40 files changed, 1936 insertions, 124 deletions

diff --git a/src/audio_core/renderer/command/command_buffer.cpp b/src/audio_core/renderer/command/command_buffer.cpp
index 2ef879ee1..8c6fe97e7 100644
--- a/src/audio_core/renderer/command/command_buffer.cpp
+++ b/src/audio_core/renderer/command/command_buffer.cpp
@@ -460,21 +460,23 @@ void CommandBuffer::GenerateDeviceSinkCommand(const s32 node_id, const s16 buffe
 
     cmd.session_id = session_id;
 
+    cmd.input_count = parameter.input_count;
+    s16 max_input{0};
+    for (u32 i = 0; i < parameter.input_count; i++) {
+        cmd.inputs[i] = buffer_offset + parameter.inputs[i];
+        max_input = std::max(max_input, cmd.inputs[i]);
+    }
+
     if (state.upsampler_info != nullptr) {
         const auto size_{state.upsampler_info->sample_count * parameter.input_count};
         const auto size_bytes{size_ * sizeof(s32)};
         const auto addr{memory_pool->Translate(state.upsampler_info->samples_pos, size_bytes)};
         cmd.sample_buffer = {reinterpret_cast<s32*>(addr),
-                             parameter.input_count * state.upsampler_info->sample_count};
+                             (max_input + 1) * state.upsampler_info->sample_count};
     } else {
         cmd.sample_buffer = samples_buffer;
     }
 
-    cmd.input_count = parameter.input_count;
-    for (u32 i = 0; i < parameter.input_count; i++) {
-        cmd.inputs[i] = buffer_offset + parameter.inputs[i];
-    }
-
     GenerateEnd<DeviceSinkCommand>(cmd);
 }
 
diff --git a/src/core/hle/kernel/hle_ipc.h b/src/core/hle/kernel/hle_ipc.h
index d87be72d6..e252b5f4b 100644
--- a/src/core/hle/kernel/hle_ipc.h
+++ b/src/core/hle/kernel/hle_ipc.h
@@ -199,7 +199,7 @@ public:
     ~HLERequestContext();
 
     /// Returns a pointer to the IPC command buffer for this request.
-    u32* CommandBuffer() {
+    [[nodiscard]] u32* CommandBuffer() {
         return cmd_buf.data();
     }
 
@@ -207,7 +207,7 @@ public:
      * Returns the session through which this request was made. This can be used as a map key to
      * access per-client data on services.
      */
-    Kernel::KServerSession* Session() {
+    [[nodiscard]] Kernel::KServerSession* Session() {
         return server_session;
     }
 
@@ -217,61 +217,61 @@ public:
     /// Writes data from this context back to the requesting process/thread.
     Result WriteToOutgoingCommandBuffer(KThread& requesting_thread);
 
-    u32_le GetHipcCommand() const {
+    [[nodiscard]] u32_le GetHipcCommand() const {
         return command;
     }
 
-    u32_le GetTipcCommand() const {
+    [[nodiscard]] u32_le GetTipcCommand() const {
         return static_cast<u32_le>(command_header->type.Value()) -
                static_cast<u32_le>(IPC::CommandType::TIPC_CommandRegion);
     }
 
-    u32_le GetCommand() const {
+    [[nodiscard]] u32_le GetCommand() const {
         return command_header->IsTipc() ? GetTipcCommand() : GetHipcCommand();
     }
 
-    bool IsTipc() const {
+    [[nodiscard]] bool IsTipc() const {
         return command_header->IsTipc();
     }
 
-    IPC::CommandType GetCommandType() const {
+    [[nodiscard]] IPC::CommandType GetCommandType() const {
         return command_header->type;
     }
 
-    u64 GetPID() const {
+    [[nodiscard]] u64 GetPID() const {
         return pid;
     }
 
-    u32 GetDataPayloadOffset() const {
+    [[nodiscard]] u32 GetDataPayloadOffset() const {
         return data_payload_offset;
     }
 
-    const std::vector<IPC::BufferDescriptorX>& BufferDescriptorX() const {
+    [[nodiscard]] const std::vector<IPC::BufferDescriptorX>& BufferDescriptorX() const {
         return buffer_x_desciptors;
     }
 
-    const std::vector<IPC::BufferDescriptorABW>& BufferDescriptorA() const {
+    [[nodiscard]] const std::vector<IPC::BufferDescriptorABW>& BufferDescriptorA() const {
         return buffer_a_desciptors;
     }
 
-    const std::vector<IPC::BufferDescriptorABW>& BufferDescriptorB() const {
+    [[nodiscard]] const std::vector<IPC::BufferDescriptorABW>& BufferDescriptorB() const {
         return buffer_b_desciptors;
     }
 
-    const std::vector<IPC::BufferDescriptorC>& BufferDescriptorC() const {
+    [[nodiscard]] const std::vector<IPC::BufferDescriptorC>& BufferDescriptorC() const {
         return buffer_c_desciptors;
     }
 
-    const IPC::DomainMessageHeader& GetDomainMessageHeader() const {
+    [[nodiscard]] const IPC::DomainMessageHeader& GetDomainMessageHeader() const {
         return domain_message_header.value();
     }
 
-    bool HasDomainMessageHeader() const {
+    [[nodiscard]] bool HasDomainMessageHeader() const {
         return domain_message_header.has_value();
     }
 
     /// Helper function to read a buffer using the appropriate buffer descriptor
-    std::vector<u8> ReadBuffer(std::size_t buffer_index = 0) const;
+    [[nodiscard]] std::vector<u8> ReadBuffer(std::size_t buffer_index = 0) const;
 
     /// Helper function to write a buffer using the appropriate buffer descriptor
     std::size_t WriteBuffer(const void* buffer, std::size_t size,
@@ -308,22 +308,34 @@ public:
     }
 
     /// Helper function to get the size of the input buffer
-    std::size_t GetReadBufferSize(std::size_t buffer_index = 0) const;
+    [[nodiscard]] std::size_t GetReadBufferSize(std::size_t buffer_index = 0) const;
 
     /// Helper function to get the size of the output buffer
-    std::size_t GetWriteBufferSize(std::size_t buffer_index = 0) const;
+    [[nodiscard]] std::size_t GetWriteBufferSize(std::size_t buffer_index = 0) const;
+
+    /// Helper function to derive the number of elements able to be contained in the read buffer
+    template <typename T>
+    [[nodiscard]] std::size_t GetReadBufferNumElements(std::size_t buffer_index = 0) const {
+        return GetReadBufferSize(buffer_index) / sizeof(T);
+    }
+
+    /// Helper function to derive the number of elements able to be contained in the write buffer
+    template <typename T>
+    [[nodiscard]] std::size_t GetWriteBufferNumElements(std::size_t buffer_index = 0) const {
+        return GetWriteBufferSize(buffer_index) / sizeof(T);
+    }
 
     /// Helper function to test whether the input buffer at buffer_index can be read
-    bool CanReadBuffer(std::size_t buffer_index = 0) const;
+    [[nodiscard]] bool CanReadBuffer(std::size_t buffer_index = 0) const;
 
     /// Helper function to test whether the output buffer at buffer_index can be written
-    bool CanWriteBuffer(std::size_t buffer_index = 0) const;
+    [[nodiscard]] bool CanWriteBuffer(std::size_t buffer_index = 0) const;
 
-    Handle GetCopyHandle(std::size_t index) const {
+    [[nodiscard]] Handle GetCopyHandle(std::size_t index) const {
         return incoming_copy_handles.at(index);
     }
 
-    Handle GetMoveHandle(std::size_t index) const {
+    [[nodiscard]] Handle GetMoveHandle(std::size_t index) const {
         return incoming_move_handles.at(index);
     }
 
@@ -348,13 +360,13 @@ public:
         manager = manager_;
     }
 
-    std::string Description() const;
+    [[nodiscard]] std::string Description() const;
 
-    KThread& GetThread() {
+    [[nodiscard]] KThread& GetThread() {
         return *thread;
     }
 
-    std::shared_ptr<SessionRequestManager> GetManager() const {
+    [[nodiscard]] std::shared_ptr<SessionRequestManager> GetManager() const {
         return manager.lock();
     }
 
diff --git a/src/core/hle/service/audio/audin_u.cpp b/src/core/hle/service/audio/audin_u.cpp
index 608925dfc..053e8f9dd 100644
--- a/src/core/hle/service/audio/audin_u.cpp
+++ b/src/core/hle/service/audio/audin_u.cpp
@@ -122,10 +122,10 @@ private:
     }
 
     void GetReleasedAudioInBuffer(Kernel::HLERequestContext& ctx) {
-        auto write_buffer_size = ctx.GetWriteBufferSize() / sizeof(u64);
-        std::vector<u64> released_buffers(write_buffer_size, 0);
+        const auto write_buffer_size = ctx.GetWriteBufferNumElements<u64>();
+        std::vector<u64> released_buffers(write_buffer_size);
 
-        auto count = impl->GetReleasedBuffers(released_buffers);
+        const auto count = impl->GetReleasedBuffers(released_buffers);
 
         [[maybe_unused]] std::string tags{};
         for (u32 i = 0; i < count; i++) {
@@ -228,7 +228,7 @@ void AudInU::ListAudioIns(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_Audio, "called");
 
     const auto write_count =
-        static_cast<u32>(ctx.GetWriteBufferSize() / sizeof(AudioDevice::AudioDeviceName));
+        static_cast<u32>(ctx.GetWriteBufferNumElements<AudioDevice::AudioDeviceName>());
     std::vector<AudioDevice::AudioDeviceName> device_names{};
 
     u32 out_count{0};
@@ -248,7 +248,7 @@ void AudInU::ListAudioInsAutoFiltered(Kernel::HLERequestContext& ctx) {
     LOG_DEBUG(Service_Audio, "called");
 
     const auto write_count =
-        static_cast<u32>(ctx.GetWriteBufferSize() / sizeof(AudioDevice::AudioDeviceName));
+        static_cast<u32>(ctx.GetWriteBufferNumElements<AudioDevice::AudioDeviceName>());
     std::vector<AudioDevice::AudioDeviceName> device_names{};
 
     u32 out_count{0};
diff --git a/src/core/hle/service/audio/audout_u.cpp b/src/core/hle/service/audio/audout_u.cpp
index 122290c6a..29751f075 100644
--- a/src/core/hle/service/audio/audout_u.cpp
+++ b/src/core/hle/service/audio/audout_u.cpp
@@ -129,16 +129,16 @@ private:
     }
 
     void GetReleasedAudioOutBuffers(Kernel::HLERequestContext& ctx) {
-        auto write_buffer_size = ctx.GetWriteBufferSize() / sizeof(u64);
-        std::vector<u64> released_buffers(write_buffer_size, 0);
+        const auto write_buffer_size = ctx.GetWriteBufferNumElements<u64>();
+        std::vector<u64> released_buffers(write_buffer_size);
 
-        auto count = impl->GetReleasedBuffers(released_buffers);
+        const auto count = impl->GetReleasedBuffers(released_buffers);
 
         [[maybe_unused]] std::string tags{};
         for (u32 i = 0; i < count; i++) {
             tags += fmt::format("{:08X}, ", released_buffers[i]);
         }
-        [[maybe_unused]] auto sessionid{impl->GetSystem().GetSessionId()};
+        [[maybe_unused]] const auto sessionid{impl->GetSystem().GetSessionId()};
         LOG_TRACE(Service_Audio, "called. Session {} released {} buffers: {}", sessionid, count,
                   tags);
 
@@ -244,7 +244,7 @@ void AudOutU::ListAudioOuts(Kernel::HLERequestContext& ctx) {
     std::scoped_lock l{impl->mutex};
 
     const auto write_count =
-        static_cast<u32>(ctx.GetWriteBufferSize() / sizeof(AudioDevice::AudioDeviceName));
+        static_cast<u32>(ctx.GetWriteBufferNumElements<AudioDevice::AudioDeviceName>());
     std::vector<AudioDevice::AudioDeviceName> device_names{};
     if (write_count > 0) {
         device_names.emplace_back("DeviceOut");
diff --git a/src/core/hle/service/audio/audren_u.cpp b/src/core/hle/service/audio/audren_u.cpp
index 13423dca6..034ee273f 100644
--- a/src/core/hle/service/audio/audren_u.cpp
+++ b/src/core/hle/service/audio/audren_u.cpp
@@ -274,7 +274,7 @@ public:
 
 private:
     void ListAudioDeviceName(Kernel::HLERequestContext& ctx) {
-        const size_t in_count = ctx.GetWriteBufferSize() / sizeof(AudioDevice::AudioDeviceName);
+        const size_t in_count = ctx.GetWriteBufferNumElements<AudioDevice::AudioDeviceName>();
 
         std::vector<AudioDevice::AudioDeviceName> out_names{};
 
@@ -335,7 +335,7 @@ private:
     }
 
     void GetActiveAudioDeviceName(Kernel::HLERequestContext& ctx) {
-        const auto write_size = ctx.GetWriteBufferSize() / sizeof(char);
+        const auto write_size = ctx.GetWriteBufferSize();
         std::string out_name{"AudioTvOutput"};
 
         LOG_DEBUG(Service_Audio, "(STUBBED) called. Name={}", out_name);
@@ -387,7 +387,7 @@ private:
     }
 
     void ListAudioOutputDeviceName(Kernel::HLERequestContext& ctx) {
-        const size_t in_count = ctx.GetWriteBufferSize() / sizeof(AudioDevice::AudioDeviceName);
+        const size_t in_count = ctx.GetWriteBufferNumElements<AudioDevice::AudioDeviceName>();
 
         std::vector<AudioDevice::AudioDeviceName> out_names{};
 
diff --git a/src/core/hle/service/audio/hwopus.cpp b/src/core/hle/service/audio/hwopus.cpp
index 8bafc3a98..825fb8bcc 100644
--- a/src/core/hle/service/audio/hwopus.cpp
+++ b/src/core/hle/service/audio/hwopus.cpp
@@ -68,7 +68,7 @@ private:
                                  ExtraBehavior extra_behavior) {
         u32 consumed = 0;
         u32 sample_count = 0;
-        std::vector<opus_int16> samples(ctx.GetWriteBufferSize() / sizeof(opus_int16));
+        std::vector<opus_int16> samples(ctx.GetWriteBufferNumElements<opus_int16>());
 
         if (extra_behavior == ExtraBehavior::ResetContext) {
             ResetDecoderContext();
diff --git a/src/core/hle/service/bcat/bcat_module.cpp b/src/core/hle/service/bcat/bcat_module.cpp
index bc08ac487..cbe690a5d 100644
--- a/src/core/hle/service/bcat/bcat_module.cpp
+++ b/src/core/hle/service/bcat/bcat_module.cpp
@@ -443,7 +443,7 @@ private:
     }
 
     void Read(Kernel::HLERequestContext& ctx) {
-        auto write_size = ctx.GetWriteBufferSize() / sizeof(DeliveryCacheDirectoryEntry);
+        auto write_size = ctx.GetWriteBufferNumElements<DeliveryCacheDirectoryEntry>();
 
         LOG_DEBUG(Service_BCAT, "called, write_size={:016X}", write_size);
 
@@ -533,7 +533,7 @@ private:
     }
 
     void EnumerateDeliveryCacheDirectory(Kernel::HLERequestContext& ctx) {
-        auto size = ctx.GetWriteBufferSize() / sizeof(DirectoryName);
+        auto size = ctx.GetWriteBufferNumElements<DirectoryName>();
 
         LOG_DEBUG(Service_BCAT, "called, size={:016X}", size);
 
diff --git a/src/core/hle/service/es/es.cpp b/src/core/hle/service/es/es.cpp
index ff9b0427c..d183e5829 100644
--- a/src/core/hle/service/es/es.cpp
+++ b/src/core/hle/service/es/es.cpp
@@ -192,12 +192,10 @@ private:
     }
 
     void ListCommonTicketRightsIds(Kernel::HLERequestContext& ctx) {
-        u32 out_entries;
-        if (keys.GetCommonTickets().empty())
-            out_entries = 0;
-        else
-            out_entries = static_cast<u32>(ctx.GetWriteBufferSize() / sizeof(u128));
-
+        size_t out_entries = 0;
+        if (!keys.GetCommonTickets().empty()) {
+            out_entries = ctx.GetWriteBufferNumElements<u128>();
+        }
         LOG_DEBUG(Service_ETicket, "called, entries={:016X}", out_entries);
 
         keys.PopulateTickets();
@@ -206,20 +204,19 @@ private:
         std::transform(tickets.begin(), tickets.end(), std::back_inserter(ids),
                        [](const auto& pair) { return pair.first; });
 
-        out_entries = static_cast<u32>(std::min<std::size_t>(ids.size(), out_entries));
+        out_entries = std::min(ids.size(), out_entries);
         ctx.WriteBuffer(ids.data(), out_entries * sizeof(u128));
 
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(ResultSuccess);
-        rb.Push<u32>(out_entries);
+        rb.Push<u32>(static_cast<u32>(out_entries));
     }
 
     void ListPersonalizedTicketRightsIds(Kernel::HLERequestContext& ctx) {
-        u32 out_entries;
-        if (keys.GetPersonalizedTickets().empty())
-            out_entries = 0;
-        else
-            out_entries = static_cast<u32>(ctx.GetWriteBufferSize() / sizeof(u128));
+        size_t out_entries = 0;
+        if (!keys.GetPersonalizedTickets().empty()) {
+            out_entries = ctx.GetWriteBufferNumElements<u128>();
+        }
 
         LOG_DEBUG(Service_ETicket, "called, entries={:016X}", out_entries);
 
@@ -229,12 +226,12 @@ private:
         std::transform(tickets.begin(), tickets.end(), std::back_inserter(ids),
                        [](const auto& pair) { return pair.first; });
 
-        out_entries = static_cast<u32>(std::min<std::size_t>(ids.size(), out_entries));
+        out_entries = std::min(ids.size(), out_entries);
         ctx.WriteBuffer(ids.data(), out_entries * sizeof(u128));
 
         IPC::ResponseBuilder rb{ctx, 3};
         rb.Push(ResultSuccess);
-        rb.Push<u32>(out_entries);
+        rb.Push<u32>(static_cast<u32>(out_entries));
     }
 
     void GetCommonTicketSize(Kernel::HLERequestContext& ctx) {
diff --git a/src/core/hle/service/filesystem/fsp_srv.cpp b/src/core/hle/service/filesystem/fsp_srv.cpp
index c08274ef9..fbb16a7da 100644
--- a/src/core/hle/service/filesystem/fsp_srv.cpp
+++ b/src/core/hle/service/filesystem/fsp_srv.cpp
@@ -277,7 +277,7 @@ private:
         LOG_DEBUG(Service_FS, "called.");
 
         // Calculate how many entries we can fit in the output buffer
-        const u64 count_entries = ctx.GetWriteBufferSize() / sizeof(FileSys::Entry);
+        const u64 count_entries = ctx.GetWriteBufferNumElements<FileSys::Entry>();
 
         // Cap at total number of entries.
         const u64 actual_entries = std::min(count_entries, entries.size() - next_entry_index);
@@ -543,7 +543,7 @@ public:
         LOG_DEBUG(Service_FS, "called");
 
         // Calculate how many entries we can fit in the output buffer
-        const u64 count_entries = ctx.GetWriteBufferSize() / sizeof(SaveDataInfo);
+        const u64 count_entries = ctx.GetWriteBufferNumElements<SaveDataInfo>();
 
         // Cap at total number of entries.
         const u64 actual_entries = std::min(count_entries, info.size() - next_entry_index);
diff --git a/src/core/hle/service/ldn/ldn.cpp b/src/core/hle/service/ldn/ldn.cpp
index 6df563136..c49c61cff 100644
--- a/src/core/hle/service/ldn/ldn.cpp
+++ b/src/core/hle/service/ldn/ldn.cpp
@@ -292,7 +292,7 @@ public:
 
     void GetNetworkInfoLatestUpdate(Kernel::HLERequestContext& ctx) {
         const std::size_t network_buffer_size = ctx.GetWriteBufferSize(0);
-        const std::size_t node_buffer_count = ctx.GetWriteBufferSize(1) / sizeof(NodeLatestUpdate);
+        const std::size_t node_buffer_count = ctx.GetWriteBufferNumElements<NodeLatestUpdate>(1);
 
         if (node_buffer_count == 0 || network_buffer_size != sizeof(NetworkInfo)) {
             LOG_ERROR(Service_LDN, "Invalid buffer, size = {}, count = {}", network_buffer_size,
@@ -333,7 +333,7 @@ public:
         const auto channel{rp.PopEnum<WifiChannel>()};
         const auto scan_filter{rp.PopRaw<ScanFilter>()};
 
-        const std::size_t network_info_size = ctx.GetWriteBufferSize() / sizeof(NetworkInfo);
+        const std::size_t network_info_size = ctx.GetWriteBufferNumElements<NetworkInfo>();
 
         if (network_info_size == 0) {
             LOG_ERROR(Service_LDN, "Invalid buffer size {}", network_info_size);
diff --git a/src/core/hle/service/nfc/nfc_user.cpp b/src/core/hle/service/nfc/nfc_user.cpp
index 0753333bf..ced2d560b 100644
--- a/src/core/hle/service/nfc/nfc_user.cpp
+++ b/src/core/hle/service/nfc/nfc_user.cpp
@@ -118,7 +118,7 @@ void IUser::ListDevices(Kernel::HLERequestContext& ctx) {
     }
 
     std::vector<u64> nfp_devices;
-    const std::size_t max_allowed_devices = ctx.GetWriteBufferSize() / sizeof(u64);
+    const std::size_t max_allowed_devices = ctx.GetWriteBufferNumElements<u64>();
 
     for (auto& device : devices) {
         if (nfp_devices.size() >= max_allowed_devices) {
diff --git a/src/core/hle/service/nfp/nfp_user.cpp b/src/core/hle/service/nfp/nfp_user.cpp
index 2fe3c0ea0..49816b4c7 100644
--- a/src/core/hle/service/nfp/nfp_user.cpp
+++ b/src/core/hle/service/nfp/nfp_user.cpp
@@ -104,9 +104,9 @@ void IUser::ListDevices(Kernel::HLERequestContext& ctx) {
     }
 
     std::vector<u64> nfp_devices;
-    const std::size_t max_allowed_devices = ctx.GetWriteBufferSize() / sizeof(u64);
+    const std::size_t max_allowed_devices = ctx.GetWriteBufferNumElements<u64>();
 
-    for (auto& device : devices) {
+    for (const auto& device : devices) {
         if (nfp_devices.size() >= max_allowed_devices) {
             continue;
         }
@@ -115,7 +115,7 @@ void IUser::ListDevices(Kernel::HLERequestContext& ctx) {
         }
     }
 
-    if (nfp_devices.size() == 0) {
+    if (nfp_devices.empty()) {
         IPC::ResponseBuilder rb{ctx, 2};
         rb.Push(DeviceNotFound);
         return;
diff --git a/src/core/hle/service/ns/iplatform_service_manager.cpp b/src/core/hle/service/ns/iplatform_service_manager.cpp
index fd047ff26..1fab2f0dd 100644
--- a/src/core/hle/service/ns/iplatform_service_manager.cpp
+++ b/src/core/hle/service/ns/iplatform_service_manager.cpp
@@ -279,13 +279,10 @@ void IPlatformServiceManager::GetSharedFontInOrderOfPriority(Kernel::HLERequestC
         font_sizes.push_back(region.size);
     }
 
-    // Resize buffers if game requests smaller size output.
-    font_codes.resize(
-        std::min<std::size_t>(font_codes.size(), ctx.GetWriteBufferSize(0) / sizeof(u32)));
-    font_offsets.resize(
-        std::min<std::size_t>(font_offsets.size(), ctx.GetWriteBufferSize(1) / sizeof(u32)));
-    font_sizes.resize(
-        std::min<std::size_t>(font_sizes.size(), ctx.GetWriteBufferSize(2) / sizeof(u32)));
+    // Resize buffers if game requests smaller size output
+    font_codes.resize(std::min(font_codes.size(), ctx.GetWriteBufferNumElements<u32>(0)));
+    font_offsets.resize(std::min(font_offsets.size(), ctx.GetWriteBufferNumElements<u32>(1)));
+    font_sizes.resize(std::min(font_sizes.size(), ctx.GetWriteBufferNumElements<u32>(2)));
 
     ctx.WriteBuffer(font_codes, 0);
     ctx.WriteBuffer(font_offsets, 1);
diff --git a/src/core/hle/service/set/set.cpp b/src/core/hle/service/set/set.cpp
index f761c2da4..4f1a8d6b7 100644
--- a/src/core/hle/service/set/set.cpp
+++ b/src/core/hle/service/set/set.cpp
@@ -83,7 +83,7 @@ void PushResponseLanguageCode(Kernel::HLERequestContext& ctx, std::size_t num_la
 }
 
 void GetAvailableLanguageCodesImpl(Kernel::HLERequestContext& ctx, std::size_t max_entries) {
-    const std::size_t requested_amount = ctx.GetWriteBufferSize() / sizeof(LanguageCode);
+    const std::size_t requested_amount = ctx.GetWriteBufferNumElements<LanguageCode>();
     const std::size_t max_amount = std::min(requested_amount, max_entries);
     const std::size_t copy_amount = std::min(available_language_codes.size(), max_amount);
     const std::size_t copy_size = copy_amount * sizeof(LanguageCode);
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt
index d7f7d336c..b03a30992 100644
--- a/src/video_core/CMakeLists.txt
+++ b/src/video_core/CMakeLists.txt
@@ -28,6 +28,10 @@ add_library(video_core STATIC
     dirty_flags.h
     dma_pusher.cpp
     dma_pusher.h
+    engines/sw_blitter/blitter.cpp
+    engines/sw_blitter/blitter.h
+    engines/sw_blitter/converter.cpp
+    engines/sw_blitter/converter.h
     engines/const_buffer_info.h
     engines/engine_interface.h
     engines/engine_upload.cpp
diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h
index 599551013..5d3a8293b 100644
--- a/src/video_core/buffer_cache/buffer_cache.h
+++ b/src/video_core/buffer_cache/buffer_cache.h
@@ -1742,12 +1742,12 @@ bool BufferCache<P>::InlineMemory(VAddr dest_address, size_t copy_size,
     SynchronizeBuffer(buffer, dest_address, static_cast<u32>(copy_size));
 
     if constexpr (USE_MEMORY_MAPS) {
+        auto upload_staging = runtime.UploadStagingBuffer(copy_size);
         std::array copies{BufferCopy{
-            .src_offset = 0,
+            .src_offset = upload_staging.offset,
             .dst_offset = buffer.Offset(dest_address),
             .size = copy_size,
         }};
-        auto upload_staging = runtime.UploadStagingBuffer(copy_size);
         u8* const src_pointer = upload_staging.mapped_span.data();
         std::memcpy(src_pointer, inlined_buffer.data(), copy_size);
         runtime.CopyBuffer(buffer, upload_staging.buffer, copies);
diff --git a/src/video_core/control/channel_state.cpp b/src/video_core/control/channel_state.cpp
index cdecc3a91..832025d75 100644
--- a/src/video_core/control/channel_state.cpp
+++ b/src/video_core/control/channel_state.cpp
@@ -20,7 +20,7 @@ void ChannelState::Init(Core::System& system, GPU& gpu) {
     ASSERT(memory_manager);
     dma_pusher = std::make_unique<Tegra::DmaPusher>(system, gpu, *memory_manager, *this);
     maxwell_3d = std::make_unique<Engines::Maxwell3D>(system, *memory_manager);
-    fermi_2d = std::make_unique<Engines::Fermi2D>();
+    fermi_2d = std::make_unique<Engines::Fermi2D>(*memory_manager);
     kepler_compute = std::make_unique<Engines::KeplerCompute>(system, *memory_manager);
     maxwell_dma = std::make_unique<Engines::MaxwellDMA>(system, *memory_manager);
     kepler_memory = std::make_unique<Engines::KeplerMemory>(system, *memory_manager);
diff --git a/src/video_core/engines/engine_upload.cpp b/src/video_core/engines/engine_upload.cpp
index a34819234..28aa85f32 100644
--- a/src/video_core/engines/engine_upload.cpp
+++ b/src/video_core/engines/engine_upload.cpp
@@ -51,11 +51,11 @@ void State::ProcessData(std::span<const u8> read_buffer) {
         } else {
             for (u32 line = 0; line < regs.line_count; ++line) {
                 const GPUVAddr dest_line = address + static_cast<size_t>(line) * regs.dest.pitch;
-                memory_manager.WriteBlockUnsafe(
-                    dest_line, read_buffer.data() + static_cast<size_t>(line) * regs.line_length_in,
-                    regs.line_length_in);
+                std::span<const u8> buffer(read_buffer.data() +
+                                               static_cast<size_t>(line) * regs.line_length_in,
+                                           regs.line_length_in);
+                rasterizer->AccelerateInlineToMemory(dest_line, regs.line_length_in, buffer);
             }
-            memory_manager.InvalidateRegion(address, regs.dest.pitch * regs.line_count);
         }
     } else {
         u32 width = regs.dest.width;
diff --git a/src/video_core/engines/fermi_2d.cpp b/src/video_core/engines/fermi_2d.cpp
index 453e0fb01..c6478ae85 100644
--- a/src/video_core/engines/fermi_2d.cpp
+++ b/src/video_core/engines/fermi_2d.cpp
@@ -3,17 +3,25 @@
 
 #include "common/assert.h"
 #include "common/logging/log.h"
+#include "common/microprofile.h"
 #include "video_core/engines/fermi_2d.h"
-#include "video_core/memory_manager.h"
+#include "video_core/engines/sw_blitter/blitter.h"
 #include "video_core/rasterizer_interface.h"
 #include "video_core/surface.h"
+#include "video_core/textures/decoders.h"
+
+MICROPROFILE_DECLARE(GPU_BlitEngine);
+MICROPROFILE_DEFINE(GPU_BlitEngine, "GPU", "Blit Engine", MP_RGB(224, 224, 128));
 
 using VideoCore::Surface::BytesPerBlock;
 using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
 
 namespace Tegra::Engines {
 
-Fermi2D::Fermi2D() {
+using namespace Texture;
+
+Fermi2D::Fermi2D(MemoryManager& memory_manager_) {
+    sw_blitter = std::make_unique<Blitter::SoftwareBlitEngine>(memory_manager_);
     // Nvidia's OpenGL driver seems to assume these values
     regs.src.depth = 1;
     regs.dst.depth = 1;
@@ -42,6 +50,7 @@ void Fermi2D::CallMultiMethod(u32 method, const u32* base_start, u32 amount, u32
 }
 
 void Fermi2D::Blit() {
+    MICROPROFILE_SCOPE(GPU_BlitEngine);
     LOG_DEBUG(HW_GPU, "called. source address=0x{:x}, destination address=0x{:x}",
               regs.src.Address(), regs.dst.Address());
 
@@ -52,9 +61,16 @@ void Fermi2D::Blit() {
     UNIMPLEMENTED_IF_MSG(regs.clip_enable != 0, "Clipped blit enabled");
 
     const auto& args = regs.pixels_from_memory;
+    constexpr s64 null_derivate = 1ULL << 32;
+    Surface src = regs.src;
+    const auto bytes_per_pixel = BytesPerBlock(PixelFormatFromRenderTargetFormat(src.format));
+    const bool delegate_to_gpu = src.width > 512 && src.height > 512 && bytes_per_pixel <= 8 &&
+                                 src.format != regs.dst.format;
     Config config{
         .operation = regs.operation,
         .filter = args.sample_mode.filter,
+        .must_accelerate =
+            args.du_dx != null_derivate || args.dv_dy != null_derivate || delegate_to_gpu,
         .dst_x0 = args.dst_x0,
         .dst_y0 = args.dst_y0,
         .dst_x1 = args.dst_x0 + args.dst_width,
@@ -64,8 +80,7 @@ void Fermi2D::Blit() {
         .src_x1 = static_cast<s32>((args.du_dx * args.dst_width + args.src_x0) >> 32),
         .src_y1 = static_cast<s32>((args.dv_dy * args.dst_height + args.src_y0) >> 32),
     };
-    Surface src = regs.src;
-    const auto bytes_per_pixel = BytesPerBlock(PixelFormatFromRenderTargetFormat(src.format));
+
     const auto need_align_to_pitch =
         src.linear == Tegra::Engines::Fermi2D::MemoryLayout::Pitch &&
         static_cast<s32>(src.width) == config.src_x1 &&
@@ -78,8 +93,9 @@ void Fermi2D::Blit() {
         config.src_x1 -= config.src_x0;
         config.src_x0 = 0;
     }
+
     if (!rasterizer->AccelerateSurfaceCopy(src, regs.dst, config)) {
-        UNIMPLEMENTED();
+        sw_blitter->Blit(src, regs.dst, config);
     }
 }
 
diff --git a/src/video_core/engines/fermi_2d.h b/src/video_core/engines/fermi_2d.h
index 1229aa35b..24b518cb5 100644
--- a/src/video_core/engines/fermi_2d.h
+++ b/src/video_core/engines/fermi_2d.h
@@ -5,6 +5,7 @@
 
 #include <array>
 #include <cstddef>
+#include <memory>
 #include "common/bit_field.h"
 #include "common/common_funcs.h"
 #include "common/common_types.h"
@@ -21,6 +22,10 @@ class RasterizerInterface;
 
 namespace Tegra::Engines {
 
+namespace Blitter {
+class SoftwareBlitEngine;
+}
+
 /**
  * This Engine is known as G80_2D. Documentation can be found in:
  * https://github.com/envytools/envytools/blob/master/rnndb/graph/g80_2d.xml
@@ -32,7 +37,7 @@ namespace Tegra::Engines {
 
 class Fermi2D final : public EngineInterface {
 public:
-    explicit Fermi2D();
+    explicit Fermi2D(MemoryManager& memory_manager_);
     ~Fermi2D() override;
 
     /// Binds a rasterizer to this engine.
@@ -286,6 +291,7 @@ public:
     struct Config {
         Operation operation;
         Filter filter;
+        bool must_accelerate;
         s32 dst_x0;
         s32 dst_y0;
         s32 dst_x1;
@@ -298,6 +304,7 @@ public:
 
 private:
     VideoCore::RasterizerInterface* rasterizer = nullptr;
+    std::unique_ptr<Blitter::SoftwareBlitEngine> sw_blitter;
 
     /// Performs the copy from the source surface to the destination surface as configured in the
     /// registers.
diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp
index 5bb1427c1..6d43e23ea 100644
--- a/src/video_core/engines/maxwell_3d.cpp
+++ b/src/video_core/engines/maxwell_3d.cpp
@@ -249,9 +249,6 @@ void Maxwell3D::ProcessMethodCall(u32 method, u32 argument, u32 nonshadow_argume
         return;
     case MAXWELL3D_REG_INDEX(fragment_barrier):
         return rasterizer->FragmentBarrier();
-    case MAXWELL3D_REG_INDEX(invalidate_texture_data_cache):
-        rasterizer->InvalidateGPUCache();
-        return rasterizer->WaitForIdle();
     case MAXWELL3D_REG_INDEX(tiled_cache_barrier):
         return rasterizer->TiledCacheBarrier();
     }
@@ -511,10 +508,7 @@ void Maxwell3D::ProcessCounterReset() {
 
 void Maxwell3D::ProcessSyncPoint() {
     const u32 sync_point = regs.sync_info.sync_point.Value();
-    const u32 cache_flush = regs.sync_info.clean_l2.Value();
-    if (cache_flush != 0) {
-        rasterizer->InvalidateGPUCache();
-    }
+    [[maybe_unused]] const u32 cache_flush = regs.sync_info.clean_l2.Value();
     rasterizer->SignalSyncPoint(sync_point);
 }
 
diff --git a/src/video_core/engines/maxwell_dma.cpp b/src/video_core/engines/maxwell_dma.cpp
index 1bf6ca2dd..334429514 100644
--- a/src/video_core/engines/maxwell_dma.cpp
+++ b/src/video_core/engines/maxwell_dma.cpp
@@ -62,7 +62,8 @@ void MaxwellDMA::Launch() {
 
         if (!is_src_pitch && !is_dst_pitch) {
             // If both the source and the destination are in block layout, assert.
-            UNIMPLEMENTED_MSG("Tiled->Tiled DMA transfers are not yet implemented");
+            CopyBlockLinearToBlockLinear();
+            ReleaseSemaphore();
             return;
         }
 
@@ -291,6 +292,70 @@ void MaxwellDMA::FastCopyBlockLinearToPitch() {
     memory_manager.WriteBlock(regs.offset_out, write_buffer.data(), dst_size);
 }
 
+void MaxwellDMA::CopyBlockLinearToBlockLinear() {
+    UNIMPLEMENTED_IF(regs.src_params.block_size.width != 0);
+
+    const bool is_remapping = regs.launch_dma.remap_enable != 0;
+
+    // Deswizzle the input and copy it over.
+    const Parameters& src = regs.src_params;
+    const Parameters& dst = regs.dst_params;
+
+    const u32 num_remap_components = regs.remap_const.num_dst_components_minus_one + 1;
+    const u32 remap_components_size = regs.remap_const.component_size_minus_one + 1;
+
+    const u32 base_bpp = !is_remapping ? 1U : num_remap_components * remap_components_size;
+
+    u32 src_width = src.width;
+    u32 dst_width = dst.width;
+    u32 x_elements = regs.line_length_in;
+    u32 src_x_offset = src.origin.x;
+    u32 dst_x_offset = dst.origin.x;
+    u32 bpp_shift = 0U;
+    if (!is_remapping) {
+        bpp_shift = Common::FoldRight(
+            4U, [](u32 x, u32 y) { return std::min(x, static_cast<u32>(std::countr_zero(y))); },
+            src_width, dst_width, x_elements, src_x_offset, dst_x_offset,
+            static_cast<u32>(regs.offset_in), static_cast<u32>(regs.offset_out));
+        src_width >>= bpp_shift;
+        dst_width >>= bpp_shift;
+        x_elements >>= bpp_shift;
+        src_x_offset >>= bpp_shift;
+        dst_x_offset >>= bpp_shift;
+    }
+
+    const u32 bytes_per_pixel = base_bpp << bpp_shift;
+    const size_t src_size = CalculateSize(true, bytes_per_pixel, src_width, src.height, src.depth,
+                                          src.block_size.height, src.block_size.depth);
+    const size_t dst_size = CalculateSize(true, bytes_per_pixel, dst_width, dst.height, dst.depth,
+                                          dst.block_size.height, dst.block_size.depth);
+
+    const u32 pitch = x_elements * bytes_per_pixel;
+    const size_t mid_buffer_size = pitch * regs.line_count;
+
+    if (read_buffer.size() < src_size) {
+        read_buffer.resize(src_size);
+    }
+    if (write_buffer.size() < dst_size) {
+        write_buffer.resize(dst_size);
+    }
+
+    intermediate_buffer.resize(mid_buffer_size);
+
+    memory_manager.ReadBlock(regs.offset_in, read_buffer.data(), src_size);
+    memory_manager.ReadBlock(regs.offset_out, write_buffer.data(), dst_size);
+
+    UnswizzleSubrect(intermediate_buffer, read_buffer, bytes_per_pixel, src_width, src.height,
+                     src.depth, src_x_offset, src.origin.y, x_elements, regs.line_count,
+                     src.block_size.height, src.block_size.depth, pitch);
+
+    SwizzleSubrect(write_buffer, intermediate_buffer, bytes_per_pixel, dst_width, dst.height,
+                   dst.depth, dst_x_offset, dst.origin.y, x_elements, regs.line_count,
+                   dst.block_size.height, dst.block_size.depth, pitch);
+
+    memory_manager.WriteBlock(regs.offset_out, write_buffer.data(), dst_size);
+}
+
 void MaxwellDMA::ReleaseSemaphore() {
     const auto type = regs.launch_dma.semaphore_type;
     const GPUVAddr address = regs.semaphore.address;
diff --git a/src/video_core/engines/maxwell_dma.h b/src/video_core/engines/maxwell_dma.h
index 953e34adc..d40d3d302 100644
--- a/src/video_core/engines/maxwell_dma.h
+++ b/src/video_core/engines/maxwell_dma.h
@@ -223,6 +223,8 @@ private:
 
     void CopyPitchToBlockLinear();
 
+    void CopyBlockLinearToBlockLinear();
+
     void FastCopyBlockLinearToPitch();
 
     void ReleaseSemaphore();
@@ -234,6 +236,7 @@ private:
 
     std::vector<u8> read_buffer;
     std::vector<u8> write_buffer;
+    std::vector<u8> intermediate_buffer;
 
     static constexpr std::size_t NUM_REGS = 0x800;
     struct Regs {
diff --git a/src/video_core/engines/puller.cpp b/src/video_core/engines/puller.cpp
index 4d2278811..c308ba3fc 100644
--- a/src/video_core/engines/puller.cpp
+++ b/src/video_core/engines/puller.cpp
@@ -118,7 +118,7 @@ void Puller::ProcessSemaphoreRelease() {
     std::function<void()> operation([this, sequence_address, payload] {
         memory_manager.Write<u32>(sequence_address, payload);
     });
-    rasterizer->SyncOperation(std::move(operation));
+    rasterizer->SignalFence(std::move(operation));
 }
 
 void Puller::ProcessSemaphoreAcquire() {
@@ -151,8 +151,8 @@ void Puller::CallPullerMethod(const MethodCall& method_call) {
     case BufferMethods::SemaphoreAddressLow:
     case BufferMethods::SemaphoreSequencePayload:
     case BufferMethods::SyncpointPayload:
-        break;
     case BufferMethods::WrcacheFlush:
+        break;
     case BufferMethods::RefCnt:
         rasterizer->SignalReference();
         break;
diff --git a/src/video_core/engines/sw_blitter/blitter.cpp b/src/video_core/engines/sw_blitter/blitter.cpp
new file mode 100644
index 000000000..2f1ea4626
--- /dev/null
+++ b/src/video_core/engines/sw_blitter/blitter.cpp
@@ -0,0 +1,238 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#include <algorithm>
+#include <cmath>
+#include <vector>
+
+#include "video_core/engines/sw_blitter/blitter.h"
+#include "video_core/engines/sw_blitter/converter.h"
+#include "video_core/memory_manager.h"
+#include "video_core/surface.h"
+#include "video_core/textures/decoders.h"
+
+namespace Tegra {
+class MemoryManager;
+}
+
+using VideoCore::Surface::BytesPerBlock;
+using VideoCore::Surface::PixelFormatFromRenderTargetFormat;
+
+namespace Tegra::Engines::Blitter {
+
+using namespace Texture;
+
+namespace {
+
+constexpr size_t ir_components = 4;
+
+void NearestNeighbor(std::span<const u8> input, std::span<u8> output, u32 src_width, u32 src_height,
+                     u32 dst_width, u32 dst_height, size_t bpp) {
+    const size_t dx_du = std::llround((static_cast<f64>(src_width) / dst_width) * (1ULL << 32));
+    const size_t dy_dv = std::llround((static_cast<f64>(src_height) / dst_height) * (1ULL << 32));
+    size_t src_y = 0;
+    for (u32 y = 0; y < dst_height; y++) {
+        size_t src_x = 0;
+        for (u32 x = 0; x < dst_width; x++) {
+            const size_t read_from = ((src_y * src_width + src_x) >> 32) * bpp;
+            const size_t write_to = (y * dst_width + x) * bpp;
+
+            std::memcpy(&output[write_to], &input[read_from], bpp);
+            src_x += dx_du;
+        }
+        src_y += dy_dv;
+    }
+}
+
+void NearestNeighborFast(std::span<const f32> input, std::span<f32> output, u32 src_width,
+                         u32 src_height, u32 dst_width, u32 dst_height) {
+    const size_t dx_du = std::llround((static_cast<f64>(src_width) / dst_width) * (1ULL << 32));
+    const size_t dy_dv = std::llround((static_cast<f64>(src_height) / dst_height) * (1ULL << 32));
+    size_t src_y = 0;
+    for (u32 y = 0; y < dst_height; y++) {
+        size_t src_x = 0;
+        for (u32 x = 0; x < dst_width; x++) {
+            const size_t read_from = ((src_y * src_width + src_x) >> 32) * ir_components;
+            const size_t write_to = (y * dst_width + x) * ir_components;
+
+            std::memcpy(&output[write_to], &input[read_from], sizeof(f32) * ir_components);
+            src_x += dx_du;
+        }
+        src_y += dy_dv;
+    }
+}
+
+void Bilinear(std::span<const f32> input, std::span<f32> output, size_t src_width,
+              size_t src_height, size_t dst_width, size_t dst_height) {
+    const auto bilinear_sample = [](std::span<const f32> x0_y0, std::span<const f32> x1_y0,
+                                    std::span<const f32> x0_y1, std::span<const f32> x1_y1,
+                                    f32 weight_x, f32 weight_y) {
+        std::array<f32, ir_components> result{};
+        for (size_t i = 0; i < ir_components; i++) {
+            const f32 a = std::lerp(x0_y0[i], x1_y0[i], weight_x);
+            const f32 b = std::lerp(x0_y1[i], x1_y1[i], weight_x);
+            result[i] = std::lerp(a, b, weight_y);
+        }
+        return result;
+    };
+    const f32 dx_du =
+        dst_width > 1 ? static_cast<f32>(src_width - 1) / static_cast<f32>(dst_width - 1) : 0.f;
+    const f32 dy_dv =
+        dst_height > 1 ? static_cast<f32>(src_height - 1) / static_cast<f32>(dst_height - 1) : 0.f;
+    for (u32 y = 0; y < dst_height; y++) {
+        for (u32 x = 0; x < dst_width; x++) {
+            const f32 x_low = std::floor(static_cast<f32>(x) * dx_du);
+            const f32 y_low = std::floor(static_cast<f32>(y) * dy_dv);
+            const f32 x_high = std::ceil(static_cast<f32>(x) * dx_du);
+            const f32 y_high = std::ceil(static_cast<f32>(y) * dy_dv);
+            const f32 weight_x = (static_cast<f32>(x) * dx_du) - x_low;
+            const f32 weight_y = (static_cast<f32>(y) * dy_dv) - y_low;
+
+            const auto read_src = [&](f32 in_x, f32 in_y) {
+                const size_t read_from =
+                    ((static_cast<size_t>(in_x) * src_width + static_cast<size_t>(in_y)) >> 32) *
+                    ir_components;
+                return std::span<const f32>(&input[read_from], ir_components);
+            };
+
+            auto x0_y0 = read_src(x_low, y_low);
+            auto x1_y0 = read_src(x_high, y_low);
+            auto x0_y1 = read_src(x_low, y_high);
+            auto x1_y1 = read_src(x_high, y_high);
+
+            const auto result = bilinear_sample(x0_y0, x1_y0, x0_y1, x1_y1, weight_x, weight_y);
+
+            const size_t write_to = (y * dst_width + x) * ir_components;
+
+            std::memcpy(&output[write_to], &result, sizeof(f32) * ir_components);
+        }
+    }
+}
+
+} // namespace
+
+struct SoftwareBlitEngine::BlitEngineImpl {
+    std::vector<u8> tmp_buffer;
+    std::vector<u8> src_buffer;
+    std::vector<u8> dst_buffer;
+    std::vector<f32> intermediate_src;
+    std::vector<f32> intermediate_dst;
+    ConverterFactory converter_factory;
+};
+
+SoftwareBlitEngine::SoftwareBlitEngine(MemoryManager& memory_manager_)
+    : memory_manager{memory_manager_} {
+    impl = std::make_unique<BlitEngineImpl>();
+}
+
+SoftwareBlitEngine::~SoftwareBlitEngine() = default;
+
+bool SoftwareBlitEngine::Blit(Fermi2D::Surface& src, Fermi2D::Surface& dst,
+                              Fermi2D::Config& config) {
+    const auto get_surface_size = [](Fermi2D::Surface& surface, u32 bytes_per_pixel) {
+        if (surface.linear == Fermi2D::MemoryLayout::BlockLinear) {
+            return CalculateSize(true, bytes_per_pixel, surface.width, surface.height,
+                                 surface.depth, surface.block_height, surface.block_depth);
+        }
+        return static_cast<size_t>(surface.pitch * surface.height);
+    };
+    const auto process_pitch_linear = [](bool unpack, std::span<const u8> input,
+                                         std::span<u8> output, u32 extent_x, u32 extent_y,
+                                         u32 pitch, u32 x0, u32 y0, size_t bpp) {
+        const size_t base_offset = x0 * bpp;
+        const size_t copy_size = extent_x * bpp;
+        for (u32 y = y0; y < extent_y; y++) {
+            const size_t first_offset = y * pitch + base_offset;
+            const size_t second_offset = y * extent_x * bpp;
+            u8* write_to = unpack ? &output[first_offset] : &output[second_offset];
+            const u8* read_from = unpack ? &input[second_offset] : &input[first_offset];
+            std::memcpy(write_to, read_from, copy_size);
+        }
+    };
+
+    const u32 src_extent_x = config.src_x1 - config.src_x0;
+    const u32 src_extent_y = config.src_y1 - config.src_y0;
+
+    const u32 dst_extent_x = config.dst_x1 - config.dst_x0;
+    const u32 dst_extent_y = config.dst_y1 - config.dst_y0;
+    const auto src_bytes_per_pixel = BytesPerBlock(PixelFormatFromRenderTargetFormat(src.format));
+    const auto dst_bytes_per_pixel = BytesPerBlock(PixelFormatFromRenderTargetFormat(dst.format));
+    const size_t src_size = get_surface_size(src, src_bytes_per_pixel);
+    impl->tmp_buffer.resize(src_size);
+    memory_manager.ReadBlock(src.Address(), impl->tmp_buffer.data(), src_size);
+
+    const size_t src_copy_size = src_extent_x * src_extent_y * src_bytes_per_pixel;
+
+    const size_t dst_copy_size = dst_extent_x * dst_extent_y * dst_bytes_per_pixel;
+
+    impl->src_buffer.resize(src_copy_size);
+
+    const bool no_passthrough =
+        src.format != dst.format || src_extent_x != dst_extent_x || src_extent_y != dst_extent_y;
+
+    const auto convertion_phase_same_format = [&]() {
+        NearestNeighbor(impl->src_buffer, impl->dst_buffer, src_extent_x, src_extent_y,
+                        dst_extent_x, dst_extent_y, dst_bytes_per_pixel);
+    };
+
+    const auto convertion_phase_ir = [&]() {
+        auto* input_converter = impl->converter_factory.GetFormatConverter(src.format);
+        impl->intermediate_src.resize((src_copy_size / src_bytes_per_pixel) * ir_components);
+        impl->intermediate_dst.resize((dst_copy_size / dst_bytes_per_pixel) * ir_components);
+        input_converter->ConvertTo(impl->src_buffer, impl->intermediate_src);
+
+        if (config.filter != Fermi2D::Filter::Bilinear) {
+            NearestNeighborFast(impl->intermediate_src, impl->intermediate_dst, src_extent_x,
+                                src_extent_y, dst_extent_x, dst_extent_y);
+        } else {
+            Bilinear(impl->intermediate_src, impl->intermediate_dst, src_extent_x, src_extent_y,
+                     dst_extent_x, dst_extent_y);
+        }
+
+        auto* output_converter = impl->converter_factory.GetFormatConverter(dst.format);
+        output_converter->ConvertFrom(impl->intermediate_dst, impl->dst_buffer);
+    };
+
+    // Do actuall Blit
+
+    impl->dst_buffer.resize(dst_copy_size);
+    if (src.linear == Fermi2D::MemoryLayout::BlockLinear) {
+        UnswizzleSubrect(impl->src_buffer, impl->tmp_buffer, src_bytes_per_pixel, src.width,
+                         src.height, src.depth, config.src_x0, config.src_y0, src_extent_x,
+                         src_extent_y, src.block_height, src.block_depth,
+                         src_extent_x * src_bytes_per_pixel);
+    } else {
+        process_pitch_linear(false, impl->tmp_buffer, impl->src_buffer, src_extent_x, src_extent_y,
+                             src.pitch, config.src_x0, config.src_y0, src_bytes_per_pixel);
+    }
+
+    // Conversion Phase
+    if (no_passthrough) {
+        if (src.format != dst.format || config.filter == Fermi2D::Filter::Bilinear) {
+            convertion_phase_ir();
+        } else {
+            convertion_phase_same_format();
+        }
+    } else {
+        impl->dst_buffer.swap(impl->src_buffer);
+    }
+
+    const size_t dst_size = get_surface_size(dst, dst_bytes_per_pixel);
+    impl->tmp_buffer.resize(dst_size);
+    memory_manager.ReadBlock(dst.Address(), impl->tmp_buffer.data(), dst_size);
+
+    if (dst.linear == Fermi2D::MemoryLayout::BlockLinear) {
+        SwizzleSubrect(impl->tmp_buffer, impl->dst_buffer, dst_bytes_per_pixel, dst.width,
+                       dst.height, dst.depth, config.dst_x0, config.dst_y0, dst_extent_x,
+                       dst_extent_y, dst.block_height, dst.block_depth,
+                       dst_extent_x * dst_bytes_per_pixel);
+    } else {
+        process_pitch_linear(true, impl->dst_buffer, impl->tmp_buffer, dst_extent_x, dst_extent_y,
+                             dst.pitch, config.dst_x0, config.dst_y0,
+                             static_cast<size_t>(dst_bytes_per_pixel));
+    }
+    memory_manager.WriteBlock(dst.Address(), impl->tmp_buffer.data(), dst_size);
+    return true;
+}
+
+} // namespace Tegra::Engines::Blitter
diff --git a/src/video_core/engines/sw_blitter/blitter.h b/src/video_core/engines/sw_blitter/blitter.h
new file mode 100644
index 000000000..85b55c836
--- /dev/null
+++ b/src/video_core/engines/sw_blitter/blitter.h
@@ -0,0 +1,27 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#pragma once
+
+#include "video_core/engines/fermi_2d.h"
+
+namespace Tegra {
+class MemoryManager;
+}
+
+namespace Tegra::Engines::Blitter {
+
+class SoftwareBlitEngine {
+public:
+    explicit SoftwareBlitEngine(MemoryManager& memory_manager_);
+    ~SoftwareBlitEngine();
+
+    bool Blit(Fermi2D::Surface& src, Fermi2D::Surface& dst, Fermi2D::Config& copy_config);
+
+private:
+    MemoryManager& memory_manager;
+    struct BlitEngineImpl;
+    std::unique_ptr<BlitEngineImpl> impl;
+};
+
+} // namespace Tegra::Engines::Blitter
diff --git a/src/video_core/engines/sw_blitter/converter.cpp b/src/video_core/engines/sw_blitter/converter.cpp
new file mode 100644
index 000000000..cd46dfd4f
--- /dev/null
+++ b/src/video_core/engines/sw_blitter/converter.cpp
@@ -0,0 +1,1234 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#include <array>
+#include <bit>
+#include <cmath>
+#include <span>
+#include <unordered_map>
+
+#include "common/assert.h"
+#include "video_core/engines/sw_blitter/converter.h"
+#include "video_core/surface.h"
+#include "video_core/textures/decoders.h"
+
+#ifdef _MSC_VER
+#define FORCE_INLINE __forceinline
+#else
+#define FORCE_INLINE inline __attribute__((always_inline))
+#endif
+
+namespace Tegra::Engines::Blitter {
+
+enum class Swizzle : size_t {
+    R = 0,
+    G = 1,
+    B = 2,
+    A = 3,
+    None,
+};
+
+enum class ComponentType : u32 {
+    SNORM = 1,
+    UNORM = 2,
+    SINT = 3,
+    UINT = 4,
+    SNORM_FORCE_FP16 = 5,
+    UNORM_FORCE_FP16 = 6,
+    FLOAT = 7,
+    SRGB = 8,
+};
+
+namespace {
+
+/*
+ * Note: Use generate_converters.py to generate the structs and searches for new render target
+ * formats and copy paste them to this file in order to update. just call "python
+ * generate_converters.py" and get the code from the output. modify the file to add new formats.
+ */
+
+constexpr std::array<f32, 256> SRGB_TO_RGB_LUT = {
+    0.000000e+00f, 3.035270e-04f, 6.070540e-04f, 9.105810e-04f, 1.214108e-03f, 1.517635e-03f,
+    1.821162e-03f, 2.124689e-03f, 2.428216e-03f, 2.731743e-03f, 3.035270e-03f, 3.346536e-03f,
+    3.676507e-03f, 4.024717e-03f, 4.391442e-03f, 4.776953e-03f, 5.181517e-03f, 5.605392e-03f,
+    6.048833e-03f, 6.512091e-03f, 6.995410e-03f, 7.499032e-03f, 8.023193e-03f, 8.568126e-03f,
+    9.134059e-03f, 9.721218e-03f, 1.032982e-02f, 1.096009e-02f, 1.161224e-02f, 1.228649e-02f,
+    1.298303e-02f, 1.370208e-02f, 1.444384e-02f, 1.520851e-02f, 1.599629e-02f, 1.680738e-02f,
+    1.764195e-02f, 1.850022e-02f, 1.938236e-02f, 2.028856e-02f, 2.121901e-02f, 2.217389e-02f,
+    2.315337e-02f, 2.415763e-02f, 2.518686e-02f, 2.624122e-02f, 2.732089e-02f, 2.842604e-02f,
+    2.955684e-02f, 3.071344e-02f, 3.189603e-02f, 3.310477e-02f, 3.433981e-02f, 3.560131e-02f,
+    3.688945e-02f, 3.820437e-02f, 3.954624e-02f, 4.091520e-02f, 4.231141e-02f, 4.373503e-02f,
+    4.518620e-02f, 4.666509e-02f, 4.817183e-02f, 4.970657e-02f, 5.126946e-02f, 5.286065e-02f,
+    5.448028e-02f, 5.612849e-02f, 5.780543e-02f, 5.951124e-02f, 6.124605e-02f, 6.301001e-02f,
+    6.480327e-02f, 6.662594e-02f, 6.847817e-02f, 7.036009e-02f, 7.227185e-02f, 7.421357e-02f,
+    7.618538e-02f, 7.818742e-02f, 8.021982e-02f, 8.228271e-02f, 8.437621e-02f, 8.650046e-02f,
+    8.865558e-02f, 9.084171e-02f, 9.305897e-02f, 9.530747e-02f, 9.758735e-02f, 9.989873e-02f,
+    1.022417e-01f, 1.046165e-01f, 1.070231e-01f, 1.094617e-01f, 1.119324e-01f, 1.144354e-01f,
+    1.169707e-01f, 1.195384e-01f, 1.221388e-01f, 1.247718e-01f, 1.274377e-01f, 1.301365e-01f,
+    1.328683e-01f, 1.356333e-01f, 1.384316e-01f, 1.412633e-01f, 1.441285e-01f, 1.470273e-01f,
+    1.499598e-01f, 1.529261e-01f, 1.559265e-01f, 1.589608e-01f, 1.620294e-01f, 1.651322e-01f,
+    1.682694e-01f, 1.714411e-01f, 1.746474e-01f, 1.778884e-01f, 1.811642e-01f, 1.844750e-01f,
+    1.878208e-01f, 1.912017e-01f, 1.946178e-01f, 1.980693e-01f, 2.015563e-01f, 2.050787e-01f,
+    2.086369e-01f, 2.122308e-01f, 2.158605e-01f, 2.195262e-01f, 2.232280e-01f, 2.269659e-01f,
+    2.307401e-01f, 2.345506e-01f, 2.383976e-01f, 2.422811e-01f, 2.462013e-01f, 2.501583e-01f,
+    2.541521e-01f, 2.581829e-01f, 2.622507e-01f, 2.663556e-01f, 2.704978e-01f, 2.746773e-01f,
+    2.788943e-01f, 2.831487e-01f, 2.874408e-01f, 2.917706e-01f, 2.961383e-01f, 3.005438e-01f,
+    3.049873e-01f, 3.094689e-01f, 3.139887e-01f, 3.185468e-01f, 3.231432e-01f, 3.277781e-01f,
+    3.324515e-01f, 3.371636e-01f, 3.419144e-01f, 3.467041e-01f, 3.515326e-01f, 3.564001e-01f,
+    3.613068e-01f, 3.662526e-01f, 3.712377e-01f, 3.762621e-01f, 3.813260e-01f, 3.864294e-01f,
+    3.915725e-01f, 3.967552e-01f, 4.019778e-01f, 4.072402e-01f, 4.125426e-01f, 4.178851e-01f,
+    4.232677e-01f, 4.286905e-01f, 4.341536e-01f, 4.396572e-01f, 4.452012e-01f, 4.507858e-01f,
+    4.564110e-01f, 4.620770e-01f, 4.677838e-01f, 4.735315e-01f, 4.793202e-01f, 4.851499e-01f,
+    4.910209e-01f, 4.969330e-01f, 5.028865e-01f, 5.088813e-01f, 5.149177e-01f, 5.209956e-01f,
+    5.271151e-01f, 5.332764e-01f, 5.394795e-01f, 5.457245e-01f, 5.520114e-01f, 5.583404e-01f,
+    5.647115e-01f, 5.711249e-01f, 5.775805e-01f, 5.840784e-01f, 5.906188e-01f, 5.972018e-01f,
+    6.038274e-01f, 6.104956e-01f, 6.172066e-01f, 6.239604e-01f, 6.307572e-01f, 6.375968e-01f,
+    6.444797e-01f, 6.514056e-01f, 6.583748e-01f, 6.653873e-01f, 6.724432e-01f, 6.795425e-01f,
+    6.866853e-01f, 6.938717e-01f, 7.011019e-01f, 7.083758e-01f, 7.156935e-01f, 7.230551e-01f,
+    7.304608e-01f, 7.379104e-01f, 7.454042e-01f, 7.529422e-01f, 7.605245e-01f, 7.681512e-01f,
+    7.758222e-01f, 7.835378e-01f, 7.912979e-01f, 7.991027e-01f, 8.069522e-01f, 8.148466e-01f,
+    8.227857e-01f, 8.307699e-01f, 8.387990e-01f, 8.468732e-01f, 8.549926e-01f, 8.631572e-01f,
+    8.713671e-01f, 8.796224e-01f, 8.879231e-01f, 8.962694e-01f, 9.046612e-01f, 9.130986e-01f,
+    9.215819e-01f, 9.301109e-01f, 9.386857e-01f, 9.473065e-01f, 9.559733e-01f, 9.646863e-01f,
+    9.734453e-01f, 9.822506e-01f, 9.911021e-01f, 1.000000e+00f};
+
+constexpr std::array<f32, 256> RGB_TO_SRGB_LUT = {
+    0.000000e+00f, 4.984009e-02f, 8.494473e-02f, 1.107021e-01f, 1.318038e-01f, 1.500052e-01f,
+    1.661857e-01f, 1.808585e-01f, 1.943532e-01f, 2.068957e-01f, 2.186491e-01f, 2.297351e-01f,
+    2.402475e-01f, 2.502604e-01f, 2.598334e-01f, 2.690152e-01f, 2.778465e-01f, 2.863614e-01f,
+    2.945889e-01f, 3.025538e-01f, 3.102778e-01f, 3.177796e-01f, 3.250757e-01f, 3.321809e-01f,
+    3.391081e-01f, 3.458689e-01f, 3.524737e-01f, 3.589320e-01f, 3.652521e-01f, 3.714419e-01f,
+    3.775084e-01f, 3.834581e-01f, 3.892968e-01f, 3.950301e-01f, 4.006628e-01f, 4.061998e-01f,
+    4.116451e-01f, 4.170030e-01f, 4.222770e-01f, 4.274707e-01f, 4.325873e-01f, 4.376298e-01f,
+    4.426010e-01f, 4.475037e-01f, 4.523403e-01f, 4.571131e-01f, 4.618246e-01f, 4.664766e-01f,
+    4.710712e-01f, 4.756104e-01f, 4.800958e-01f, 4.845292e-01f, 4.889122e-01f, 4.932462e-01f,
+    4.975329e-01f, 5.017734e-01f, 5.059693e-01f, 5.101216e-01f, 5.142317e-01f, 5.183006e-01f,
+    5.223295e-01f, 5.263194e-01f, 5.302714e-01f, 5.341862e-01f, 5.380651e-01f, 5.419087e-01f,
+    5.457181e-01f, 5.494938e-01f, 5.532369e-01f, 5.569480e-01f, 5.606278e-01f, 5.642771e-01f,
+    5.678965e-01f, 5.714868e-01f, 5.750484e-01f, 5.785821e-01f, 5.820884e-01f, 5.855680e-01f,
+    5.890211e-01f, 5.924487e-01f, 5.958509e-01f, 5.992285e-01f, 6.025819e-01f, 6.059114e-01f,
+    6.092176e-01f, 6.125010e-01f, 6.157619e-01f, 6.190008e-01f, 6.222180e-01f, 6.254140e-01f,
+    6.285890e-01f, 6.317436e-01f, 6.348780e-01f, 6.379926e-01f, 6.410878e-01f, 6.441637e-01f,
+    6.472208e-01f, 6.502595e-01f, 6.532799e-01f, 6.562824e-01f, 6.592672e-01f, 6.622347e-01f,
+    6.651851e-01f, 6.681187e-01f, 6.710356e-01f, 6.739363e-01f, 6.768209e-01f, 6.796897e-01f,
+    6.825429e-01f, 6.853807e-01f, 6.882034e-01f, 6.910111e-01f, 6.938041e-01f, 6.965826e-01f,
+    6.993468e-01f, 7.020969e-01f, 7.048331e-01f, 7.075556e-01f, 7.102645e-01f, 7.129600e-01f,
+    7.156424e-01f, 7.183118e-01f, 7.209683e-01f, 7.236121e-01f, 7.262435e-01f, 7.288625e-01f,
+    7.314693e-01f, 7.340640e-01f, 7.366470e-01f, 7.392181e-01f, 7.417776e-01f, 7.443256e-01f,
+    7.468624e-01f, 7.493880e-01f, 7.519025e-01f, 7.544061e-01f, 7.568989e-01f, 7.593810e-01f,
+    7.618526e-01f, 7.643137e-01f, 7.667645e-01f, 7.692052e-01f, 7.716358e-01f, 7.740564e-01f,
+    7.764671e-01f, 7.788681e-01f, 7.812595e-01f, 7.836413e-01f, 7.860138e-01f, 7.883768e-01f,
+    7.907307e-01f, 7.930754e-01f, 7.954110e-01f, 7.977377e-01f, 8.000556e-01f, 8.023647e-01f,
+    8.046651e-01f, 8.069569e-01f, 8.092403e-01f, 8.115152e-01f, 8.137818e-01f, 8.160402e-01f,
+    8.182903e-01f, 8.205324e-01f, 8.227665e-01f, 8.249926e-01f, 8.272109e-01f, 8.294214e-01f,
+    8.316242e-01f, 8.338194e-01f, 8.360070e-01f, 8.381871e-01f, 8.403597e-01f, 8.425251e-01f,
+    8.446831e-01f, 8.468339e-01f, 8.489776e-01f, 8.511142e-01f, 8.532437e-01f, 8.553662e-01f,
+    8.574819e-01f, 8.595907e-01f, 8.616927e-01f, 8.637881e-01f, 8.658767e-01f, 8.679587e-01f,
+    8.700342e-01f, 8.721032e-01f, 8.741657e-01f, 8.762218e-01f, 8.782716e-01f, 8.803151e-01f,
+    8.823524e-01f, 8.843835e-01f, 8.864085e-01f, 8.884274e-01f, 8.904402e-01f, 8.924471e-01f,
+    8.944480e-01f, 8.964431e-01f, 8.984324e-01f, 9.004158e-01f, 9.023935e-01f, 9.043654e-01f,
+    9.063318e-01f, 9.082925e-01f, 9.102476e-01f, 9.121972e-01f, 9.141413e-01f, 9.160800e-01f,
+    9.180133e-01f, 9.199412e-01f, 9.218637e-01f, 9.237810e-01f, 9.256931e-01f, 9.276000e-01f,
+    9.295017e-01f, 9.313982e-01f, 9.332896e-01f, 9.351761e-01f, 9.370575e-01f, 9.389339e-01f,
+    9.408054e-01f, 9.426719e-01f, 9.445336e-01f, 9.463905e-01f, 9.482424e-01f, 9.500897e-01f,
+    9.519322e-01f, 9.537700e-01f, 9.556032e-01f, 9.574316e-01f, 9.592555e-01f, 9.610748e-01f,
+    9.628896e-01f, 9.646998e-01f, 9.665055e-01f, 9.683068e-01f, 9.701037e-01f, 9.718961e-01f,
+    9.736842e-01f, 9.754679e-01f, 9.772474e-01f, 9.790225e-01f, 9.807934e-01f, 9.825601e-01f,
+    9.843225e-01f, 9.860808e-01f, 9.878350e-01f, 9.895850e-01f, 9.913309e-01f, 9.930727e-01f,
+    9.948106e-01f, 9.965444e-01f, 9.982741e-01f, 1.000000e+00f};
+
+} // namespace
+
+struct R32G32B32A32_FLOATTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::FLOAT, ComponentType::FLOAT, ComponentType::FLOAT, ComponentType::FLOAT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32, 32, 32, 32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::A};
+};
+
+struct R32G32B32A32_SINTTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SINT, ComponentType::SINT, ComponentType::SINT, ComponentType::SINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32, 32, 32, 32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::A};
+};
+
+struct R32G32B32A32_UINTTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UINT, ComponentType::UINT, ComponentType::UINT, ComponentType::UINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32, 32, 32, 32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::A};
+};
+
+struct R32G32B32X32_FLOATTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::FLOAT, ComponentType::FLOAT, ComponentType::FLOAT, ComponentType::FLOAT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32, 32, 32, 32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::None};
+};
+
+struct R32G32B32X32_SINTTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SINT, ComponentType::SINT, ComponentType::SINT, ComponentType::SINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32, 32, 32, 32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::None};
+};
+
+struct R32G32B32X32_UINTTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UINT, ComponentType::UINT, ComponentType::UINT, ComponentType::UINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32, 32, 32, 32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::None};
+};
+
+struct R16G16B16A16_UNORMTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {16, 16, 16, 16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::A};
+};
+
+struct R16G16B16A16_SNORMTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SNORM, ComponentType::SNORM, ComponentType::SNORM, ComponentType::SNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {16, 16, 16, 16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::A};
+};
+
+struct R16G16B16A16_SINTTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SINT, ComponentType::SINT, ComponentType::SINT, ComponentType::SINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {16, 16, 16, 16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::A};
+};
+
+struct R16G16B16A16_UINTTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UINT, ComponentType::UINT, ComponentType::UINT, ComponentType::UINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {16, 16, 16, 16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::A};
+};
+
+struct R16G16B16A16_FLOATTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::FLOAT, ComponentType::FLOAT, ComponentType::FLOAT, ComponentType::FLOAT};
+    static constexpr std::array<size_t, num_components> component_sizes = {16, 16, 16, 16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::A};
+};
+
+struct R32G32_FLOATTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::FLOAT, ComponentType::FLOAT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32, 32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct R32G32_SINTTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SINT, ComponentType::SINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32, 32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct R32G32_UINTTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UINT, ComponentType::UINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32, 32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct R16G16B16X16_FLOATTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::FLOAT, ComponentType::FLOAT, ComponentType::FLOAT, ComponentType::FLOAT};
+    static constexpr std::array<size_t, num_components> component_sizes = {16, 16, 16, 16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B, Swizzle::None};
+};
+
+struct A8R8G8B8_UNORMTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8, 8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::A, Swizzle::R, Swizzle::G, Swizzle::B};
+};
+
+struct A8R8G8B8_SRGBTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SRGB, ComponentType::SRGB, ComponentType::SRGB, ComponentType::SRGB};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8, 8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::A, Swizzle::R, Swizzle::G, Swizzle::B};
+};
+
+struct A2B10G10R10_UNORMTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {2, 10, 10, 10};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::A, Swizzle::B, Swizzle::G, Swizzle::R};
+};
+
+struct A2B10G10R10_UINTTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UINT, ComponentType::UINT, ComponentType::UINT, ComponentType::UINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {2, 10, 10, 10};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::A, Swizzle::B, Swizzle::G, Swizzle::R};
+};
+
+struct A2R10G10B10_UNORMTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {2, 10, 10, 10};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::A, Swizzle::R, Swizzle::G, Swizzle::B};
+};
+
+struct A8B8G8R8_UNORMTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8, 8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::A, Swizzle::B, Swizzle::G, Swizzle::R};
+};
+
+struct A8B8G8R8_SRGBTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SRGB, ComponentType::SRGB, ComponentType::SRGB, ComponentType::SRGB};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8, 8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::A, Swizzle::B, Swizzle::G, Swizzle::R};
+};
+
+struct A8B8G8R8_SNORMTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SNORM, ComponentType::SNORM, ComponentType::SNORM, ComponentType::SNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8, 8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::A, Swizzle::B, Swizzle::G, Swizzle::R};
+};
+
+struct A8B8G8R8_SINTTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SINT, ComponentType::SINT, ComponentType::SINT, ComponentType::SINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8, 8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::A, Swizzle::B, Swizzle::G, Swizzle::R};
+};
+
+struct A8B8G8R8_UINTTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UINT, ComponentType::UINT, ComponentType::UINT, ComponentType::UINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8, 8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::A, Swizzle::B, Swizzle::G, Swizzle::R};
+};
+
+struct R16G16_UNORMTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {16, 16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct R16G16_SNORMTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SNORM, ComponentType::SNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {16, 16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct R16G16_SINTTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SINT, ComponentType::SINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {16, 16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct R16G16_UINTTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UINT, ComponentType::UINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {16, 16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct R16G16_FLOATTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::FLOAT, ComponentType::FLOAT};
+    static constexpr std::array<size_t, num_components> component_sizes = {16, 16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct B10G11R11_FLOATTraits {
+    static constexpr size_t num_components = 3;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::FLOAT, ComponentType::FLOAT, ComponentType::FLOAT};
+    static constexpr std::array<size_t, num_components> component_sizes = {10, 11, 11};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::B, Swizzle::G, Swizzle::R};
+};
+
+struct R32_SINTTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct R32_UINTTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct R32_FLOATTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::FLOAT};
+    static constexpr std::array<size_t, num_components> component_sizes = {32};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct X8R8G8B8_UNORMTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8, 8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::None, Swizzle::R, Swizzle::G, Swizzle::B};
+};
+
+struct X8R8G8B8_SRGBTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SRGB, ComponentType::SRGB, ComponentType::SRGB, ComponentType::SRGB};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8, 8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::None, Swizzle::R, Swizzle::G, Swizzle::B};
+};
+
+struct R5G6B5_UNORMTraits {
+    static constexpr size_t num_components = 3;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {5, 6, 5};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::R, Swizzle::G, Swizzle::B};
+};
+
+struct A1R5G5B5_UNORMTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {1, 5, 5, 5};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::A, Swizzle::R, Swizzle::G, Swizzle::B};
+};
+
+struct R8G8_UNORMTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct R8G8_SNORMTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SNORM, ComponentType::SNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct R8G8_SINTTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SINT, ComponentType::SINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct R8G8_UINTTraits {
+    static constexpr size_t num_components = 2;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UINT, ComponentType::UINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R,
+                                                                              Swizzle::G};
+};
+
+struct R16_UNORMTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct R16_SNORMTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct R16_SINTTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct R16_UINTTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct R16_FLOATTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::FLOAT};
+    static constexpr std::array<size_t, num_components> component_sizes = {16};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct R8_UNORMTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct R8_SNORMTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct R8_SINTTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct R8_UINTTraits {
+    static constexpr size_t num_components = 1;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UINT};
+    static constexpr std::array<size_t, num_components> component_sizes = {8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {Swizzle::R};
+};
+
+struct X1R5G5B5_UNORMTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {1, 5, 5, 5};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::None, Swizzle::R, Swizzle::G, Swizzle::B};
+};
+
+struct X8B8G8R8_UNORMTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM, ComponentType::UNORM};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8, 8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::None, Swizzle::B, Swizzle::G, Swizzle::R};
+};
+
+struct X8B8G8R8_SRGBTraits {
+    static constexpr size_t num_components = 4;
+    static constexpr std::array<ComponentType, num_components> component_types = {
+        ComponentType::SRGB, ComponentType::SRGB, ComponentType::SRGB, ComponentType::SRGB};
+    static constexpr std::array<size_t, num_components> component_sizes = {8, 8, 8, 8};
+    static constexpr std::array<Swizzle, num_components> component_swizzle = {
+        Swizzle::None, Swizzle::B, Swizzle::G, Swizzle::R};
+};
+
+template <class ConverterTraits>
+class ConverterImpl : public Converter {
+private:
+    static constexpr size_t num_components = ConverterTraits::num_components;
+    static constexpr std::array<ComponentType, num_components> component_types =
+        ConverterTraits::component_types;
+    static constexpr std::array<size_t, num_components> component_sizes =
+        ConverterTraits::component_sizes;
+    static constexpr std::array<Swizzle, num_components> component_swizzle =
+        ConverterTraits::component_swizzle;
+
+    static constexpr size_t CalculateByteSize() {
+        size_t size = 0;
+        for (const size_t component_size : component_sizes) {
+            size += component_size;
+        }
+        const size_t power = (sizeof(size_t) * 8) - std::countl_zero(size) - 1ULL;
+        const size_t base_size = 1ULL << power;
+        const size_t mask = base_size - 1ULL;
+        return ((size & mask) != 0 ? base_size << 1ULL : base_size) / 8;
+    }
+
+    static constexpr size_t total_bytes_per_pixel = CalculateByteSize();
+    static constexpr size_t total_words_per_pixel =
+        (total_bytes_per_pixel + sizeof(u32) - 1U) / sizeof(u32);
+    static constexpr size_t components_per_ir_rep = 4;
+
+    template <bool get_offsets>
+    static constexpr std::array<size_t, num_components> GetBoundWordsOffsets() {
+        std::array<size_t, num_components> result;
+        result.fill(0);
+        constexpr size_t total_bits_per_word = sizeof(u32) * 8;
+        size_t accumulated_size = 0;
+        size_t count = 0;
+        for (size_t i = 0; i < num_components; i++) {
+            if constexpr (get_offsets) {
+                result[i] = accumulated_size;
+            } else {
+                result[i] = count;
+            }
+            accumulated_size += component_sizes[i];
+            if (accumulated_size > total_bits_per_word) {
+                if constexpr (get_offsets) {
+                    result[i] = 0;
+                } else {
+                    result[i]++;
+                }
+                count++;
+                accumulated_size = component_sizes[i];
+            }
+        }
+        return result;
+    }
+
+    static constexpr std::array<size_t, num_components> bound_words = GetBoundWordsOffsets<false>();
+    static constexpr std::array<size_t, num_components> bound_offsets =
+        GetBoundWordsOffsets<true>();
+
+    static constexpr std::array<u32, num_components> GetComponentsMask() {
+        std::array<u32, num_components> result;
+        for (size_t i = 0; i < num_components; i++) {
+            result[i] = (((u32)~0) >> (8 * sizeof(u32) - component_sizes[i])) << bound_offsets[i];
+        }
+        return result;
+    }
+
+    static constexpr std::array<u32, num_components> component_mask = GetComponentsMask();
+
+    // We are forcing inline so the compiler can SIMD the conversations, since it may do 4 function
+    // calls, it may fail to detect the benefit of inlining.
+    template <size_t which_component>
+    FORCE_INLINE void ConvertToComponent(u32 which_word, f32& out_component) {
+        const u32 value = (which_word >> bound_offsets[which_component]) &
+                          static_cast<u32>((1ULL << component_sizes[which_component]) - 1ULL);
+        const auto sign_extend = [](u32 base_value, size_t bits) {
+            const size_t shift_amount = sizeof(u32) * 8 - bits;
+            s32 shifted_value = static_cast<s32>(base_value << shift_amount);
+            return shifted_value >> shift_amount;
+        };
+        const auto force_to_fp16 = [](f32 base_value) {
+            u32 tmp = std::bit_cast<u32>(base_value);
+            constexpr size_t fp32_mantissa_bits = 23;
+            constexpr size_t fp16_mantissa_bits = 10;
+            constexpr size_t mantissa_mask =
+                ~((1ULL << (fp32_mantissa_bits - fp16_mantissa_bits)) - 1ULL);
+            tmp = tmp & static_cast<u32>(mantissa_mask);
+            // TODO: force the exponent within the range of half float. Not needed in UNORM / SNORM
+            return std::bit_cast<f32>(tmp);
+        };
+        const auto from_fp_n = [&sign_extend](u32 base_value, size_t bits, size_t mantissa) {
+            constexpr size_t fp32_mantissa_bits = 23;
+            size_t shift_towards = fp32_mantissa_bits - mantissa;
+            const u32 new_value =
+                static_cast<u32>(sign_extend(base_value, bits) << shift_towards) & (~(1U << 31));
+            return std::bit_cast<f32>(new_value);
+        };
+        const auto calculate_snorm = [&]() {
+            return static_cast<f32>(
+                static_cast<f32>(sign_extend(value, component_sizes[which_component])) /
+                static_cast<f32>((1ULL << (component_sizes[which_component] - 1ULL)) - 1ULL));
+        };
+        const auto calculate_unorm = [&]() {
+            return static_cast<f32>(
+                static_cast<f32>(value) /
+                static_cast<f32>((1ULL << (component_sizes[which_component])) - 1ULL));
+        };
+        if constexpr (component_types[which_component] == ComponentType::SNORM) {
+            out_component = calculate_snorm();
+        } else if constexpr (component_types[which_component] == ComponentType::UNORM) {
+            out_component = calculate_unorm();
+        } else if constexpr (component_types[which_component] == ComponentType::SINT) {
+            out_component = static_cast<f32>(
+                static_cast<s32>(sign_extend(value, component_sizes[which_component])));
+        } else if constexpr (component_types[which_component] == ComponentType::UINT) {
+            out_component = static_cast<f32>(
+                static_cast<s32>(sign_extend(value, component_sizes[which_component])));
+        } else if constexpr (component_types[which_component] == ComponentType::SNORM_FORCE_FP16) {
+            out_component = calculate_snorm();
+            out_component = force_to_fp16(out_component);
+        } else if constexpr (component_types[which_component] == ComponentType::UNORM_FORCE_FP16) {
+            out_component = calculate_unorm();
+            out_component = force_to_fp16(out_component);
+        } else if constexpr (component_types[which_component] == ComponentType::FLOAT) {
+            if constexpr (component_sizes[which_component] == 32) {
+                out_component = std::bit_cast<f32>(value);
+            } else if constexpr (component_sizes[which_component] == 16) {
+                static constexpr u32 sign_mask = 0x8000;
+                static constexpr u32 mantissa_mask = 0x8000;
+                out_component = std::bit_cast<f32>(((value & sign_mask) << 16) |
+                                                   (((value & 0x7c00) + 0x1C000) << 13) |
+                                                   ((value & mantissa_mask) << 13));
+            } else {
+                out_component = from_fp_n(value, component_sizes[which_component],
+                                          component_sizes[which_component] - 5);
+            }
+        } else if constexpr (component_types[which_component] == ComponentType::SRGB) {
+            if constexpr (component_swizzle[which_component] == Swizzle::A) {
+                out_component = calculate_unorm();
+            } else if constexpr (component_sizes[which_component] == 8) {
+                out_component = SRGB_TO_RGB_LUT[value];
+            } else {
+                out_component = calculate_unorm();
+                UNIMPLEMENTED_MSG("SRGB Conversion with component sizes of {} is unimplemented",
+                                  component_sizes[which_component]);
+            }
+        }
+    }
+
+    // We are forcing inline so the compiler can SIMD the conversations, since it may do 4 function
+    // calls, it may fail to detect the benefit of inlining.
+    template <size_t which_component>
+    FORCE_INLINE void ConvertFromComponent(u32& which_word, f32 in_component) {
+        const auto insert_to_word = [&]<typename T>(T new_word) {
+            which_word |= (static_cast<u32>(new_word) << bound_offsets[which_component]) &
+                          component_mask[which_component];
+        };
+        const auto to_fp_n = [](f32 base_value, size_t bits, size_t mantissa) {
+            constexpr size_t fp32_mantissa_bits = 23;
+            u32 tmp_value = std::bit_cast<u32>(std::max(base_value, 0.0f));
+            size_t shift_towards = fp32_mantissa_bits - mantissa;
+            return tmp_value >> shift_towards;
+        };
+        const auto calculate_unorm = [&]() {
+            return static_cast<u32>(
+                static_cast<f32>(in_component) *
+                static_cast<f32>((1ULL << (component_sizes[which_component])) - 1ULL));
+        };
+        if constexpr (component_types[which_component] == ComponentType::SNORM ||
+                      component_types[which_component] == ComponentType::SNORM_FORCE_FP16) {
+            s32 tmp_word = static_cast<s32>(
+                static_cast<f32>(in_component) *
+                static_cast<f32>((1ULL << (component_sizes[which_component] - 1ULL)) - 1ULL));
+            insert_to_word(tmp_word);
+
+        } else if constexpr (component_types[which_component] == ComponentType::UNORM ||
+                             component_types[which_component] == ComponentType::UNORM_FORCE_FP16) {
+            u32 tmp_word = calculate_unorm();
+            insert_to_word(tmp_word);
+        } else if constexpr (component_types[which_component] == ComponentType::SINT) {
+            s32 tmp_word = static_cast<s32>(in_component);
+            insert_to_word(tmp_word);
+        } else if constexpr (component_types[which_component] == ComponentType::UINT) {
+            u32 tmp_word = static_cast<u32>(in_component);
+            insert_to_word(tmp_word);
+        } else if constexpr (component_types[which_component] == ComponentType::FLOAT) {
+            if constexpr (component_sizes[which_component] == 32) {
+                u32 tmp_word = std::bit_cast<u32>(in_component);
+                insert_to_word(tmp_word);
+            } else if constexpr (component_sizes[which_component] == 16) {
+                static constexpr u32 sign_mask = 0x8000;
+                static constexpr u32 mantissa_mask = 0x03ff;
+                static constexpr u32 exponent_mask = 0x7c00;
+                const u32 tmp_word = std::bit_cast<u32>(in_component);
+                const u32 half = ((tmp_word >> 16) & sign_mask) |
+                                 ((((tmp_word & 0x7f800000) - 0x38000000) >> 13) & exponent_mask) |
+                                 ((tmp_word >> 13) & mantissa_mask);
+                insert_to_word(half);
+            } else {
+                insert_to_word(to_fp_n(in_component, component_sizes[which_component],
+                                       component_sizes[which_component] - 5));
+            }
+        } else if constexpr (component_types[which_component] == ComponentType::SRGB) {
+            if constexpr (component_swizzle[which_component] != Swizzle::A) {
+                if constexpr (component_sizes[which_component] == 8) {
+                    const u32 index = calculate_unorm();
+                    in_component = RGB_TO_SRGB_LUT[index];
+                } else {
+                    UNIMPLEMENTED_MSG("SRGB Conversion with component sizes of {} is unimplemented",
+                                      component_sizes[which_component]);
+                }
+            }
+            const u32 tmp_word = calculate_unorm();
+            insert_to_word(tmp_word);
+        }
+    }
+
+public:
+    void ConvertTo(std::span<const u8> input, std::span<f32> output) override {
+        const size_t num_pixels = output.size() / components_per_ir_rep;
+        for (size_t pixel = 0; pixel < num_pixels; pixel++) {
+            std::array<u32, total_words_per_pixel> words{};
+
+            std::memcpy(words.data(), &input[pixel * total_bytes_per_pixel], total_bytes_per_pixel);
+            std::span<f32> new_components(&output[pixel * components_per_ir_rep],
+                                          components_per_ir_rep);
+            if constexpr (component_swizzle[0] != Swizzle::None) {
+                ConvertToComponent<0>(words[bound_words[0]],
+                                      new_components[static_cast<size_t>(component_swizzle[0])]);
+            } else {
+                new_components[0] = 0.0f;
+            }
+            if constexpr (num_components >= 2) {
+                if constexpr (component_swizzle[1] != Swizzle::None) {
+                    ConvertToComponent<1>(
+                        words[bound_words[1]],
+                        new_components[static_cast<size_t>(component_swizzle[1])]);
+                } else {
+                    new_components[1] = 0.0f;
+                }
+            } else {
+                new_components[1] = 0.0f;
+            }
+            if constexpr (num_components >= 3) {
+                if constexpr (component_swizzle[2] != Swizzle::None) {
+                    ConvertToComponent<2>(
+                        words[bound_words[2]],
+                        new_components[static_cast<size_t>(component_swizzle[2])]);
+                } else {
+                    new_components[2] = 0.0f;
+                }
+            } else {
+                new_components[2] = 0.0f;
+            }
+            if constexpr (num_components >= 4) {
+                if constexpr (component_swizzle[3] != Swizzle::None) {
+                    ConvertToComponent<3>(
+                        words[bound_words[3]],
+                        new_components[static_cast<size_t>(component_swizzle[3])]);
+                } else {
+                    new_components[3] = 0.0f;
+                }
+            } else {
+                new_components[3] = 0.0f;
+            }
+        }
+    }
+
+    void ConvertFrom(std::span<const f32> input, std::span<u8> output) override {
+        const size_t num_pixels = output.size() / total_bytes_per_pixel;
+        for (size_t pixel = 0; pixel < num_pixels; pixel++) {
+            std::span<const f32> old_components(&input[pixel * components_per_ir_rep],
+                                                components_per_ir_rep);
+            std::array<u32, total_words_per_pixel> words{};
+            if constexpr (component_swizzle[0] != Swizzle::None) {
+                ConvertFromComponent<0>(words[bound_words[0]],
+                                        old_components[static_cast<size_t>(component_swizzle[0])]);
+            }
+            if constexpr (num_components >= 2) {
+                if constexpr (component_swizzle[1] != Swizzle::None) {
+                    ConvertFromComponent<1>(
+                        words[bound_words[1]],
+                        old_components[static_cast<size_t>(component_swizzle[1])]);
+                }
+            }
+            if constexpr (num_components >= 3) {
+                if constexpr (component_swizzle[2] != Swizzle::None) {
+                    ConvertFromComponent<2>(
+                        words[bound_words[2]],
+                        old_components[static_cast<size_t>(component_swizzle[2])]);
+                }
+            }
+            if constexpr (num_components >= 4) {
+                if constexpr (component_swizzle[3] != Swizzle::None) {
+                    ConvertFromComponent<3>(
+                        words[bound_words[3]],
+                        old_components[static_cast<size_t>(component_swizzle[3])]);
+                }
+            }
+            std::memcpy(&output[pixel * total_bytes_per_pixel], words.data(),
+                        total_bytes_per_pixel);
+        }
+    }
+
+    ConverterImpl() = default;
+    ~ConverterImpl() override = default;
+};
+
+struct ConverterFactory::ConverterFactoryImpl {
+    std::unordered_map<RenderTargetFormat, std::unique_ptr<Converter>> converters_cache;
+};
+
+ConverterFactory::ConverterFactory() {
+    impl = std::make_unique<ConverterFactoryImpl>();
+}
+
+ConverterFactory::~ConverterFactory() = default;
+
+Converter* ConverterFactory::GetFormatConverter(RenderTargetFormat format) {
+    auto it = impl->converters_cache.find(format);
+    if (it == impl->converters_cache.end()) [[unlikely]] {
+        return BuildConverter(format);
+    }
+    return it->second.get();
+}
+
+class NullConverter : public Converter {
+public:
+    void ConvertTo([[maybe_unused]] std::span<const u8> input, std::span<f32> output) override {
+        std::fill(output.begin(), output.end(), 0.0f);
+    }
+    void ConvertFrom([[maybe_unused]] std::span<const f32> input, std::span<u8> output) override {
+        const u8 fill_value = 0U;
+        std::fill(output.begin(), output.end(), fill_value);
+    }
+    NullConverter() = default;
+    ~NullConverter() = default;
+};
+
+Converter* ConverterFactory::BuildConverter(RenderTargetFormat format) {
+    switch (format) {
+    case RenderTargetFormat::R32G32B32A32_FLOAT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32G32B32A32_FLOATTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R32G32B32A32_SINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32G32B32A32_SINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R32G32B32A32_UINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32G32B32A32_UINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R32G32B32X32_FLOAT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32G32B32X32_FLOATTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R32G32B32X32_SINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32G32B32X32_SINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R32G32B32X32_UINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32G32B32X32_UINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16G16B16A16_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16G16B16A16_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16G16B16A16_SNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16G16B16A16_SNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16G16B16A16_SINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16G16B16A16_SINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16G16B16A16_UINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16G16B16A16_UINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16G16B16A16_FLOAT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16G16B16A16_FLOATTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R32G32_FLOAT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32G32_FLOATTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R32G32_SINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32G32_SINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R32G32_UINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32G32_UINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16G16B16X16_FLOAT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16G16B16X16_FLOATTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::A8R8G8B8_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<A8R8G8B8_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::A8R8G8B8_SRGB:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<A8R8G8B8_SRGBTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::A2B10G10R10_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<A2B10G10R10_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::A2B10G10R10_UINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<A2B10G10R10_UINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::A2R10G10B10_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<A2R10G10B10_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::A8B8G8R8_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<A8B8G8R8_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::A8B8G8R8_SRGB:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<A8B8G8R8_SRGBTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::A8B8G8R8_SNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<A8B8G8R8_SNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::A8B8G8R8_SINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<A8B8G8R8_SINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::A8B8G8R8_UINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<A8B8G8R8_UINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16G16_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16G16_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16G16_SNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16G16_SNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16G16_SINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16G16_SINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16G16_UINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16G16_UINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16G16_FLOAT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16G16_FLOATTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::B10G11R11_FLOAT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<B10G11R11_FLOATTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R32_SINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32_SINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R32_UINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32_UINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R32_FLOAT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R32_FLOATTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::X8R8G8B8_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<X8R8G8B8_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::X8R8G8B8_SRGB:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<X8R8G8B8_SRGBTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R5G6B5_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R5G6B5_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::A1R5G5B5_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<A1R5G5B5_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R8G8_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R8G8_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R8G8_SNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R8G8_SNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R8G8_SINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R8G8_SINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R8G8_UINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R8G8_UINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16_SNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16_SNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16_SINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16_SINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16_UINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16_UINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R16_FLOAT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R16_FLOATTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R8_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R8_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R8_SNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R8_SNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R8_SINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R8_SINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::R8_UINT:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<R8_UINTTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::X1R5G5B5_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<X1R5G5B5_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::X8B8G8R8_UNORM:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<X8B8G8R8_UNORMTraits>>())
+            .first->second.get();
+        break;
+    case RenderTargetFormat::X8B8G8R8_SRGB:
+        return impl->converters_cache
+            .emplace(format, std::make_unique<ConverterImpl<X8B8G8R8_SRGBTraits>>())
+            .first->second.get();
+        break;
+    default: {
+        UNIMPLEMENTED_MSG("This format {} converter is not implemented", format);
+        return impl->converters_cache.emplace(format, std::make_unique<NullConverter>())
+            .first->second.get();
+    }
+    }
+}
+
+} // namespace Tegra::Engines::Blitter
diff --git a/src/video_core/engines/sw_blitter/converter.h b/src/video_core/engines/sw_blitter/converter.h
new file mode 100644
index 000000000..f9bdc516e
--- /dev/null
+++ b/src/video_core/engines/sw_blitter/converter.h
@@ -0,0 +1,36 @@
+// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#pragma once
+
+#include <memory>
+#include <span>
+
+#include "common/common_types.h"
+
+#include "video_core/gpu.h"
+
+namespace Tegra::Engines::Blitter {
+
+class Converter {
+public:
+    virtual void ConvertTo(std::span<const u8> input, std::span<f32> output) = 0;
+    virtual void ConvertFrom(std::span<const f32> input, std::span<u8> output) = 0;
+    virtual ~Converter() = default;
+};
+
+class ConverterFactory {
+public:
+    ConverterFactory();
+    ~ConverterFactory();
+
+    Converter* GetFormatConverter(RenderTargetFormat format);
+
+private:
+    Converter* BuildConverter(RenderTargetFormat format);
+
+    struct ConverterFactoryImpl;
+    std::unique_ptr<ConverterFactoryImpl> impl;
+};
+
+} // namespace Tegra::Engines::Blitter
diff --git a/src/video_core/engines/sw_blitter/generate_converters.py b/src/video_core/engines/sw_blitter/generate_converters.py
new file mode 100644
index 000000000..f641564f7
--- /dev/null
+++ b/src/video_core/engines/sw_blitter/generate_converters.py
@@ -0,0 +1,136 @@
+# SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+import re
+
+class Format:
+    def __init__(self, string_value):
+        self.name = string_value
+        tmp = string_value.split('_')
+        self.component_type = tmp[1]
+        component_data = re.findall(r"\w\d+", tmp[0])
+        self.num_components = len(component_data)
+        sizes = []
+        swizzle = []
+        for data in component_data:
+            swizzle.append(data[0])
+            sizes.append(int(data[1:]))
+        self.sizes = sizes
+        self.swizzle = swizzle
+
+    def build_component_type_array(self):
+        result = "{ "
+        b = False
+        for i in range(0, self.num_components):
+            if b:
+                result += ", "
+            b = True
+            result += "ComponentType::" + self.component_type
+        result += " }"
+        return result
+
+    def build_component_sizes_array(self):
+        result = "{ "
+        b = False
+        for i in range(0, self.num_components):
+            if b:
+                result += ", "
+            b = True
+            result += str(self.sizes[i])
+        result += " }"
+        return result
+
+    def build_component_swizzle_array(self):
+        result = "{ "
+        b = False
+        for i in range(0, self.num_components):
+            if b:
+                result += ", "
+            b = True
+            swizzle = self.swizzle[i]
+            if swizzle == "X":
+                swizzle = "None"
+            result += "Swizzle::" + swizzle
+        result += " }"
+        return result
+
+    def print_declaration(self):
+        print("struct " + self.name + "Traits {")
+        print("  static constexpr size_t num_components = " + str(self.num_components) + ";")
+        print("  static constexpr std::array<ComponentType, num_components> component_types = " + self.build_component_type_array() + ";")
+        print("  static constexpr std::array<size_t, num_components> component_sizes = " + self.build_component_sizes_array() + ";")
+        print("  static constexpr std::array<Swizzle, num_components> component_swizzle = " + self.build_component_swizzle_array() + ";")
+        print("};\n")
+
+    def print_case(self):
+        print("case RenderTargetFormat::" + self.name + ":")
+        print("  return impl->converters_cache")
+        print("    .emplace(format, std::make_unique<ConverterImpl<" + self.name + "Traits>>())")
+        print("    .first->second.get();")
+        print("  break;")
+
+txt = """
+R32G32B32A32_FLOAT
+R32G32B32A32_SINT
+R32G32B32A32_UINT
+R32G32B32X32_FLOAT
+R32G32B32X32_SINT
+R32G32B32X32_UINT
+R16G16B16A16_UNORM
+R16G16B16A16_SNORM
+R16G16B16A16_SINT
+R16G16B16A16_UINT
+R16G16B16A16_FLOAT
+R32G32_FLOAT
+R32G32_SINT
+R32G32_UINT
+R16G16B16X16_FLOAT
+A8R8G8B8_UNORM
+A8R8G8B8_SRGB
+A2B10G10R10_UNORM
+A2B10G10R10_UINT
+A2R10G10B10_UNORM
+A8B8G8R8_UNORM
+A8B8G8R8_SRGB
+A8B8G8R8_SNORM
+A8B8G8R8_SINT
+A8B8G8R8_UINT
+R16G16_UNORM
+R16G16_SNORM
+R16G16_SINT
+R16G16_UINT
+R16G16_FLOAT
+B10G11R11_FLOAT
+R32_SINT
+R32_UINT
+R32_FLOAT
+X8R8G8B8_UNORM
+X8R8G8B8_SRGB
+R5G6B5_UNORM
+A1R5G5B5_UNORM
+R8G8_UNORM
+R8G8_SNORM
+R8G8_SINT
+R8G8_UINT
+R16_UNORM
+R16_SNORM
+R16_SINT
+R16_UINT
+R16_FLOAT
+R8_UNORM
+R8_SNORM
+R8_SINT
+R8_UINT
+X1R5G5B5_UNORM
+X8B8G8R8_UNORM
+X8B8G8R8_SRGB
+"""
+
+x = txt.split()
+y = list(map(lambda a: Format(a), x))
+formats = list(y)
+for format in formats:
+  format.print_declaration()
+
+for format in formats:
+  format.print_case()
diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h
index d0709dc69..8a871593a 100644
--- a/src/video_core/gpu.h
+++ b/src/video_core/gpu.h
@@ -27,12 +27,12 @@ struct CommandList;
 // TODO: Implement the commented ones
 enum class RenderTargetFormat : u32 {
     NONE = 0x0,
-    R32B32G32A32_FLOAT = 0xC0,
+    R32G32B32A32_FLOAT = 0xC0,
     R32G32B32A32_SINT = 0xC1,
     R32G32B32A32_UINT = 0xC2,
-    // R32G32B32X32_FLOAT = 0xC3,
-    // R32G32B32X32_SINT = 0xC4,
-    // R32G32B32X32_UINT = 0xC5,
+    R32G32B32X32_FLOAT = 0xC3,
+    R32G32B32X32_SINT = 0xC4,
+    R32G32B32X32_UINT = 0xC5,
     R16G16B16A16_UNORM = 0xC6,
     R16G16B16A16_SNORM = 0xC7,
     R16G16B16A16_SINT = 0xC8,
@@ -56,13 +56,13 @@ enum class RenderTargetFormat : u32 {
     R16G16_SINT = 0xDC,
     R16G16_UINT = 0xDD,
     R16G16_FLOAT = 0xDE,
-    // A2R10G10B10_UNORM = 0xDF,
+    A2R10G10B10_UNORM = 0xDF,
     B10G11R11_FLOAT = 0xE0,
     R32_SINT = 0xE3,
     R32_UINT = 0xE4,
     R32_FLOAT = 0xE5,
-    // X8R8G8B8_UNORM = 0xE6,
-    // X8R8G8B8_SRGB = 0xE7,
+    X8R8G8B8_UNORM = 0xE6,
+    X8R8G8B8_SRGB = 0xE7,
     R5G6B5_UNORM = 0xE8,
     A1R5G5B5_UNORM = 0xE9,
     R8G8_UNORM = 0xEA,
@@ -79,11 +79,11 @@ enum class RenderTargetFormat : u32 {
     R8_SINT = 0xF5,
     R8_UINT = 0xF6,
 
-    /*
-    A8_UNORM = 0xF7,
+    // A8_UNORM = 0xF7,
     X1R5G5B5_UNORM = 0xF8,
     X8B8G8R8_UNORM = 0xF9,
     X8B8G8R8_SRGB = 0xFA,
+    /*
     Z1R5G5B5_UNORM = 0xFB,
     O1R5G5B5_UNORM = 0xFC,
     Z8R8G8B8_UNORM = 0xFD,
diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp
index 354c6e429..f71a316b6 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp
@@ -466,8 +466,7 @@ bool RasterizerOpenGL::AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Surf
                                              const Tegra::Engines::Fermi2D::Config& copy_config) {
     MICROPROFILE_SCOPE(OpenGL_Blits);
     std::scoped_lock lock{texture_cache.mutex};
-    texture_cache.BlitImage(dst, src, copy_config);
-    return true;
+    return texture_cache.BlitImage(dst, src, copy_config);
 }
 
 Tegra::Engines::AccelerateDMAInterface& RasterizerOpenGL::AccessAccelerateDMA() {
diff --git a/src/video_core/renderer_opengl/maxwell_to_gl.h b/src/video_core/renderer_opengl/maxwell_to_gl.h
index e14f9b2db..ef1190e1f 100644
--- a/src/video_core/renderer_opengl/maxwell_to_gl.h
+++ b/src/video_core/renderer_opengl/maxwell_to_gl.h
@@ -28,6 +28,7 @@ constexpr std::array<FormatTuple, VideoCore::Surface::MaxPixelFormat> FORMAT_TAB
     {GL_RGB5_A1, GL_BGRA, GL_UNSIGNED_SHORT_1_5_5_5_REV},             // A1R5G5B5_UNORM
     {GL_RGB10_A2, GL_RGBA, GL_UNSIGNED_INT_2_10_10_10_REV},           // A2B10G10R10_UNORM
     {GL_RGB10_A2UI, GL_RGBA_INTEGER, GL_UNSIGNED_INT_2_10_10_10_REV}, // A2B10G10R10_UINT
+    {GL_RGB10_A2, GL_BGRA, GL_UNSIGNED_INT_2_10_10_10_REV},           // A2R10G10B10_UNORM
     {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_1_5_5_5_REV},             // A1B5G5R5_UNORM
     {GL_RGB5_A1, GL_RGBA, GL_UNSIGNED_SHORT_5_5_5_1},                 // A5B5G5R1_UNORM
     {GL_R8, GL_RED, GL_UNSIGNED_BYTE},                                // R8_UNORM
diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
index b4f5ee665..430a84272 100644
--- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
+++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp
@@ -125,6 +125,7 @@ struct FormatTuple {
     {VK_FORMAT_A1R5G5B5_UNORM_PACK16, Attachable},              // A1R5G5B5_UNORM
     {VK_FORMAT_A2B10G10R10_UNORM_PACK32, Attachable | Storage}, // A2B10G10R10_UNORM
     {VK_FORMAT_A2B10G10R10_UINT_PACK32, Attachable | Storage},  // A2B10G10R10_UINT
+    {VK_FORMAT_A2R10G10B10_UNORM_PACK32, Attachable | Storage}, // A2R10G10B10_UNORM
     {VK_FORMAT_A1R5G5B5_UNORM_PACK16, Attachable},         // A1B5G5R5_UNORM (flipped with swizzle)
     {VK_FORMAT_R5G5B5A1_UNORM_PACK16},                     // A5B5G5R1_UNORM (specially swizzled)
     {VK_FORMAT_R8_UNORM, Attachable | Storage},            // R8_UNORM
diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
index 12b13cc59..d8ad8815c 100644
--- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp
+++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp
@@ -542,8 +542,7 @@ bool RasterizerVulkan::AccelerateSurfaceCopy(const Tegra::Engines::Fermi2D::Surf
                                              const Tegra::Engines::Fermi2D::Surface& dst,
                                              const Tegra::Engines::Fermi2D::Config& copy_config) {
     std::scoped_lock lock{texture_cache.mutex};
-    texture_cache.BlitImage(dst, src, copy_config);
-    return true;
+    return texture_cache.BlitImage(dst, src, copy_config);
 }
 
 Tegra::Engines::AccelerateDMAInterface& RasterizerVulkan::AccessAccelerateDMA() {
diff --git a/src/video_core/surface.cpp b/src/video_core/surface.cpp
index 6bd133d10..b618e1a25 100644
--- a/src/video_core/surface.cpp
+++ b/src/video_core/surface.cpp
@@ -93,11 +93,14 @@ PixelFormat PixelFormatFromDepthFormat(Tegra::DepthFormat format) {
 
 PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format) {
     switch (format) {
-    case Tegra::RenderTargetFormat::R32B32G32A32_FLOAT:
+    case Tegra::RenderTargetFormat::R32G32B32A32_FLOAT:
+    case Tegra::RenderTargetFormat::R32G32B32X32_FLOAT:
         return PixelFormat::R32G32B32A32_FLOAT;
     case Tegra::RenderTargetFormat::R32G32B32A32_SINT:
+    case Tegra::RenderTargetFormat::R32G32B32X32_SINT:
         return PixelFormat::R32G32B32A32_SINT;
     case Tegra::RenderTargetFormat::R32G32B32A32_UINT:
+    case Tegra::RenderTargetFormat::R32G32B32X32_UINT:
         return PixelFormat::R32G32B32A32_UINT;
     case Tegra::RenderTargetFormat::R16G16B16A16_UNORM:
         return PixelFormat::R16G16B16A16_UNORM;
@@ -118,16 +121,22 @@ PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format)
     case Tegra::RenderTargetFormat::R16G16B16X16_FLOAT:
         return PixelFormat::R16G16B16X16_FLOAT;
     case Tegra::RenderTargetFormat::A8R8G8B8_UNORM:
+    case Tegra::RenderTargetFormat::X8R8G8B8_UNORM:
         return PixelFormat::B8G8R8A8_UNORM;
     case Tegra::RenderTargetFormat::A8R8G8B8_SRGB:
+    case Tegra::RenderTargetFormat::X8R8G8B8_SRGB:
         return PixelFormat::B8G8R8A8_SRGB;
     case Tegra::RenderTargetFormat::A2B10G10R10_UNORM:
         return PixelFormat::A2B10G10R10_UNORM;
     case Tegra::RenderTargetFormat::A2B10G10R10_UINT:
         return PixelFormat::A2B10G10R10_UINT;
+    case Tegra::RenderTargetFormat::A2R10G10B10_UNORM:
+        return PixelFormat::A2R10G10B10_UNORM;
     case Tegra::RenderTargetFormat::A8B8G8R8_UNORM:
+    case Tegra::RenderTargetFormat::X8B8G8R8_UNORM:
         return PixelFormat::A8B8G8R8_UNORM;
     case Tegra::RenderTargetFormat::A8B8G8R8_SRGB:
+    case Tegra::RenderTargetFormat::X8B8G8R8_SRGB:
         return PixelFormat::A8B8G8R8_SRGB;
     case Tegra::RenderTargetFormat::A8B8G8R8_SNORM:
         return PixelFormat::A8B8G8R8_SNORM;
@@ -156,6 +165,7 @@ PixelFormat PixelFormatFromRenderTargetFormat(Tegra::RenderTargetFormat format)
     case Tegra::RenderTargetFormat::R5G6B5_UNORM:
         return PixelFormat::R5G6B5_UNORM;
     case Tegra::RenderTargetFormat::A1R5G5B5_UNORM:
+    case Tegra::RenderTargetFormat::X1R5G5B5_UNORM:
         return PixelFormat::A1R5G5B5_UNORM;
     case Tegra::RenderTargetFormat::R8G8_UNORM:
         return PixelFormat::R8G8_UNORM;
diff --git a/src/video_core/surface.h b/src/video_core/surface.h
index 57ca7f597..44b79af20 100644
--- a/src/video_core/surface.h
+++ b/src/video_core/surface.h
@@ -23,6 +23,7 @@ enum class PixelFormat {
     A1R5G5B5_UNORM,
     A2B10G10R10_UNORM,
     A2B10G10R10_UINT,
+    A2R10G10B10_UNORM,
     A1B5G5R5_UNORM,
     A5B5G5R1_UNORM,
     R8_UNORM,
@@ -159,6 +160,7 @@ constexpr std::array<u8, MaxPixelFormat> BLOCK_WIDTH_TABLE = {{
     1,  // A1R5G5B5_UNORM
     1,  // A2B10G10R10_UNORM
     1,  // A2B10G10R10_UINT
+    1,  // A2R10G10B10_UNORM
     1,  // A1B5G5R5_UNORM
     1,  // A5B5G5R1_UNORM
     1,  // R8_UNORM
@@ -264,6 +266,7 @@ constexpr std::array<u8, MaxPixelFormat> BLOCK_HEIGHT_TABLE = {{
     1,  // A1R5G5B5_UNORM
     1,  // A2B10G10R10_UNORM
     1,  // A2B10G10R10_UINT
+    1,  // A2R10G10B10_UNORM
     1,  // A1B5G5R5_UNORM
     1,  // A5B5G5R1_UNORM
     1,  // R8_UNORM
@@ -369,6 +372,7 @@ constexpr std::array<u8, MaxPixelFormat> BITS_PER_BLOCK_TABLE = {{
     16,  // A1R5G5B5_UNORM
     32,  // A2B10G10R10_UNORM
     32,  // A2B10G10R10_UINT
+    32,  // A2R10G10B10_UNORM
     16,  // A1B5G5R5_UNORM
     16,  // A5B5G5R1_UNORM
     8,   // R8_UNORM
diff --git a/src/video_core/texture_cache/formatter.h b/src/video_core/texture_cache/formatter.h
index acc854715..f1f0a057b 100644
--- a/src/video_core/texture_cache/formatter.h
+++ b/src/video_core/texture_cache/formatter.h
@@ -35,6 +35,8 @@ struct fmt::formatter<VideoCore::Surface::PixelFormat> : fmt::formatter<fmt::str
                 return "A2B10G10R10_UNORM";
             case PixelFormat::A2B10G10R10_UINT:
                 return "A2B10G10R10_UINT";
+            case PixelFormat::A2R10G10B10_UNORM:
+                return "A2R10G10B10_UNORM";
             case PixelFormat::A1B5G5R5_UNORM:
                 return "A1B5G5R5_UNORM";
             case PixelFormat::A5B5G5R1_UNORM:
diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h
index 8ef75fe73..8e68a2e53 100644
--- a/src/video_core/texture_cache/texture_cache.h
+++ b/src/video_core/texture_cache/texture_cache.h
@@ -506,10 +506,14 @@ void TextureCache<P>::UnmapGPUMemory(size_t as_id, GPUVAddr gpu_addr, size_t siz
 }
 
 template <class P>
-void TextureCache<P>::BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
+bool TextureCache<P>::BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
                                 const Tegra::Engines::Fermi2D::Surface& src,
                                 const Tegra::Engines::Fermi2D::Config& copy) {
-    const BlitImages images = GetBlitImages(dst, src, copy);
+    const auto result = GetBlitImages(dst, src, copy);
+    if (!result) {
+        return false;
+    }
+    const BlitImages images = *result;
     const ImageId dst_id = images.dst_id;
     const ImageId src_id = images.src_id;
 
@@ -596,6 +600,7 @@ void TextureCache<P>::BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
         runtime.BlitImage(dst_framebuffer, dst_view, src_view, dst_region, src_region, copy.filter,
                           copy.operation);
     }
+    return true;
 }
 
 template <class P>
@@ -1133,7 +1138,7 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA
 }
 
 template <class P>
-typename TextureCache<P>::BlitImages TextureCache<P>::GetBlitImages(
+std::optional<typename TextureCache<P>::BlitImages> TextureCache<P>::GetBlitImages(
     const Tegra::Engines::Fermi2D::Surface& dst, const Tegra::Engines::Fermi2D::Surface& src,
     const Tegra::Engines::Fermi2D::Config& copy) {
 
@@ -1154,6 +1159,20 @@ typename TextureCache<P>::BlitImages TextureCache<P>::GetBlitImages(
         has_deleted_images = false;
         src_id = FindImage(src_info, src_addr, try_options);
         dst_id = FindImage(dst_info, dst_addr, try_options);
+        if (!copy.must_accelerate) {
+            do {
+                if (!src_id && !dst_id) {
+                    return std::nullopt;
+                }
+                if (src_id && True(slot_images[src_id].flags & ImageFlagBits::GpuModified)) {
+                    break;
+                }
+                if (dst_id && True(slot_images[dst_id].flags & ImageFlagBits::GpuModified)) {
+                    break;
+                }
+                return std::nullopt;
+            } while (false);
+        }
         const ImageBase* const src_image = src_id ? &slot_images[src_id] : nullptr;
         if (src_image && src_image->info.num_samples > 1) {
             RelaxedOptions find_options{FIND_OPTIONS | RelaxedOptions::ForceBrokenViews};
@@ -1194,12 +1213,12 @@ typename TextureCache<P>::BlitImages TextureCache<P>::GetBlitImages(
             dst_id = FindOrInsertImage(dst_info, dst_addr, RelaxedOptions{});
         } while (has_deleted_images);
     }
-    return BlitImages{
+    return {BlitImages{
         .dst_id = dst_id,
         .src_id = src_id,
         .dst_format = dst_info.format,
         .src_format = src_info.format,
-    };
+    }};
 }
 
 template <class P>
diff --git a/src/video_core/texture_cache/texture_cache_base.h b/src/video_core/texture_cache/texture_cache_base.h
index 2fa8445eb..9db7195bf 100644
--- a/src/video_core/texture_cache/texture_cache_base.h
+++ b/src/video_core/texture_cache/texture_cache_base.h
@@ -174,7 +174,7 @@ public:
     void UnmapGPUMemory(size_t as_id, GPUVAddr gpu_addr, size_t size);
 
     /// Blit an image with the given parameters
-    void BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
+    bool BlitImage(const Tegra::Engines::Fermi2D::Surface& dst,
                    const Tegra::Engines::Fermi2D::Surface& src,
                    const Tegra::Engines::Fermi2D::Config& copy);
 
@@ -285,9 +285,9 @@ private:
     [[nodiscard]] ImageId JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VAddr cpu_addr);
 
     /// Return a blit image pair from the given guest blit parameters
-    [[nodiscard]] BlitImages GetBlitImages(const Tegra::Engines::Fermi2D::Surface& dst,
-                                           const Tegra::Engines::Fermi2D::Surface& src,
-                                           const Tegra::Engines::Fermi2D::Config& copy);
+    [[nodiscard]] std::optional<BlitImages> GetBlitImages(
+        const Tegra::Engines::Fermi2D::Surface& dst, const Tegra::Engines::Fermi2D::Surface& src,
+        const Tegra::Engines::Fermi2D::Config& copy);
 
     /// Find or create a sampler from a guest descriptor sampler
     [[nodiscard]] SamplerId FindSampler(const TSCEntry& config);
diff --git a/src/yuzu/main.ui b/src/yuzu/main.ui
index e670acc30..013ba0ceb 100644
--- a/src/yuzu/main.ui
+++ b/src/yuzu/main.ui
@@ -231,6 +231,9 @@
    <property name="text">
     <string>Con&amp;figure...</string>
    </property>
+   <property name="menuRole">
+    <enum>QAction::PreferencesRole</enum>
+   </property>
   </action>
   <action name="action_Display_Dock_Widget_Headers">
    <property name="checkable">
@@ -363,6 +366,9 @@
    <property name="text">
     <string>&amp;Configure TAS...</string>
    </property>
+   <property name="menuRole">
+    <enum>QAction::NoRole</enum>
+   </property>
   </action>
   <action name="action_Configure_Current_Game">
    <property name="enabled">
@@ -371,6 +377,9 @@
    <property name="text">
     <string>Configure C&amp;urrent Game...</string>
    </property>
+   <property name="menuRole">
+    <enum>QAction::NoRole</enum>
+   </property>
   </action>
   <action name="action_TAS_Start">
    <property name="enabled">