cache/buffer_cache.h

@@ -58,22 +58,17 @@ void BufferCache<P>::RunGarbageCollector() {
     const bool aggressive_gc = total_used_memory >= critical_memory;
     const u64 ticks_to_destroy = aggressive_gc ? 60 : 120;
     int num_iterations = aggressive_gc ? 64 : 32;
-    const u64 threshold = frame_tick - ticks_to_destroy;
-    boost::container::small_vector<BufferId, 64> expired;
-    for (auto [id, buffer] : slot_buffers) {
-        if (buffer->GetFrameTick() < threshold) {
-            expired.push_back(id);
-        }
-    }
-    for (const auto buffer_id : expired) {
+    const auto clean_up = [this, &num_iterations](BufferId buffer_id) {
         if (num_iterations == 0) {
-            break;
+            return true;
         }
         --num_iterations;
         auto& buffer = slot_buffers[buffer_id];
         DownloadBufferMemory(buffer);
         DeleteBuffer(buffer_id);
-    }
+        return false;
+    };
+    lru_cache.ForEachItemBelow(frame_tick - ticks_to_destroy, clean_up);
 }
 
 template <class P>
@@ -1596,9 +1591,10 @@ void BufferCache<P>::ChangeRegister(BufferId buffer_id) {
     const auto size = buffer.SizeBytes();
     if (insert) {
         total_used_memory += Common::AlignUp(size, 1024);
-        buffer.SetFrameTick(frame_tick);
+        buffer.setLRUID(lru_cache.Insert(buffer_id, frame_tick));
     } else {
         total_used_memory -= Common::AlignUp(size, 1024);
+        lru_cache.Free(buffer.getLRUID());
     }
     const DAddr device_addr_begin = buffer.CpuAddr();
     const DAddr device_addr_end = device_addr_begin + size;
@@ -1616,7 +1612,7 @@ void BufferCache<P>::ChangeRegister(BufferId buffer_id) {
 template <class P>
 void BufferCache<P>::TouchBuffer(Buffer& buffer, BufferId buffer_id) noexcept {
     if (buffer_id != NULL_BUFFER_ID) {
-        buffer.SetFrameTick(frame_tick);
+        lru_cache.Touch(buffer.getLRUID(), frame_tick);
     }
 }
 

src/video_core/buffer_cache/buffer_cache_base.h

@@ -23,6 +23,7 @@
 #include "common/common_types.h"
 #include "common/div_ceil.h"
 #include "common/literals.h"
+#include "common/lru_cache.h"
 #include "common/range_sets.h"
 #include "common/scope_exit.h"
 #include "common/settings.h"
@@ -505,6 +506,11 @@ private:
     size_t immediate_buffer_capacity = 0;
     Common::ScratchBuffer<u8> immediate_buffer_alloc;
 
+    struct LRUItemParams {
+        using ObjectType = BufferId;
+        using TickType = u64;
+    };
+    Common::LeastRecentlyUsedCache<LRUItemParams> lru_cache;
     u64 frame_tick = 0;
     u64 total_used_memory = 0;
     u64 minimum_memory = 0;

src/video_core/dma_pusher.cpp

@@ -117,35 +117,7 @@ void DmaPusher::ProcessCommands(std::span<const CommandHeader> commands) {
             dma_state.is_last_call = true;
             index += max_write;
         } else if (dma_state.method_count) {
-            if (!dma_state.non_incrementing && !dma_increment_once &&
-                dma_state.method >= non_puller_methods) {
-                auto subchannel = subchannels[dma_state.subchannel];
-                const u32 available = u32(std::min<size_t>(
-                    index + dma_state.method_count, commands.size()) - index);
-                u32 batch = 0;
-                u32 method = dma_state.method;
-                while (batch < available) {
-                    const bool needs_exec =
-                        (method < Engines::EngineInterface::EXECUTION_MASK_TABLE_SIZE)
-                            ? subchannel->execution_mask[method]
-                            : subchannel->execution_mask_default;
-                    if (needs_exec) break;
-                    batch++;
-                    method++;
-                }
-                if (batch > 0) {
-                    auto& sink = subchannel->method_sink;
-                    sink.reserve(sink.size() + batch);
-                    for (u32 j = 0; j < batch; j++) {
-                        sink.emplace_back(dma_state.method + j, commands[index + j].argument);
-                    }
-                    dma_state.method += batch;
-                    dma_state.method_count -= batch;
-                    index += batch;
-                    continue;
-                }
-            }
-            auto const command_header = commands[index];
+            auto const command_header = commands[index]; //can copy
             dma_state.dma_word_offset = u32(index * sizeof(u32));
             dma_state.is_last_call = dma_state.method_count <= 1;
             CallMethod(command_header.argument);
@@ -204,11 +176,7 @@ void DmaPusher::CallMethod(u32 argument) {
         });
     } else {
         auto subchannel = subchannels[dma_state.subchannel];
-        const bool needs_execution =
-            (dma_state.method < Engines::EngineInterface::EXECUTION_MASK_TABLE_SIZE)
-                ? subchannel->execution_mask[dma_state.method]
-                : subchannel->execution_mask_default;
-        if (!needs_execution) {
+        if (!subchannel->execution_mask[dma_state.method]) {
             subchannel->method_sink.emplace_back(dma_state.method, argument);
         } else {
             subchannel->ConsumeSink(system);

src/video_core/engines/engine_interface.h

@@ -6,8 +6,9 @@
 
 #pragma once
 
-#include <array>
-#include <boost/container/small_vector.hpp>
+#include <bitset>
+#include <limits>
+#include <vector>
 
 #include "common/common_types.h"
 
@@ -42,15 +43,10 @@ public:
         }
     }
 
-    static constexpr size_t EXECUTION_MASK_TABLE_SIZE = 0xE00;
-
-    std::array<u8, EXECUTION_MASK_TABLE_SIZE> execution_mask{};
-    bool execution_mask_default{};
-    boost::container::small_vector<std::pair<u32, u32>, 64> method_sink{};
+    std::bitset<(std::numeric_limits<u16>::max)()> execution_mask{};
+    std::vector<std::pair<u32, u32>> method_sink{};
     GPUVAddr current_dma_segment;
-    /// @brief Indicates whether the current DMA segment is dirty.
     bool current_dirty{};
-
 protected:
     virtual void ConsumeSinkImpl(Core::System& system) {
         for (auto [method, value] : method_sink) {

src/video_core/engines/fermi_2d.cpp

@@ -26,7 +26,7 @@ Fermi2D::Fermi2D(MemoryManager& memory_manager_) : memory_manager{memory_manager
     regs.src.depth = 1;
     regs.dst.depth = 1;
 
-    execution_mask.fill(0);
+    execution_mask.reset();
     execution_mask[FERMI2D_REG_INDEX(pixels_from_memory.src_y0) + 1] = true;
 }
 

src/video_core/engines/kepler_compute.cpp

@@ -20,7 +20,7 @@ KeplerCompute::KeplerCompute(MemoryManager& memory_manager_)
     : memory_manager{memory_manager_}
     , upload_state{memory_manager, regs.upload}
 {
-    execution_mask.fill(0);
+    execution_mask.reset();
     execution_mask[KEPLER_COMPUTE_REG_INDEX(exec_upload)] = true;
     execution_mask[KEPLER_COMPUTE_REG_INDEX(data_upload)] = true;
     execution_mask[KEPLER_COMPUTE_REG_INDEX(launch)] = true;

src/video_core/engines/kepler_memory.cpp

@@ -23,7 +23,7 @@ KeplerMemory::~KeplerMemory() = default;
 void KeplerMemory::BindRasterizer(VideoCore::RasterizerInterface* rasterizer_) {
     upload_state.BindRasterizer(rasterizer_);
 
-    execution_mask.fill(0);
+    execution_mask.reset();
     execution_mask[KEPLERMEMORY_REG_INDEX(exec)] = true;
     execution_mask[KEPLERMEMORY_REG_INDEX(data)] = true;
 }

src/video_core/engines/maxwell_3d.cpp

@@ -4,14 +4,8 @@
 // SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
-#include <algorithm>
 #include <cstring>
 #include <optional>
-
-#if defined(_MSC_VER) && !defined(__clang__)
-#include <intrin.h>
-#endif
-
 #include "common/assert.h"
 #include "common/bit_util.h"
 #include "common/scope_exit.h"
@@ -28,16 +22,6 @@
 
 namespace Tegra::Engines {
 
-namespace {
-inline void PrefetchLine(const void* addr) {
-#if defined(_MSC_VER) && !defined(__clang__)
-    _mm_prefetch(static_cast<const char*>(addr), _MM_HINT_T0);
-#else
-    __builtin_prefetch(addr, 0, 1);
-#endif
-}
-} // namespace
-
 /// First register id that is actually a Macro call.
 constexpr u32 MacroRegistersStart = 0xE00;
 
@@ -53,10 +37,9 @@ Maxwell3D::Maxwell3D(MemoryManager& memory_manager_)
 {
     dirty.flags.flip();
     InitializeRegisterDefaults();
-    execution_mask.fill(0);
-    for (size_t i = 0; i < EXECUTION_MASK_TABLE_SIZE; i++)
+    execution_mask.reset();
+    for (size_t i = 0; i < execution_mask.size(); i++)
         execution_mask[i] = IsMethodExecutable(u32(i));
-    execution_mask_default = true;
 }
 
 Maxwell3D::~Maxwell3D() = default;
@@ -299,44 +282,18 @@ u32 Maxwell3D::ProcessShadowRam(u32 method, u32 argument) {
 }
 
 void Maxwell3D::ConsumeSinkImpl(Core::System& system) {
-    std::stable_sort(method_sink.begin(), method_sink.end(),
-                     [](const auto& a, const auto& b) { return a.first < b.first; });
-
-    const auto sink_size = method_sink.size();
     const auto control = shadow_state.shadow_ram_control;
     if (control == Regs::ShadowRamControl::Track || control == Regs::ShadowRamControl::TrackWithFilter) {
-        for (size_t i = 0; i < sink_size; ++i) {
-            const auto [method, value] = method_sink[i];
-            if (i + 1 < sink_size) {
-                const u32 next = method_sink[i + 1].first;
-                PrefetchLine(&regs.reg_array[next]);
-                PrefetchLine(&shadow_state.reg_array[next]);
-                PrefetchLine(&dirty.tables[0][next]);
-            }
+        for (auto [method, value] : method_sink) {
             shadow_state.reg_array[method] = value;
             ProcessDirtyRegisters(method, value);
         }
     } else if (control == Regs::ShadowRamControl::Replay) {
-        for (size_t i = 0; i < sink_size; ++i) {
-            const auto [method, value] = method_sink[i];
-            if (i + 1 < sink_size) {
-                const u32 next = method_sink[i + 1].first;
-                PrefetchLine(&regs.reg_array[next]);
-                PrefetchLine(&shadow_state.reg_array[next]);
-                PrefetchLine(&dirty.tables[0][next]);
-            }
+        for (auto [method, value] : method_sink)
             ProcessDirtyRegisters(method, shadow_state.reg_array[method]);
-        }
     } else {
-        for (size_t i = 0; i < sink_size; ++i) {
-            const auto [method, value] = method_sink[i];
-            if (i + 1 < sink_size) {
-                const u32 next = method_sink[i + 1].first;
-                PrefetchLine(&regs.reg_array[next]);
-                PrefetchLine(&dirty.tables[0][next]);
-            }
+        for (auto [method, value] : method_sink)
             ProcessDirtyRegisters(method, value);
-        }
     }
     method_sink.clear();
 }

src/video_core/engines/maxwell_dma.cpp

@@ -24,7 +24,7 @@ using namespace Texture;
 MaxwellDMA::MaxwellDMA(MemoryManager& memory_manager_)
     : memory_manager{memory_manager_}
 {
-    execution_mask.fill(0);
+    execution_mask.reset();
     execution_mask[offsetof(Regs, launch_dma) / sizeof(u32)] = true;
 }
 

src/video_core/host_shaders/block_linear_unswizzle_2d.comp

@@ -11,9 +11,8 @@
 #define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
 #define END_PUSH_CONSTANTS };
 #define UNIFORM(n)
-#define BINDING_SWIZZLE_BUFFER 0
-#define BINDING_INPUT_BUFFER 1
-#define BINDING_OUTPUT_IMAGE 2
+#define BINDING_INPUT_BUFFER 0
+#define BINDING_OUTPUT_IMAGE 1
 
 #else // ^^^ Vulkan ^^^ // vvv OpenGL vvv
 
@@ -26,8 +25,7 @@
 #define BEGIN_PUSH_CONSTANTS
 #define END_PUSH_CONSTANTS
 #define UNIFORM(n) layout (location = n) uniform
-#define BINDING_SWIZZLE_BUFFER 0
-#define BINDING_INPUT_BUFFER 1
+#define BINDING_INPUT_BUFFER 0
 #define BINDING_OUTPUT_IMAGE 0
 
 #endif
@@ -43,10 +41,6 @@ UNIFORM(6) uint block_height;
 UNIFORM(7) uint block_height_mask;
 END_PUSH_CONSTANTS
 
-layout(binding = BINDING_SWIZZLE_BUFFER, std430) readonly buffer SwizzleTable {
-    uint swizzle_table[];
-};
-
 #if HAS_EXTENDED_TYPES
 layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
 layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
@@ -71,9 +65,19 @@ const uint GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHI
 
 const uvec2 SWIZZLE_MASK = uvec2(GOB_SIZE_X - 1, GOB_SIZE_Y - 1);
 
+uint SwizzleTable(uint pos) {
+    const uint t[8] = uint[](
+        0x12100200, 0x13110301, 0x16140604, 0x17150705,
+        0x1a180a08, 0x1b190b09, 0x1e1c0e0c, 0x1f1d0f0d
+    );
+    const uint i = pos >> 4;
+    const uint h = (t[i / 4] >> ((i % 4) * 8)) & 0xff;
+    return (h << 4) | (pos & 0xf);
+}
+
 uint SwizzleOffset(uvec2 pos) {
     pos = pos & SWIZZLE_MASK;
-    return swizzle_table[pos.y * 64 + pos.x];
+    return SwizzleTable(pos.y * 64 + pos.x);
 }
 
 uvec4 ReadTexel(uint offset) {

src/video_core/host_shaders/block_linear_unswizzle_3d.comp

@@ -11,9 +11,8 @@
 #define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
 #define END_PUSH_CONSTANTS };
 #define UNIFORM(n)
-#define BINDING_SWIZZLE_BUFFER 0
-#define BINDING_INPUT_BUFFER 1
-#define BINDING_OUTPUT_IMAGE 2
+#define BINDING_INPUT_BUFFER 0
+#define BINDING_OUTPUT_IMAGE 1
 
 #else // ^^^ Vulkan ^^^ // vvv OpenGL vvv
 
@@ -26,8 +25,7 @@
 #define BEGIN_PUSH_CONSTANTS
 #define END_PUSH_CONSTANTS
 #define UNIFORM(n) layout (location = n) uniform
-#define BINDING_SWIZZLE_BUFFER 0
-#define BINDING_INPUT_BUFFER 1
+#define BINDING_INPUT_BUFFER 0
 #define BINDING_OUTPUT_IMAGE 0
 
 #endif
@@ -45,10 +43,6 @@ UNIFORM(8) uint block_depth;
 UNIFORM(9) uint block_depth_mask;
 END_PUSH_CONSTANTS
 
-layout(binding = BINDING_SWIZZLE_BUFFER, std430) readonly buffer SwizzleTable {
-    uint swizzle_table[];
-};
-
 #if HAS_EXTENDED_TYPES
 layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8 { uint8_t u8data[]; };
 layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16 { uint16_t u16data[]; };
@@ -73,9 +67,19 @@ const uint GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHI
 
 const uvec2 SWIZZLE_MASK = uvec2(GOB_SIZE_X - 1, GOB_SIZE_Y - 1);
 
+uint SwizzleTable(uint pos) {
+    const uint t[8] = uint[](
+        0x12100200, 0x13110301, 0x16140604, 0x17150705,
+        0x1a180a08, 0x1b190b09, 0x1e1c0e0c, 0x1f1d0f0d
+    );
+    const uint i = pos >> 4;
+    const uint h = (t[i / 4] >> ((i % 4) * 8)) & 0xff;
+    return (h << 4) | (pos & 0xf);
+}
+
 uint SwizzleOffset(uvec2 pos) {
     pos = pos & SWIZZLE_MASK;
-    return swizzle_table[pos.y * 64 + pos.x];
+    return SwizzleTable(pos.y * 64 + pos.x);
 }
 
 uvec4 ReadTexel(uint offset) {

src/video_core/host_shaders/block_linear_unswizzle_3d_bcn.comp

@@ -10,9 +10,8 @@
     #define BEGIN_PUSH_CONSTANTS layout(push_constant) uniform PushConstants {
     #define END_PUSH_CONSTANTS };
     #define UNIFORM(n)
-    #define BINDING_SWIZZLE_BUFFER 0
-    #define BINDING_INPUT_BUFFER   1
-    #define BINDING_OUTPUT_BUFFER  2
+    #define BINDING_INPUT_BUFFER   0
+    #define BINDING_OUTPUT_BUFFER  1
 #else
     #extension GL_NV_gpu_shader5 : enable
     #ifdef GL_NV_gpu_shader5
@@ -23,7 +22,6 @@
     #define BEGIN_PUSH_CONSTANTS
     #define END_PUSH_CONSTANTS
     #define UNIFORM(n) layout(location = n) uniform
-    #define BINDING_SWIZZLE_BUFFER 0
     #define BINDING_INPUT_BUFFER   1
     #define BINDING_OUTPUT_BUFFER  0
 #endif
@@ -66,13 +64,9 @@ END_PUSH_CONSTANTS
 #endif
 
 // --- Buffers ---
-layout(binding = BINDING_SWIZZLE_BUFFER, std430) readonly buffer SwizzleTable {
-    uint swizzle_table[];
-};
-
 #if HAS_EXTENDED_TYPES
-    layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8   { uint8_t  u8data[];  };
-    layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16  { uint16_t u16data[]; };
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU8   { uint8_t  u8data[];  };
+layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU16  { uint16_t u16data[]; };
 #endif
 layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU32  { uint   u32data[];  };
 layout(binding = BINDING_INPUT_BUFFER, std430) buffer InputBufferU64  { uvec2  u64data[];  };
@@ -96,10 +90,20 @@ const uint GOB_SIZE_Z_SHIFT = 0;
 const uint GOB_SIZE_SHIFT   = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_SHIFT;
 const uvec2 SWIZZLE_MASK = uvec2(GOB_SIZE_X - 1u, GOB_SIZE_Y - 1u);
 
+uint SwizzleTable(uint pos) {
+    const uint t[8] = uint[](
+        0x12100200, 0x13110301, 0x16140604, 0x17150705,
+        0x1a180a08, 0x1b190b09, 0x1e1c0e0c, 0x1f1d0f0d
+    );
+    const uint i = pos >> 4;
+    const uint h = (t[i / 4] >> ((i % 4) * 8)) & 0xff;
+    return (h << 4) | (pos & 0xf);
+}
+
 // --- Helpers ---
 uint SwizzleOffset(uvec2 pos) {
     pos &= SWIZZLE_MASK;
-    return swizzle_table[pos.y * 64u + pos.x];
+    return SwizzleTable(pos.y * 64u + pos.x);
 }
 
 uvec4 ReadTexel(uint offset) {

src/video_core/query_cache/query_cache.h

@@ -287,7 +287,6 @@ void QueryCacheBase<Traits>::CounterReport(GPUVAddr addr, QueryType counter_type
             u32 value = static_cast<u32>(query_base->value);
             std::memcpy(pointer, &value, sizeof(value));
         }
-        query_base->flags |= QueryFlagBits::IsGuestSynced;
         if (!is_synced) [[likely]] {
             impl->pending_unregister.push_back(query_location);
         }
@@ -570,12 +569,10 @@ bool QueryCacheBase<Traits>::SemiFlushQueryDirty(QueryCacheBase<Traits>::QueryLo
         auto* ptr = impl->device_memory.template GetPointer<u8>(query_base->guest_address);
         if (True(query_base->flags & QueryFlagBits::HasTimestamp)) {
             std::memcpy(ptr, &query_base->value, sizeof(query_base->value));
-            query_base->flags |= QueryFlagBits::IsGuestSynced;
             return false;
         }
         u32 value_l = static_cast<u32>(query_base->value);
         std::memcpy(ptr, &value_l, sizeof(value_l));
-        query_base->flags |= QueryFlagBits::IsGuestSynced;
         return false;
     }
     return True(query_base->flags & QueryFlagBits::IsHostManaged) &&

src/video_core/renderer_opengl/gl_shader_cache.cpp

@@ -245,16 +245,31 @@ ShaderCache::ShaderCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
               std::min<u32>(device.GetMaxUserClipDistances(), Maxwell::Regs::NumClipDistances),
       },
       host_info{
-          .support_float64 = true,
-          .support_float16 = false,
-          .support_int64 = device.HasShaderInt64(),
-          .needs_demote_reorder = device.IsAmd(),
-          .support_snorm_render_buffer = false,
-          .support_viewport_index_layer = device.HasVertexViewportLayer(),
-          .min_ssbo_alignment = static_cast<u32>(device.GetShaderStorageBufferAlignment()),
-          .support_geometry_shader_passthrough = device.HasGeometryShaderPassthrough(),
-          .support_conditional_barrier = device.SupportsConditionalBarriers(),
+        .min_ssbo_alignment = static_cast<u32>(device.GetShaderStorageBufferAlignment()),
+        .max_per_stage_descriptor_sampled_images =
+            Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_per_stage_resources = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_samplers = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_uniform_buffers = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_uniform_buffers_dynamic =
+            Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_storage_buffers = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_storage_buffers_dynamic =
+            Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_sampled_images = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_storage_images = Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .max_descriptor_set_input_attachements =
+            Shader::HostTranslateInfo::DEFAULT_DESCRIPTOR_LIMIT,
+        .support_float64 = true,
+        .support_float16 = false,
+        .support_int64 = device.HasShaderInt64(),
+        .needs_demote_reorder = device.IsAmd(),
+        .support_snorm_render_buffer = false,
+        .support_viewport_index_layer = device.HasVertexViewportLayer(),
+        .support_geometry_shader_passthrough = device.HasGeometryShaderPassthrough(),
+        .support_conditional_barrier = device.SupportsConditionalBarriers(),
       } {
+    host_info.ApplyDescriptorLimitPolicy();
     if (use_asynchronous_shaders) {
         workers = CreateWorkers();
     }

src/video_core/renderer_opengl/util_shaders.cpp

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -56,10 +59,8 @@ UtilShaders::UtilShaders(ProgramManager& program_manager_)
       copy_bc4_program(MakeProgram(OPENGL_COPY_BC4_COMP)),
       convert_s8d24_program(MakeProgram(OPENGL_CONVERT_S8D24_COMP)),
       convert_ms_to_nonms_program(MakeProgram(CONVERT_MSAA_TO_NON_MSAA_COMP)),
-      convert_nonms_to_ms_program(MakeProgram(CONVERT_NON_MSAA_TO_MSAA_COMP)) {
-    const auto swizzle_table = Tegra::Texture::MakeSwizzleTable();
-    swizzle_table_buffer.Create();
-    glNamedBufferStorage(swizzle_table_buffer.handle, sizeof(swizzle_table), &swizzle_table, 0);
+      convert_nonms_to_ms_program(MakeProgram(CONVERT_NON_MSAA_TO_MSAA_COMP))
+{
 }
 
 UtilShaders::~UtilShaders() = default;
@@ -116,13 +117,11 @@ void UtilShaders::ASTCDecode(Image& image, const StagingBufferMap& map,
 void UtilShaders::BlockLinearUpload2D(Image& image, const StagingBufferMap& map,
                                       std::span<const SwizzleParameters> swizzles) {
     static constexpr Extent3D WORKGROUP_SIZE{32, 32, 1};
-    static constexpr GLuint BINDING_SWIZZLE_BUFFER = 0;
-    static constexpr GLuint BINDING_INPUT_BUFFER = 1;
+    static constexpr GLuint BINDING_INPUT_BUFFER = 0;
     static constexpr GLuint BINDING_OUTPUT_IMAGE = 0;
 
     program_manager.BindComputeProgram(block_linear_unswizzle_2d_program.handle);
     glFlushMappedNamedBufferRange(map.buffer, map.offset, image.guest_size_bytes);
-    glBindBufferBase(GL_SHADER_STORAGE_BUFFER, BINDING_SWIZZLE_BUFFER, swizzle_table_buffer.handle);
 
     const GLenum store_format = StoreFormat(BytesPerBlock(image.info.format));
     for (const SwizzleParameters& swizzle : swizzles) {
@@ -153,14 +152,11 @@ void UtilShaders::BlockLinearUpload2D(Image& image, const StagingBufferMap& map,
 void UtilShaders::BlockLinearUpload3D(Image& image, const StagingBufferMap& map,
                                       std::span<const SwizzleParameters> swizzles) {
     static constexpr Extent3D WORKGROUP_SIZE{16, 8, 8};
-
-    static constexpr GLuint BINDING_SWIZZLE_BUFFER = 0;
-    static constexpr GLuint BINDING_INPUT_BUFFER = 1;
+    static constexpr GLuint BINDING_INPUT_BUFFER = 0;
     static constexpr GLuint BINDING_OUTPUT_IMAGE = 0;
 
     glFlushMappedNamedBufferRange(map.buffer, map.offset, image.guest_size_bytes);
     program_manager.BindComputeProgram(block_linear_unswizzle_3d_program.handle);
-    glBindBufferBase(GL_SHADER_STORAGE_BUFFER, BINDING_SWIZZLE_BUFFER, swizzle_table_buffer.handle);
 
     const GLenum store_format = StoreFormat(BytesPerBlock(image.info.format));
     for (const SwizzleParameters& swizzle : swizzles) {

src/video_core/renderer_opengl/util_shaders.h

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -45,9 +48,6 @@ public:
 
 private:
     ProgramManager& program_manager;
-
-    OGLBuffer swizzle_table_buffer;
-
     OGLProgram astc_decoder_program;
     OGLProgram block_linear_unswizzle_2d_program;
     OGLProgram block_linear_unswizzle_3d_program;

src/video_core/renderer_vulkan/fixed_pipeline_state.cpp

@@ -62,7 +62,6 @@ void FixedPipelineState::Refresh(Tegra::Engines::Maxwell3D& maxwell3d, DynamicFe
     extended_dynamic_state_2_logic_op.Assign(features.has_extended_dynamic_state_2_logic_op ? 1 : 0);
     extended_dynamic_state_3_blend.Assign(features.has_extended_dynamic_state_3_blend ? 1 : 0);
     extended_dynamic_state_3_enables.Assign(features.has_extended_dynamic_state_3_enables ? 1 : 0);
-    dynamic_state3_depth_clamp_enable.Assign(features.has_dynamic_state3_depth_clamp_enable ? 1 : 0);
     dynamic_vertex_input.Assign(features.has_dynamic_vertex_input ? 1 : 0);
     xfb_enabled.Assign(regs.transform_feedback_enabled != 0);
     ndc_minus_one_to_one.Assign(regs.depth_mode == Maxwell::DepthMode::MinusOneToOne ? 1 : 0);

src/video_core/renderer_vulkan/fixed_pipeline_state.h

@@ -208,7 +208,6 @@ struct FixedPipelineState {
         BitField<12, 2, u32> tessellation_spacing;
         BitField<14, 1, u32> tessellation_clockwise;
         BitField<15, 5, u32> patch_control_points_minus_one;
-        BitField<20, 1, u32> dynamic_state3_depth_clamp_enable;
 
         BitField<24, 4, Maxwell::PrimitiveTopology> topology;
         BitField<28, 4, Tegra::Texture::MsaaMode> msaa_mode;

src/video_core/renderer_vulkan/pipeline_helper.h

@@ -22,6 +22,15 @@ namespace Vulkan {
 
 using Shader::Backend::SPIRV::NUM_TEXTURE_AND_IMAGE_SCALING_WORDS;
 
+[[nodiscard]] inline u32 NumDescriptorEntries(const Shader::Info& info) {
+    return Shader::NumDescriptors(info.constant_buffer_descriptors) +
+           Shader::NumDescriptors(info.storage_buffers_descriptors) +
+           Shader::NumDescriptors(info.texture_buffer_descriptors) +
+           Shader::NumDescriptors(info.image_buffer_descriptors) +
+           Shader::NumDescriptors(info.texture_descriptors) +
+           Shader::NumDescriptors(info.image_descriptors);
+}
+
 class DescriptorLayoutBuilder {
 public:
     DescriptorLayoutBuilder(const Device& device_) : device{&device_} {}
@@ -194,7 +203,11 @@ inline void PushImageDescriptors(TextureCache& texture_cache,
             const VideoCommon::ImageViewId image_view_id{(views++)->id};
             const VideoCommon::SamplerId sampler_id{*(samplers++)};
             ImageView& image_view{texture_cache.GetImageView(image_view_id)};
-            const VkImageView vk_image_view{image_view.Handle(desc.type)};
+            VkImageView vk_image_view{image_view.Handle(desc.type)};
+            if (vk_image_view == VK_NULL_HANDLE) {
+                const VkImageView null_image_view{texture_cache.GetImageView(VideoCommon::NULL_IMAGE_VIEW_ID).Handle(desc.type)};
+                if (null_image_view != VK_NULL_HANDLE) vk_image_view = null_image_view;
+            }
             const Sampler& sampler{texture_cache.GetSampler(sampler_id)};
             const bool use_fallback_sampler{sampler.HasAddedAnisotropy() &&
                                             !image_view.SupportsAnisotropy()};

src/video_core/renderer_vulkan/renderer_vulkan.cpp

@@ -169,6 +169,7 @@ try
 }
 
 RendererVulkan::~RendererVulkan() {
+    scheduler.WaitWorker();
     scheduler.RegisterOnSubmit([] {});
     void(device.GetLogical().WaitIdle());
 }

src/video_core/renderer_vulkan/vk_compute_pass.cpp

@@ -25,6 +25,9 @@
 #include "video_core/host_shaders/vulkan_quad_indexed_comp_spv.h"
 #include "video_core/host_shaders/vulkan_uint8_comp_spv.h"
 #include "video_core/host_shaders/block_linear_unswizzle_3d_bcn_comp_spv.h"
+#include "video_core/host_shaders/block_linear_unswizzle_2d_comp_spv.h"
+#include "video_core/host_shaders/block_linear_unswizzle_3d_comp_spv.h"
+#include "video_core/host_shaders/pitch_unswizzle_comp_spv.h"
 #include "video_core/renderer_vulkan/vk_compute_pass.h"
 #include "video_core/renderer_vulkan/vk_descriptor_pool.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"
@@ -232,6 +235,26 @@ struct QueriesPrefixScanPushConstants {
 struct ConditionalRenderingResolvePushConstants {
     u32 compare_to_zero;
 };
+
+struct BlockLinear3DImagePushConstants {
+    alignas(16) std::array<u32, 3> origin;
+    alignas(16) std::array<s32, 3> destination;
+    u32 bytes_per_block_log2;
+    u32 slice_size;
+    u32 block_size;
+    u32 x_shift;
+    u32 block_height;
+    u32 block_height_mask;
+    u32 block_depth;
+    u32 block_depth_mask;
+};
+
+struct PitchUnswizzlePushConstants {
+    std::array<u32, 2> origin;
+    std::array<s32, 2> destination;
+    u32 bytes_per_block;
+    u32 pitch;
+};
 } // Anonymous namespace
 
 ComputePass::ComputePass(const Device& device_, Scheduler& scheduler, DescriptorPool& descriptor_pool,
@@ -326,7 +349,7 @@ std::pair<VkBuffer, VkDeviceSize> Uint8Pass::Assemble(u32 num_vertices, VkBuffer
     const u32 staging_size = static_cast<u32>(num_vertices * sizeof(u16));
     const auto staging = staging_buffer_pool.Request(staging_size, MemoryUsage::DeviceLocal);
 
-    compute_pass_descriptor_queue.Acquire();
+    compute_pass_descriptor_queue.Acquire(scheduler, 2);
     compute_pass_descriptor_queue.AddBuffer(src_buffer, src_offset, num_vertices);
     compute_pass_descriptor_queue.AddBuffer(staging.buffer, staging.offset, staging_size);
     const void* const descriptor_data{compute_pass_descriptor_queue.UpdateData()};
@@ -384,7 +407,7 @@ std::pair<VkBuffer, VkDeviceSize> QuadIndexedPass::Assemble(
     const std::size_t staging_size = num_tri_vertices * sizeof(u32);
     const auto staging = staging_buffer_pool.Request(staging_size, MemoryUsage::DeviceLocal);
 
-    compute_pass_descriptor_queue.Acquire();
+    compute_pass_descriptor_queue.Acquire(scheduler, 2);
     compute_pass_descriptor_queue.AddBuffer(src_buffer, src_offset, input_size);
     compute_pass_descriptor_queue.AddBuffer(staging.buffer, staging.offset, staging_size);
     const void* const descriptor_data{compute_pass_descriptor_queue.UpdateData()};
@@ -429,7 +452,7 @@ void ConditionalRenderingResolvePass::Resolve(VkBuffer dst_buffer, VkBuffer src_
     }
     const size_t compare_size = compare_to_zero ? 8 : 24;
 
-    compute_pass_descriptor_queue.Acquire();
+    compute_pass_descriptor_queue.Acquire(scheduler, 2);
     compute_pass_descriptor_queue.AddBuffer(src_buffer, src_offset, compare_size);
     compute_pass_descriptor_queue.AddBuffer(dst_buffer, 0, sizeof(u32));
     const void* const descriptor_data{compute_pass_descriptor_queue.UpdateData()};
@@ -498,7 +521,7 @@ void QueriesPrefixScanPass::Run(VkBuffer accumulation_buffer, VkBuffer dst_buffe
         static constexpr size_t DISPATCH_SIZE = 2048U;
         size_t runs_to_do = std::min<size_t>(current_runs, DISPATCH_SIZE);
         current_runs -= runs_to_do;
-        compute_pass_descriptor_queue.Acquire();
+        compute_pass_descriptor_queue.Acquire(scheduler, 3);
         compute_pass_descriptor_queue.AddBuffer(src_buffer, 0, number_of_sums * sizeof(u64));
         compute_pass_descriptor_queue.AddBuffer(dst_buffer, 0, number_of_sums * sizeof(u64));
         compute_pass_descriptor_queue.AddBuffer(accumulation_buffer, 0, sizeof(u64));
@@ -600,7 +623,7 @@ void ASTCDecoderPass::Assemble(Image& image, const StagingBufferRef& map,
         const u32 num_dispatches_y = Common::DivCeil(swizzle.num_tiles.height, 8U);
         const u32 num_dispatches_z = image.info.resources.layers;
 
-        compute_pass_descriptor_queue.Acquire();
+        compute_pass_descriptor_queue.Acquire(scheduler, 2);
         compute_pass_descriptor_queue.AddBuffer(map.buffer, input_offset,
                                                 image.guest_size_bytes - swizzle.buffer_offset);
         compute_pass_descriptor_queue.AddImage(image.StorageImageView(swizzle.level));
@@ -653,71 +676,311 @@ void ASTCDecoderPass::Assemble(Image& image, const StagingBufferRef& map,
     scheduler.Finish();
 }
 
-constexpr u32 BL3D_BINDING_SWIZZLE_TABLE = 0;
-constexpr u32 BL3D_BINDING_INPUT_BUFFER  = 1;
-constexpr u32 BL3D_BINDING_OUTPUT_BUFFER = 2;
+BlockLinearUnswizzleImage2DPass::BlockLinearUnswizzleImage2DPass(
+    const Device& device_, Scheduler& scheduler_, DescriptorPool& descriptor_pool_,
+    StagingBufferPool& staging_buffer_pool_,
+    ComputePassDescriptorQueue& compute_pass_descriptor_queue_)
+    : ComputePass(device_, scheduler_, descriptor_pool_, ASTC_DESCRIPTOR_SET_BINDINGS,
+                  ASTC_PASS_DESCRIPTOR_UPDATE_TEMPLATE_ENTRY, ASTC_BANK_INFO,
+                  COMPUTE_PUSH_CONSTANT_RANGE<sizeof(BlockLinearSwizzle2DParams)>,
+                  BLOCK_LINEAR_UNSWIZZLE_2D_COMP_SPV),
+      scheduler{scheduler_}, staging_buffer_pool{staging_buffer_pool_},
+      compute_pass_descriptor_queue{compute_pass_descriptor_queue_} {}
 
-constexpr std::array<VkDescriptorSetLayoutBinding, 3> BL3D_DESCRIPTOR_SET_BINDINGS{{
-    {
-        .binding = BL3D_BINDING_SWIZZLE_TABLE,
-        .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER, // swizzle_table[]
-        .descriptorCount = 1,
-        .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
-        .pImmutableSamplers = nullptr,
-    },
-    {
-        .binding = BL3D_BINDING_INPUT_BUFFER,
-        .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER, // block-linear input
-        .descriptorCount = 1,
-        .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
-        .pImmutableSamplers = nullptr,
-    },
-    {
-        .binding = BL3D_BINDING_OUTPUT_BUFFER,
-        .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
-        .descriptorCount = 1,
-        .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
-        .pImmutableSamplers = nullptr,
-    },
-}};
+BlockLinearUnswizzleImage2DPass::~BlockLinearUnswizzleImage2DPass() = default;
 
-constexpr DescriptorBankInfo BL3D_BANK_INFO{
-    .uniform_buffers = 0,
-    .storage_buffers = 3,
-    .texture_buffers = 0,
-    .image_buffers = 0,
-    .textures = 0,
-    .images = 0,
-    .score = 3,
-};
+void BlockLinearUnswizzleImage2DPass::Unswizzle(
+    Image& image, const StagingBufferRef& map,
+    std::span<const VideoCommon::SwizzleParameters> swizzles) {
+    using namespace VideoCommon::Accelerated;
+    scheduler.RequestOutsideRenderPassOperationContext();
+    const VkPipeline vk_pipeline = *pipeline;
+    const VkImageAspectFlags aspect_mask = image.AspectMask();
+    const VkImage vk_image = image.Handle();
+    const bool is_initialized = image.ExchangeInitialization();
+    scheduler.Record([vk_pipeline, vk_image, aspect_mask,
+                      is_initialized](vk::CommandBuffer cmdbuf) {
+        const VkImageMemoryBarrier image_barrier{
+            .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+            .pNext = nullptr,
+            .srcAccessMask = static_cast<VkAccessFlags>(is_initialized ? VK_ACCESS_SHADER_WRITE_BIT
+                                                                       : VK_ACCESS_NONE),
+            .dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT,
+            .oldLayout = is_initialized ? VK_IMAGE_LAYOUT_GENERAL : VK_IMAGE_LAYOUT_UNDEFINED,
+            .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+            .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .image = vk_image,
+            .subresourceRange{
+                .aspectMask = aspect_mask,
+                .baseMipLevel = 0,
+                .levelCount = VK_REMAINING_MIP_LEVELS,
+                .baseArrayLayer = 0,
+                .layerCount = VK_REMAINING_ARRAY_LAYERS,
+            },
+        };
+        cmdbuf.PipelineBarrier(is_initialized ? vk::PIPELINE_STAGE_GRAPHICS_COMPUTE_TRANSFER
+                              : VkPipelineStageFlags(VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT),
+                       VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0, image_barrier);
+        cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, vk_pipeline);
+    });
+    for (const VideoCommon::SwizzleParameters& swizzle : swizzles) {
+        const size_t input_offset = swizzle.buffer_offset + map.offset;
+        const u32 num_dispatches_x = Common::DivCeil(swizzle.num_tiles.width, 32U);
+        const u32 num_dispatches_y = Common::DivCeil(swizzle.num_tiles.height, 32U);
+        const u32 num_dispatches_z = image.info.resources.layers;
 
-constexpr std::array<VkDescriptorUpdateTemplateEntry, 3>
-    BL3D_DESCRIPTOR_UPDATE_TEMPLATE_ENTRY{{
-        {
-            .dstBinding = BL3D_BINDING_SWIZZLE_TABLE,
-            .dstArrayElement = 0,
-            .descriptorCount = 1,
-            .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
-            .offset = BL3D_BINDING_SWIZZLE_TABLE * sizeof(DescriptorUpdateEntry),
-            .stride = sizeof(DescriptorUpdateEntry),
-        },
-        {
-            .dstBinding = BL3D_BINDING_INPUT_BUFFER,
-            .dstArrayElement = 0,
-            .descriptorCount = 1,
-            .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
-            .offset = BL3D_BINDING_INPUT_BUFFER * sizeof(DescriptorUpdateEntry),
-            .stride = sizeof(DescriptorUpdateEntry),
-        },
-        {
-            .dstBinding = BL3D_BINDING_OUTPUT_BUFFER,
-            .dstArrayElement = 0,
-            .descriptorCount = 1,
-            .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
-            .offset = BL3D_BINDING_OUTPUT_BUFFER * sizeof(DescriptorUpdateEntry),
-            .stride = sizeof(DescriptorUpdateEntry),
-        }
-    }};
+        compute_pass_descriptor_queue.Acquire(scheduler, 2);
+        compute_pass_descriptor_queue.AddBuffer(map.buffer, input_offset,
+                                                image.guest_size_bytes - swizzle.buffer_offset);
+        compute_pass_descriptor_queue.AddImage(image.StorageImageView(swizzle.level));
+        const void* const descriptor_data{compute_pass_descriptor_queue.UpdateData()};
+
+        const auto params = MakeBlockLinearSwizzle2DParams(swizzle, image.info);
+        scheduler.Record([this, num_dispatches_x, num_dispatches_y, num_dispatches_z, params,
+                          descriptor_data](vk::CommandBuffer cmdbuf) {
+            const VkDescriptorSet set = descriptor_allocator.Commit();
+            device.GetLogical().UpdateDescriptorSet(set, *descriptor_template, descriptor_data);
+            cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, *layout, 0, set, {});
+            cmdbuf.PushConstants(*layout, VK_SHADER_STAGE_COMPUTE_BIT, params);
+            cmdbuf.Dispatch(num_dispatches_x, num_dispatches_y, num_dispatches_z);
+        });
+    }
+    scheduler.Record([vk_image, aspect_mask](vk::CommandBuffer cmdbuf) {
+        const VkImageMemoryBarrier image_barrier{
+            .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+            .pNext = nullptr,
+            .srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT,
+            .dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT |
+                             VK_ACCESS_TRANSFER_READ_BIT | VK_ACCESS_TRANSFER_WRITE_BIT,
+            .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+            .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+            .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .image = vk_image,
+            .subresourceRange{
+                .aspectMask = aspect_mask,
+                .baseMipLevel = 0,
+                .levelCount = VK_REMAINING_MIP_LEVELS,
+                .baseArrayLayer = 0,
+                .layerCount = VK_REMAINING_ARRAY_LAYERS,
+            },
+        };
+        cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
+                       vk::PIPELINE_STAGE_GRAPHICS_COMPUTE_TRANSFER, 0, image_barrier);
+    });
+    scheduler.Finish();
+}
+
+BlockLinearUnswizzleImage3DPass::BlockLinearUnswizzleImage3DPass(
+    const Device& device_, Scheduler& scheduler_, DescriptorPool& descriptor_pool_,
+    StagingBufferPool& staging_buffer_pool_,
+    ComputePassDescriptorQueue& compute_pass_descriptor_queue_)
+    : ComputePass(device_, scheduler_, descriptor_pool_, ASTC_DESCRIPTOR_SET_BINDINGS,
+                  ASTC_PASS_DESCRIPTOR_UPDATE_TEMPLATE_ENTRY, ASTC_BANK_INFO,
+                  COMPUTE_PUSH_CONSTANT_RANGE<sizeof(BlockLinear3DImagePushConstants)>,
+                  BLOCK_LINEAR_UNSWIZZLE_3D_COMP_SPV),
+      scheduler{scheduler_}, staging_buffer_pool{staging_buffer_pool_},
+      compute_pass_descriptor_queue{compute_pass_descriptor_queue_} {}
+
+BlockLinearUnswizzleImage3DPass::~BlockLinearUnswizzleImage3DPass() = default;
+
+void BlockLinearUnswizzleImage3DPass::Unswizzle(
+    Image& image, const StagingBufferRef& map,
+    std::span<const VideoCommon::SwizzleParameters> swizzles) {
+    using namespace VideoCommon::Accelerated;
+    scheduler.RequestOutsideRenderPassOperationContext();
+    const VkPipeline vk_pipeline = *pipeline;
+    const VkImageAspectFlags aspect_mask = image.AspectMask();
+    const VkImage vk_image = image.Handle();
+    const bool is_initialized = image.ExchangeInitialization();
+    scheduler.Record([vk_pipeline, vk_image, aspect_mask,
+                      is_initialized](vk::CommandBuffer cmdbuf) {
+        const VkImageMemoryBarrier image_barrier{
+            .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+            .pNext = nullptr,
+            .srcAccessMask = static_cast<VkAccessFlags>(is_initialized ? VK_ACCESS_SHADER_WRITE_BIT
+                                                                       : VK_ACCESS_NONE),
+            .dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT,
+            .oldLayout = is_initialized ? VK_IMAGE_LAYOUT_GENERAL : VK_IMAGE_LAYOUT_UNDEFINED,
+            .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+            .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .image = vk_image,
+            .subresourceRange{
+                .aspectMask = aspect_mask,
+                .baseMipLevel = 0,
+                .levelCount = VK_REMAINING_MIP_LEVELS,
+                .baseArrayLayer = 0,
+                .layerCount = VK_REMAINING_ARRAY_LAYERS,
+            },
+        };
+        cmdbuf.PipelineBarrier(is_initialized ? vk::PIPELINE_STAGE_GRAPHICS_COMPUTE_TRANSFER
+                              : VkPipelineStageFlags(VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT),
+                       VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0, image_barrier);
+        cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, vk_pipeline);
+    });
+    for (const VideoCommon::SwizzleParameters& swizzle : swizzles) {
+        const size_t input_offset = swizzle.buffer_offset + map.offset;
+        const u32 num_dispatches_x = Common::DivCeil(swizzle.num_tiles.width, 16U);
+        const u32 num_dispatches_y = Common::DivCeil(swizzle.num_tiles.height, 8U);
+        const u32 num_dispatches_z = Common::DivCeil(swizzle.num_tiles.depth, 8U);
+
+        compute_pass_descriptor_queue.Acquire(scheduler, 2);
+        compute_pass_descriptor_queue.AddBuffer(map.buffer, input_offset,
+                                                image.guest_size_bytes - swizzle.buffer_offset);
+        compute_pass_descriptor_queue.AddImage(image.StorageImageView(swizzle.level));
+        const void* const descriptor_data{compute_pass_descriptor_queue.UpdateData()};
+
+        const auto p = MakeBlockLinearSwizzle3DParams(swizzle, image.info);
+        const BlockLinear3DImagePushConstants params{
+            .origin = p.origin,
+            .destination = p.destination,
+            .bytes_per_block_log2 = p.bytes_per_block_log2,
+            .slice_size = p.slice_size,
+            .block_size = p.block_size,
+            .x_shift = p.x_shift,
+            .block_height = p.block_height,
+            .block_height_mask = p.block_height_mask,
+            .block_depth = p.block_depth,
+            .block_depth_mask = p.block_depth_mask,
+        };
+        scheduler.Record([this, num_dispatches_x, num_dispatches_y, num_dispatches_z, params,
+                          descriptor_data](vk::CommandBuffer cmdbuf) {
+            const VkDescriptorSet set = descriptor_allocator.Commit();
+            device.GetLogical().UpdateDescriptorSet(set, *descriptor_template, descriptor_data);
+            cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, *layout, 0, set, {});
+            cmdbuf.PushConstants(*layout, VK_SHADER_STAGE_COMPUTE_BIT, params);
+            cmdbuf.Dispatch(num_dispatches_x, num_dispatches_y, num_dispatches_z);
+        });
+    }
+    scheduler.Record([vk_image, aspect_mask](vk::CommandBuffer cmdbuf) {
+        const VkImageMemoryBarrier image_barrier{
+            .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+            .pNext = nullptr,
+            .srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT,
+            .dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT |
+                             VK_ACCESS_TRANSFER_READ_BIT | VK_ACCESS_TRANSFER_WRITE_BIT,
+            .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+            .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+            .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .image = vk_image,
+            .subresourceRange{
+                .aspectMask = aspect_mask,
+                .baseMipLevel = 0,
+                .levelCount = VK_REMAINING_MIP_LEVELS,
+                .baseArrayLayer = 0,
+                .layerCount = VK_REMAINING_ARRAY_LAYERS,
+            },
+        };
+        cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
+                       vk::PIPELINE_STAGE_GRAPHICS_COMPUTE_TRANSFER, 0, image_barrier);
+    });
+    scheduler.Finish();
+}
+
+PitchUnswizzlePass::PitchUnswizzlePass(const Device& device_, Scheduler& scheduler_,
+                                       DescriptorPool& descriptor_pool_,
+                                       StagingBufferPool& staging_buffer_pool_,
+                                       ComputePassDescriptorQueue& compute_pass_descriptor_queue_)
+    : ComputePass(device_, scheduler_, descriptor_pool_, ASTC_DESCRIPTOR_SET_BINDINGS,
+                  ASTC_PASS_DESCRIPTOR_UPDATE_TEMPLATE_ENTRY, ASTC_BANK_INFO,
+                  COMPUTE_PUSH_CONSTANT_RANGE<sizeof(PitchUnswizzlePushConstants)>,
+                  PITCH_UNSWIZZLE_COMP_SPV),
+      scheduler{scheduler_}, staging_buffer_pool{staging_buffer_pool_},
+      compute_pass_descriptor_queue{compute_pass_descriptor_queue_} {}
+
+PitchUnswizzlePass::~PitchUnswizzlePass() = default;
+
+void PitchUnswizzlePass::Unswizzle(Image& image, const StagingBufferRef& map,
+                                   std::span<const VideoCommon::SwizzleParameters> swizzles) {
+    scheduler.RequestOutsideRenderPassOperationContext();
+    const VkPipeline vk_pipeline = *pipeline;
+    const VkImageAspectFlags aspect_mask = image.AspectMask();
+    const VkImage vk_image = image.Handle();
+    const bool is_initialized = image.ExchangeInitialization();
+    scheduler.Record([vk_pipeline, vk_image, aspect_mask,
+                      is_initialized](vk::CommandBuffer cmdbuf) {
+        const VkImageMemoryBarrier image_barrier{
+            .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+            .pNext = nullptr,
+            .srcAccessMask = static_cast<VkAccessFlags>(is_initialized ? VK_ACCESS_SHADER_WRITE_BIT
+                                                                       : VK_ACCESS_NONE),
+            .dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT,
+            .oldLayout = is_initialized ? VK_IMAGE_LAYOUT_GENERAL : VK_IMAGE_LAYOUT_UNDEFINED,
+            .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+            .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .image = vk_image,
+            .subresourceRange{
+                .aspectMask = aspect_mask,
+                .baseMipLevel = 0,
+                .levelCount = VK_REMAINING_MIP_LEVELS,
+                .baseArrayLayer = 0,
+                .layerCount = VK_REMAINING_ARRAY_LAYERS,
+            },
+        };
+        cmdbuf.PipelineBarrier(is_initialized ? vk::PIPELINE_STAGE_GRAPHICS_COMPUTE_TRANSFER
+                              : VkPipelineStageFlags(VK_PIPELINE_STAGE_TOP_OF_PIPE_BIT),
+                       VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0, image_barrier);
+        cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, vk_pipeline);
+    });
+    const u32 bytes_per_block = VideoCore::Surface::BytesPerBlock(image.info.format);
+    for (const VideoCommon::SwizzleParameters& swizzle : swizzles) {
+        const size_t input_offset = swizzle.buffer_offset + map.offset;
+        const u32 num_dispatches_x = Common::DivCeil(swizzle.num_tiles.width, 32U);
+        const u32 num_dispatches_y = Common::DivCeil(swizzle.num_tiles.height, 32U);
+
+        compute_pass_descriptor_queue.Acquire(scheduler, 2);
+        compute_pass_descriptor_queue.AddBuffer(map.buffer, input_offset,
+                                                image.guest_size_bytes - swizzle.buffer_offset);
+        compute_pass_descriptor_queue.AddImage(image.StorageImageView(swizzle.level));
+        const void* const descriptor_data{compute_pass_descriptor_queue.UpdateData()};
+
+        const PitchUnswizzlePushConstants params{
+            .origin = {0, 0},
+            .destination = {0, 0},
+            .bytes_per_block = bytes_per_block,
+            .pitch = image.info.pitch,
+        };
+        scheduler.Record([this, num_dispatches_x, num_dispatches_y, params,
+                          descriptor_data](vk::CommandBuffer cmdbuf) {
+            const VkDescriptorSet set = descriptor_allocator.Commit();
+            device.GetLogical().UpdateDescriptorSet(set, *descriptor_template, descriptor_data);
+            cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, *layout, 0, set, {});
+            cmdbuf.PushConstants(*layout, VK_SHADER_STAGE_COMPUTE_BIT, params);
+            cmdbuf.Dispatch(num_dispatches_x, num_dispatches_y, 1);
+        });
+    }
+    scheduler.Record([vk_image, aspect_mask](vk::CommandBuffer cmdbuf) {
+        const VkImageMemoryBarrier image_barrier{
+            .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+            .pNext = nullptr,
+            .srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT,
+            .dstAccessMask = VK_ACCESS_SHADER_READ_BIT | VK_ACCESS_SHADER_WRITE_BIT |
+                             VK_ACCESS_TRANSFER_READ_BIT | VK_ACCESS_TRANSFER_WRITE_BIT,
+            .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+            .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+            .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .image = vk_image,
+            .subresourceRange{
+                .aspectMask = aspect_mask,
+                .baseMipLevel = 0,
+                .levelCount = VK_REMAINING_MIP_LEVELS,
+                .baseArrayLayer = 0,
+                .layerCount = VK_REMAINING_ARRAY_LAYERS,
+            },
+        };
+        cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
+                       vk::PIPELINE_STAGE_GRAPHICS_COMPUTE_TRANSFER, 0, image_barrier);
+    });
+    scheduler.Finish();
+}
+
+constexpr u32 BL3D_BINDING_INPUT_BUFFER  = 0;
+constexpr u32 BL3D_BINDING_OUTPUT_BUFFER = 1;
 
 struct alignas(16) BlockLinearUnswizzle3DPushConstants {
     u32 blocks_dim[3];           // Offset 0
@@ -745,11 +1008,50 @@ BlockLinearUnswizzle3DPass::BlockLinearUnswizzle3DPass(
     DescriptorPool& descriptor_pool_,
     StagingBufferPool& staging_buffer_pool_,
     ComputePassDescriptorQueue& compute_pass_descriptor_queue_)
-    : ComputePass(
-          device_, scheduler_, descriptor_pool_,
-          BL3D_DESCRIPTOR_SET_BINDINGS,
-          BL3D_DESCRIPTOR_UPDATE_TEMPLATE_ENTRY,
-          BL3D_BANK_INFO,
+    : ComputePass(device_, scheduler_, descriptor_pool_,
+        std::array<VkDescriptorSetLayoutBinding, 2>{{
+    {
+        .binding = BL3D_BINDING_INPUT_BUFFER,
+        .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER, // block-linear input
+        .descriptorCount = 1,
+        .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
+        .pImmutableSamplers = nullptr,
+    },
+    {
+        .binding = BL3D_BINDING_OUTPUT_BUFFER,
+        .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
+        .descriptorCount = 1,
+        .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
+        .pImmutableSamplers = nullptr,
+    },
+        }},
+        std::array<VkDescriptorUpdateTemplateEntry, 2>{{
+        {
+            .dstBinding = BL3D_BINDING_INPUT_BUFFER,
+            .dstArrayElement = 0,
+            .descriptorCount = 1,
+            .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
+            .offset = BL3D_BINDING_INPUT_BUFFER * sizeof(DescriptorUpdateEntry),
+            .stride = sizeof(DescriptorUpdateEntry),
+        },
+        {
+            .dstBinding = BL3D_BINDING_OUTPUT_BUFFER,
+            .dstArrayElement = 0,
+            .descriptorCount = 1,
+            .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_BUFFER,
+            .offset = BL3D_BINDING_OUTPUT_BUFFER * sizeof(DescriptorUpdateEntry),
+            .stride = sizeof(DescriptorUpdateEntry),
+        }
+        }},
+        DescriptorBankInfo{
+            .uniform_buffers = 0,
+            .storage_buffers = 2,
+            .texture_buffers = 0,
+            .image_buffers = 0,
+            .textures = 0,
+            .images = 0,
+            .score = 2,
+        },
           COMPUTE_PUSH_CONSTANT_RANGE<sizeof(BlockLinearUnswizzle3DPushConstants)>,
           BLOCK_LINEAR_UNSWIZZLE_3D_BCN_COMP_SPV),
       scheduler{scheduler_},
@@ -821,9 +1123,7 @@ void BlockLinearUnswizzle3DPass::UnswizzleChunk(
     pc.blocks_dim[1] = blocks_y;
     pc.blocks_dim[2] = z_count; // Only process the count
 
-    compute_pass_descriptor_queue.Acquire();
-    compute_pass_descriptor_queue.AddBuffer(*image.runtime->swizzle_table_buffer, 0,
-                                           image.runtime->swizzle_table_size);
+    compute_pass_descriptor_queue.Acquire(scheduler, 3);
     compute_pass_descriptor_queue.AddBuffer(swizzled.buffer,
                                            sw.buffer_offset + swizzled.offset,
                                            image.guest_size_bytes - sw.buffer_offset);
@@ -859,6 +1159,23 @@ void BlockLinearUnswizzle3DPass::UnswizzleChunk(
             return;
         }
 
+        if (!is_first_chunk) {
+            const VkBufferMemoryBarrier reuse_barrier{
+                .sType = VK_STRUCTURE_TYPE_BUFFER_MEMORY_BARRIER,
+                .pNext = nullptr,
+                .srcAccessMask = VK_ACCESS_TRANSFER_READ_BIT,
+                .dstAccessMask = VK_ACCESS_SHADER_WRITE_BIT,
+                .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+                .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+                .buffer = out_buffer,
+                .offset = 0,
+                .size = VK_WHOLE_SIZE,
+            };
+            cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT,
+                                   VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0, nullptr,
+                                   reuse_barrier, nullptr);
+        }
+
         device.GetLogical().UpdateDescriptorSet(set, *descriptor_template, descriptor_data);
         cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, *pipeline);
         cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, *layout, 0, set, {});
@@ -989,7 +1306,7 @@ void MSAACopyPass::CopyImage(Image& dst_image, Image& src_image,
         ASSERT(copy.dst_subresource.base_layer == 0);
         ASSERT(copy.dst_subresource.num_layers == 1);
 
-        compute_pass_descriptor_queue.Acquire();
+        compute_pass_descriptor_queue.Acquire(scheduler, 2);
         compute_pass_descriptor_queue.AddImage(
             src_image.StorageImageView(copy.src_subresource.base_level));
         compute_pass_descriptor_queue.AddImage(

src/video_core/renderer_vulkan/vk_compute_pass.h

@@ -25,6 +25,7 @@ struct SwizzleParameters;
 
 namespace Vulkan {
 
+using VideoCommon::Accelerated::BlockLinearSwizzle2DParams;
 using VideoCommon::Accelerated::BlockLinearSwizzle3DParams;
 
 class Device;
@@ -164,6 +165,56 @@ private:
     ComputePassDescriptorQueue& compute_pass_descriptor_queue;
 };
 
+class BlockLinearUnswizzleImage2DPass final : public ComputePass {
+public:
+    explicit BlockLinearUnswizzleImage2DPass(const Device& device_, Scheduler& scheduler_,
+                                             DescriptorPool& descriptor_pool_,
+                                             StagingBufferPool& staging_buffer_pool_,
+                                             ComputePassDescriptorQueue& compute_pass_descriptor_queue_);
+    ~BlockLinearUnswizzleImage2DPass();
+
+    void Unswizzle(Image& image, const StagingBufferRef& map,
+                   std::span<const VideoCommon::SwizzleParameters> swizzles);
+
+private:
+    Scheduler& scheduler;
+    StagingBufferPool& staging_buffer_pool;
+    ComputePassDescriptorQueue& compute_pass_descriptor_queue;
+};
+
+class BlockLinearUnswizzleImage3DPass final : public ComputePass {
+public:
+    explicit BlockLinearUnswizzleImage3DPass(const Device& device_, Scheduler& scheduler_,
+                                             DescriptorPool& descriptor_pool_,
+                                             StagingBufferPool& staging_buffer_pool_,
+                                             ComputePassDescriptorQueue& compute_pass_descriptor_queue_);
+    ~BlockLinearUnswizzleImage3DPass();
+
+    void Unswizzle(Image& image, const StagingBufferRef& map,
+                   std::span<const VideoCommon::SwizzleParameters> swizzles);
+
+private:
+    Scheduler& scheduler;
+    StagingBufferPool& staging_buffer_pool;
+    ComputePassDescriptorQueue& compute_pass_descriptor_queue;
+};
+
+class PitchUnswizzlePass final : public ComputePass {
+public:
+    explicit PitchUnswizzlePass(const Device& device_, Scheduler& scheduler_,
+                                DescriptorPool& descriptor_pool_,
+                                StagingBufferPool& staging_buffer_pool_,
+                                ComputePassDescriptorQueue& compute_pass_descriptor_queue_);
+    ~PitchUnswizzlePass();
+
+    void Unswizzle(Image& image, const StagingBufferRef& map,
+                   std::span<const VideoCommon::SwizzleParameters> swizzles);
+
+private:
+    Scheduler& scheduler;
+    StagingBufferPool& staging_buffer_pool;
+    ComputePassDescriptorQueue& compute_pass_descriptor_queue;
+};
 
 class MSAACopyPass final : public ComputePass {
 public:

src/video_core/renderer_vulkan/vk_compute_pipeline.cpp

@@ -45,6 +45,7 @@ ComputePipeline::ComputePipeline(const Device& device_, Scheduler& scheduler, vk
     }
     std::copy_n(info.constant_buffer_used_sizes.begin(), uniform_buffer_sizes.size(),
                 uniform_buffer_sizes.begin());
+    num_descriptor_entries = NumDescriptorEntries(info);
 
     auto func{[this, &scheduler, &descriptor_pool, shader_notify, pipeline_statistics] {
         DescriptorLayoutBuilder builder{device};
@@ -113,7 +114,7 @@ ComputePipeline::ComputePipeline(const Device& device_, Scheduler& scheduler, vk
 void ComputePipeline::Configure(Tegra::Engines::KeplerCompute& kepler_compute,
                                 Tegra::MemoryManager& gpu_memory, Scheduler& scheduler,
                                 BufferCache& buffer_cache, TextureCache& texture_cache) {
-    guest_descriptor_queue.Acquire();
+    guest_descriptor_queue.Acquire(scheduler, num_descriptor_entries);
 
     buffer_cache.SetComputeUniformBufferState(info.constant_buffer_mask, &uniform_buffer_sizes);
     buffer_cache.UnbindComputeStorageBuffers();

src/video_core/renderer_vulkan/vk_compute_pipeline.h

@@ -53,6 +53,7 @@ private:
     vk::PipelineCache& pipeline_cache;
     GuestDescriptorQueue& guest_descriptor_queue;
     Shader::Info info;
+    u32 num_descriptor_entries{};
 
     VideoCommon::ComputeUniformBufferSizes uniform_buffer_sizes{};
 

src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp

@@ -268,6 +268,7 @@ GraphicsPipeline::GraphicsPipeline(
         num_textures += Shader::NumDescriptors(info->texture_descriptors);
         num_image_elements += Shader::NumDescriptors(info->texture_descriptors);
         num_image_elements += Shader::NumDescriptors(info->image_descriptors);
+        num_descriptor_entries += NumDescriptorEntries(*info);
     }
     fragment_has_color0_output = stage_infos[NUM_STAGES - 1].stores_frag_color[0];
     auto func{[this, shader_notify, &render_pass_cache, &descriptor_pool, pipeline_statistics] {
@@ -473,7 +474,7 @@ bool GraphicsPipeline::ConfigureImpl(bool is_indexed) {
     buffer_cache.UpdateGraphicsBuffers(is_indexed);
     buffer_cache.BindHostGeometryBuffers(is_indexed);
 
-    guest_descriptor_queue.Acquire();
+    guest_descriptor_queue.Acquire(scheduler, num_descriptor_entries);
 
     RescalingPushConstant rescaling;
     RenderAreaPushConstant render_area;
@@ -907,7 +908,7 @@ void GraphicsPipeline::MakePipeline(VkRenderPass render_pass) {
 
     // EDS3 - Enables (composite: per-feature)
     if (key.state.extended_dynamic_state_3_enables) {
-        if (key.state.dynamic_state3_depth_clamp_enable != 0) {
+        if (device.SupportsDynamicState3DepthClampEnable()) {
             dynamic_states.push_back(VK_DYNAMIC_STATE_DEPTH_CLAMP_ENABLE_EXT);
         }
         if (device.SupportsDynamicState3LogicOpEnable()) {

src/video_core/renderer_vulkan/vk_graphics_pipeline.h

@@ -159,6 +159,7 @@ private:
     std::array<Shader::Info, NUM_STAGES> stage_infos;
     std::array<u32, 5> enabled_uniform_buffer_masks{};
     VideoCommon::UniformBufferSizes uniform_buffer_sizes{};
+    u32 num_descriptor_entries{};
     size_t num_image_elements{};
     u32 num_textures{};
     bool fragment_has_color0_output{};

src/video_core/renderer_vulkan/vk_pipeline_cache.cpp

@@ -439,10 +439,21 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
         .has_broken_robust =
             device.IsNvidia() && device.GetNvidiaArch() <= NvidiaArchitecture::Arch_Pascal,
         .min_ssbo_alignment = device.GetStorageBufferAlignment(),
-        .max_user_clip_distances = device.GetMaxUserClipDistances(),
+        .max_user_clip_distances = device.GetMaxUserClipDistances()
     };
 
     host_info = Shader::HostTranslateInfo{
+        .min_ssbo_alignment = device.GetStorageBufferAlignment(),
+        .max_per_stage_descriptor_sampled_images = device.GetMaxPerStageDescriptorSampledImages(),
+        .max_per_stage_resources = device.GetMaxPerStageResources(),
+        .max_descriptor_set_samplers = device.GetMaxDescriptorSetSamplers(),
+        .max_descriptor_set_uniform_buffers = device.GetMaxDescriptorSetUniformBuffers(),
+        .max_descriptor_set_uniform_buffers_dynamic = device.GetMaxDescriptorSetUniformBuffersDynamic(),
+        .max_descriptor_set_storage_buffers = device.GetMaxDescriptorSetStorageBuffers(),
+        .max_descriptor_set_storage_buffers_dynamic = device.GetMaxDescriptorSetStorageBuffersDynamic(),
+        .max_descriptor_set_sampled_images = device.GetMaxDescriptorSetSampledImages(),
+        .max_descriptor_set_storage_images = device.GetMaxDescriptorSetStorageImages(),
+        .max_descriptor_set_input_attachements = device.GetMaxDescriptorSetInputAttachments(),
         .support_float64 = device.IsFloat64Supported(),
         .support_float16 = device.IsFloat16Supported(),
         .support_int64 = device.IsShaderInt64Supported(),
@@ -451,13 +462,10 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
                                 driver_id == VK_DRIVER_ID_SAMSUNG_PROPRIETARY,
         .support_snorm_render_buffer = true,
         .support_viewport_index_layer = device.IsExtShaderViewportIndexLayerSupported(),
-        .min_ssbo_alignment = static_cast<u32>(device.GetStorageBufferAlignment()),
-        .max_per_stage_descriptor_sampled_images = device.GetMaxPerStageDescriptorSampledImages(),
-        .max_per_stage_resources = device.GetMaxPerStageResources(),
-        .max_descriptor_set_sampled_images = device.GetMaxDescriptorSetSampledImages(),
         .support_geometry_shader_passthrough = device.IsNvGeometryShaderPassthroughSupported(),
         .support_conditional_barrier = device.SupportsConditionalBarriers(),
     };
+    host_info.ApplyDescriptorLimitPolicy();
 
     if (device.GetMaxVertexInputAttributes() < Maxwell::NumVertexAttributes) {
         LOG_WARNING(Render_Vulkan, "maxVertexInputAttributes is too low: {} < {}",
@@ -496,8 +504,10 @@ PipelineCache::PipelineCache(Tegra::MaxwellDeviceMemoryManager& device_memory_,
         dynamic_features.has_extended_dynamic_state_3_enables &&
         device.SupportsDynamicState3DepthClampEnable();
     dynamic_features.has_dynamic_state3_logic_op_enable =
+        dynamic_features.has_extended_dynamic_state_3_enables &&
         device.SupportsDynamicState3LogicOpEnable();
     dynamic_features.has_dynamic_state3_line_stipple_enable =
+        dynamic_features.has_extended_dynamic_state_3_enables &&
         device.SupportsDynamicState3LineStippleEnable();
 
     // VIDS: Independent toggle (not affected by dyna_state levels)

src/video_core/renderer_vulkan/vk_pipeline_cache.h

@@ -113,10 +113,6 @@ public:
 
     [[nodiscard]] ComputePipeline* CurrentComputePipeline();
 
-    [[nodiscard]] bool SupportsDynamicState3DepthClampEnable() const {
-        return dynamic_features.has_dynamic_state3_depth_clamp_enable;
-    }
-
     void LoadDiskResources(u64 title_id, std::stop_token stop_loading,
                            const VideoCore::DiskResourceLoadCallback& callback);
 

src/video_core/renderer_vulkan/vk_rasterizer.cpp

@@ -203,7 +203,7 @@ RasterizerVulkan::RasterizerVulkan(Core::Frontend::EmuWindow& emu_window_, Tegra
     : gpu{gpu_}, device_memory{device_memory_}, device{device_},
       memory_allocator{memory_allocator_}, state_tracker{state_tracker_}, scheduler{scheduler_},
       staging_pool(device, memory_allocator, scheduler), descriptor_pool(device, scheduler),
-      guest_descriptor_queue(device, scheduler), compute_pass_descriptor_queue(device, scheduler),
+      guest_descriptor_queue(device), compute_pass_descriptor_queue(device),
       blit_image(device, scheduler, state_tracker, descriptor_pool), render_pass_cache(device),
       texture_cache_runtime{
           device,     scheduler,         memory_allocator, staging_pool,
@@ -1578,7 +1578,7 @@ void RasterizerVulkan::UpdateDepthClampEnable(Tegra::Engines::Maxwell3D::Regs& r
     if (!state_tracker.TouchDepthClampEnable()) {
         return;
     }
-    if (!pipeline_cache.SupportsDynamicState3DepthClampEnable()) {
+    if (!device.SupportsDynamicState3DepthClampEnable()) {
         return;
     }
     bool is_enabled = !(regs.viewport_clip_control.geometry_clip ==

src/video_core/renderer_vulkan/vk_scheduler.cpp

@@ -155,15 +155,14 @@ void Scheduler::WaitWorker() {
 }
 
 void Scheduler::DispatchWork() {
-    if (chunk->Empty()) {
-        return;
+    if (chunk && !chunk->Empty()) {
+        {
+            std::scoped_lock ql{queue_mutex};
+            work_queue.push(std::move(chunk));
+        }
+        event_cv.notify_all();
+        AcquireNewChunk();
     }
-    {
-        std::scoped_lock ql{queue_mutex};
-        work_queue.push(std::move(chunk));
-    }
-    event_cv.notify_all();
-    AcquireNewChunk();
 }
 
 void Scheduler::RequestRenderpass(const Framebuffer* framebuffer) {

src/video_core/renderer_vulkan/vk_scheduler.h

@@ -6,7 +6,6 @@
 
 #pragma once
 
-#include <atomic>
 #include <condition_variable>
 #include <cstddef>
 #include <functional>
@@ -93,12 +92,10 @@ public:
         requires std::is_invocable_v<T, vk::CommandBuffer, vk::CommandBuffer>
     void RecordWithUploadBuffer(T&& command) {
         if (chunk->Record(command)) {
-            record_serial.fetch_add(1, std::memory_order_relaxed);
             return;
         }
         DispatchWork();
         (void)chunk->Record(command);
-        record_serial.fetch_add(1, std::memory_order_relaxed);
     }
 
     template <typename T>
@@ -120,11 +117,6 @@ public:
         return master_semaphore->IsFree(tick);
     }
 
-    /// Returns a monotonic serial incremented for every recorded command callback.
-    [[nodiscard]] u64 CurrentRecordSerial() const noexcept {
-        return record_serial.load(std::memory_order_relaxed);
-    }
-
     /// Waits for the given GPU tick, optionally pacing frames.
     void Wait(u64 tick, double target_fps = 0.0) {
         if (tick > 0) {
@@ -306,7 +298,6 @@ private:
     u64 frame_counter{};
 
     u64 last_submitted_tick = 0;
-    std::atomic<u64> record_serial{0};
 };
 
 } // namespace Vulkan

src/video_core/renderer_vulkan/vk_texture_cache.cpp

@@ -195,8 +195,42 @@ constexpr VkBorderColor ConvertBorderColor(const std::array<float, 4>& color) {
     return allocator.CreateImage(image_ci);
 }
 
+[[nodiscard]] VkFormat UnswizzleStorageFormat(u32 bytes_per_block) {
+    switch (bytes_per_block) {
+    case 1:
+        return VK_FORMAT_R8_UINT;
+    case 2:
+        return VK_FORMAT_R16_UINT;
+    case 4:
+        return VK_FORMAT_R32_UINT;
+    case 8:
+        return VK_FORMAT_R32G32_UINT;
+    case 16:
+        return VK_FORMAT_R32G32B32A32_UINT;
+    default:
+        ASSERT_MSG(false, "Invalid bytes_per_block={} for accelerated unswizzle", bytes_per_block);
+        return VK_FORMAT_R32_UINT;
+    }
+}
+
+[[nodiscard]] bool IsUnswizzleStorageFormatSupported(const Device& device, u32 bytes_per_block) {
+    switch (bytes_per_block) {
+    case 1:
+        return device.IsStorageBuffer8BitAccessSupported();
+    case 2:
+        return device.IsStorageBuffer16BitAccessSupported();
+    case 4:
+    case 8:
+    case 16:
+        return true;
+    default:
+        return false;
+    }
+}
+
 [[nodiscard]] vk::ImageView MakeStorageView(const vk::Device& device, u32 level, VkImage image,
-                                            VkFormat format) {
+                                            VkFormat format,
+                                            VkImageViewType view_type = VK_IMAGE_VIEW_TYPE_2D_ARRAY) {
     static constexpr VkImageViewUsageCreateInfo storage_image_view_usage_create_info{
         .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_USAGE_CREATE_INFO,
         .pNext = nullptr,
@@ -207,7 +241,7 @@ constexpr VkBorderColor ConvertBorderColor(const std::array<float, 4>& color) {
         .pNext = &storage_image_view_usage_create_info,
         .flags = 0,
         .image = image,
-        .viewType = VK_IMAGE_VIEW_TYPE_2D_ARRAY,
+        .viewType = view_type,
         .format = format,
         .components{
             .r = VK_COMPONENT_SWIZZLE_IDENTITY,
@@ -887,6 +921,12 @@ TextureCacheRuntime::TextureCacheRuntime(const Device& device_, Scheduler& sched
     if (device.IsStorageImageMultisampleSupported()) {
         msaa_copy_pass.emplace(device, scheduler, descriptor_pool, staging_buffer_pool, compute_pass_descriptor_queue);
     }
+    bl_unswizzle_2d_pass.emplace(device, scheduler, descriptor_pool, staging_buffer_pool,
+                                 compute_pass_descriptor_queue);
+    bl_unswizzle_3d_pass.emplace(device, scheduler, descriptor_pool, staging_buffer_pool,
+                                 compute_pass_descriptor_queue);
+    pitch_unswizzle_pass.emplace(device, scheduler, descriptor_pool, staging_buffer_pool,
+                                 compute_pass_descriptor_queue);
     if (!device.IsKhrImageFormatListSupported()) {
         return;
     }
@@ -894,6 +934,11 @@ TextureCacheRuntime::TextureCacheRuntime(const Device& device_, Scheduler& sched
         const auto image_format = static_cast<PixelFormat>(index_a);
         if (IsPixelFormatASTC(image_format) && !device.IsOptimalAstcSupported()) {
             view_formats[index_a].push_back(VK_FORMAT_A8B8G8R8_UNORM_PACK32);
+        } else if (!IsPixelFormatASTC(image_format) && !IsPixelFormatBCn(image_format)) {
+            const u32 bpp = VideoCore::Surface::BytesPerBlock(image_format);
+            if (IsUnswizzleStorageFormatSupported(device, bpp)) {
+                view_formats[index_a].push_back(UnswizzleStorageFormat(bpp));
+            }
         }
         for (size_t index_b = 0; index_b < VideoCore::Surface::MaxPixelFormat; index_b++) {
             const auto view_format = static_cast<PixelFormat>(index_b);
@@ -909,40 +954,6 @@ TextureCacheRuntime::TextureCacheRuntime(const Device& device_, Scheduler& sched
         bl3d_unswizzle_pass.emplace(device, scheduler, descriptor_pool,
                                    staging_buffer_pool, compute_pass_descriptor_queue);
     }
-
-    // --- Create swizzle table buffer ---
-    {
-        auto table = Tegra::Texture::MakeSwizzleTable();
-
-        swizzle_table_size = static_cast<VkDeviceSize>(table.size() * sizeof(table[0]));
-
-        auto staging = staging_buffer_pool.Request(swizzle_table_size, MemoryUsage::Upload);
-        std::memcpy(staging.mapped_span.data(), table.data(), static_cast<size_t>(swizzle_table_size));
-
-        VkBufferCreateInfo ci{
-            .sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
-            .size = swizzle_table_size,
-            .usage = VK_BUFFER_USAGE_STORAGE_BUFFER_BIT |
-                     VK_BUFFER_USAGE_TRANSFER_DST_BIT |
-                     VK_BUFFER_USAGE_TRANSFER_SRC_BIT,
-            .sharingMode = VK_SHARING_MODE_EXCLUSIVE,
-        };
-        swizzle_table_buffer = memory_allocator.CreateBuffer(ci, MemoryUsage::DeviceLocal);
-
-        scheduler.RequestOutsideRenderPassOperationContext();
-        scheduler.Record([staging_buf = staging.buffer,
-                          dst_buf = *swizzle_table_buffer,
-                          size = swizzle_table_size,
-                          src_off = staging.offset](vk::CommandBuffer cmdbuf) {
-
-            const VkBufferCopy region{
-                .srcOffset = src_off,
-                .dstOffset = 0,
-                .size = size,
-            };
-            cmdbuf.CopyBuffer(staging_buf, dst_buf, region);
-        });
-    }
 }
 
 void TextureCacheRuntime::Finish() {
@@ -1614,6 +1625,16 @@ Image::Image(TextureCacheRuntime& runtime_, const ImageInfo& info_, GPUVAddr gpu
         flags |= VideoCommon::ImageFlagBits::Converted;
         flags |= VideoCommon::ImageFlagBits::CostlyLoad;
     }
+    if (!IsPixelFormatASTC(info.format) && !IsPixelFormatBCn(info.format) &&
+        VideoCore::Surface::GetFormatType(info.format) ==
+            VideoCore::Surface::SurfaceType::ColorTexture &&
+        (info.type == ImageType::e2D || info.type == ImageType::e3D ||
+         info.type == ImageType::Linear)) {
+        if (IsUnswizzleStorageFormatSupported(runtime->device,
+                                              VideoCore::Surface::BytesPerBlock(info.format))) {
+            flags |= VideoCommon::ImageFlagBits::AcceleratedUpload;
+        }
+    }
     if (runtime->device.HasDebuggingToolAttached()) {
         original_image.SetObjectNameEXT(VideoCommon::Name(*this).c_str());
     }
@@ -1627,6 +1648,19 @@ Image::Image(TextureCacheRuntime& runtime_, const ImageInfo& info_, GPUVAddr gpu
             storage_image_views[level] =
                 MakeStorageView(device, level, *original_image, VK_FORMAT_A8B8G8R8_UNORM_PACK32);
         }
+    } else if (True(flags & VideoCommon::ImageFlagBits::AcceleratedUpload)) {
+        const auto& device = runtime->device.GetLogical();
+        const VkFormat storage_format =
+            UnswizzleStorageFormat(VideoCore::Surface::BytesPerBlock(info.format));
+        const VkImageViewType storage_view_type = info.type == ImageType::e3D
+                                                       ? VK_IMAGE_VIEW_TYPE_3D
+                                                   : info.type == ImageType::Linear
+                                                       ? VK_IMAGE_VIEW_TYPE_2D
+                                                       : VK_IMAGE_VIEW_TYPE_2D_ARRAY;
+        for (s32 level = 0; level < info.resources.levels; ++level) {
+            storage_image_views[level] =
+                MakeStorageView(device, level, *original_image, storage_format, storage_view_type);
+        }
     }
 }
 
@@ -2482,21 +2516,24 @@ void TextureCacheRuntime::AccelerateImageUpload(
         return astc_decoder_pass->Assemble(image, map, swizzles);
     }
 
-    if (!Settings::values.gpu_unswizzle_enabled.GetValue() || !bl3d_unswizzle_pass) {
-        if (IsPixelFormatBCn(image.info.format) && image.info.type == ImageType::e3D) {
-            ASSERT_MSG(false, "GPU unswizzle is disabled for BCn 3D texture");
+    if (IsPixelFormatBCn(image.info.format)) {
+        if (Settings::values.gpu_unswizzle_enabled.GetValue() && bl3d_unswizzle_pass &&
+            image.info.type == ImageType::e3D && image.info.resources.levels == 1 &&
+            image.info.resources.layers == 1) {
+            return bl3d_unswizzle_pass->Unswizzle(image, map, swizzles, z_start, z_count);
         }
-        ASSERT(false);
+        ASSERT(false && "GPU unswizzle is disabled for BCn 3D texture");
         return;
     }
 
-    if (bl3d_unswizzle_pass &&
-        IsPixelFormatBCn(image.info.format) &&
-        image.info.type == ImageType::e3D &&
-        image.info.resources.levels == 1 &&
-        image.info.resources.layers == 1) {
-
-        return bl3d_unswizzle_pass->Unswizzle(image, map, swizzles, z_start, z_count);
+    if (image.info.type == ImageType::e2D) {
+        return bl_unswizzle_2d_pass->Unswizzle(image, map, swizzles);
+    }
+    if (image.info.type == ImageType::e3D) {
+        return bl_unswizzle_3d_pass->Unswizzle(image, map, swizzles);
+    }
+    if (image.info.type == ImageType::Linear) {
+        return pitch_unswizzle_pass->Unswizzle(image, map, swizzles);
     }
 
     ASSERT(false);

src/video_core/renderer_vulkan/vk_texture_cache.h

@@ -1,4 +1,4 @@
-// SPDX-FileCopyrightText: Copyright 2025 Eden Emulator Project
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
 // SPDX-License-Identifier: GPL-3.0-or-later
 
 // SPDX-FileCopyrightText: Copyright 2019 yuzu Emulator Project
@@ -130,9 +130,9 @@ public:
     std::optional<ASTCDecoderPass> astc_decoder_pass;
 
     std::optional<BlockLinearUnswizzle3DPass> bl3d_unswizzle_pass;
-    vk::Buffer swizzle_table_buffer;
-    VkDeviceSize swizzle_table_size = 0;
-
+    std::optional<BlockLinearUnswizzleImage2DPass> bl_unswizzle_2d_pass;
+    std::optional<BlockLinearUnswizzleImage3DPass> bl_unswizzle_3d_pass;
+    std::optional<PitchUnswizzlePass> pitch_unswizzle_pass;
     std::optional<MSAACopyPass> msaa_copy_pass;
     const Settings::ResolutionScalingInfo& resolution;
     std::array<std::vector<VkFormat>, VideoCore::Surface::MaxPixelFormat> view_formats;

src/video_core/renderer_vulkan/vk_update_descriptor.cpp

@@ -7,6 +7,7 @@
 #include <variant>
 #include <boost/container/static_vector.hpp>
 
+#include "common/assert.h"
 #include "common/logging.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/vk_update_descriptor.h"
@@ -15,8 +16,9 @@
 
 namespace Vulkan {
 
-UpdateDescriptorQueue::UpdateDescriptorQueue(const Device& device_, Scheduler& scheduler_)
-    : device{device_}, scheduler{scheduler_} {
+UpdateDescriptorQueue::UpdateDescriptorQueue(const Device& device_)
+    : device{device_}
+{
     payload_start = payload.data();
     payload_cursor = payload.data();
 }
@@ -31,13 +33,15 @@ void UpdateDescriptorQueue::TickFrame() {
     payload_cursor = payload_start;
 }
 
-void UpdateDescriptorQueue::Acquire() {
-    // Minimum number of entries required.
-    // This is the maximum number of entries a single draw call might use.
-    static constexpr size_t MIN_ENTRIES = 0x400;
-
-    if (std::distance(payload_start, payload_cursor) + MIN_ENTRIES >= FRAME_PAYLOAD_SIZE) {
-        LOG_WARNING(Render_Vulkan, "Payload overflow, waiting for worker thread");
+void UpdateDescriptorQueue::Acquire(Scheduler& scheduler, size_t required_entries) {
+    static constexpr size_t DEFAULT_REQUIRED_ENTRIES = 0x400;
+    const size_t reserve = required_entries > 0 ? required_entries : DEFAULT_REQUIRED_ENTRIES;
+    ASSERT_MSG(reserve < FRAME_PAYLOAD_SIZE, "Descriptor reservation {} >= frame capacity {}",
+               reserve, FRAME_PAYLOAD_SIZE);
+    const size_t used = static_cast<size_t>(std::distance(payload_start, payload_cursor));
+    if (used + reserve >= FRAME_PAYLOAD_SIZE) {
+        LOG_WARNING(Render_Vulkan, "Payload overflow (used={}, reserve={}, capacity={})",
+                    used, reserve, FRAME_PAYLOAD_SIZE);
         scheduler.WaitWorker();
         payload_cursor = payload_start;
     }

src/video_core/renderer_vulkan/vk_update_descriptor.h

@@ -34,12 +34,11 @@ class UpdateDescriptorQueue final {
     static constexpr size_t PAYLOAD_SIZE = FRAME_PAYLOAD_SIZE * FRAMES_IN_FLIGHT;
 
 public:
-    explicit UpdateDescriptorQueue(const Device& device_, Scheduler& scheduler_);
+    explicit UpdateDescriptorQueue(const Device& device_);
     ~UpdateDescriptorQueue();
 
     void TickFrame();
-
-    void Acquire();
+    void Acquire(Scheduler& scheduler, size_t required_entries = 0);
 
     const DescriptorUpdateEntry* UpdateData() const noexcept {
         return upload_start;
@@ -75,8 +74,6 @@ public:
 
 private:
     const Device& device;
-    Scheduler& scheduler;
-
     size_t frame_index{0};
     DescriptorUpdateEntry* payload_cursor = nullptr;
     DescriptorUpdateEntry* payload_start = nullptr;

src/video_core/shader_environment.cpp

@@ -255,8 +255,7 @@ std::optional<u64> GenericEnvironment::TryFindSize() {
     static constexpr u64 SELF_BRANCH_A = 0xE2400FFFFF87000FULL;
     static constexpr u64 SELF_BRANCH_B = 0xE2400FFFFF07000FULL;
 
-    static constexpr u64 MESA_EXIT_MASK = 0xFFF00000000F001FULL;
-    static constexpr u64 MESA_EXIT_VALUE = (0xE30ULL << 52) | (0x7ULL << 16) | 0xFULL;
+    static constexpr u64 EXIT_VALUE = 0xE30000000007000FULL;
 
     code.resize(MAXIMUM_SIZE / INST_SIZE);
 
@@ -271,7 +270,7 @@ std::optional<u64> GenericEnvironment::TryFindSize() {
             if (inst == SELF_BRANCH_A || inst == SELF_BRANCH_B) {
                 return offset + index;
             }
-            if ((inst & MESA_EXIT_MASK) == MESA_EXIT_VALUE) {
+            if (!is_proprietary_driver && inst == EXIT_VALUE) {
                 return offset + index + INST_SIZE;
             }
         }

src/video_core/texture_cache/image_base.h

@@ -1,6 +1,3 @@
-// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
-// SPDX-License-Identifier: GPL-3.0-or-later
-
 // SPDX-FileCopyrightText: Copyright 2020 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -105,7 +102,7 @@ struct ImageBase {
     VAddr cpu_addr_end = 0;
 
     u64 modification_tick = 0;
-    u64 last_use_tick = 0;
+    size_t lru_index = SIZE_MAX;
 
     std::array<u32, MAX_MIP_LEVELS> mip_level_offsets{};
 

src/video_core/texture_cache/texture_cache.h

@@ -159,54 +159,11 @@ void TextureCache<P>::RunGarbageCollector() {
         }
         return false;
     };
-
-    const auto CollectBelow = [this](u64 threshold) {
-        boost::container::small_vector<ImageId, 64> expired;
-        for (auto [id, image] : slot_images) {
-            if (image->last_use_tick < threshold) {
-                expired.push_back(id);
-            }
-        }
-        return expired;
-    };
-
-    // Aggressively clear massive sparse textures
-    if (total_used_memory >= expected_memory) {
-        auto candidates = CollectBelow(frame_tick);
-        for (const auto image_id : candidates) {
-            auto& image = slot_images[image_id];
-            if (image.info.is_sparse &&
-                image.guest_size_bytes >= 256_MiB &&
-                image.allocation_tick < frame_tick - 3) {
-                LOG_DEBUG(HW_GPU, "GC targeting old sparse texture at 0x{:X} ({} MiB, age: {} frames)",
-                         image.gpu_addr, image.guest_size_bytes / (1024 * 1024),
-                         frame_tick - image.allocation_tick);
-                if (Cleanup(image_id)) {
-                    break;
-                }
-            }
-        }
-    }
-
     Configure(false);
-    {
-        auto expired = CollectBelow(frame_tick - ticks_to_destroy);
-        for (const auto image_id : expired) {
-            if (Cleanup(image_id)) {
-                break;
-            }
-        }
-    }
-
-    // If pressure is still too high, prune aggressively.
+    lru_cache.ForEachItemBelow(frame_tick - ticks_to_destroy, Cleanup);
     if (total_used_memory >= critical_memory) {
         Configure(true);
-        auto expired = CollectBelow(frame_tick - ticks_to_destroy);
-        for (const auto image_id : expired) {
-            if (Cleanup(image_id)) {
-                break;
-            }
-        }
+        lru_cache.ForEachItemBelow(frame_tick - ticks_to_destroy, Cleanup);
     }
 }
 
@@ -1908,7 +1865,7 @@ std::pair<u32, u32> TextureCache<P>::PrepareDmaImage(ImageId dst_id, GPUVAddr ba
     const auto base = image.TryFindBase(base_addr);
     PrepareImage(dst_id, mark_as_modified, false);
     const auto& new_image = slot_images[dst_id];
-    new_image.last_use_tick = frame_tick;
+    lru_cache.Touch(new_image.lru_index, frame_tick);
     return std::make_pair(base->level, base->layer);
 }
 
@@ -2235,7 +2192,7 @@ void TextureCache<P>::RegisterImage(ImageId image_id) {
         tentative_size = TranscodedAstcSize(tentative_size, image.info.format);
     }
     total_used_memory += Common::AlignUp(tentative_size, 1024);
-    image.last_use_tick = frame_tick;
+    image.lru_index = lru_cache.Insert(image_id, frame_tick);
 
     ForEachGPUPage(image.gpu_addr, image.guest_size_bytes, [this, image_id](u64 page) {
         (*channel_state->gpu_page_table)[page].push_back(image_id);
@@ -2269,7 +2226,7 @@ void TextureCache<P>::UnregisterImage(ImageId image_id) {
                "Trying to unregister an already registered image");
     image.flags &= ~ImageFlagBits::Registered;
     image.flags &= ~ImageFlagBits::BadOverlap;
-
+    lru_cache.Free(image.lru_index);
     const auto& clear_page_table =
         [image_id](u64 page, ankerl::unordered_dense::map<u64, std::vector<ImageId>, Common::IdentityHash<u64>>& selected_page_table) {
             const auto page_it = selected_page_table.find(page);
@@ -2597,7 +2554,7 @@ void TextureCache<P>::PrepareImage(ImageId image_id, bool is_modification, bool
     if (is_modification) {
         MarkModification(image);
     }
-    image.last_use_tick = frame_tick;
+    lru_cache.Touch(image.lru_index, frame_tick);
 }
 
 template <class P>

src/video_core/texture_cache/texture_cache_base.h

@@ -23,7 +23,7 @@
 #include "common/common_types.h"
 #include "common/hash.h"
 #include "common/literals.h"
-
+#include "common/lru_cache.h"
 #include <ranges>
 #include "common/scratch_buffer.h"
 #include "common/slot_vector.h"
@@ -485,7 +485,11 @@ private:
     std::deque<std::vector<AsyncBuffer>> async_buffers;
     std::deque<AsyncBuffer> async_buffers_death_ring;
 
-
+    struct LRUItemParams {
+        using ObjectType = ImageId;
+        using TickType = u64;
+    };
+    Common::LeastRecentlyUsedCache<LRUItemParams> lru_cache;
 
  #ifdef YUZU_LEGACY
     static constexpr size_t TICKS_TO_DESTROY = 6;

src/video_core/textures/decoders.h

@@ -1,3 +1,6 @@
+// SPDX-FileCopyrightText: Copyright 2026 Eden Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
 // SPDX-FileCopyrightText: Copyright 2018 yuzu Emulator Project
 // SPDX-License-Identifier: GPL-2.0-or-later
 
@@ -23,24 +26,6 @@ constexpr u32 GOB_SIZE_SHIFT = GOB_SIZE_X_SHIFT + GOB_SIZE_Y_SHIFT + GOB_SIZE_Z_
 constexpr u32 SWIZZLE_X_BITS = 0b100101111;
 constexpr u32 SWIZZLE_Y_BITS = 0b011010000;
 
-using SwizzleTable = std::array<std::array<u32, GOB_SIZE_X>, GOB_SIZE_Y>;
-
-/**
- * This table represents the internal swizzle of a gob, in format 16 bytes x 2 sector packing.
- * Calculates the offset of an (x, y) position within a swizzled texture.
- * Taken from the Tegra X1 Technical Reference Manual. pages 1187-1188
- */
-constexpr SwizzleTable MakeSwizzleTable() {
-    SwizzleTable table{};
-    for (u32 y = 0; y < table.size(); ++y) {
-        for (u32 x = 0; x < table[0].size(); ++x) {
-            table[y][x] = ((x % 64) / 32) * 256 + ((y % 8) / 2) * 64 + ((x % 32) / 16) * 32 +
-                          (y % 2) * 16 + (x % 16);
-        }
-    }
-    return table;
-}
-
 /// Unswizzles a block linear texture into linear memory.
 void UnswizzleTexture(std::span<u8> output, std::span<const u8> input, u32 bytes_per_pixel,
                       u32 width, u32 height, u32 depth, u32 block_height, u32 block_depth,

src/video_core/vulkan_common/vulkan_device.h

@@ -321,32 +321,23 @@ public:
         return properties.properties.limits.maxPushConstantsSize;
     }
 
-    /// Returns the maximum size for shared memory.
-    u32 GetMaxComputeSharedMemorySize() const {
-        return properties.properties.limits.maxComputeSharedMemorySize;
-    }
-
-    /// Returns the maximum number of dynamic storage buffer descriptors per set.
-    u32 GetMaxDescriptorSetStorageBuffersDynamic() const {
-        return properties.properties.limits.maxDescriptorSetStorageBuffersDynamic;
-    }
-
-    /// Returns the maximum number of dynamic uniform buffer descriptors per set.
-    u32 GetMaxDescriptorSetUniformBuffersDynamic() const {
-        return properties.properties.limits.maxDescriptorSetUniformBuffersDynamic;
-    }
-
-    u32 GetMaxPerStageDescriptorSampledImages() const {
-        return properties.properties.limits.maxPerStageDescriptorSampledImages;
-    }
-
-    u32 GetMaxPerStageResources() const {
-        return properties.properties.limits.maxPerStageResources;
-    }
-
-    u32 GetMaxDescriptorSetSampledImages() const {
-        return properties.properties.limits.maxDescriptorSetSampledImages;
-    }
+#define FN_MAX_LIMIT_LIST \
+    FN_MAX_LIMIT_ELEM(ComputeSharedMemorySize) \
+    FN_MAX_LIMIT_ELEM(PerStageDescriptorSampledImages) \
+    FN_MAX_LIMIT_ELEM(PerStageResources) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetSamplers) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetUniformBuffers) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetUniformBuffersDynamic) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetStorageBuffers) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetStorageBuffersDynamic) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetSampledImages) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetStorageImages) \
+    FN_MAX_LIMIT_ELEM(DescriptorSetInputAttachments)
+#define FN_MAX_LIMIT_ELEM(name) \
+    u32 GetMax##name() const { return properties.properties.limits.max##name; }
+FN_MAX_LIMIT_LIST
+#undef FN_MAX_LIMIT_ELEM
+#undef FN_MAX_LIMIT_LIST
 
     /// Returns float control properties of the device.
     const VkPhysicalDeviceFloatControlsPropertiesKHR& FloatControlProperties() const {
@@ -897,6 +888,16 @@ public:
                features.bit16_storage.storageBuffer16BitAccess;
     }
 
+    /// Returns true if the device supports reading 8-bit values from a storage buffer.
+    bool IsStorageBuffer8BitAccessSupported() const {
+        return features.bit8_storage.storageBuffer8BitAccess;
+    }
+
+    /// Returns true if the device supports reading 16-bit values from a storage buffer.
+    bool IsStorageBuffer16BitAccessSupported() const {
+        return features.bit16_storage.storageBuffer16BitAccess;
+    }
+
     [[nodiscard]] static constexpr bool CheckBrokenCompute(VkDriverId driver_id,
                                                            u32 driver_version) {
         if (driver_id == VK_DRIVER_ID_INTEL_PROPRIETARY_WINDOWS) {

tools/maxwell-ir/main.cpp

@@ -44,6 +44,7 @@ int IrShaderRecompilerImpl(int argc, char *argv[]) {
     host_info.support_geometry_shader_passthrough = true;
     host_info.support_conditional_barrier = true;
     host_info.min_ssbo_alignment = 0;
+    host_info.ApplyDescriptorLimitPolicy();
     auto program = Shader::Maxwell::TranslateProgram(inst_pool, block_pool, env, cfg, host_info);
     auto const dumped_ir = Shader::IR::DumpProgram(program);
     std::printf("%s\n", dumped_ir.c_str());

tools/maxwell-spirv/spirv_recompiler_impl.cpp

@@ -52,6 +52,7 @@ int SpirvShaderRecompilerImpl(int argc, char *argv[]) {
     host_info.support_geometry_shader_passthrough = true;
     host_info.support_conditional_barrier = true;
     host_info.min_ssbo_alignment = 0;
+    host_info.ApplyDescriptorLimitPolicy();
     auto program = Shader::Maxwell::TranslateProgram(inst_pool, block_pool, env, cfg, host_info);
 
     // IR::Program TranslateProgram(ObjectPool<IR::Inst>& inst_pool, ObjectPool<IR::Block>& block_pool,