39 files changed, 281 insertions, 117 deletions

diff --git a/src/core/hle/kernel/hle_ipc.cpp b/src/core/hle/kernel/hle_ipc.cpp
index 91d94025c..ba0eac4c2 100644
--- a/src/core/hle/kernel/hle_ipc.cpp
+++ b/src/core/hle/kernel/hle_ipc.cpp
@@ -13,7 +13,6 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/logging/log.h"
-#include "core/core.h"
 #include "core/hle/ipc_helpers.h"
 #include "core/hle/kernel/handle_table.h"
 #include "core/hle/kernel/hle_ipc.h"
@@ -57,7 +56,6 @@ std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread(
             return true;
         });
 
-    auto& kernel = Core::System::GetInstance().Kernel();
     if (!writable_event) {
         // Create event if not provided
         const auto pair = WritableEvent::CreateEventPair(kernel, "HLE Pause Event: " + reason);
@@ -79,9 +77,11 @@ std::shared_ptr<WritableEvent> HLERequestContext::SleepClientThread(
     return writable_event;
 }
 
-HLERequestContext::HLERequestContext(std::shared_ptr<Kernel::ServerSession> server_session,
+HLERequestContext::HLERequestContext(KernelCore& kernel, Core::Memory::Memory& memory,
+                                     std::shared_ptr<ServerSession> server_session,
                                      std::shared_ptr<Thread> thread)
-    : server_session(std::move(server_session)), thread(std::move(thread)) {
+    : server_session(std::move(server_session)),
+      thread(std::move(thread)), kernel{kernel}, memory{memory} {
     cmd_buf[0] = 0;
 }
 
@@ -216,7 +216,6 @@ ResultCode HLERequestContext::PopulateFromIncomingCommandBuffer(const HandleTabl
 ResultCode HLERequestContext::WriteToOutgoingCommandBuffer(Thread& thread) {
     auto& owner_process = *thread.GetOwnerProcess();
     auto& handle_table = owner_process.GetHandleTable();
-    auto& memory = Core::System::GetInstance().Memory();
 
     std::array<u32, IPC::COMMAND_BUFFER_LENGTH> dst_cmdbuf;
     memory.ReadBlock(owner_process, thread.GetTLSAddress(), dst_cmdbuf.data(),
@@ -286,7 +285,6 @@ std::vector<u8> HLERequestContext::ReadBuffer(std::size_t buffer_index) const {
     std::vector<u8> buffer;
     const bool is_buffer_a{BufferDescriptorA().size() > buffer_index &&
                            BufferDescriptorA()[buffer_index].Size()};
-    auto& memory = Core::System::GetInstance().Memory();
 
     if (is_buffer_a) {
         ASSERT_MSG(BufferDescriptorA().size() > buffer_index,
@@ -319,7 +317,6 @@ std::size_t HLERequestContext::WriteBuffer(const void* buffer, std::size_t size,
         size = buffer_size; // TODO(bunnei): This needs to be HW tested
     }
 
-    auto& memory = Core::System::GetInstance().Memory();
     if (is_buffer_b) {
         ASSERT_MSG(BufferDescriptorB().size() > buffer_index,
                    "BufferDescriptorB invalid buffer_index {}", buffer_index);
diff --git a/src/core/hle/kernel/hle_ipc.h b/src/core/hle/kernel/hle_ipc.h
index af3330297..b31673928 100644
--- a/src/core/hle/kernel/hle_ipc.h
+++ b/src/core/hle/kernel/hle_ipc.h
@@ -19,6 +19,10 @@
 
 union ResultCode;
 
+namespace Core::Memory {
+class Memory;
+}
+
 namespace Service {
 class ServiceFrameworkBase;
 }
@@ -28,6 +32,7 @@ namespace Kernel {
 class Domain;
 class HandleTable;
 class HLERequestContext;
+class KernelCore;
 class Process;
 class ServerSession;
 class Thread;
@@ -98,7 +103,8 @@ protected:
  */
 class HLERequestContext {
 public:
-    explicit HLERequestContext(std::shared_ptr<ServerSession> session,
+    explicit HLERequestContext(KernelCore& kernel, Core::Memory::Memory& memory,
+                               std::shared_ptr<ServerSession> session,
                                std::shared_ptr<Thread> thread);
     ~HLERequestContext();
 
@@ -305,6 +311,9 @@ private:
 
     std::vector<std::shared_ptr<SessionRequestHandler>> domain_request_handlers;
     bool is_thread_waiting{};
+
+    KernelCore& kernel;
+    Core::Memory::Memory& memory;
 };
 
 } // namespace Kernel
diff --git a/src/core/hle/kernel/memory/address_space_info.cpp b/src/core/hle/kernel/memory/address_space_info.cpp
index 27fae05e7..a523a2502 100644
--- a/src/core/hle/kernel/memory/address_space_info.cpp
+++ b/src/core/hle/kernel/memory/address_space_info.cpp
@@ -2,8 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-// This file references various implementation details from Atmosphère, an open-source firmware for
-// the Nintendo Switch. Copyright 2018-2020 Atmosphère-NX.
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
 
 #include <array>
 
@@ -49,18 +49,18 @@ constexpr bool IsAllowedIndexForAddress(std::size_t index) {
     return index < std::size(AddressSpaceInfos) && AddressSpaceInfos[index].GetAddress() != Invalid;
 }
 
-constexpr std::size_t
-    AddressSpaceIndices32Bit[static_cast<std::size_t>(AddressSpaceInfo::Type::Count)]{
+constexpr std::array<std::size_t, static_cast<std::size_t>(AddressSpaceInfo::Type::Count)>
+    AddressSpaceIndices32Bit{
         0, 1, 0, 2, 0, 3,
     };
 
-constexpr std::size_t
-    AddressSpaceIndices36Bit[static_cast<std::size_t>(AddressSpaceInfo::Type::Count)]{
+constexpr std::array<std::size_t, static_cast<std::size_t>(AddressSpaceInfo::Type::Count)>
+    AddressSpaceIndices36Bit{
         4, 5, 4, 6, 4, 7,
     };
 
-constexpr std::size_t
-    AddressSpaceIndices39Bit[static_cast<std::size_t>(AddressSpaceInfo::Type::Count)]{
+constexpr std::array<std::size_t, static_cast<std::size_t>(AddressSpaceInfo::Type::Count)>
+    AddressSpaceIndices39Bit{
         9, 8, 8, 10, 12, 11,
     };
 
diff --git a/src/core/hle/kernel/memory/address_space_info.h b/src/core/hle/kernel/memory/address_space_info.h
index cc9a6421e..c479890be 100644
--- a/src/core/hle/kernel/memory/address_space_info.h
+++ b/src/core/hle/kernel/memory/address_space_info.h
@@ -2,12 +2,11 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-// This file references various implementation details from Atmosphère, an open-source firmware for
-// the Nintendo Switch. Copyright 2018-2020 Atmosphère-NX.
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
 
 #pragma once
 
-#include "common/common_funcs.h"
 #include "common/common_types.h"
 
 namespace Kernel::Memory {
diff --git a/src/core/hle/kernel/memory/memory_block.h b/src/core/hle/kernel/memory/memory_block.h
index 9db1f7b39..9d7839d08 100644
--- a/src/core/hle/kernel/memory/memory_block.h
+++ b/src/core/hle/kernel/memory/memory_block.h
@@ -2,8 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-// This file references various implementation details from Atmosphère, an open-source firmware for
-// the Nintendo Switch. Copyright 2018-2020 Atmosphère-NX.
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
 
 #pragma once
 
diff --git a/src/core/hle/kernel/memory/memory_block_manager.cpp b/src/core/hle/kernel/memory/memory_block_manager.cpp
index 900395c37..0732fa5a1 100644
--- a/src/core/hle/kernel/memory/memory_block_manager.cpp
+++ b/src/core/hle/kernel/memory/memory_block_manager.cpp
@@ -67,7 +67,6 @@ void MemoryBlockManager::Update(VAddr addr, std::size_t num_pages, MemoryState p
                                 MemoryPermission prev_perm, MemoryAttribute prev_attribute,
                                 MemoryState state, MemoryPermission perm,
                                 MemoryAttribute attribute) {
-    const std::size_t prev_count{memory_block_tree.size()};
     const VAddr end_addr{addr + num_pages * PageSize};
     iterator node{memory_block_tree.begin()};
 
@@ -109,7 +108,6 @@ void MemoryBlockManager::Update(VAddr addr, std::size_t num_pages, MemoryState p
 
 void MemoryBlockManager::Update(VAddr addr, std::size_t num_pages, MemoryState state,
                                 MemoryPermission perm, MemoryAttribute attribute) {
-    const std::size_t prev_count{memory_block_tree.size()};
     const VAddr end_addr{addr + num_pages * PageSize};
     iterator node{memory_block_tree.begin()};
 
@@ -145,7 +143,6 @@ void MemoryBlockManager::Update(VAddr addr, std::size_t num_pages, MemoryState s
 
 void MemoryBlockManager::UpdateLock(VAddr addr, std::size_t num_pages, LockFunc&& lock_func,
                                     MemoryPermission perm) {
-    const std::size_t prev_count{memory_block_tree.size()};
     const VAddr end_addr{addr + num_pages * PageSize};
     iterator node{memory_block_tree.begin()};
 
diff --git a/src/core/hle/kernel/memory/memory_block_manager.h b/src/core/hle/kernel/memory/memory_block_manager.h
index 9451b5df6..6e1d41075 100644
--- a/src/core/hle/kernel/memory/memory_block_manager.h
+++ b/src/core/hle/kernel/memory/memory_block_manager.h
@@ -6,7 +6,6 @@
 
 #include <functional>
 #include <list>
-#include <memory>
 
 #include "common/common_types.h"
 #include "core/hle/kernel/memory/memory_block.h"
diff --git a/src/core/hle/kernel/memory/memory_manager.cpp b/src/core/hle/kernel/memory/memory_manager.cpp
index 3cd4f9e85..6b432e1b2 100644
--- a/src/core/hle/kernel/memory/memory_manager.cpp
+++ b/src/core/hle/kernel/memory/memory_manager.cpp
@@ -104,9 +104,9 @@ ResultCode MemoryManager::Allocate(PageLinkedList& page_list, std::size_t num_pa
     // Ensure that we don't leave anything un-freed
     auto group_guard = detail::ScopeExit([&] {
         for (const auto& it : page_list.Nodes()) {
-            const auto num_pages{std::min(
+            const auto min_num_pages{std::min(
                 it.GetNumPages(), (chosen_manager.GetEndAddress() - it.GetAddress()) / PageSize)};
-            chosen_manager.Free(it.GetAddress(), num_pages);
+            chosen_manager.Free(it.GetAddress(), min_num_pages);
         }
     });
 
@@ -165,9 +165,9 @@ ResultCode MemoryManager::Free(PageLinkedList& page_list, std::size_t num_pages,
 
     // Free all of the pages
     for (const auto& it : page_list.Nodes()) {
-        const auto num_pages{std::min(
+        const auto min_num_pages{std::min(
             it.GetNumPages(), (chosen_manager.GetEndAddress() - it.GetAddress()) / PageSize)};
-        chosen_manager.Free(it.GetAddress(), num_pages);
+        chosen_manager.Free(it.GetAddress(), min_num_pages);
     }
 
     return RESULT_SUCCESS;
diff --git a/src/core/hle/kernel/memory/memory_manager.h b/src/core/hle/kernel/memory/memory_manager.h
index b078d7a5e..3cf444857 100644
--- a/src/core/hle/kernel/memory/memory_manager.h
+++ b/src/core/hle/kernel/memory/memory_manager.h
@@ -7,7 +7,6 @@
 #include <array>
 #include <mutex>
 
-#include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "core/hle/kernel/memory/page_heap.h"
 #include "core/hle/result.h"
diff --git a/src/core/hle/kernel/memory/page_heap.cpp b/src/core/hle/kernel/memory/page_heap.cpp
index efcbb3cad..0ab1f7205 100644
--- a/src/core/hle/kernel/memory/page_heap.cpp
+++ b/src/core/hle/kernel/memory/page_heap.cpp
@@ -2,8 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-// This file references various implementation details from Atmosphère, an open-source firmware for
-// the Nintendo Switch. Copyright 2018-2020 Atmosphère-NX.
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
 
 #include "core/core.h"
 #include "core/hle/kernel/memory/page_heap.h"
diff --git a/src/core/hle/kernel/memory/page_heap.h b/src/core/hle/kernel/memory/page_heap.h
index 380c3f5a1..22b0de860 100644
--- a/src/core/hle/kernel/memory/page_heap.h
+++ b/src/core/hle/kernel/memory/page_heap.h
@@ -2,8 +2,8 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-// This file references various implementation details from Atmosphère, an open-source firmware for
-// the Nintendo Switch. Copyright 2018-2020 Atmosphère-NX.
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
 
 #pragma once
 
diff --git a/src/core/hle/kernel/memory/page_linked_list.h b/src/core/hle/kernel/memory/page_linked_list.h
index 0668d00c6..45dc13eaf 100644
--- a/src/core/hle/kernel/memory/page_linked_list.h
+++ b/src/core/hle/kernel/memory/page_linked_list.h
@@ -7,7 +7,6 @@
 #include <list>
 
 #include "common/assert.h"
-#include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "core/hle/kernel/memory/memory_types.h"
 #include "core/hle/result.h"
diff --git a/src/core/hle/kernel/memory/page_table.cpp b/src/core/hle/kernel/memory/page_table.cpp
index 3281611f8..5d6aac00f 100644
--- a/src/core/hle/kernel/memory/page_table.cpp
+++ b/src/core/hle/kernel/memory/page_table.cpp
@@ -6,7 +6,6 @@
 #include "common/assert.h"
 #include "common/scope_exit.h"
 #include "core/core.h"
-#include "core/device_memory.h"
 #include "core/hle/kernel/errors.h"
 #include "core/hle/kernel/kernel.h"
 #include "core/hle/kernel/memory/address_space_info.h"
diff --git a/src/core/hle/kernel/memory/page_table.h b/src/core/hle/kernel/memory/page_table.h
index a867aa050..ce0d38849 100644
--- a/src/core/hle/kernel/memory/page_table.h
+++ b/src/core/hle/kernel/memory/page_table.h
@@ -4,16 +4,15 @@
 
 #pragma once
 
-#include <list>
 #include <memory>
 #include <mutex>
 
-#include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/page_table.h"
 #include "core/file_sys/program_metadata.h"
 #include "core/hle/kernel/memory/memory_block.h"
 #include "core/hle/kernel/memory/memory_manager.h"
+#include "core/hle/result.h"
 
 namespace Core {
 class System;
diff --git a/src/core/hle/kernel/memory/slab_heap.h b/src/core/hle/kernel/memory/slab_heap.h
index be95fc3f7..465eaddb3 100644
--- a/src/core/hle/kernel/memory/slab_heap.h
+++ b/src/core/hle/kernel/memory/slab_heap.h
@@ -2,15 +2,14 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-// This file references various implementation details from Atmosphère, an open-source firmware for
-// the Nintendo Switch. Copyright 2018-2020 Atmosphère-NX.
+// This file references various implementation details from Atmosphere, an open-source firmware for
+// the Nintendo Switch. Copyright 2018-2020 Atmosphere-NX.
 
 #pragma once
 
 #include <atomic>
 
 #include "common/assert.h"
-#include "common/common_funcs.h"
 #include "common/common_types.h"
 
 namespace Kernel::Memory {
diff --git a/src/core/hle/kernel/memory/system_control.cpp b/src/core/hle/kernel/memory/system_control.cpp
index 9cae3c6cb..2f98e9c4c 100644
--- a/src/core/hle/kernel/memory/system_control.cpp
+++ b/src/core/hle/kernel/memory/system_control.cpp
@@ -2,8 +2,6 @@
 // Licensed under GPLv2 or any later version
 // Refer to the license.txt file included.
 
-#pragma once
-
 #include <random>
 
 #include "core/hle/kernel/memory/system_control.h"
diff --git a/src/core/hle/kernel/server_session.cpp b/src/core/hle/kernel/server_session.cpp
index 0f102ca44..25438b86b 100644
--- a/src/core/hle/kernel/server_session.cpp
+++ b/src/core/hle/kernel/server_session.cpp
@@ -137,8 +137,8 @@ ResultCode ServerSession::HandleDomainSyncRequest(Kernel::HLERequestContext& con
 ResultCode ServerSession::QueueSyncRequest(std::shared_ptr<Thread> thread,
                                            Core::Memory::Memory& memory) {
     u32* cmd_buf{reinterpret_cast<u32*>(memory.GetPointer(thread->GetTLSAddress()))};
-    std::shared_ptr<Kernel::HLERequestContext> context{
-        std::make_shared<Kernel::HLERequestContext>(SharedFrom(this), std::move(thread))};
+    auto context =
+        std::make_shared<HLERequestContext>(kernel, memory, SharedFrom(this), std::move(thread));
 
     context->PopulateFromIncomingCommandBuffer(kernel.CurrentProcess()->GetHandleTable(), cmd_buf);
     request_queue.Push(std::move(context));
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index ff53282c9..d23c53843 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -8,6 +8,7 @@ add_library(video_core STATIC
     dma_pusher.h
     engines/const_buffer_engine_interface.h
     engines/const_buffer_info.h
+    engines/engine_interface.h
     engines/engine_upload.cpp
     engines/engine_upload.h
     engines/fermi_2d.cpp
diff --git a/src/video_core/dma_pusher.cpp b/src/video_core/dma_pusher.cpp
index 16311f05e..bdc023d54 100644
--- a/src/video_core/dma_pusher.cpp
+++ b/src/video_core/dma_pusher.cpp
@@ -27,6 +27,8 @@ void DmaPusher::DispatchCalls() {
 
     dma_pushbuffer_subindex = 0;
 
+    dma_state.is_last_call = true;
+
     while (system.IsPoweredOn()) {
         if (!Step()) {
             break;
@@ -82,9 +84,11 @@ bool DmaPusher::Step() {
                     index);
                 CallMultiMethod(&command_header.argument, max_write);
                 dma_state.method_count -= max_write;
+                dma_state.is_last_call = true;
                 index += max_write;
                 continue;
             } else {
+                dma_state.is_last_call = dma_state.method_count <= 1;
                 CallMethod(command_header.argument);
             }
 
@@ -144,12 +148,22 @@ void DmaPusher::SetState(const CommandHeader& command_header) {
 }
 
 void DmaPusher::CallMethod(u32 argument) const {
-    gpu.CallMethod({dma_state.method, argument, dma_state.subchannel, dma_state.method_count});
+    if (dma_state.method < non_puller_methods) {
+        gpu.CallMethod({dma_state.method, argument, dma_state.subchannel, dma_state.method_count});
+    } else {
+        subchannels[dma_state.subchannel]->CallMethod(dma_state.method, argument,
+                                                      dma_state.is_last_call);
+    }
 }
 
 void DmaPusher::CallMultiMethod(const u32* base_start, u32 num_methods) const {
-    gpu.CallMultiMethod(dma_state.method, dma_state.subchannel, base_start, num_methods,
-                        dma_state.method_count);
+    if (dma_state.method < non_puller_methods) {
+        gpu.CallMultiMethod(dma_state.method, dma_state.subchannel, base_start, num_methods,
+                            dma_state.method_count);
+    } else {
+        subchannels[dma_state.subchannel]->CallMultiMethod(dma_state.method, base_start,
+                                                           num_methods, dma_state.method_count);
+    }
 }
 
 } // namespace Tegra
diff --git a/src/video_core/dma_pusher.h b/src/video_core/dma_pusher.h
index 6cef71306..e8b714e94 100644
--- a/src/video_core/dma_pusher.h
+++ b/src/video_core/dma_pusher.h
@@ -4,11 +4,13 @@
 
 #pragma once
 
+#include <array>
 #include <vector>
 #include <queue>
 
 #include "common/bit_field.h"
 #include "common/common_types.h"
+#include "video_core/engines/engine_interface.h"
 
 namespace Core {
 class System;
@@ -69,7 +71,13 @@ public:
 
     void DispatchCalls();
 
+    void BindSubchannel(Tegra::Engines::EngineInterface* engine, u32 subchannel_id) {
+        subchannels[subchannel_id] = engine;
+    }
+
 private:
+    static constexpr u32 non_puller_methods = 0x40;
+    static constexpr u32 max_subchannels = 8;
     bool Step();
 
     void SetState(const CommandHeader& command_header);
@@ -88,6 +96,7 @@ private:
         u32 method_count;      ///< Current method count
         u32 length_pending;    ///< Large NI command length pending
         bool non_incrementing; ///< Current command's NI flag
+        bool is_last_call;
     };
 
     DmaState dma_state{};
@@ -96,6 +105,8 @@ private:
     GPUVAddr dma_mget{};  ///< main pushbuffer last read address
     bool ib_enable{true}; ///< IB mode enabled
 
+    std::array<Tegra::Engines::EngineInterface*, max_subchannels> subchannels{};
+
     GPU& gpu;
     Core::System& system;
 };
diff --git a/src/video_core/engines/engine_interface.h b/src/video_core/engines/engine_interface.h
new file mode 100644
index 000000000..18a9db7e6
--- /dev/null
+++ b/src/video_core/engines/engine_interface.h
@@ -0,0 +1,22 @@
+// Copyright 2020 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include <type_traits>
+#include "common/common_types.h"
+
+namespace Tegra::Engines {
+
+class EngineInterface {
+public:
+    /// Write the value to the register identified by method.
+    virtual void CallMethod(u32 method, u32 method_argument, bool is_last_call) = 0;
+
+    /// Write multiple values to the register identified by method.
+    virtual void CallMultiMethod(u32 method, const u32* base_start, u32 amount,
+                                 u32 methods_pending) = 0;
+};
+
+} // namespace Tegra::Engines
diff --git a/src/video_core/engines/fermi_2d.cpp b/src/video_core/engines/fermi_2d.cpp
index 8a47614d2..ff10ff40d 100644
--- a/src/video_core/engines/fermi_2d.cpp
+++ b/src/video_core/engines/fermi_2d.cpp
@@ -12,13 +12,13 @@ namespace Tegra::Engines {
 
 Fermi2D::Fermi2D(VideoCore::RasterizerInterface& rasterizer) : rasterizer{rasterizer} {}
 
-void Fermi2D::CallMethod(const GPU::MethodCall& method_call) {
-    ASSERT_MSG(method_call.method < Regs::NUM_REGS,
+void Fermi2D::CallMethod(u32 method, u32 method_argument, bool is_last_call) {
+    ASSERT_MSG(method < Regs::NUM_REGS,
                "Invalid Fermi2D register, increase the size of the Regs structure");
 
-    regs.reg_array[method_call.method] = method_call.argument;
+    regs.reg_array[method] = method_argument;
 
-    switch (method_call.method) {
+    switch (method) {
     // Trigger the surface copy on the last register write. This is blit_src_y, but this is 64-bit,
     // so trigger on the second 32-bit write.
     case FERMI2D_REG_INDEX(blit_src_y) + 1: {
@@ -30,7 +30,7 @@ void Fermi2D::CallMethod(const GPU::MethodCall& method_call) {
 
 void Fermi2D::CallMultiMethod(u32 method, const u32* base_start, u32 amount, u32 methods_pending) {
     for (std::size_t i = 0; i < amount; i++) {
-        CallMethod({method, base_start[i], 0, methods_pending - static_cast<u32>(i)});
+        CallMethod(method, base_start[i], methods_pending - static_cast<u32>(i) <= 1);
     }
 }
 
diff --git a/src/video_core/engines/fermi_2d.h b/src/video_core/engines/fermi_2d.h
index 939a5966d..8f37d053f 100644
--- a/src/video_core/engines/fermi_2d.h
+++ b/src/video_core/engines/fermi_2d.h
@@ -10,6 +10,7 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "common/math_util.h"
+#include "video_core/engines/engine_interface.h"
 #include "video_core/gpu.h"
 
 namespace Tegra {
@@ -31,16 +32,17 @@ namespace Tegra::Engines {
 #define FERMI2D_REG_INDEX(field_name)                                                              \
     (offsetof(Tegra::Engines::Fermi2D::Regs, field_name) / sizeof(u32))
 
-class Fermi2D final {
+class Fermi2D final : public EngineInterface {
 public:
     explicit Fermi2D(VideoCore::RasterizerInterface& rasterizer);
     ~Fermi2D() = default;
 
     /// Write the value to the register identified by method.
-    void CallMethod(const GPU::MethodCall& method_call);
+    void CallMethod(u32 method, u32 method_argument, bool is_last_call) override;
 
     /// Write multiple values to the register identified by method.
-    void CallMultiMethod(u32 method, const u32* base_start, u32 amount, u32 methods_pending);
+    void CallMultiMethod(u32 method, const u32* base_start, u32 amount,
+                         u32 methods_pending) override;
 
     enum class Origin : u32 {
         Center = 0,
diff --git a/src/video_core/engines/kepler_compute.cpp b/src/video_core/engines/kepler_compute.cpp
index 00a12175f..f6237fc6a 100644
--- a/src/video_core/engines/kepler_compute.cpp
+++ b/src/video_core/engines/kepler_compute.cpp
@@ -24,20 +24,19 @@ KeplerCompute::KeplerCompute(Core::System& system, VideoCore::RasterizerInterfac
 
 KeplerCompute::~KeplerCompute() = default;
 
-void KeplerCompute::CallMethod(const GPU::MethodCall& method_call) {
-    ASSERT_MSG(method_call.method < Regs::NUM_REGS,
+void KeplerCompute::CallMethod(u32 method, u32 method_argument, bool is_last_call) {
+    ASSERT_MSG(method < Regs::NUM_REGS,
                "Invalid KeplerCompute register, increase the size of the Regs structure");
 
-    regs.reg_array[method_call.method] = method_call.argument;
+    regs.reg_array[method] = method_argument;
 
-    switch (method_call.method) {
+    switch (method) {
     case KEPLER_COMPUTE_REG_INDEX(exec_upload): {
         upload_state.ProcessExec(regs.exec_upload.linear != 0);
         break;
     }
     case KEPLER_COMPUTE_REG_INDEX(data_upload): {
-        const bool is_last_call = method_call.IsLastCall();
-        upload_state.ProcessData(method_call.argument, is_last_call);
+        upload_state.ProcessData(method_argument, is_last_call);
         if (is_last_call) {
             system.GPU().Maxwell3D().OnMemoryWrite();
         }
@@ -54,7 +53,7 @@ void KeplerCompute::CallMethod(const GPU::MethodCall& method_call) {
 void KeplerCompute::CallMultiMethod(u32 method, const u32* base_start, u32 amount,
                                     u32 methods_pending) {
     for (std::size_t i = 0; i < amount; i++) {
-        CallMethod({method, base_start[i], 0, methods_pending - static_cast<u32>(i)});
+        CallMethod(method, base_start[i], methods_pending - static_cast<u32>(i) <= 1);
     }
 }
 
diff --git a/src/video_core/engines/kepler_compute.h b/src/video_core/engines/kepler_compute.h
index fe55fdfd0..18ceedfaf 100644
--- a/src/video_core/engines/kepler_compute.h
+++ b/src/video_core/engines/kepler_compute.h
@@ -11,6 +11,7 @@
 #include "common/common_funcs.h"
 #include "common/common_types.h"
 #include "video_core/engines/const_buffer_engine_interface.h"
+#include "video_core/engines/engine_interface.h"
 #include "video_core/engines/engine_upload.h"
 #include "video_core/engines/shader_type.h"
 #include "video_core/gpu.h"
@@ -39,7 +40,7 @@ namespace Tegra::Engines {
 #define KEPLER_COMPUTE_REG_INDEX(field_name)                                                       \
     (offsetof(Tegra::Engines::KeplerCompute::Regs, field_name) / sizeof(u32))
 
-class KeplerCompute final : public ConstBufferEngineInterface {
+class KeplerCompute final : public ConstBufferEngineInterface, public EngineInterface {
 public:
     explicit KeplerCompute(Core::System& system, VideoCore::RasterizerInterface& rasterizer,
                            MemoryManager& memory_manager);
@@ -200,10 +201,11 @@ public:
                   "KeplerCompute LaunchParams has wrong size");
 
     /// Write the value to the register identified by method.
-    void CallMethod(const GPU::MethodCall& method_call);
+    void CallMethod(u32 method, u32 method_argument, bool is_last_call) override;
 
     /// Write multiple values to the register identified by method.
-    void CallMultiMethod(u32 method, const u32* base_start, u32 amount, u32 methods_pending);
+    void CallMultiMethod(u32 method, const u32* base_start, u32 amount,
+                         u32 methods_pending) override;
 
     Texture::FullTextureInfo GetTexture(std::size_t offset) const;
 
diff --git a/src/video_core/engines/kepler_memory.cpp b/src/video_core/engines/kepler_memory.cpp
index 586ff15dc..dc71b2eec 100644
--- a/src/video_core/engines/kepler_memory.cpp
+++ b/src/video_core/engines/kepler_memory.cpp
@@ -19,20 +19,19 @@ KeplerMemory::KeplerMemory(Core::System& system, MemoryManager& memory_manager)
 
 KeplerMemory::~KeplerMemory() = default;
 
-void KeplerMemory::CallMethod(const GPU::MethodCall& method_call) {
-    ASSERT_MSG(method_call.method < Regs::NUM_REGS,
+void KeplerMemory::CallMethod(u32 method, u32 method_argument, bool is_last_call) {
+    ASSERT_MSG(method < Regs::NUM_REGS,
                "Invalid KeplerMemory register, increase the size of the Regs structure");
 
-    regs.reg_array[method_call.method] = method_call.argument;
+    regs.reg_array[method] = method_argument;
 
-    switch (method_call.method) {
+    switch (method) {
     case KEPLERMEMORY_REG_INDEX(exec): {
         upload_state.ProcessExec(regs.exec.linear != 0);
         break;
     }
     case KEPLERMEMORY_REG_INDEX(data): {
-        const bool is_last_call = method_call.IsLastCall();
-        upload_state.ProcessData(method_call.argument, is_last_call);
+        upload_state.ProcessData(method_argument, is_last_call);
         if (is_last_call) {
             system.GPU().Maxwell3D().OnMemoryWrite();
         }
@@ -44,7 +43,7 @@ void KeplerMemory::CallMethod(const GPU::MethodCall& method_call) {
 void KeplerMemory::CallMultiMethod(u32 method, const u32* base_start, u32 amount,
                                    u32 methods_pending) {
     for (std::size_t i = 0; i < amount; i++) {
-        CallMethod({method, base_start[i], 0, methods_pending - static_cast<u32>(i)});
+        CallMethod(method, base_start[i], methods_pending - static_cast<u32>(i) <= 1);
     }
 }
 
diff --git a/src/video_core/engines/kepler_memory.h b/src/video_core/engines/kepler_memory.h
index bb26fb030..5b7f71a00 100644
--- a/src/video_core/engines/kepler_memory.h
+++ b/src/video_core/engines/kepler_memory.h
@@ -10,6 +10,7 @@
 #include "common/bit_field.h"
 #include "common/common_funcs.h"
 #include "common/common_types.h"
+#include "video_core/engines/engine_interface.h"
 #include "video_core/engines/engine_upload.h"
 #include "video_core/gpu.h"
 
@@ -32,16 +33,17 @@ namespace Tegra::Engines {
 #define KEPLERMEMORY_REG_INDEX(field_name)                                                         \
     (offsetof(Tegra::Engines::KeplerMemory::Regs, field_name) / sizeof(u32))
 
-class KeplerMemory final {
+class KeplerMemory final : public EngineInterface {
 public:
     KeplerMemory(Core::System& system, MemoryManager& memory_manager);
     ~KeplerMemory();
 
     /// Write the value to the register identified by method.
-    void CallMethod(const GPU::MethodCall& method_call);
+    void CallMethod(u32 method, u32 method_argument, bool is_last_call) override;
 
     /// Write multiple values to the register identified by method.
-    void CallMultiMethod(u32 method, const u32* base_start, u32 amount, u32 methods_pending);
+    void CallMultiMethod(u32 method, const u32* base_start, u32 amount,
+                         u32 methods_pending) override;
 
     struct Regs {
         static constexpr size_t NUM_REGS = 0x7F;
diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp
index 8155f6e2e..024c9e43b 100644
--- a/src/video_core/engines/maxwell_3d.cpp
+++ b/src/video_core/engines/maxwell_3d.cpp
@@ -131,12 +131,10 @@ void Maxwell3D::CallMacroMethod(u32 method, std::size_t num_parameters, const u3
     }
 }
 
-void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
-    const u32 method = method_call.method;
-
+void Maxwell3D::CallMethod(u32 method, u32 method_argument, bool is_last_call) {
     if (method == cb_data_state.current) {
-        regs.reg_array[method] = method_call.argument;
-        ProcessCBData(method_call.argument);
+        regs.reg_array[method] = method_argument;
+        ProcessCBData(method_argument);
         return;
     } else if (cb_data_state.current != null_cb_data) {
         FinishCBData();
@@ -159,10 +157,10 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
             executing_macro = method;
         }
 
-        macro_params.push_back(method_call.argument);
+        macro_params.push_back(method_argument);
 
         // Call the macro when there are no more parameters in the command buffer
-        if (method_call.IsLastCall()) {
+        if (is_last_call) {
             CallMacroMethod(executing_macro, macro_params.size(), macro_params.data());
             macro_params.clear();
         }
@@ -172,7 +170,7 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
     ASSERT_MSG(method < Regs::NUM_REGS,
                "Invalid Maxwell3D register, increase the size of the Regs structure");
 
-    u32 arg = method_call.argument;
+    u32 arg = method_argument;
     // Keep track of the register value in shadow_state when requested.
     if (shadow_state.shadow_ram_control == Regs::ShadowRamControl::Track ||
         shadow_state.shadow_ram_control == Regs::ShadowRamControl::TrackWithFilter) {
@@ -195,7 +193,7 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
         break;
     }
     case MAXWELL3D_REG_INDEX(shadow_ram_control): {
-        shadow_state.shadow_ram_control = static_cast<Regs::ShadowRamControl>(method_call.argument);
+        shadow_state.shadow_ram_control = static_cast<Regs::ShadowRamControl>(method_argument);
         break;
     }
     case MAXWELL3D_REG_INDEX(macros.data): {
@@ -278,7 +276,6 @@ void Maxwell3D::CallMethod(const GPU::MethodCall& method_call) {
         break;
     }
     case MAXWELL3D_REG_INDEX(data_upload): {
-        const bool is_last_call = method_call.IsLastCall();
         upload_state.ProcessData(arg, is_last_call);
         if (is_last_call) {
             OnMemoryWrite();
@@ -336,7 +333,7 @@ void Maxwell3D::CallMultiMethod(u32 method, const u32* base_start, u32 amount,
     }
     default: {
         for (std::size_t i = 0; i < amount; i++) {
-            CallMethod({method, base_start[i], 0, methods_pending - static_cast<u32>(i)});
+            CallMethod(method, base_start[i], methods_pending - static_cast<u32>(i) <= 1);
         }
     }
     }
@@ -366,16 +363,15 @@ void Maxwell3D::StepInstance(const MMEDrawMode expected_mode, const u32 count) {
     StepInstance(expected_mode, count);
 }
 
-void Maxwell3D::CallMethodFromMME(const GPU::MethodCall& method_call) {
-    const u32 method = method_call.method;
+void Maxwell3D::CallMethodFromMME(u32 method, u32 method_argument) {
     if (mme_inline[method]) {
-        regs.reg_array[method] = method_call.argument;
+        regs.reg_array[method] = method_argument;
         if (method == MAXWELL3D_REG_INDEX(vertex_buffer.count) ||
             method == MAXWELL3D_REG_INDEX(index_array.count)) {
             const MMEDrawMode expected_mode = method == MAXWELL3D_REG_INDEX(vertex_buffer.count)
                                                   ? MMEDrawMode::Array
                                                   : MMEDrawMode::Indexed;
-            StepInstance(expected_mode, method_call.argument);
+            StepInstance(expected_mode, method_argument);
         } else if (method == MAXWELL3D_REG_INDEX(draw.vertex_begin_gl)) {
             mme_draw.instance_mode =
                 (regs.draw.instance_next != 0) || (regs.draw.instance_cont != 0);
@@ -387,7 +383,7 @@ void Maxwell3D::CallMethodFromMME(const GPU::MethodCall& method_call) {
         if (mme_draw.current_mode != MMEDrawMode::Undefined) {
             FlushMMEInlineDraw();
         }
-        CallMethod(method_call);
+        CallMethod(method, method_argument, true);
     }
 }
 
diff --git a/src/video_core/engines/maxwell_3d.h b/src/video_core/engines/maxwell_3d.h
index 89e29a0d3..05dd6b39b 100644
--- a/src/video_core/engines/maxwell_3d.h
+++ b/src/video_core/engines/maxwell_3d.h
@@ -19,6 +19,7 @@
 #include "common/math_util.h"
 #include "video_core/engines/const_buffer_engine_interface.h"
 #include "video_core/engines/const_buffer_info.h"
+#include "video_core/engines/engine_interface.h"
 #include "video_core/engines/engine_upload.h"
 #include "video_core/engines/shader_type.h"
 #include "video_core/gpu.h"
@@ -48,7 +49,7 @@ namespace Tegra::Engines {
 #define MAXWELL3D_REG_INDEX(field_name)                                                            \
     (offsetof(Tegra::Engines::Maxwell3D::Regs, field_name) / sizeof(u32))
 
-class Maxwell3D final : public ConstBufferEngineInterface {
+class Maxwell3D final : public ConstBufferEngineInterface, public EngineInterface {
 public:
     explicit Maxwell3D(Core::System& system, VideoCore::RasterizerInterface& rasterizer,
                        MemoryManager& memory_manager);
@@ -1378,13 +1379,14 @@ public:
     u32 GetRegisterValue(u32 method) const;
 
     /// Write the value to the register identified by method.
-    void CallMethod(const GPU::MethodCall& method_call);
+    void CallMethod(u32 method, u32 method_argument, bool is_last_call) override;
 
     /// Write multiple values to the register identified by method.
-    void CallMultiMethod(u32 method, const u32* base_start, u32 amount, u32 methods_pending);
+    void CallMultiMethod(u32 method, const u32* base_start, u32 amount,
+                         u32 methods_pending) override;
 
     /// Write the value to the register identified by method.
-    void CallMethodFromMME(const GPU::MethodCall& method_call);
+    void CallMethodFromMME(u32 method, u32 method_argument);
 
     void FlushMMEInlineDraw();
 
diff --git a/src/video_core/engines/maxwell_dma.cpp b/src/video_core/engines/maxwell_dma.cpp
index 6630005b0..01d7df405 100644
--- a/src/video_core/engines/maxwell_dma.cpp
+++ b/src/video_core/engines/maxwell_dma.cpp
@@ -17,16 +17,16 @@ namespace Tegra::Engines {
 MaxwellDMA::MaxwellDMA(Core::System& system, MemoryManager& memory_manager)
     : system{system}, memory_manager{memory_manager} {}
 
-void MaxwellDMA::CallMethod(const GPU::MethodCall& method_call) {
-    ASSERT_MSG(method_call.method < Regs::NUM_REGS,
+void MaxwellDMA::CallMethod(u32 method, u32 method_argument, bool is_last_call) {
+    ASSERT_MSG(method < Regs::NUM_REGS,
                "Invalid MaxwellDMA register, increase the size of the Regs structure");
 
-    regs.reg_array[method_call.method] = method_call.argument;
+    regs.reg_array[method] = method_argument;
 
 #define MAXWELLDMA_REG_INDEX(field_name)                                                           \
     (offsetof(Tegra::Engines::MaxwellDMA::Regs, field_name) / sizeof(u32))
 
-    switch (method_call.method) {
+    switch (method) {
     case MAXWELLDMA_REG_INDEX(exec): {
         HandleCopy();
         break;
@@ -39,7 +39,7 @@ void MaxwellDMA::CallMethod(const GPU::MethodCall& method_call) {
 void MaxwellDMA::CallMultiMethod(u32 method, const u32* base_start, u32 amount,
                                  u32 methods_pending) {
     for (std::size_t i = 0; i < amount; i++) {
-        CallMethod({method, base_start[i], 0, methods_pending - static_cast<u32>(i)});
+        CallMethod(method, base_start[i], methods_pending - static_cast<u32>(i) <= 1);
     }
 }
 
@@ -90,7 +90,47 @@ void MaxwellDMA::HandleCopy() {
     ASSERT(regs.exec.enable_2d == 1);
 
     if (regs.exec.is_dst_linear && !regs.exec.is_src_linear) {
+
         ASSERT(regs.src_params.BlockDepth() == 0);
+        // Optimized path for micro copies.
+        if (regs.dst_pitch * regs.y_count < Texture::GetGOBSize() && regs.dst_pitch <= 64) {
+            const u32 bytes_per_pixel = regs.dst_pitch / regs.x_count;
+            const std::size_t src_size = Texture::GetGOBSize();
+            const std::size_t dst_size = regs.dst_pitch * regs.y_count;
+            u32 pos_x = regs.src_params.pos_x;
+            u32 pos_y = regs.src_params.pos_y;
+            const u64 offset =
+                Texture::GetGOBOffset(regs.src_params.size_x, regs.src_params.size_y, pos_x, pos_y,
+                                      regs.src_params.BlockDepth(), bytes_per_pixel);
+            const u32 x_in_gob = 64 / bytes_per_pixel;
+            pos_x = pos_x % x_in_gob;
+            pos_y = pos_y % 8;
+
+            if (read_buffer.size() < src_size) {
+                read_buffer.resize(src_size);
+            }
+
+            if (write_buffer.size() < dst_size) {
+                write_buffer.resize(dst_size);
+            }
+
+            if (Settings::IsGPULevelExtreme()) {
+                memory_manager.ReadBlock(source + offset, read_buffer.data(), src_size);
+                memory_manager.ReadBlock(dest, write_buffer.data(), dst_size);
+            } else {
+                memory_manager.ReadBlockUnsafe(source + offset, read_buffer.data(), src_size);
+                memory_manager.ReadBlockUnsafe(dest, write_buffer.data(), dst_size);
+            }
+
+            Texture::UnswizzleSubrect(regs.x_count, regs.y_count, regs.dst_pitch,
+                                      regs.src_params.size_x, bytes_per_pixel, read_buffer.data(),
+                                      write_buffer.data(), regs.src_params.BlockHeight(), pos_x,
+                                      pos_y);
+
+            memory_manager.WriteBlock(dest, write_buffer.data(), dst_size);
+
+            return;
+        }
         // If the input is tiled and the output is linear, deswizzle the input and copy it over.
         const u32 bytes_per_pixel = regs.dst_pitch / regs.x_count;
         const std::size_t src_size = Texture::CalculateSize(
diff --git a/src/video_core/engines/maxwell_dma.h b/src/video_core/engines/maxwell_dma.h
index c43ed8194..502dd8509 100644
--- a/src/video_core/engines/maxwell_dma.h
+++ b/src/video_core/engines/maxwell_dma.h
@@ -10,6 +10,7 @@
 #include "common/bit_field.h"
 #include "common/common_funcs.h"
 #include "common/common_types.h"
+#include "video_core/engines/engine_interface.h"
 #include "video_core/gpu.h"
 
 namespace Core {
@@ -27,16 +28,17 @@ namespace Tegra::Engines {
  * https://github.com/envytools/envytools/blob/master/rnndb/fifo/gk104_copy.xml
  */
 
-class MaxwellDMA final {
+class MaxwellDMA final : public EngineInterface {
 public:
     explicit MaxwellDMA(Core::System& system, MemoryManager& memory_manager);
     ~MaxwellDMA() = default;
 
     /// Write the value to the register identified by method.
-    void CallMethod(const GPU::MethodCall& method_call);
+    void CallMethod(u32 method, u32 method_argument, bool is_last_call) override;
 
     /// Write multiple values to the register identified by method.
-    void CallMultiMethod(u32 method, const u32* base_start, u32 amount, u32 methods_pending);
+    void CallMultiMethod(u32 method, const u32* base_start, u32 amount,
+                         u32 methods_pending) override;
 
     struct Regs {
         static constexpr std::size_t NUM_REGS = 0x1D6;
diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp
index b87fd873d..8eb017f65 100644
--- a/src/video_core/gpu.cpp
+++ b/src/video_core/gpu.cpp
@@ -299,19 +299,21 @@ void GPU::CallEngineMethod(const MethodCall& method_call) {
 
     switch (engine) {
     case EngineID::FERMI_TWOD_A:
-        fermi_2d->CallMethod(method_call);
+        fermi_2d->CallMethod(method_call.method, method_call.argument, method_call.IsLastCall());
         break;
     case EngineID::MAXWELL_B:
-        maxwell_3d->CallMethod(method_call);
+        maxwell_3d->CallMethod(method_call.method, method_call.argument, method_call.IsLastCall());
         break;
     case EngineID::KEPLER_COMPUTE_B:
-        kepler_compute->CallMethod(method_call);
+        kepler_compute->CallMethod(method_call.method, method_call.argument,
+                                   method_call.IsLastCall());
         break;
     case EngineID::MAXWELL_DMA_COPY_A:
-        maxwell_dma->CallMethod(method_call);
+        maxwell_dma->CallMethod(method_call.method, method_call.argument, method_call.IsLastCall());
         break;
     case EngineID::KEPLER_INLINE_TO_MEMORY_B:
-        kepler_memory->CallMethod(method_call);
+        kepler_memory->CallMethod(method_call.method, method_call.argument,
+                                  method_call.IsLastCall());
         break;
     default:
         UNIMPLEMENTED_MSG("Unimplemented engine");
@@ -347,7 +349,27 @@ void GPU::ProcessBindMethod(const MethodCall& method_call) {
     // Bind the current subchannel to the desired engine id.
     LOG_DEBUG(HW_GPU, "Binding subchannel {} to engine {}", method_call.subchannel,
               method_call.argument);
-    bound_engines[method_call.subchannel] = static_cast<EngineID>(method_call.argument);
+    const auto engine_id = static_cast<EngineID>(method_call.argument);
+    bound_engines[method_call.subchannel] = static_cast<EngineID>(engine_id);
+    switch (engine_id) {
+    case EngineID::FERMI_TWOD_A:
+        dma_pusher->BindSubchannel(fermi_2d.get(), method_call.subchannel);
+        break;
+    case EngineID::MAXWELL_B:
+        dma_pusher->BindSubchannel(maxwell_3d.get(), method_call.subchannel);
+        break;
+    case EngineID::KEPLER_COMPUTE_B:
+        dma_pusher->BindSubchannel(kepler_compute.get(), method_call.subchannel);
+        break;
+    case EngineID::MAXWELL_DMA_COPY_A:
+        dma_pusher->BindSubchannel(maxwell_dma.get(), method_call.subchannel);
+        break;
+    case EngineID::KEPLER_INLINE_TO_MEMORY_B:
+        dma_pusher->BindSubchannel(kepler_memory.get(), method_call.subchannel);
+        break;
+    default:
+        UNIMPLEMENTED_MSG("Unimplemented engine {:04X}", static_cast<u32>(engine_id));
+    }
 }
 
 void GPU::ProcessSemaphoreTriggerMethod() {
diff --git a/src/video_core/macro_interpreter.cpp b/src/video_core/macro_interpreter.cpp
index 42031d80a..947364928 100644
--- a/src/video_core/macro_interpreter.cpp
+++ b/src/video_core/macro_interpreter.cpp
@@ -328,7 +328,7 @@ void MacroInterpreter::SetMethodAddress(u32 address) {
 }
 
 void MacroInterpreter::Send(u32 value) {
-    maxwell3d.CallMethodFromMME({method_address.address, value});
+    maxwell3d.CallMethodFromMME(method_address.address, value);
     // Increment the method address by the method increment.
     method_address.address.Assign(method_address.address.Value() +
                                   method_address.increment.Value());
diff --git a/src/video_core/renderer_vulkan/vk_device.cpp b/src/video_core/renderer_vulkan/vk_device.cpp
index e1f65e3a7..170cdaed0 100644
--- a/src/video_core/renderer_vulkan/vk_device.cpp
+++ b/src/video_core/renderer_vulkan/vk_device.cpp
@@ -297,6 +297,17 @@ bool VKDevice::Create() {
         LOG_INFO(Render_Vulkan, "Device doesn't support transform feedbacks");
     }
 
+    VkPhysicalDeviceCustomBorderColorFeaturesEXT custom_border;
+    if (ext_custom_border_color) {
+        custom_border.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_CUSTOM_BORDER_COLOR_FEATURES_EXT;
+        custom_border.pNext = nullptr;
+        custom_border.customBorderColors = VK_TRUE;
+        custom_border.customBorderColorWithoutFormat = VK_TRUE;
+        SetNext(next, custom_border);
+    } else {
+        LOG_INFO(Render_Vulkan, "Device doesn't support custom border colors");
+    }
+
     if (!ext_depth_range_unrestricted) {
         LOG_INFO(Render_Vulkan, "Device doesn't support depth range unrestricted");
     }
@@ -524,6 +535,7 @@ std::vector<const char*> VKDevice::LoadExtensions() {
     bool has_khr_shader_float16_int8{};
     bool has_ext_subgroup_size_control{};
     bool has_ext_transform_feedback{};
+    bool has_ext_custom_border_color{};
     for (const auto& extension : physical.EnumerateDeviceExtensionProperties()) {
         Test(extension, nv_viewport_swizzle, VK_NV_VIEWPORT_SWIZZLE_EXTENSION_NAME, true);
         Test(extension, khr_uniform_buffer_standard_layout,
@@ -539,6 +551,8 @@ std::vector<const char*> VKDevice::LoadExtensions() {
              false);
         Test(extension, has_ext_transform_feedback, VK_EXT_TRANSFORM_FEEDBACK_EXTENSION_NAME,
              false);
+        Test(extension, has_ext_custom_border_color, VK_EXT_CUSTOM_BORDER_COLOR_EXTENSION_NAME,
+             false);
         if (Settings::values.renderer_debug) {
             Test(extension, nv_device_diagnostics_config,
                  VK_NV_DEVICE_DIAGNOSTICS_CONFIG_EXTENSION_NAME, true);
@@ -611,6 +625,19 @@ std::vector<const char*> VKDevice::LoadExtensions() {
         }
     }
 
+    if (has_ext_custom_border_color) {
+        VkPhysicalDeviceCustomBorderColorFeaturesEXT border_features;
+        border_features.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_CUSTOM_BORDER_COLOR_FEATURES_EXT;
+        border_features.pNext = nullptr;
+        features.pNext = &border_features;
+        physical.GetFeatures2KHR(features);
+
+        if (border_features.customBorderColors && border_features.customBorderColorWithoutFormat) {
+            extensions.push_back(VK_EXT_CUSTOM_BORDER_COLOR_EXTENSION_NAME);
+            ext_custom_border_color = true;
+        }
+    }
+
     return extensions;
 }
 
diff --git a/src/video_core/renderer_vulkan/vk_device.h b/src/video_core/renderer_vulkan/vk_device.h
index 12b05651b..6b9227b09 100644
--- a/src/video_core/renderer_vulkan/vk_device.h
+++ b/src/video_core/renderer_vulkan/vk_device.h
@@ -177,6 +177,11 @@ public:
         return ext_transform_feedback;
     }
 
+    /// Returns true if the device supports VK_EXT_custom_border_color.
+    bool IsExtCustomBorderColorSupported() const {
+        return ext_custom_border_color;
+    }
+
     /// Returns the vendor name reported from Vulkan.
     std::string_view GetVendorName() const {
         return vendor_name;
@@ -233,6 +238,7 @@ private:
     bool ext_depth_range_unrestricted{};       ///< Support for VK_EXT_depth_range_unrestricted.
     bool ext_shader_viewport_index_layer{};    ///< Support for VK_EXT_shader_viewport_index_layer.
     bool ext_transform_feedback{};             ///< Support for VK_EXT_transform_feedback.
+    bool ext_custom_border_color{};            ///< Support for VK_EXT_custom_border_color.
     bool nv_device_diagnostics_config{};       ///< Support for VK_NV_device_diagnostics_config.
 
     // Telemetry parameters
diff --git a/src/video_core/renderer_vulkan/vk_sampler_cache.cpp b/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
index 2687d8d95..e6f2fa553 100644
--- a/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
+++ b/src/video_core/renderer_vulkan/vk_sampler_cache.cpp
@@ -39,9 +39,18 @@ VKSamplerCache::VKSamplerCache(const VKDevice& device) : device{device} {}
 VKSamplerCache::~VKSamplerCache() = default;
 
 vk::Sampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc) const {
+    const bool arbitrary_borders = device.IsExtCustomBorderColorSupported();
+    const std::array color = tsc.GetBorderColor();
+
+    VkSamplerCustomBorderColorCreateInfoEXT border;
+    border.sType = VK_STRUCTURE_TYPE_SAMPLER_CUSTOM_BORDER_COLOR_CREATE_INFO_EXT;
+    border.pNext = nullptr;
+    border.format = VK_FORMAT_UNDEFINED;
+    std::memcpy(&border.customBorderColor, color.data(), sizeof(color));
+
     VkSamplerCreateInfo ci;
     ci.sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO;
-    ci.pNext = nullptr;
+    ci.pNext = arbitrary_borders ? &border : nullptr;
     ci.flags = 0;
     ci.magFilter = MaxwellToVK::Sampler::Filter(tsc.mag_filter);
     ci.minFilter = MaxwellToVK::Sampler::Filter(tsc.min_filter);
@@ -56,7 +65,7 @@ vk::Sampler VKSamplerCache::CreateSampler(const Tegra::Texture::TSCEntry& tsc) c
     ci.compareOp = MaxwellToVK::Sampler::DepthCompareFunction(tsc.depth_compare_func);
     ci.minLod = tsc.GetMinLod();
     ci.maxLod = tsc.GetMaxLod();
-    ci.borderColor = ConvertBorderColor(tsc.GetBorderColor());
+    ci.borderColor = arbitrary_borders ? VK_BORDER_COLOR_INT_CUSTOM_EXT : ConvertBorderColor(color);
     ci.unnormalizedCoordinates = VK_FALSE;
     return device.GetLogical().CreateSampler(ci);
 }
diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp
index fae8638ec..548e4c3fe 100644
--- a/src/video_core/textures/decoders.cpp
+++ b/src/video_core/textures/decoders.cpp
@@ -382,4 +382,18 @@ std::size_t CalculateSize(bool tiled, u32 bytes_per_pixel, u32 width, u32 height
     }
 }
 
+u64 GetGOBOffset(u32 width, u32 height, u32 dst_x, u32 dst_y, u32 block_height,
+                 u32 bytes_per_pixel) {
+    auto div_ceil = [](const u32 x, const u32 y) { return ((x + y - 1) / y); };
+    const u32 gobs_in_block = 1 << block_height;
+    const u32 y_blocks = gob_size_y << block_height;
+    const u32 x_per_gob = gob_size_x / bytes_per_pixel;
+    const u32 x_blocks = div_ceil(width, x_per_gob);
+    const u32 block_size = gob_size * gobs_in_block;
+    const u32 stride = block_size * x_blocks;
+    const u32 base = (dst_y / y_blocks) * stride + (dst_x / x_per_gob) * block_size;
+    const u32 relative_y = dst_y % y_blocks;
+    return base + (relative_y / gob_size_y) * gob_size;
+}
+
 } // namespace Tegra::Texture
diff --git a/src/video_core/textures/decoders.h b/src/video_core/textures/decoders.h
index 9f2d6d308..06f3ebf87 100644
--- a/src/video_core/textures/decoders.h
+++ b/src/video_core/textures/decoders.h
@@ -59,4 +59,8 @@ void UnswizzleSubrect(u32 subrect_width, u32 subrect_height, u32 dest_pitch, u32
 void SwizzleKepler(u32 width, u32 height, u32 dst_x, u32 dst_y, u32 block_height,
                    std::size_t copy_size, const u8* source_data, u8* swizzle_data);
 
+/// Obtains the offset of the gob for positions 'dst_x' & 'dst_y'
+u64 GetGOBOffset(u32 width, u32 height, u32 dst_x, u32 dst_y, u32 block_height,
+                 u32 bytes_per_pixel);
+
 } // namespace Tegra::Texture
diff --git a/src/yuzu/main.cpp b/src/yuzu/main.cpp
index 0a6839b2d..86e8a1d49 100644
--- a/src/yuzu/main.cpp
+++ b/src/yuzu/main.cpp
@@ -1026,7 +1026,6 @@ void GMainWindow::BootGame(const QString& filename) {
         mouse_hide_timer.start();
         setMouseTracking(true);
         ui.centralwidget->setMouseTracking(true);
-        ui.menubar->setMouseTracking(true);
     }
 
     const u64 title_id = Core::System::GetInstance().CurrentProcess()->GetTitleID();
@@ -1099,7 +1098,6 @@ void GMainWindow::ShutdownGame() {
 
     setMouseTracking(false);
     ui.centralwidget->setMouseTracking(false);
-    ui.menubar->setMouseTracking(false);
 
     UpdateWindowTitle();
 
@@ -1861,12 +1859,10 @@ void GMainWindow::OnConfigure() {
     if (UISettings::values.hide_mouse && emulation_running) {
         setMouseTracking(true);
         ui.centralwidget->setMouseTracking(true);
-        ui.menubar->setMouseTracking(true);
         mouse_hide_timer.start();
     } else {
         setMouseTracking(false);
         ui.centralwidget->setMouseTracking(false);
-        ui.menubar->setMouseTracking(false);
     }
 
     dock_status_button->setChecked(Settings::values.use_docked_mode);