diff options
Diffstat (limited to 'src')
46 files changed, 1037 insertions, 221 deletions
diff --git a/src/audio_core/renderer/adsp/audio_renderer.cpp b/src/audio_core/renderer/adsp/audio_renderer.cpp index 503f40349..1cbeed302 100644 --- a/src/audio_core/renderer/adsp/audio_renderer.cpp +++ b/src/audio_core/renderer/adsp/audio_renderer.cpp @@ -154,6 +154,11 @@ void AudioRenderer::ThreadFunc() { return; case RenderMessage::AudioRenderer_Render: { + if (system.IsShuttingDown()) [[unlikely]] { + std::this_thread::sleep_for(std::chrono::milliseconds(5)); + mailbox->ADSPSendMessage(RenderMessage::AudioRenderer_RenderResponse); + continue; + } std::array<bool, MaxRendererSessions> buffers_reset{}; std::array<u64, MaxRendererSessions> render_times_taken{}; const auto start_time{system.CoreTiming().GetClockTicks()}; diff --git a/src/audio_core/renderer/system_manager.cpp b/src/audio_core/renderer/system_manager.cpp index 07d8ed093..300ecdbf1 100644 --- a/src/audio_core/renderer/system_manager.cpp +++ b/src/audio_core/renderer/system_manager.cpp @@ -27,7 +27,7 @@ bool SystemManager::InitializeUnsafe() { if (!active) { if (adsp.Start()) { active = true; - thread = std::jthread([this](std::stop_token stop_token) { ThreadFunc(); }); + thread = std::jthread([this](std::stop_token stop_token) { ThreadFunc(stop_token); }); } } @@ -39,8 +39,7 @@ void SystemManager::Stop() { return; } active = false; - update.store(true); - update.notify_all(); + thread.request_stop(); thread.join(); adsp.Stop(); } @@ -85,12 +84,12 @@ bool SystemManager::Remove(System& system_) { return true; } -void SystemManager::ThreadFunc() { +void SystemManager::ThreadFunc(std::stop_token stop_token) { static constexpr char name[]{"AudioRenderSystemManager"}; MicroProfileOnThreadCreate(name); Common::SetCurrentThreadName(name); Common::SetCurrentThreadPriority(Common::ThreadPriority::High); - while (active) { + while (active && !stop_token.stop_requested()) { { std::scoped_lock l{mutex1}; diff --git a/src/audio_core/renderer/system_manager.h b/src/audio_core/renderer/system_manager.h index 1f0bbd8b4..9681fd121 100644 --- a/src/audio_core/renderer/system_manager.h +++ b/src/audio_core/renderer/system_manager.h @@ -66,13 +66,7 @@ private: /** * Main thread responsible for command generation. */ - void ThreadFunc(); - - enum class StreamState { - Filling, - Steady, - Draining, - }; + void ThreadFunc(std::stop_token stop_token); /// Core system Core::System& core; @@ -90,8 +84,6 @@ private: ADSP::ADSP& adsp; /// AudioRenderer mailbox for communication ADSP::AudioRenderer_Mailbox* mailbox{}; - /// Atomic for main thread to wait on - std::atomic<bool> update{}; }; } // namespace AudioCore::AudioRenderer diff --git a/src/audio_core/sink/sink_stream.cpp b/src/audio_core/sink/sink_stream.cpp index 13ba26e74..9bbb54162 100644 --- a/src/audio_core/sink/sink_stream.cpp +++ b/src/audio_core/sink/sink_stream.cpp @@ -271,8 +271,8 @@ u64 SinkStream::GetExpectedPlayedSampleCount() { void SinkStream::WaitFreeSpace() { std::unique_lock lk{release_mutex}; - release_cv.wait( - lk, [this]() { return queued_buffers < max_queue_size || system.IsShuttingDown(); }); + release_cv.wait_for(lk, std::chrono::milliseconds(5), + [this]() { return queued_buffers < max_queue_size; }); } } // namespace AudioCore::Sink diff --git a/src/common/settings.cpp b/src/common/settings.cpp index ba617aea1..ff53e80bb 100644 --- a/src/common/settings.cpp +++ b/src/common/settings.cpp @@ -61,6 +61,7 @@ void LogSettings() { log_setting("Renderer_NvdecEmulation", values.nvdec_emulation.GetValue()); log_setting("Renderer_AccelerateASTC", values.accelerate_astc.GetValue()); log_setting("Renderer_AsyncASTC", values.async_astc.GetValue()); + log_setting("Renderer_AstcRecompression", values.astc_recompression.GetValue()); log_setting("Renderer_UseVsync", values.vsync_mode.GetValue()); log_setting("Renderer_UseReactiveFlushing", values.use_reactive_flushing.GetValue()); log_setting("Renderer_ShaderBackend", values.shader_backend.GetValue()); @@ -224,6 +225,7 @@ void RestoreGlobalState(bool is_powered_on) { values.nvdec_emulation.SetGlobal(true); values.accelerate_astc.SetGlobal(true); values.async_astc.SetGlobal(true); + values.astc_recompression.SetGlobal(true); values.use_reactive_flushing.SetGlobal(true); values.shader_backend.SetGlobal(true); values.use_asynchronous_shaders.SetGlobal(true); diff --git a/src/common/settings.h b/src/common/settings.h index 36ffcd693..7f865b2a7 100644 --- a/src/common/settings.h +++ b/src/common/settings.h @@ -90,6 +90,12 @@ enum class AntiAliasing : u32 { LastAA = Smaa, }; +enum class AstcRecompression : u32 { + Uncompressed = 0, + Bc1 = 1, + Bc3 = 2, +}; + struct ResolutionScalingInfo { u32 up_scale{1}; u32 down_shift{0}; @@ -473,6 +479,9 @@ struct Values { SwitchableSetting<bool> use_vulkan_driver_pipeline_cache{true, "use_vulkan_driver_pipeline_cache"}; SwitchableSetting<bool> enable_compute_pipelines{false, "enable_compute_pipelines"}; + SwitchableSetting<AstcRecompression, true> astc_recompression{ + AstcRecompression::Uncompressed, AstcRecompression::Uncompressed, AstcRecompression::Bc3, + "astc_recompression"}; SwitchableSetting<u8> bg_red{0, "bg_red"}; SwitchableSetting<u8> bg_green{0, "bg_green"}; diff --git a/src/core/hid/emulated_controller.cpp b/src/core/hid/emulated_controller.cpp index 366880711..bbfea7117 100644 --- a/src/core/hid/emulated_controller.cpp +++ b/src/core/hid/emulated_controller.cpp @@ -1283,9 +1283,14 @@ bool EmulatedController::HasNfc() const { } bool EmulatedController::WriteNfc(const std::vector<u8>& data) { - auto& nfc_output_device = output_devices[3]; + auto& nfc_output_device = output_devices[static_cast<std::size_t>(DeviceIndex::Right)]; + auto& nfc_virtual_output_device = output_devices[3]; + + if (nfc_output_device->SupportsNfc() != Common::Input::NfcState::NotSupported) { + return nfc_output_device->WriteNfcData(data) == Common::Input::NfcState::Success; + } - return nfc_output_device->WriteNfcData(data) == Common::Input::NfcState::Success; + return nfc_virtual_output_device->WriteNfcData(data) == Common::Input::NfcState::Success; } void EmulatedController::SetLedPattern() { diff --git a/src/core/hle/kernel/k_memory_block_manager.h b/src/core/hle/kernel/k_memory_block_manager.h index 7c0bd16f0..96496e990 100644 --- a/src/core/hle/kernel/k_memory_block_manager.h +++ b/src/core/hle/kernel/k_memory_block_manager.h @@ -144,14 +144,10 @@ private: class KScopedMemoryBlockManagerAuditor { public: - explicit KScopedMemoryBlockManagerAuditor(KMemoryBlockManager* m) : m_manager(m) { - ASSERT(m_manager->CheckState()); - } + explicit KScopedMemoryBlockManagerAuditor(KMemoryBlockManager* m) : m_manager(m) {} explicit KScopedMemoryBlockManagerAuditor(KMemoryBlockManager& m) : KScopedMemoryBlockManagerAuditor(std::addressof(m)) {} - ~KScopedMemoryBlockManagerAuditor() { - ASSERT(m_manager->CheckState()); - } + ~KScopedMemoryBlockManagerAuditor() = default; private: KMemoryBlockManager* m_manager; diff --git a/src/core/hle/service/nfc/common/amiibo_crypto.cpp b/src/core/hle/service/nfc/common/amiibo_crypto.cpp index f3901ee8d..b2bcb68c3 100644 --- a/src/core/hle/service/nfc/common/amiibo_crypto.cpp +++ b/src/core/hle/service/nfc/common/amiibo_crypto.cpp @@ -52,9 +52,6 @@ bool IsAmiiboValid(const EncryptedNTAG215File& ntag_file) { if (ntag_file.compability_container != 0xEEFF10F1U) { return false; } - if (amiibo_data.constant_value != 0xA5) { - return false; - } if (amiibo_data.model_info.tag_type != NFC::PackedTagType::Type2) { return false; } diff --git a/src/core/hle/service/nfc/common/device.cpp b/src/core/hle/service/nfc/common/device.cpp index 322bde2ed..0bd7900e1 100644 --- a/src/core/hle/service/nfc/common/device.cpp +++ b/src/core/hle/service/nfc/common/device.cpp @@ -119,18 +119,31 @@ bool NfcDevice::LoadNfcTag(std::span<const u8> data) { memcpy(&tag_data, data.data(), sizeof(NFP::EncryptedNTAG215File)); is_plain_amiibo = NFP::AmiiboCrypto::IsAmiiboValid(tag_data); + is_write_protected = false; + device_state = DeviceState::TagFound; + deactivate_event->GetReadableEvent().Clear(); + activate_event->Signal(); + + // Fallback for plain amiibos if (is_plain_amiibo) { - encrypted_tag_data = NFP::AmiiboCrypto::EncodedDataToNfcData(tag_data); LOG_INFO(Service_NFP, "Using plain amiibo"); - } else { - tag_data = {}; + encrypted_tag_data = NFP::AmiiboCrypto::EncodedDataToNfcData(tag_data); + return true; + } + + // Fallback for encrypted amiibos without keys + if (!NFP::AmiiboCrypto::IsKeyAvailable()) { + LOG_INFO(Service_NFC, "Loading amiibo without keys"); memcpy(&encrypted_tag_data, data.data(), sizeof(NFP::EncryptedNTAG215File)); + BuildAmiiboWithoutKeys(); + is_plain_amiibo = true; + is_write_protected = true; + return true; } - device_state = DeviceState::TagFound; - deactivate_event->GetReadableEvent().Clear(); - activate_event->Signal(); + tag_data = {}; + memcpy(&encrypted_tag_data, data.data(), sizeof(NFP::EncryptedNTAG215File)); return true; } @@ -346,23 +359,15 @@ Result NfcDevice::Mount(NFP::ModelType model_type, NFP::MountTarget mount_target return ResultWrongDeviceState; } - // The loaded amiibo is not encrypted - if (is_plain_amiibo) { - device_state = DeviceState::TagMounted; - mount_target = mount_target_; - return ResultSuccess; - } - if (!NFP::AmiiboCrypto::IsAmiiboValid(encrypted_tag_data)) { LOG_ERROR(Service_NFP, "Not an amiibo"); return ResultNotAnAmiibo; } - // Mark amiibos as read only when keys are missing - if (!NFP::AmiiboCrypto::IsKeyAvailable()) { - LOG_ERROR(Service_NFP, "No keys detected"); + // The loaded amiibo is not encrypted + if (is_plain_amiibo) { device_state = DeviceState::TagMounted; - mount_target = NFP::MountTarget::Rom; + mount_target = mount_target_; return ResultSuccess; } @@ -421,11 +426,11 @@ Result NfcDevice::Flush() { tag_data.write_counter++; - FlushWithBreak(NFP::BreakType::Normal); + const auto result = FlushWithBreak(NFP::BreakType::Normal); is_data_moddified = false; - return ResultSuccess; + return result; } Result NfcDevice::FlushDebug() { @@ -444,11 +449,11 @@ Result NfcDevice::FlushDebug() { tag_data.write_counter++; - FlushWithBreak(NFP::BreakType::Normal); + const auto result = FlushWithBreak(NFP::BreakType::Normal); is_data_moddified = false; - return ResultSuccess; + return result; } Result NfcDevice::FlushWithBreak(NFP::BreakType break_type) { @@ -457,6 +462,11 @@ Result NfcDevice::FlushWithBreak(NFP::BreakType break_type) { return ResultWrongDeviceState; } + if (is_write_protected) { + LOG_ERROR(Service_NFP, "No keys available skipping write request"); + return ResultSuccess; + } + std::vector<u8> data(sizeof(NFP::EncryptedNTAG215File)); if (is_plain_amiibo) { memcpy(data.data(), &tag_data, sizeof(tag_data)); @@ -1033,7 +1043,6 @@ Result NfcDevice::GetAll(NFP::NfpData& data) const { } NFP::CommonInfo common_info{}; - Service::Mii::MiiManager manager; const u64 application_id = tag_data.application_id; GetCommonInfo(common_info); @@ -1249,6 +1258,28 @@ void NfcDevice::UpdateRegisterInfoCrc() { tag_data.register_info_crc = crc.checksum(); } +void NfcDevice::BuildAmiiboWithoutKeys() { + Service::Mii::MiiManager manager; + auto& settings = tag_data.settings; + + tag_data = NFP::AmiiboCrypto::NfcDataToEncodedData(encrypted_tag_data); + + // Common info + tag_data.write_counter = 0; + tag_data.amiibo_version = 0; + settings.write_date = GetAmiiboDate(GetCurrentPosixTime()); + + // Register info + SetAmiiboName(settings, {'y', 'u', 'z', 'u', 'A', 'm', 'i', 'i', 'b', 'o'}); + settings.settings.font_region.Assign(0); + settings.init_date = GetAmiiboDate(GetCurrentPosixTime()); + tag_data.owner_mii = manager.BuildFromStoreData(manager.BuildDefault(0)); + + // Admin info + settings.settings.amiibo_initialized.Assign(1); + settings.settings.appdata_initialized.Assign(0); +} + u64 NfcDevice::GetHandle() const { // Generate a handle based of the npad id return static_cast<u64>(npad_id); diff --git a/src/core/hle/service/nfc/common/device.h b/src/core/hle/service/nfc/common/device.h index 98e1945c1..6a37e8458 100644 --- a/src/core/hle/service/nfc/common/device.h +++ b/src/core/hle/service/nfc/common/device.h @@ -110,6 +110,8 @@ private: void UpdateSettingsCrc(); void UpdateRegisterInfoCrc(); + void BuildAmiiboWithoutKeys(); + bool is_controller_set{}; int callback_key; const Core::HID::NpadIdType npad_id; @@ -128,6 +130,7 @@ private: bool is_data_moddified{}; bool is_app_area_open{}; bool is_plain_amiibo{}; + bool is_write_protected{}; NFP::MountTarget mount_target{NFP::MountTarget::None}; NFP::NTAG215File tag_data{}; diff --git a/src/input_common/drivers/joycon.cpp b/src/input_common/drivers/joycon.cpp index 653862a72..b2b5677c8 100644 --- a/src/input_common/drivers/joycon.cpp +++ b/src/input_common/drivers/joycon.cpp @@ -291,9 +291,13 @@ Common::Input::NfcState Joycons::SupportsNfc(const PadIdentifier& identifier_) c return Common::Input::NfcState::Success; }; -Common::Input::NfcState Joycons::WriteNfcData(const PadIdentifier& identifier_, +Common::Input::NfcState Joycons::WriteNfcData(const PadIdentifier& identifier, const std::vector<u8>& data) { - return Common::Input::NfcState::NotSupported; + auto handle = GetHandle(identifier); + if (handle->WriteNfcData(data) != Joycon::DriverResult::Success) { + return Common::Input::NfcState::WriteFailed; + } + return Common::Input::NfcState::Success; }; Common::Input::DriverResult Joycons::SetPollingMode(const PadIdentifier& identifier, diff --git a/src/input_common/helpers/joycon_driver.cpp b/src/input_common/helpers/joycon_driver.cpp index 83429a336..95106f16d 100644 --- a/src/input_common/helpers/joycon_driver.cpp +++ b/src/input_common/helpers/joycon_driver.cpp @@ -492,6 +492,26 @@ DriverResult JoyconDriver::SetRingConMode() { return result; } +DriverResult JoyconDriver::WriteNfcData(std::span<const u8> data) { + std::scoped_lock lock{mutex}; + disable_input_thread = true; + + if (!supported_features.nfc) { + return DriverResult::NotSupported; + } + if (!nfc_protocol->IsEnabled()) { + return DriverResult::Disabled; + } + if (!amiibo_detected) { + return DriverResult::ErrorWritingData; + } + + const auto result = nfc_protocol->WriteAmiibo(data); + + disable_input_thread = false; + return result; +} + bool JoyconDriver::IsConnected() const { std::scoped_lock lock{mutex}; return is_connected.load(); diff --git a/src/input_common/helpers/joycon_driver.h b/src/input_common/helpers/joycon_driver.h index 72a9e71dc..e9b2fccbb 100644 --- a/src/input_common/helpers/joycon_driver.h +++ b/src/input_common/helpers/joycon_driver.h @@ -49,6 +49,7 @@ public: DriverResult SetIrMode(); DriverResult SetNfcMode(); DriverResult SetRingConMode(); + DriverResult WriteNfcData(std::span<const u8> data); void SetCallbacks(const JoyconCallbacks& callbacks); diff --git a/src/input_common/helpers/joycon_protocol/joycon_types.h b/src/input_common/helpers/joycon_protocol/joycon_types.h index 353dc744d..5007b0e18 100644 --- a/src/input_common/helpers/joycon_protocol/joycon_types.h +++ b/src/input_common/helpers/joycon_protocol/joycon_types.h @@ -23,6 +23,7 @@ constexpr std::array<u8, 8> DefaultVibrationBuffer{0x0, 0x1, 0x40, 0x40, 0x0, 0x using MacAddress = std::array<u8, 6>; using SerialNumber = std::array<u8, 15>; +using TagUUID = std::array<u8, 7>; enum class ControllerType : u8 { None = 0x00, @@ -276,12 +277,13 @@ enum class MCUPacketFlag : u8 { LastCommandPacket = 0x08, }; -enum class NFCReadCommand : u8 { +enum class NFCCommand : u8 { CancelAll = 0x00, StartPolling = 0x01, StopPolling = 0x02, StartWaitingRecieve = 0x04, - Ntag = 0x06, + ReadNtag = 0x06, + WriteNtag = 0x08, Mifare = 0x0F, }; @@ -292,14 +294,19 @@ enum class NFCTagType : u8 { enum class NFCPages { Block0 = 0, + Block3 = 3, Block45 = 45, Block135 = 135, Block231 = 231, }; enum class NFCStatus : u8 { + Ready = 0x00, + Polling = 0x01, LastPackage = 0x04, + WriteDone = 0x05, TagLost = 0x07, + WriteReady = 0x09, }; enum class IrsMode : u8 { @@ -559,13 +566,32 @@ static_assert(sizeof(NFCReadBlockCommand) == 0x9, "NFCReadBlockCommand is an inv struct NFCReadCommandData { u8 unknown; u8 uuid_length; - u8 unknown_2; - std::array<u8, 6> uid; + TagUUID uid; NFCTagType tag_type; NFCReadBlockCommand read_block; }; static_assert(sizeof(NFCReadCommandData) == 0x13, "NFCReadCommandData is an invalid size"); +#pragma pack(push, 1) +struct NFCWriteCommandData { + u8 unknown; + u8 uuid_length; + TagUUID uid; + NFCTagType tag_type; + u8 unknown2; + u8 unknown3; + u8 unknown4; + u8 unknown5; + u8 unknown6; + u8 unknown7; + u8 unknown8; + u8 magic; + u16_be write_count; + u8 amiibo_version; +}; +static_assert(sizeof(NFCWriteCommandData) == 0x15, "NFCWriteCommandData is an invalid size"); +#pragma pack(pop) + struct NFCPollingCommandData { u8 enable_mifare; u8 unknown_1; @@ -576,8 +602,8 @@ struct NFCPollingCommandData { static_assert(sizeof(NFCPollingCommandData) == 0x05, "NFCPollingCommandData is an invalid size"); struct NFCRequestState { - NFCReadCommand command_argument; - INSERT_PADDING_BYTES(0x1); + NFCCommand command_argument; + u8 block_id; u8 packet_id; MCUPacketFlag packet_flag; u8 data_length; @@ -591,6 +617,18 @@ struct NFCRequestState { }; static_assert(sizeof(NFCRequestState) == 0x26, "NFCRequestState is an invalid size"); +struct NFCDataChunk { + u8 nfc_page; + u8 data_size; + std::array<u8, 0xFF> data; +}; + +struct NFCWritePackage { + NFCWriteCommandData command_data; + u8 number_of_chunks; + std::array<NFCDataChunk, 4> data_chunks; +}; + struct IrsConfigure { MCUCommand command; MCUSubCommand sub_command; diff --git a/src/input_common/helpers/joycon_protocol/nfc.cpp b/src/input_common/helpers/joycon_protocol/nfc.cpp index 46c9e9489..3b7a628e5 100644 --- a/src/input_common/helpers/joycon_protocol/nfc.cpp +++ b/src/input_common/helpers/joycon_protocol/nfc.cpp @@ -34,6 +34,12 @@ DriverResult NfcProtocol::EnableNfc() { result = ConfigureMCU(config); } + if (result == DriverResult::Success) { + result = WaitSetMCUMode(ReportMode::NFC_IR_MODE_60HZ, MCUMode::NFC); + } + if (result == DriverResult::Success) { + result = WaitUntilNfcIs(NFCStatus::Ready); + } return result; } @@ -56,27 +62,20 @@ DriverResult NfcProtocol::StartNFCPollingMode() { LOG_DEBUG(Input, "Start NFC pooling Mode"); ScopedSetBlocking sb(this); DriverResult result{DriverResult::Success}; - TagFoundData tag_data{}; if (result == DriverResult::Success) { - result = WaitSetMCUMode(ReportMode::NFC_IR_MODE_60HZ, MCUMode::NFC); - } - if (result == DriverResult::Success) { - result = WaitUntilNfcIsReady(); - } - if (result == DriverResult::Success) { MCUCommandResponse output{}; result = SendStopPollingRequest(output); } if (result == DriverResult::Success) { - result = WaitUntilNfcIsReady(); + result = WaitUntilNfcIs(NFCStatus::Ready); } if (result == DriverResult::Success) { MCUCommandResponse output{}; result = SendStartPollingRequest(output); } if (result == DriverResult::Success) { - result = WaitUntilNfcIsPolling(); + result = WaitUntilNfcIs(NFCStatus::Polling); } if (result == DriverResult::Success) { is_enabled = true; @@ -112,6 +111,49 @@ DriverResult NfcProtocol::ScanAmiibo(std::vector<u8>& data) { return result; } +DriverResult NfcProtocol::WriteAmiibo(std::span<const u8> data) { + LOG_DEBUG(Input, "Write amiibo"); + ScopedSetBlocking sb(this); + DriverResult result{DriverResult::Success}; + TagUUID tag_uuid = GetTagUUID(data); + TagFoundData tag_data{}; + + if (result == DriverResult::Success) { + result = IsTagInRange(tag_data, 7); + } + if (result == DriverResult::Success) { + if (tag_data.uuid != tag_uuid) { + result = DriverResult::InvalidParameters; + } + } + if (result == DriverResult::Success) { + MCUCommandResponse output{}; + result = SendStopPollingRequest(output); + } + if (result == DriverResult::Success) { + result = WaitUntilNfcIs(NFCStatus::Ready); + } + if (result == DriverResult::Success) { + MCUCommandResponse output{}; + result = SendStartPollingRequest(output, true); + } + if (result == DriverResult::Success) { + result = WaitUntilNfcIs(NFCStatus::WriteReady); + } + if (result == DriverResult::Success) { + result = WriteAmiiboData(tag_uuid, data); + } + if (result == DriverResult::Success) { + result = WaitUntilNfcIs(NFCStatus::WriteDone); + } + if (result == DriverResult::Success) { + MCUCommandResponse output{}; + result = SendStopPollingRequest(output); + } + + return result; +} + bool NfcProtocol::HasAmiibo() { if (update_counter++ < AMIIBO_UPDATE_DELAY) { return true; @@ -129,7 +171,7 @@ bool NfcProtocol::HasAmiibo() { return result == DriverResult::Success; } -DriverResult NfcProtocol::WaitUntilNfcIsReady() { +DriverResult NfcProtocol::WaitUntilNfcIs(NFCStatus status) { constexpr std::size_t timeout_limit = 10; MCUCommandResponse output{}; std::size_t tries = 0; @@ -145,28 +187,7 @@ DriverResult NfcProtocol::WaitUntilNfcIsReady() { } } while (output.mcu_report != MCUReport::NFCState || (output.mcu_data[1] << 8) + output.mcu_data[0] != 0x0500 || - output.mcu_data[5] != 0x31 || output.mcu_data[6] != 0x00); - - return DriverResult::Success; -} - -DriverResult NfcProtocol::WaitUntilNfcIsPolling() { - constexpr std::size_t timeout_limit = 10; - MCUCommandResponse output{}; - std::size_t tries = 0; - - do { - auto result = SendNextPackageRequest(output, {}); - - if (result != DriverResult::Success) { - return result; - } - if (tries++ > timeout_limit) { - return DriverResult::Timeout; - } - } while (output.mcu_report != MCUReport::NFCState || - (output.mcu_data[1] << 8) + output.mcu_data[0] != 0x0500 || - output.mcu_data[5] != 0x31 || output.mcu_data[6] != 0x01); + output.mcu_data[5] != 0x31 || output.mcu_data[6] != static_cast<u8>(status)); return DriverResult::Success; } @@ -188,7 +209,7 @@ DriverResult NfcProtocol::IsTagInRange(TagFoundData& data, std::size_t timeout_l (output.mcu_data[6] != 0x09 && output.mcu_data[6] != 0x04)); data.type = output.mcu_data[12]; - data.uuid.resize(output.mcu_data[14]); + data.uuid_size = std::min(output.mcu_data[14], static_cast<u8>(sizeof(TagUUID))); memcpy(data.uuid.data(), output.mcu_data.data() + 15, data.uuid.size()); return DriverResult::Success; @@ -245,17 +266,94 @@ DriverResult NfcProtocol::GetAmiiboData(std::vector<u8>& ntag_data) { return DriverResult::Timeout; } -DriverResult NfcProtocol::SendStartPollingRequest(MCUCommandResponse& output) { +DriverResult NfcProtocol::WriteAmiiboData(const TagUUID& tag_uuid, std::span<const u8> data) { + constexpr std::size_t timeout_limit = 60; + const auto nfc_data = MakeAmiiboWritePackage(tag_uuid, data); + const std::vector<u8> nfc_buffer_data = SerializeWritePackage(nfc_data); + std::span<const u8> buffer(nfc_buffer_data); + MCUCommandResponse output{}; + u8 block_id = 1; + u8 package_index = 0; + std::size_t tries = 0; + std::size_t current_position = 0; + + LOG_INFO(Input, "Writing amiibo data"); + + auto result = SendWriteAmiiboRequest(output, tag_uuid); + + if (result != DriverResult::Success) { + return result; + } + + // Read Tag data but ignore the actual sent data + while (tries++ < timeout_limit) { + result = SendNextPackageRequest(output, package_index); + const auto nfc_status = static_cast<NFCStatus>(output.mcu_data[6]); + + if (result != DriverResult::Success) { + return result; + } + + if ((output.mcu_report == MCUReport::NFCReadData || + output.mcu_report == MCUReport::NFCState) && + nfc_status == NFCStatus::TagLost) { + return DriverResult::ErrorReadingData; + } + + if (output.mcu_report == MCUReport::NFCReadData && output.mcu_data[1] == 0x07) { + package_index++; + continue; + } + + if (output.mcu_report == MCUReport::NFCState && nfc_status == NFCStatus::LastPackage) { + LOG_INFO(Input, "Finished reading amiibo"); + break; + } + } + + // Send Data. Nfc buffer size is 31, Send the data in smaller packages + while (current_position < buffer.size() && tries++ < timeout_limit) { + const std::size_t next_position = + std::min(current_position + sizeof(NFCRequestState::raw_data), buffer.size()); + const std::size_t block_size = next_position - current_position; + const bool is_last_packet = block_size < sizeof(NFCRequestState::raw_data); + + SendWriteDataAmiiboRequest(output, block_id, is_last_packet, + buffer.subspan(current_position, block_size)); + + const auto nfc_status = static_cast<NFCStatus>(output.mcu_data[6]); + + if ((output.mcu_report == MCUReport::NFCReadData || + output.mcu_report == MCUReport::NFCState) && + nfc_status == NFCStatus::TagLost) { + return DriverResult::ErrorReadingData; + } + + // Increase position when data is confirmed by the joycon + if (output.mcu_report == MCUReport::NFCState && + (output.mcu_data[1] << 8) + output.mcu_data[0] == 0x0500 && + output.mcu_data[3] == block_id) { + block_id++; + current_position = next_position; + } + } + + return result; +} + +DriverResult NfcProtocol::SendStartPollingRequest(MCUCommandResponse& output, + bool is_second_attempt) { NFCRequestState request{ - .command_argument = NFCReadCommand::StartPolling, - .packet_id = 0x0, + .command_argument = NFCCommand::StartPolling, + .block_id = {}, + .packet_id = {}, .packet_flag = MCUPacketFlag::LastCommandPacket, .data_length = sizeof(NFCPollingCommandData), .nfc_polling = { - .enable_mifare = 0x01, - .unknown_1 = 0x00, - .unknown_2 = 0x00, + .enable_mifare = 0x00, + .unknown_1 = static_cast<u8>(is_second_attempt ? 0xe8 : 0x00), + .unknown_2 = static_cast<u8>(is_second_attempt ? 0x03 : 0x00), .unknown_3 = 0x2c, .unknown_4 = 0x01, }, @@ -271,10 +369,11 @@ DriverResult NfcProtocol::SendStartPollingRequest(MCUCommandResponse& output) { DriverResult NfcProtocol::SendStopPollingRequest(MCUCommandResponse& output) { NFCRequestState request{ - .command_argument = NFCReadCommand::StopPolling, - .packet_id = 0x0, + .command_argument = NFCCommand::StopPolling, + .block_id = {}, + .packet_id = {}, .packet_flag = MCUPacketFlag::LastCommandPacket, - .data_length = 0, + .data_length = {}, .raw_data = {}, .crc = {}, }; @@ -288,10 +387,11 @@ DriverResult NfcProtocol::SendStopPollingRequest(MCUCommandResponse& output) { DriverResult NfcProtocol::SendNextPackageRequest(MCUCommandResponse& output, u8 packet_id) { NFCRequestState request{ - .command_argument = NFCReadCommand::StartWaitingRecieve, + .command_argument = NFCCommand::StartWaitingRecieve, + .block_id = {}, .packet_id = packet_id, .packet_flag = MCUPacketFlag::LastCommandPacket, - .data_length = 0, + .data_length = {}, .raw_data = {}, .crc = {}, }; @@ -305,17 +405,17 @@ DriverResult NfcProtocol::SendNextPackageRequest(MCUCommandResponse& output, u8 DriverResult NfcProtocol::SendReadAmiiboRequest(MCUCommandResponse& output, NFCPages ntag_pages) { NFCRequestState request{ - .command_argument = NFCReadCommand::Ntag, - .packet_id = 0x0, + .command_argument = NFCCommand::ReadNtag, + .block_id = {}, + .packet_id = {}, .packet_flag = MCUPacketFlag::LastCommandPacket, .data_length = sizeof(NFCReadCommandData), .nfc_read = { .unknown = 0xd0, - .uuid_length = 0x07, - .unknown_2 = 0x00, + .uuid_length = sizeof(NFCReadCommandData::uid), .uid = {}, - .tag_type = NFCTagType::AllTags, + .tag_type = NFCTagType::Ntag215, .read_block = GetReadBlockCommand(ntag_pages), }, .crc = {}, @@ -328,12 +428,135 @@ DriverResult NfcProtocol::SendReadAmiiboRequest(MCUCommandResponse& output, NFCP output); } +DriverResult NfcProtocol::SendWriteAmiiboRequest(MCUCommandResponse& output, + const TagUUID& tag_uuid) { + NFCRequestState request{ + .command_argument = NFCCommand::ReadNtag, + .block_id = {}, + .packet_id = {}, + .packet_flag = MCUPacketFlag::LastCommandPacket, + .data_length = sizeof(NFCReadCommandData), + .nfc_read = + { + .unknown = 0xd0, + .uuid_length = sizeof(NFCReadCommandData::uid), + .uid = tag_uuid, + .tag_type = NFCTagType::Ntag215, + .read_block = GetReadBlockCommand(NFCPages::Block3), + }, + .crc = {}, + }; + + std::array<u8, sizeof(NFCRequestState)> request_data{}; + memcpy(request_data.data(), &request, sizeof(NFCRequestState)); + request_data[36] = CalculateMCU_CRC8(request_data.data(), 36); + return SendMCUData(ReportMode::NFC_IR_MODE_60HZ, MCUSubCommand::ReadDeviceMode, request_data, + output); +} + +DriverResult NfcProtocol::SendWriteDataAmiiboRequest(MCUCommandResponse& output, u8 block_id, + bool is_last_packet, + std::span<const u8> data) { + const auto data_size = std::min(data.size(), sizeof(NFCRequestState::raw_data)); + NFCRequestState request{ + .command_argument = NFCCommand::WriteNtag, + .block_id = block_id, + .packet_id = {}, + .packet_flag = + is_last_packet ? MCUPacketFlag::LastCommandPacket : MCUPacketFlag::MorePacketsRemaining, + .data_length = static_cast<u8>(data_size), + .raw_data = {}, + .crc = {}, + }; + memcpy(request.raw_data.data(), data.data(), data_size); + + std::array<u8, sizeof(NFCRequestState)> request_data{}; + memcpy(request_data.data(), &request, sizeof(NFCRequestState)); + request_data[36] = CalculateMCU_CRC8(request_data.data(), 36); + return SendMCUData(ReportMode::NFC_IR_MODE_60HZ, MCUSubCommand::ReadDeviceMode, request_data, + output); +} + +std::vector<u8> NfcProtocol::SerializeWritePackage(const NFCWritePackage& package) const { + const std::size_t header_size = + sizeof(NFCWriteCommandData) + sizeof(NFCWritePackage::number_of_chunks); + std::vector<u8> serialized_data(header_size); + std::size_t start_index = 0; + + memcpy(serialized_data.data(), &package, header_size); + start_index += header_size; + + for (const auto& data_chunk : package.data_chunks) { + const std::size_t chunk_size = + sizeof(NFCDataChunk::nfc_page) + sizeof(NFCDataChunk::data_size) + data_chunk.data_size; + + serialized_data.resize(start_index + chunk_size); + memcpy(serialized_data.data() + start_index, &data_chunk, chunk_size); + start_index += chunk_size; + } + + return serialized_data; +} + +NFCWritePackage NfcProtocol::MakeAmiiboWritePackage(const TagUUID& tag_uuid, + std::span<const u8> data) const { + return { + .command_data{ + .unknown = 0xd0, + .uuid_length = sizeof(NFCReadCommandData::uid), + .uid = tag_uuid, + .tag_type = NFCTagType::Ntag215, + .unknown2 = 0x00, + .unknown3 = 0x01, + .unknown4 = 0x04, + .unknown5 = 0xff, + .unknown6 = 0xff, + .unknown7 = 0xff, + .unknown8 = 0xff, + .magic = data[16], + .write_count = static_cast<u16>((data[17] << 8) + data[18]), + .amiibo_version = data[19], + }, + .number_of_chunks = 3, + .data_chunks = + { + MakeAmiiboChunk(0x05, 0x20, data), + MakeAmiiboChunk(0x20, 0xf0, data), + MakeAmiiboChunk(0x5c, 0x98, data), + }, + }; +} + +NFCDataChunk NfcProtocol::MakeAmiiboChunk(u8 page, u8 size, std::span<const u8> data) const { + constexpr u8 PAGE_SIZE = 4; + + if (static_cast<std::size_t>(page * PAGE_SIZE) + size >= data.size()) { + return {}; + } + + NFCDataChunk chunk{ + .nfc_page = page, + .data_size = size, + .data = {}, + }; + std::memcpy(chunk.data.data(), data.data() + (page * PAGE_SIZE), size); + return chunk; +} + NFCReadBlockCommand NfcProtocol::GetReadBlockCommand(NFCPages pages) const { switch (pages) { case NFCPages::Block0: return { .block_count = 1, }; + case NFCPages::Block3: + return { + .block_count = 1, + .blocks = + { + NFCReadBlock{0x03, 0x03}, + }, + }; case NFCPages::Block45: return { .block_count = 1, @@ -368,6 +591,17 @@ NFCReadBlockCommand NfcProtocol::GetReadBlockCommand(NFCPages pages) const { }; } +TagUUID NfcProtocol::GetTagUUID(std::span<const u8> data) const { + if (data.size() < 10) { + return {}; + } + + // crc byte 3 is omitted in this operation + return { + data[0], data[1], data[2], data[4], data[5], data[6], data[7], + }; +} + bool NfcProtocol::IsEnabled() const { return is_enabled; } diff --git a/src/input_common/helpers/joycon_protocol/nfc.h b/src/input_common/helpers/joycon_protocol/nfc.h index c9e9af03f..eb58c427d 100644 --- a/src/input_common/helpers/joycon_protocol/nfc.h +++ b/src/input_common/helpers/joycon_protocol/nfc.h @@ -27,6 +27,8 @@ public: DriverResult ScanAmiibo(std::vector<u8>& data); + DriverResult WriteAmiibo(std::span<const u8> data); + bool HasAmiibo(); bool IsEnabled() const; @@ -37,18 +39,20 @@ private: struct TagFoundData { u8 type; - std::vector<u8> uuid; + u8 uuid_size; + TagUUID uuid; }; - DriverResult WaitUntilNfcIsReady(); - - DriverResult WaitUntilNfcIsPolling(); + DriverResult WaitUntilNfcIs(NFCStatus status); DriverResult IsTagInRange(TagFoundData& data, std::size_t timeout_limit = 1); DriverResult GetAmiiboData(std::vector<u8>& data); - DriverResult SendStartPollingRequest(MCUCommandResponse& output); + DriverResult WriteAmiiboData(const TagUUID& tag_uuid, std::span<const u8> data); + + DriverResult SendStartPollingRequest(MCUCommandResponse& output, + bool is_second_attempt = false); DriverResult SendStopPollingRequest(MCUCommandResponse& output); @@ -56,8 +60,21 @@ private: DriverResult SendReadAmiiboRequest(MCUCommandResponse& output, NFCPages ntag_pages); + DriverResult SendWriteAmiiboRequest(MCUCommandResponse& output, const TagUUID& tag_uuid); + + DriverResult SendWriteDataAmiiboRequest(MCUCommandResponse& output, u8 block_id, + bool is_last_packet, std::span<const u8> data); + + std::vector<u8> SerializeWritePackage(const NFCWritePackage& package) const; + + NFCWritePackage MakeAmiiboWritePackage(const TagUUID& tag_uuid, std::span<const u8> data) const; + + NFCDataChunk MakeAmiiboChunk(u8 page, u8 size, std::span<const u8> data) const; + NFCReadBlockCommand GetReadBlockCommand(NFCPages pages) const; + TagUUID GetTagUUID(std::span<const u8> data) const; + bool is_enabled{}; std::size_t update_counter{}; }; diff --git a/src/shader_recompiler/frontend/maxwell/translate/impl/integer_funnel_shift.cpp b/src/shader_recompiler/frontend/maxwell/translate/impl/integer_funnel_shift.cpp index 442365a26..c2a0ee6f1 100644 --- a/src/shader_recompiler/frontend/maxwell/translate/impl/integer_funnel_shift.cpp +++ b/src/shader_recompiler/frontend/maxwell/translate/impl/integer_funnel_shift.cpp @@ -30,7 +30,7 @@ void SHF(TranslatorVisitor& v, u64 insn, const IR::U32& shift, const IR::U32& hi union { u64 insn; BitField<0, 8, IR::Reg> dest_reg; - BitField<0, 8, IR::Reg> lo_bits_reg; + BitField<8, 8, IR::Reg> lo_bits_reg; BitField<37, 2, MaxShift> max_shift; BitField<47, 1, u64> cc; BitField<48, 2, u64> x_mode; diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt index a0009a36f..308d013d6 100644 --- a/src/video_core/CMakeLists.txt +++ b/src/video_core/CMakeLists.txt @@ -246,10 +246,14 @@ add_library(video_core STATIC texture_cache/util.h textures/astc.h textures/astc.cpp + textures/bcn.cpp + textures/bcn.h textures/decoders.cpp textures/decoders.h textures/texture.cpp textures/texture.h + textures/workers.cpp + textures/workers.h transform_feedback.cpp transform_feedback.h video_core.cpp @@ -275,7 +279,7 @@ add_library(video_core STATIC create_target_directory_groups(video_core) target_link_libraries(video_core PUBLIC common core) -target_link_libraries(video_core PUBLIC glad shader_recompiler) +target_link_libraries(video_core PUBLIC glad shader_recompiler stb) if (YUZU_USE_BUNDLED_FFMPEG AND NOT WIN32) add_dependencies(video_core ffmpeg-build) diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h index 98756e4da..65494097b 100644 --- a/src/video_core/buffer_cache/buffer_cache.h +++ b/src/video_core/buffer_cache/buffer_cache.h @@ -30,8 +30,8 @@ BufferCache<P>::BufferCache(VideoCore::RasterizerInterface& rasterizer_, } const s64 device_memory = static_cast<s64>(runtime.GetDeviceLocalMemory()); - const s64 min_spacing_expected = device_memory - 1_GiB - 512_MiB; - const s64 min_spacing_critical = device_memory - 1_GiB; + const s64 min_spacing_expected = device_memory - 1_GiB; + const s64 min_spacing_critical = device_memory - 512_MiB; const s64 mem_threshold = std::min(device_memory, TARGET_THRESHOLD); const s64 min_vacancy_expected = (6 * mem_threshold) / 10; const s64 min_vacancy_critical = (3 * mem_threshold) / 10; @@ -1664,7 +1664,7 @@ typename BufferCache<P>::Binding BufferCache<P>::StorageBufferBinding(GPUVAddr s // cbufs, which do not store the sizes adjacent to the addresses, so use the fully // mapped buffer size for now. const u32 memory_layout_size = static_cast<u32>(gpu_memory->GetMemoryLayoutSize(gpu_addr)); - return memory_layout_size; + return std::min(memory_layout_size, static_cast<u32>(8_MiB)); }(); const std::optional<VAddr> cpu_addr = gpu_memory->GpuToCpuAddress(gpu_addr); if (!cpu_addr || size == 0) { diff --git a/src/video_core/renderer_opengl/gl_texture_cache.cpp b/src/video_core/renderer_opengl/gl_texture_cache.cpp index 31118886f..1e0823836 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.cpp +++ b/src/video_core/renderer_opengl/gl_texture_cache.cpp @@ -233,6 +233,8 @@ void ApplySwizzle(GLuint handle, PixelFormat format, std::array<SwizzleSource, 4 const VideoCommon::ImageInfo& info) { if (IsPixelFormatASTC(info.format) && info.size.depth == 1 && !runtime.HasNativeASTC()) { return Settings::values.accelerate_astc.GetValue() && + Settings::values.astc_recompression.GetValue() == + Settings::AstcRecompression::Uncompressed && !Settings::values.async_astc.GetValue(); } // Disable other accelerated uploads for now as they don't implement swizzled uploads @@ -437,6 +439,19 @@ OGLTexture MakeImage(const VideoCommon::ImageInfo& info, GLenum gl_internal_form return GL_R32UI; } +[[nodiscard]] GLenum SelectAstcFormat(PixelFormat format, bool is_srgb) { + switch (Settings::values.astc_recompression.GetValue()) { + case Settings::AstcRecompression::Bc1: + return is_srgb ? GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT1_EXT : GL_COMPRESSED_RGBA_S3TC_DXT1_EXT; + break; + case Settings::AstcRecompression::Bc3: + return is_srgb ? GL_COMPRESSED_SRGB_ALPHA_S3TC_DXT5_EXT : GL_COMPRESSED_RGBA_S3TC_DXT5_EXT; + break; + default: + return is_srgb ? GL_SRGB8_ALPHA8 : GL_RGBA8; + } +} + } // Anonymous namespace ImageBufferMap::~ImageBufferMap() { @@ -739,9 +754,16 @@ Image::Image(TextureCacheRuntime& runtime_, const VideoCommon::ImageInfo& info_, if (IsConverted(runtime->device, info.format, info.type)) { flags |= ImageFlagBits::Converted; flags |= ImageFlagBits::CostlyLoad; - gl_internal_format = IsPixelFormatSRGB(info.format) ? GL_SRGB8_ALPHA8 : GL_RGBA8; + + const bool is_srgb = IsPixelFormatSRGB(info.format); + gl_internal_format = is_srgb ? GL_SRGB8_ALPHA8 : GL_RGBA8; gl_format = GL_RGBA; gl_type = GL_UNSIGNED_INT_8_8_8_8_REV; + + if (IsPixelFormatASTC(info.format)) { + gl_internal_format = SelectAstcFormat(info.format, is_srgb); + gl_format = GL_NONE; + } } else { const auto& tuple = MaxwellToGL::GetFormatTuple(info.format); gl_internal_format = tuple.internal_format; @@ -1130,7 +1152,12 @@ ImageView::ImageView(TextureCacheRuntime& runtime, const VideoCommon::ImageViewI views{runtime.null_image_views} { const Device& device = runtime.device; if (True(image.flags & ImageFlagBits::Converted)) { - internal_format = IsPixelFormatSRGB(info.format) ? GL_SRGB8_ALPHA8 : GL_RGBA8; + const bool is_srgb = IsPixelFormatSRGB(info.format); + internal_format = is_srgb ? GL_SRGB8_ALPHA8 : GL_RGBA8; + + if (IsPixelFormatASTC(info.format)) { + internal_format = SelectAstcFormat(info.format, is_srgb); + } } else { internal_format = MaxwellToGL::GetFormatTuple(format).internal_format; } diff --git a/src/video_core/renderer_opengl/gl_texture_cache.h b/src/video_core/renderer_opengl/gl_texture_cache.h index 1190999a8..3e9b3302b 100644 --- a/src/video_core/renderer_opengl/gl_texture_cache.h +++ b/src/video_core/renderer_opengl/gl_texture_cache.h @@ -144,6 +144,10 @@ public: return state_tracker; } + void BarrierFeedbackLoop() const noexcept { + // OpenGL does not require a barrier for attachment feedback loops. + } + private: struct StagingBuffers { explicit StagingBuffers(GLenum storage_flags_, GLenum map_flags_); diff --git a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp index 8853cf0f7..b75d7220d 100644 --- a/src/video_core/renderer_vulkan/maxwell_to_vk.cpp +++ b/src/video_core/renderer_vulkan/maxwell_to_vk.cpp @@ -6,6 +6,7 @@ #include "common/assert.h" #include "common/common_types.h" #include "common/logging/log.h" +#include "common/settings.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/renderer_vulkan/maxwell_to_vk.h" #include "video_core/surface.h" @@ -237,14 +238,25 @@ FormatInfo SurfaceFormat(const Device& device, FormatType format_type, bool with PixelFormat pixel_format) { ASSERT(static_cast<size_t>(pixel_format) < std::size(tex_format_tuples)); FormatTuple tuple = tex_format_tuples[static_cast<size_t>(pixel_format)]; - // Use A8B8G8R8_UNORM on hardware that doesn't support ASTC natively + // Transcode on hardware that doesn't support ASTC natively if (!device.IsOptimalAstcSupported() && VideoCore::Surface::IsPixelFormatASTC(pixel_format)) { const bool is_srgb = with_srgb && VideoCore::Surface::IsPixelFormatSRGB(pixel_format); - if (is_srgb) { - tuple.format = VK_FORMAT_A8B8G8R8_SRGB_PACK32; - } else { - tuple.format = VK_FORMAT_A8B8G8R8_UNORM_PACK32; - tuple.usage |= Storage; + + switch (Settings::values.astc_recompression.GetValue()) { + case Settings::AstcRecompression::Uncompressed: + if (is_srgb) { + tuple.format = VK_FORMAT_A8B8G8R8_SRGB_PACK32; + } else { + tuple.format = VK_FORMAT_A8B8G8R8_UNORM_PACK32; + tuple.usage |= Storage; + } + break; + case Settings::AstcRecompression::Bc1: + tuple.format = is_srgb ? VK_FORMAT_BC1_RGBA_SRGB_BLOCK : VK_FORMAT_BC1_RGBA_UNORM_BLOCK; + break; + case Settings::AstcRecompression::Bc3: + tuple.format = is_srgb ? VK_FORMAT_BC3_SRGB_BLOCK : VK_FORMAT_BC3_UNORM_BLOCK; + break; } } const bool attachable = (tuple.usage & Attachable) != 0; diff --git a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp index f1bcd5cd6..506b78f08 100644 --- a/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp +++ b/src/video_core/renderer_vulkan/vk_graphics_pipeline.cpp @@ -481,12 +481,13 @@ void GraphicsPipeline::ConfigureImpl(bool is_indexed) { if constexpr (Spec::enabled_stages[4]) { prepare_stage(4); } + texture_cache.UpdateRenderTargets(false); + texture_cache.CheckFeedbackLoop(views); ConfigureDraw(rescaling, render_area); } void GraphicsPipeline::ConfigureDraw(const RescalingPushConstant& rescaling, const RenderAreaPushConstant& render_area) { - texture_cache.UpdateRenderTargets(false); scheduler.RequestRenderpass(texture_cache.GetFramebuffer()); if (!is_built.load(std::memory_order::relaxed)) { diff --git a/src/video_core/renderer_vulkan/vk_master_semaphore.cpp b/src/video_core/renderer_vulkan/vk_master_semaphore.cpp index 47c74e4d8..8b65aeaeb 100644 --- a/src/video_core/renderer_vulkan/vk_master_semaphore.cpp +++ b/src/video_core/renderer_vulkan/vk_master_semaphore.cpp @@ -10,11 +10,16 @@ namespace Vulkan { +constexpr u64 FENCE_RESERVE_SIZE = 8; + MasterSemaphore::MasterSemaphore(const Device& device_) : device(device_) { if (!device.HasTimelineSemaphore()) { static constexpr VkFenceCreateInfo fence_ci{ .sType = VK_STRUCTURE_TYPE_FENCE_CREATE_INFO, .pNext = nullptr, .flags = 0}; - fence = device.GetLogical().CreateFence(fence_ci); + free_queue.resize(FENCE_RESERVE_SIZE); + std::ranges::generate(free_queue, + [&] { return device.GetLogical().CreateFence(fence_ci); }); + wait_thread = std::jthread([this](std::stop_token token) { WaitThread(token); }); return; } @@ -167,16 +172,53 @@ VkResult MasterSemaphore::SubmitQueueFence(vk::CommandBuffer& cmdbuf, VkSemaphor .pSignalSemaphores = &signal_semaphore, }; + auto fence = GetFreeFence(); auto result = device.GetGraphicsQueue().Submit(submit_info, *fence); if (result == VK_SUCCESS) { + std::scoped_lock lock{wait_mutex}; + wait_queue.emplace(host_tick, std::move(fence)); + wait_cv.notify_one(); + } + + return result; +} + +void MasterSemaphore::WaitThread(std::stop_token token) { + while (!token.stop_requested()) { + u64 host_tick; + vk::Fence fence; + { + std::unique_lock lock{wait_mutex}; + Common::CondvarWait(wait_cv, lock, token, [this] { return !wait_queue.empty(); }); + if (token.stop_requested()) { + return; + } + std::tie(host_tick, fence) = std::move(wait_queue.front()); + wait_queue.pop(); + } + fence.Wait(); fence.Reset(); gpu_tick.store(host_tick); gpu_tick.notify_all(); + + std::scoped_lock lock{free_mutex}; + free_queue.push_front(std::move(fence)); } +} - return result; +vk::Fence MasterSemaphore::GetFreeFence() { + std::scoped_lock lock{free_mutex}; + if (free_queue.empty()) { + static constexpr VkFenceCreateInfo fence_ci{ + .sType = VK_STRUCTURE_TYPE_FENCE_CREATE_INFO, .pNext = nullptr, .flags = 0}; + return device.GetLogical().CreateFence(fence_ci); + } + + auto fence = std::move(free_queue.back()); + free_queue.pop_back(); + return fence; } } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/vk_master_semaphore.h b/src/video_core/renderer_vulkan/vk_master_semaphore.h index f2f61f781..1e7c90215 100644 --- a/src/video_core/renderer_vulkan/vk_master_semaphore.h +++ b/src/video_core/renderer_vulkan/vk_master_semaphore.h @@ -5,8 +5,10 @@ #include <atomic> #include <condition_variable> +#include <deque> #include <mutex> #include <thread> +#include <queue> #include "common/common_types.h" #include "common/polyfill_thread.h" @@ -17,6 +19,8 @@ namespace Vulkan { class Device; class MasterSemaphore { + using Waitable = std::pair<u64, vk::Fence>; + public: explicit MasterSemaphore(const Device& device); ~MasterSemaphore(); @@ -57,13 +61,22 @@ private: VkResult SubmitQueueFence(vk::CommandBuffer& cmdbuf, VkSemaphore signal_semaphore, VkSemaphore wait_semaphore, u64 host_tick); + void WaitThread(std::stop_token token); + + vk::Fence GetFreeFence(); + private: const Device& device; ///< Device. - vk::Fence fence; ///< Fence. vk::Semaphore semaphore; ///< Timeline semaphore. std::atomic<u64> gpu_tick{0}; ///< Current known GPU tick. std::atomic<u64> current_tick{1}; ///< Current logical tick. + std::mutex wait_mutex; + std::mutex free_mutex; + std::condition_variable_any wait_cv; + std::queue<Waitable> wait_queue; ///< Queue for the fences to be waited on by the wait thread. + std::deque<vk::Fence> free_queue; ///< Holds available fences for submission. std::jthread debug_thread; ///< Debug thread to workaround validation layer bugs. + std::jthread wait_thread; ///< Helper thread that waits for submitted fences. }; } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.cpp b/src/video_core/renderer_vulkan/vk_texture_cache.cpp index 4d0481f2a..8711e2a87 100644 --- a/src/video_core/renderer_vulkan/vk_texture_cache.cpp +++ b/src/video_core/renderer_vulkan/vk_texture_cache.cpp @@ -861,6 +861,10 @@ VkBuffer TextureCacheRuntime::GetTemporaryBuffer(size_t needed_size) { return *buffers[level]; } +void TextureCacheRuntime::BarrierFeedbackLoop() { + scheduler.RequestOutsideRenderPassOperationContext(); +} + void TextureCacheRuntime::ReinterpretImage(Image& dst, Image& src, std::span<const VideoCommon::ImageCopy> copies) { std::vector<VkBufferImageCopy> vk_in_copies(copies.size()); @@ -1268,7 +1272,9 @@ Image::Image(TextureCacheRuntime& runtime_, const ImageInfo& info_, GPUVAddr gpu if (IsPixelFormatASTC(info.format) && !runtime->device.IsOptimalAstcSupported()) { if (Settings::values.async_astc.GetValue()) { flags |= VideoCommon::ImageFlagBits::AsynchronousDecode; - } else if (Settings::values.accelerate_astc.GetValue() && info.size.depth == 1) { + } else if (Settings::values.astc_recompression.GetValue() == + Settings::AstcRecompression::Uncompressed && + Settings::values.accelerate_astc.GetValue() && info.size.depth == 1) { flags |= VideoCommon::ImageFlagBits::AcceleratedUpload; } flags |= VideoCommon::ImageFlagBits::Converted; @@ -1283,7 +1289,9 @@ Image::Image(TextureCacheRuntime& runtime_, const ImageInfo& info_, GPUVAddr gpu .usage = VK_IMAGE_USAGE_STORAGE_BIT, }; current_image = *original_image; - if (IsPixelFormatASTC(info.format) && !runtime->device.IsOptimalAstcSupported()) { + if (IsPixelFormatASTC(info.format) && !runtime->device.IsOptimalAstcSupported() && + Settings::values.astc_recompression.GetValue() == + Settings::AstcRecompression::Uncompressed) { const auto& device = runtime->device.GetLogical(); storage_image_views.reserve(info.resources.levels); for (s32 level = 0; level < info.resources.levels; ++level) { diff --git a/src/video_core/renderer_vulkan/vk_texture_cache.h b/src/video_core/renderer_vulkan/vk_texture_cache.h index 4166b3d20..0f7a5ffd4 100644 --- a/src/video_core/renderer_vulkan/vk_texture_cache.h +++ b/src/video_core/renderer_vulkan/vk_texture_cache.h @@ -103,6 +103,8 @@ public: [[nodiscard]] VkBuffer GetTemporaryBuffer(size_t needed_size); + void BarrierFeedbackLoop(); + const Device& device; Scheduler& scheduler; MemoryAllocator& memory_allocator; diff --git a/src/video_core/texture_cache/image_base.cpp b/src/video_core/texture_cache/image_base.cpp index 91512022f..d79594ce5 100644 --- a/src/video_core/texture_cache/image_base.cpp +++ b/src/video_core/texture_cache/image_base.cpp @@ -155,7 +155,7 @@ void ImageBase::CheckAliasState() { flags &= ~ImageFlagBits::Alias; } -void AddImageAlias(ImageBase& lhs, ImageBase& rhs, ImageId lhs_id, ImageId rhs_id) { +bool AddImageAlias(ImageBase& lhs, ImageBase& rhs, ImageId lhs_id, ImageId rhs_id) { static constexpr auto OPTIONS = RelaxedOptions::Size | RelaxedOptions::Format; ASSERT(lhs.info.type == rhs.info.type); std::optional<SubresourceBase> base; @@ -169,7 +169,7 @@ void AddImageAlias(ImageBase& lhs, ImageBase& rhs, ImageId lhs_id, ImageId rhs_i } if (!base) { LOG_ERROR(HW_GPU, "Image alias should have been flipped"); - return; + return false; } const PixelFormat lhs_format = lhs.info.format; const PixelFormat rhs_format = rhs.info.format; @@ -248,12 +248,13 @@ void AddImageAlias(ImageBase& lhs, ImageBase& rhs, ImageId lhs_id, ImageId rhs_i } ASSERT(lhs_alias.copies.empty() == rhs_alias.copies.empty()); if (lhs_alias.copies.empty()) { - return; + return false; } lhs.aliased_images.push_back(std::move(lhs_alias)); rhs.aliased_images.push_back(std::move(rhs_alias)); lhs.flags &= ~ImageFlagBits::IsRescalable; rhs.flags &= ~ImageFlagBits::IsRescalable; + return true; } } // namespace VideoCommon diff --git a/src/video_core/texture_cache/image_base.h b/src/video_core/texture_cache/image_base.h index 329396bb6..1b8a17ee8 100644 --- a/src/video_core/texture_cache/image_base.h +++ b/src/video_core/texture_cache/image_base.h @@ -142,6 +142,6 @@ struct ImageAllocBase { std::vector<ImageId> images; }; -void AddImageAlias(ImageBase& lhs, ImageBase& rhs, ImageId lhs_id, ImageId rhs_id); +bool AddImageAlias(ImageBase& lhs, ImageBase& rhs, ImageId lhs_id, ImageId rhs_id); } // namespace VideoCommon diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h index b24086fce..2cf082c5d 100644 --- a/src/video_core/texture_cache/texture_cache.h +++ b/src/video_core/texture_cache/texture_cache.h @@ -49,8 +49,8 @@ TextureCache<P>::TextureCache(Runtime& runtime_, VideoCore::RasterizerInterface& if constexpr (HAS_DEVICE_MEMORY_INFO) { const s64 device_memory = static_cast<s64>(runtime.GetDeviceLocalMemory()); - const s64 min_spacing_expected = device_memory - 1_GiB - 512_MiB; - const s64 min_spacing_critical = device_memory - 1_GiB; + const s64 min_spacing_expected = device_memory - 1_GiB; + const s64 min_spacing_critical = device_memory - 512_MiB; const s64 mem_threshold = std::min(device_memory, TARGET_THRESHOLD); const s64 min_vacancy_expected = (6 * mem_threshold) / 10; const s64 min_vacancy_critical = (3 * mem_threshold) / 10; @@ -86,10 +86,12 @@ void TextureCache<P>::RunGarbageCollector() { // used by the async decoder thread. return false; } + if (!aggressive_mode && True(image.flags & ImageFlagBits::CostlyLoad)) { + return false; + } const bool must_download = image.IsSafeDownload() && False(image.flags & ImageFlagBits::BadOverlap); - if (!high_priority_mode && - (must_download || True(image.flags & ImageFlagBits::CostlyLoad))) { + if (!high_priority_mode && must_download) { return false; } if (must_download) { @@ -137,7 +139,6 @@ void TextureCache<P>::TickFrame() { TickAsyncDecode(); runtime.TickFrame(); - critical_gc = 0; ++frame_tick; if constexpr (IMPLEMENTS_ASYNC_DOWNLOADS) { @@ -184,6 +185,42 @@ void TextureCache<P>::FillComputeImageViews(std::span<ImageViewInOut> views) { } template <class P> +void TextureCache<P>::CheckFeedbackLoop(std::span<const ImageViewInOut> views) { + const bool requires_barrier = [&] { + for (const auto& view : views) { + if (!view.id) { + continue; + } + auto& image_view = slot_image_views[view.id]; + + // Check color targets + for (const auto& ct_view_id : render_targets.color_buffer_ids) { + if (ct_view_id) { + auto& ct_view = slot_image_views[ct_view_id]; + if (image_view.image_id == ct_view.image_id) { + return true; + } + } + } + + // Check zeta target + if (render_targets.depth_buffer_id) { + auto& zt_view = slot_image_views[render_targets.depth_buffer_id]; + if (image_view.image_id == zt_view.image_id) { + return true; + } + } + } + + return false; + }(); + + if (requires_barrier) { + runtime.BarrierFeedbackLoop(); + } +} + +template <class P> typename P::Sampler* TextureCache<P>::GetGraphicsSampler(u32 index) { if (index > channel_state->graphics_sampler_table.Limit()) { LOG_DEBUG(HW_GPU, "Invalid sampler index={}", index); @@ -1274,17 +1311,18 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA const size_t size_bytes = CalculateGuestSizeInBytes(new_info); const bool broken_views = runtime.HasBrokenTextureViewFormats(); const bool native_bgr = runtime.HasNativeBgr(); - boost::container::small_vector<ImageId, 4> overlap_ids; - std::unordered_set<ImageId> overlaps_found; - boost::container::small_vector<ImageId, 4> left_aliased_ids; - boost::container::small_vector<ImageId, 4> right_aliased_ids; - std::unordered_set<ImageId> ignore_textures; - boost::container::small_vector<ImageId, 4> bad_overlap_ids; - boost::container::small_vector<ImageId, 4> all_siblings; + join_overlap_ids.clear(); + join_overlaps_found.clear(); + join_left_aliased_ids.clear(); + join_right_aliased_ids.clear(); + join_ignore_textures.clear(); + join_bad_overlap_ids.clear(); + join_copies_to_do.clear(); + join_alias_indices.clear(); const bool this_is_linear = info.type == ImageType::Linear; const auto region_check = [&](ImageId overlap_id, ImageBase& overlap) { if (True(overlap.flags & ImageFlagBits::Remapped)) { - ignore_textures.insert(overlap_id); + join_ignore_textures.insert(overlap_id); return; } const bool overlap_is_linear = overlap.info.type == ImageType::Linear; @@ -1294,11 +1332,11 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA if (this_is_linear && overlap_is_linear) { if (info.pitch == overlap.info.pitch && gpu_addr == overlap.gpu_addr) { // Alias linear images with the same pitch - left_aliased_ids.push_back(overlap_id); + join_left_aliased_ids.push_back(overlap_id); } return; } - overlaps_found.insert(overlap_id); + join_overlaps_found.insert(overlap_id); static constexpr bool strict_size = true; const std::optional<OverlapResult> solution = ResolveOverlap( new_info, gpu_addr, cpu_addr, overlap, strict_size, broken_views, native_bgr); @@ -1306,33 +1344,33 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA gpu_addr = solution->gpu_addr; cpu_addr = solution->cpu_addr; new_info.resources = solution->resources; - overlap_ids.push_back(overlap_id); - all_siblings.push_back(overlap_id); + join_overlap_ids.push_back(overlap_id); + join_copies_to_do.emplace_back(JoinCopy{false, overlap_id}); return; } static constexpr auto options = RelaxedOptions::Size | RelaxedOptions::Format; const ImageBase new_image_base(new_info, gpu_addr, cpu_addr); if (IsSubresource(new_info, overlap, gpu_addr, options, broken_views, native_bgr)) { - left_aliased_ids.push_back(overlap_id); + join_left_aliased_ids.push_back(overlap_id); overlap.flags |= ImageFlagBits::Alias; - all_siblings.push_back(overlap_id); + join_copies_to_do.emplace_back(JoinCopy{true, overlap_id}); } else if (IsSubresource(overlap.info, new_image_base, overlap.gpu_addr, options, broken_views, native_bgr)) { - right_aliased_ids.push_back(overlap_id); + join_right_aliased_ids.push_back(overlap_id); overlap.flags |= ImageFlagBits::Alias; - all_siblings.push_back(overlap_id); + join_copies_to_do.emplace_back(JoinCopy{true, overlap_id}); } else { - bad_overlap_ids.push_back(overlap_id); + join_bad_overlap_ids.push_back(overlap_id); } }; ForEachImageInRegion(cpu_addr, size_bytes, region_check); const auto region_check_gpu = [&](ImageId overlap_id, ImageBase& overlap) { - if (!overlaps_found.contains(overlap_id)) { + if (!join_overlaps_found.contains(overlap_id)) { if (True(overlap.flags & ImageFlagBits::Remapped)) { - ignore_textures.insert(overlap_id); + join_ignore_textures.insert(overlap_id); } if (overlap.gpu_addr == gpu_addr && overlap.guest_size_bytes == size_bytes) { - ignore_textures.insert(overlap_id); + join_ignore_textures.insert(overlap_id); } } }; @@ -1340,11 +1378,11 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA bool can_rescale = info.rescaleable; bool any_rescaled = false; - for (const ImageId sibling_id : all_siblings) { + for (const auto& copy : join_copies_to_do) { if (!can_rescale) { break; } - Image& sibling = slot_images[sibling_id]; + Image& sibling = slot_images[copy.id]; can_rescale &= ImageCanRescale(sibling); any_rescaled |= True(sibling.flags & ImageFlagBits::Rescaled); } @@ -1352,13 +1390,13 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA can_rescale &= any_rescaled; if (can_rescale) { - for (const ImageId sibling_id : all_siblings) { - Image& sibling = slot_images[sibling_id]; + for (const auto& copy : join_copies_to_do) { + Image& sibling = slot_images[copy.id]; ScaleUp(sibling); } } else { - for (const ImageId sibling_id : all_siblings) { - Image& sibling = slot_images[sibling_id]; + for (const auto& copy : join_copies_to_do) { + Image& sibling = slot_images[copy.id]; ScaleDown(sibling); } } @@ -1370,7 +1408,7 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA new_image.flags |= ImageFlagBits::Sparse; } - for (const ImageId overlap_id : ignore_textures) { + for (const ImageId overlap_id : join_ignore_textures) { Image& overlap = slot_images[overlap_id]; if (True(overlap.flags & ImageFlagBits::GpuModified)) { UNIMPLEMENTED(); @@ -1391,14 +1429,60 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA ScaleDown(new_image); } - std::ranges::sort(overlap_ids, [this](const ImageId lhs, const ImageId rhs) { - const ImageBase& lhs_image = slot_images[lhs]; - const ImageBase& rhs_image = slot_images[rhs]; + std::ranges::sort(join_copies_to_do, [this](const JoinCopy& lhs, const JoinCopy& rhs) { + const ImageBase& lhs_image = slot_images[lhs.id]; + const ImageBase& rhs_image = slot_images[rhs.id]; return lhs_image.modification_tick < rhs_image.modification_tick; }); - for (const ImageId overlap_id : overlap_ids) { - Image& overlap = slot_images[overlap_id]; + ImageBase& new_image_base = new_image; + for (const ImageId aliased_id : join_right_aliased_ids) { + ImageBase& aliased = slot_images[aliased_id]; + size_t alias_index = new_image_base.aliased_images.size(); + if (!AddImageAlias(new_image_base, aliased, new_image_id, aliased_id)) { + continue; + } + join_alias_indices.emplace(aliased_id, alias_index); + new_image.flags |= ImageFlagBits::Alias; + } + for (const ImageId aliased_id : join_left_aliased_ids) { + ImageBase& aliased = slot_images[aliased_id]; + size_t alias_index = new_image_base.aliased_images.size(); + if (!AddImageAlias(aliased, new_image_base, aliased_id, new_image_id)) { + continue; + } + join_alias_indices.emplace(aliased_id, alias_index); + new_image.flags |= ImageFlagBits::Alias; + } + for (const ImageId aliased_id : join_bad_overlap_ids) { + ImageBase& aliased = slot_images[aliased_id]; + aliased.overlapping_images.push_back(new_image_id); + new_image.overlapping_images.push_back(aliased_id); + if (aliased.info.resources.levels == 1 && aliased.info.block.depth == 0 && + aliased.overlapping_images.size() > 1) { + aliased.flags |= ImageFlagBits::BadOverlap; + } + if (new_image.info.resources.levels == 1 && new_image.info.block.depth == 0 && + new_image.overlapping_images.size() > 1) { + new_image.flags |= ImageFlagBits::BadOverlap; + } + } + + for (const auto& copy_object : join_copies_to_do) { + Image& overlap = slot_images[copy_object.id]; + if (copy_object.is_alias) { + if (!overlap.IsSafeDownload()) { + continue; + } + const auto alias_pointer = join_alias_indices.find(copy_object.id); + if (alias_pointer == join_alias_indices.end()) { + continue; + } + const AliasedImage& aliased = new_image.aliased_images[alias_pointer->second]; + CopyImage(new_image_id, aliased.id, aliased.copies); + new_image.modification_tick = overlap.modification_tick; + continue; + } if (True(overlap.flags & ImageFlagBits::GpuModified)) { new_image.flags |= ImageFlagBits::GpuModified; const auto& resolution = Settings::values.resolution_info; @@ -1411,35 +1495,15 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, VA } else { runtime.CopyImage(new_image, overlap, std::move(copies)); } + new_image.modification_tick = overlap.modification_tick; } if (True(overlap.flags & ImageFlagBits::Tracked)) { - UntrackImage(overlap, overlap_id); - } - UnregisterImage(overlap_id); - DeleteImage(overlap_id); - } - ImageBase& new_image_base = new_image; - for (const ImageId aliased_id : right_aliased_ids) { - ImageBase& aliased = slot_images[aliased_id]; - AddImageAlias(new_image_base, aliased, new_image_id, aliased_id); - new_image.flags |= ImageFlagBits::Alias; - } - for (const ImageId aliased_id : left_aliased_ids) { - ImageBase& aliased = slot_images[aliased_id]; - AddImageAlias(aliased, new_image_base, aliased_id, new_image_id); - new_image.flags |= ImageFlagBits::Alias; - } - for (const ImageId aliased_id : bad_overlap_ids) { - ImageBase& aliased = slot_images[aliased_id]; - aliased.overlapping_images.push_back(new_image_id); - new_image.overlapping_images.push_back(aliased_id); - if (aliased.info.resources.levels == 1 && aliased.overlapping_images.size() > 1) { - aliased.flags |= ImageFlagBits::BadOverlap; - } - if (new_image.info.resources.levels == 1 && new_image.overlapping_images.size() > 1) { - new_image.flags |= ImageFlagBits::BadOverlap; + UntrackImage(overlap, copy_object.id); } + UnregisterImage(copy_object.id); + DeleteImage(copy_object.id); } + RegisterImage(new_image_id); return new_image_id; } @@ -1469,7 +1533,7 @@ std::optional<typename TextureCache<P>::BlitImages> TextureCache<P>::GetBlitImag if (!copy.must_accelerate) { do { if (!src_id && !dst_id) { - break; + return std::nullopt; } if (src_id && True(slot_images[src_id].flags & ImageFlagBits::GpuModified)) { break; @@ -1847,10 +1911,6 @@ void TextureCache<P>::RegisterImage(ImageId image_id) { tentative_size = EstimatedDecompressedSize(tentative_size, image.info.format); } total_used_memory += Common::AlignUp(tentative_size, 1024); - if (total_used_memory > critical_memory && critical_gc < GC_EMERGENCY_COUNTS) { - RunGarbageCollector(); - critical_gc++; - } image.lru_index = lru_cache.Insert(image_id, frame_tick); ForEachGPUPage(image.gpu_addr, image.guest_size_bytes, [this, image_id](u64 page) { diff --git a/src/video_core/texture_cache/texture_cache_base.h b/src/video_core/texture_cache/texture_cache_base.h index 0720494e5..3bfa92154 100644 --- a/src/video_core/texture_cache/texture_cache_base.h +++ b/src/video_core/texture_cache/texture_cache_base.h @@ -10,7 +10,9 @@ #include <span> #include <type_traits> #include <unordered_map> +#include <unordered_set> #include <vector> +#include <boost/container/small_vector.hpp> #include <queue> #include "common/common_types.h" @@ -148,6 +150,9 @@ public: /// Fill image_view_ids with the compute images in indices void FillComputeImageViews(std::span<ImageViewInOut> views); + /// Handle feedback loops during draws. + void CheckFeedbackLoop(std::span<const ImageViewInOut> views); + /// Get the sampler from the graphics descriptor table in the specified index Sampler* GetGraphicsSampler(u32 index); @@ -424,7 +429,6 @@ private: u64 minimum_memory; u64 expected_memory; u64 critical_memory; - size_t critical_gc; struct BufferDownload { GPUVAddr address; @@ -474,6 +478,20 @@ private: Common::ThreadWorker texture_decode_worker{1, "TextureDecoder"}; std::vector<std::unique_ptr<AsyncDecodeContext>> async_decodes; + + // Join caching + boost::container::small_vector<ImageId, 4> join_overlap_ids; + std::unordered_set<ImageId> join_overlaps_found; + boost::container::small_vector<ImageId, 4> join_left_aliased_ids; + boost::container::small_vector<ImageId, 4> join_right_aliased_ids; + std::unordered_set<ImageId> join_ignore_textures; + boost::container::small_vector<ImageId, 4> join_bad_overlap_ids; + struct JoinCopy { + bool is_alias; + ImageId id; + }; + boost::container::small_vector<JoinCopy, 4> join_copies_to_do; + std::unordered_map<ImageId, size_t> join_alias_indices; }; } // namespace VideoCommon diff --git a/src/video_core/texture_cache/util.cpp b/src/video_core/texture_cache/util.cpp index f1071aa23..95a5b47d8 100644 --- a/src/video_core/texture_cache/util.cpp +++ b/src/video_core/texture_cache/util.cpp @@ -18,6 +18,8 @@ #include "common/bit_util.h" #include "common/common_types.h" #include "common/div_ceil.h" +#include "common/scratch_buffer.h" +#include "common/settings.h" #include "video_core/compatible_formats.h" #include "video_core/engines/maxwell_3d.h" #include "video_core/memory_manager.h" @@ -28,6 +30,7 @@ #include "video_core/texture_cache/samples_helper.h" #include "video_core/texture_cache/util.h" #include "video_core/textures/astc.h" +#include "video_core/textures/bcn.h" #include "video_core/textures/decoders.h" namespace VideoCommon { @@ -120,7 +123,9 @@ template <u32 GOB_EXTENT> return { .width = AdjustMipBlockSize<GOB_SIZE_X>(num_tiles.width, block_size.width, level), .height = AdjustMipBlockSize<GOB_SIZE_Y>(num_tiles.height, block_size.height, level), - .depth = AdjustMipBlockSize<GOB_SIZE_Z>(num_tiles.depth, block_size.depth, level), + .depth = level == 0 + ? block_size.depth + : AdjustMipBlockSize<GOB_SIZE_Z>(num_tiles.depth, block_size.depth, level), }; } @@ -162,6 +167,13 @@ template <u32 GOB_EXTENT> } [[nodiscard]] constexpr Extent3D TileShift(const LevelInfo& info, u32 level) { + if (level == 0) { + return Extent3D{ + .width = info.block.width, + .height = info.block.height, + .depth = info.block.depth, + }; + } const Extent3D blocks = NumLevelBlocks(info, level); return Extent3D{ .width = AdjustTileSize(info.block.width, GOB_SIZE_X, blocks.width), @@ -585,6 +597,21 @@ u32 CalculateConvertedSizeBytes(const ImageInfo& info) noexcept { return info.size.width * BytesPerBlock(info.format); } static constexpr Extent2D TILE_SIZE{1, 1}; + if (IsPixelFormatASTC(info.format) && Settings::values.astc_recompression.GetValue() != + Settings::AstcRecompression::Uncompressed) { + const u32 bpp_div = + Settings::values.astc_recompression.GetValue() == Settings::AstcRecompression::Bc1 ? 2 + : 1; + // NumBlocksPerLayer doesn't account for this correctly, so we have to do it manually. + u32 output_size = 0; + for (s32 i = 0; i < info.resources.levels; i++) { + const auto mip_size = AdjustMipSize(info.size, i); + const u32 plane_dim = + Common::AlignUp(mip_size.width, 4U) * Common::AlignUp(mip_size.height, 4U); + output_size += (plane_dim * info.size.depth * info.resources.layers) / bpp_div; + } + return output_size; + } return NumBlocksPerLayer(info, TILE_SIZE) * info.resources.layers * CONVERTED_BYTES_PER_BLOCK; } @@ -885,6 +912,7 @@ BufferCopy UploadBufferCopy(Tegra::MemoryManager& gpu_memory, GPUVAddr gpu_addr, void ConvertImage(std::span<const u8> input, const ImageInfo& info, std::span<u8> output, std::span<BufferImageCopy> copies) { u32 output_offset = 0; + Common::ScratchBuffer<u8> decode_scratch; const Extent2D tile_size = DefaultBlockSize(info.format); for (BufferImageCopy& copy : copies) { @@ -895,22 +923,58 @@ void ConvertImage(std::span<const u8> input, const ImageInfo& info, std::span<u8 ASSERT(copy.image_extent == mip_size); ASSERT(copy.buffer_row_length == Common::AlignUp(mip_size.width, tile_size.width)); ASSERT(copy.buffer_image_height == Common::AlignUp(mip_size.height, tile_size.height)); - if (IsPixelFormatASTC(info.format)) { + + const auto input_offset = input.subspan(copy.buffer_offset); + copy.buffer_offset = output_offset; + copy.buffer_row_length = mip_size.width; + copy.buffer_image_height = mip_size.height; + + const auto recompression_setting = Settings::values.astc_recompression.GetValue(); + const bool astc = IsPixelFormatASTC(info.format); + + if (astc && recompression_setting == Settings::AstcRecompression::Uncompressed) { Tegra::Texture::ASTC::Decompress( - input.subspan(copy.buffer_offset), copy.image_extent.width, - copy.image_extent.height, + input_offset, copy.image_extent.width, copy.image_extent.height, copy.image_subresource.num_layers * copy.image_extent.depth, tile_size.width, tile_size.height, output.subspan(output_offset)); + + output_offset += copy.image_extent.width * copy.image_extent.height * + copy.image_subresource.num_layers * CONVERTED_BYTES_PER_BLOCK; + } else if (astc) { + // BC1 uses 0.5 bytes per texel + // BC3 uses 1 byte per texel + const auto compress = recompression_setting == Settings::AstcRecompression::Bc1 + ? Tegra::Texture::BCN::CompressBC1 + : Tegra::Texture::BCN::CompressBC3; + const auto bpp_div = recompression_setting == Settings::AstcRecompression::Bc1 ? 2 : 1; + + const u32 plane_dim = copy.image_extent.width * copy.image_extent.height; + const u32 level_size = plane_dim * copy.image_extent.depth * + copy.image_subresource.num_layers * CONVERTED_BYTES_PER_BLOCK; + decode_scratch.resize_destructive(level_size); + + Tegra::Texture::ASTC::Decompress( + input_offset, copy.image_extent.width, copy.image_extent.height, + copy.image_subresource.num_layers * copy.image_extent.depth, tile_size.width, + tile_size.height, decode_scratch); + + compress(decode_scratch, copy.image_extent.width, copy.image_extent.height, + copy.image_subresource.num_layers * copy.image_extent.depth, + output.subspan(output_offset)); + + const u32 aligned_plane_dim = Common::AlignUp(copy.image_extent.width, 4) * + Common::AlignUp(copy.image_extent.height, 4); + + copy.buffer_size = + (aligned_plane_dim * copy.image_extent.depth * copy.image_subresource.num_layers) / + bpp_div; + output_offset += static_cast<u32>(copy.buffer_size); } else { - DecompressBC4(input.subspan(copy.buffer_offset), copy.image_extent, - output.subspan(output_offset)); - } - copy.buffer_offset = output_offset; - copy.buffer_row_length = mip_size.width; - copy.buffer_image_height = mip_size.height; + DecompressBC4(input_offset, copy.image_extent, output.subspan(output_offset)); - output_offset += copy.image_extent.width * copy.image_extent.height * - copy.image_subresource.num_layers * CONVERTED_BYTES_PER_BLOCK; + output_offset += copy.image_extent.width * copy.image_extent.height * + copy.image_subresource.num_layers * CONVERTED_BYTES_PER_BLOCK; + } } } @@ -1233,7 +1297,9 @@ u32 MapSizeBytes(const ImageBase& image) { static_assert(CalculateLevelSize(LevelInfo{{1920, 1080, 1}, {0, 2, 0}, {1, 1}, 2, 0}, 0) == 0x7f8000); -static_assert(CalculateLevelSize(LevelInfo{{32, 32, 1}, {0, 0, 4}, {1, 1}, 4, 0}, 0) == 0x4000); +static_assert(CalculateLevelSize(LevelInfo{{32, 32, 1}, {0, 0, 4}, {1, 1}, 4, 0}, 0) == 0x40000); + +static_assert(CalculateLevelSize(LevelInfo{{128, 8, 1}, {0, 4, 0}, {1, 1}, 4, 0}, 0) == 0x40000); static_assert(CalculateLevelOffset(PixelFormat::R8_SINT, {1920, 1080, 1}, {0, 2, 0}, 0, 7) == 0x2afc00); diff --git a/src/video_core/textures/astc.cpp b/src/video_core/textures/astc.cpp index a68bc0d77..fef0be31d 100644 --- a/src/video_core/textures/astc.cpp +++ b/src/video_core/textures/astc.cpp @@ -16,8 +16,8 @@ #include "common/alignment.h" #include "common/common_types.h" #include "common/polyfill_ranges.h" -#include "common/thread_worker.h" #include "video_core/textures/astc.h" +#include "video_core/textures/workers.h" class InputBitStream { public: @@ -1656,8 +1656,7 @@ void Decompress(std::span<const uint8_t> data, uint32_t width, uint32_t height, const u32 rows = Common::DivideUp(height, block_height); const u32 cols = Common::DivideUp(width, block_width); - static Common::ThreadWorker workers{std::max(std::thread::hardware_concurrency(), 2U) / 2, - "ASTCDecompress"}; + Common::ThreadWorker& workers{GetThreadWorkers()}; for (u32 z = 0; z < depth; ++z) { const u32 depth_offset = z * height * width * 4; diff --git a/src/video_core/textures/bcn.cpp b/src/video_core/textures/bcn.cpp new file mode 100644 index 000000000..671212a49 --- /dev/null +++ b/src/video_core/textures/bcn.cpp @@ -0,0 +1,87 @@ +// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#include <stb_dxt.h> +#include <string.h> + +#include "common/alignment.h" +#include "video_core/textures/bcn.h" +#include "video_core/textures/workers.h" + +namespace Tegra::Texture::BCN { + +using BCNCompressor = void(u8* block_output, const u8* block_input, bool any_alpha); + +template <u32 BytesPerBlock, bool ThresholdAlpha = false> +void CompressBCN(std::span<const uint8_t> data, uint32_t width, uint32_t height, uint32_t depth, + std::span<uint8_t> output, BCNCompressor f) { + constexpr u8 alpha_threshold = 128; + constexpr u32 bytes_per_px = 4; + const u32 plane_dim = width * height; + + Common::ThreadWorker& workers{GetThreadWorkers()}; + + for (u32 z = 0; z < depth; z++) { + for (u32 y = 0; y < height; y += 4) { + auto compress_row = [z, y, width, height, plane_dim, f, data, output]() { + for (u32 x = 0; x < width; x += 4) { + // Gather 4x4 block of RGBA texels + u8 input_colors[4][4][4]; + bool any_alpha = false; + + for (u32 j = 0; j < 4; j++) { + for (u32 i = 0; i < 4; i++) { + const size_t coord = + (z * plane_dim + (y + j) * width + (x + i)) * bytes_per_px; + + if ((x + i < width) && (y + j < height)) { + if constexpr (ThresholdAlpha) { + if (data[coord + 3] >= alpha_threshold) { + input_colors[j][i][0] = data[coord + 0]; + input_colors[j][i][1] = data[coord + 1]; + input_colors[j][i][2] = data[coord + 2]; + input_colors[j][i][3] = 255; + } else { + any_alpha = true; + memset(input_colors[j][i], 0, bytes_per_px); + } + } else { + memcpy(input_colors[j][i], &data[coord], bytes_per_px); + } + } else { + memset(input_colors[j][i], 0, bytes_per_px); + } + } + } + + const u32 bytes_per_row = BytesPerBlock * Common::DivideUp(width, 4U); + const u32 bytes_per_plane = bytes_per_row * Common::DivideUp(height, 4U); + f(output.data() + z * bytes_per_plane + (y / 4) * bytes_per_row + + (x / 4) * BytesPerBlock, + reinterpret_cast<u8*>(input_colors), any_alpha); + } + }; + workers.QueueWork(std::move(compress_row)); + } + workers.WaitForRequests(); + } +} + +void CompressBC1(std::span<const uint8_t> data, uint32_t width, uint32_t height, uint32_t depth, + std::span<uint8_t> output) { + CompressBCN<8, true>(data, width, height, depth, output, + [](u8* block_output, const u8* block_input, bool any_alpha) { + stb_compress_bc1_block(block_output, block_input, any_alpha, + STB_DXT_NORMAL); + }); +} + +void CompressBC3(std::span<const uint8_t> data, uint32_t width, uint32_t height, uint32_t depth, + std::span<uint8_t> output) { + CompressBCN<16, false>(data, width, height, depth, output, + [](u8* block_output, const u8* block_input, bool any_alpha) { + stb_compress_bc3_block(block_output, block_input, STB_DXT_NORMAL); + }); +} + +} // namespace Tegra::Texture::BCN diff --git a/src/video_core/textures/bcn.h b/src/video_core/textures/bcn.h new file mode 100644 index 000000000..6464af885 --- /dev/null +++ b/src/video_core/textures/bcn.h @@ -0,0 +1,17 @@ +// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include <span> +#include <stdint.h> + +namespace Tegra::Texture::BCN { + +void CompressBC1(std::span<const uint8_t> data, uint32_t width, uint32_t height, uint32_t depth, + std::span<uint8_t> output); + +void CompressBC3(std::span<const uint8_t> data, uint32_t width, uint32_t height, uint32_t depth, + std::span<uint8_t> output); + +} // namespace Tegra::Texture::BCN diff --git a/src/video_core/textures/workers.cpp b/src/video_core/textures/workers.cpp new file mode 100644 index 000000000..a71c305f4 --- /dev/null +++ b/src/video_core/textures/workers.cpp @@ -0,0 +1,15 @@ +// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#include "video_core/textures/workers.h" + +namespace Tegra::Texture { + +Common::ThreadWorker& GetThreadWorkers() { + static Common::ThreadWorker workers{std::max(std::thread::hardware_concurrency(), 2U) / 2, + "ImageTranscode"}; + + return workers; +} + +} // namespace Tegra::Texture diff --git a/src/video_core/textures/workers.h b/src/video_core/textures/workers.h new file mode 100644 index 000000000..008dd05b3 --- /dev/null +++ b/src/video_core/textures/workers.h @@ -0,0 +1,12 @@ +// SPDX-FileCopyrightText: Copyright 2023 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include "common/thread_worker.h" + +namespace Tegra::Texture { + +Common::ThreadWorker& GetThreadWorkers(); + +} diff --git a/src/video_core/vulkan_common/vulkan_device.cpp b/src/video_core/vulkan_common/vulkan_device.cpp index c03f4a56b..aea677cb3 100644 --- a/src/video_core/vulkan_common/vulkan_device.cpp +++ b/src/video_core/vulkan_common/vulkan_device.cpp @@ -1002,6 +1002,11 @@ u64 Device::GetDeviceMemoryUsage() const { } void Device::CollectPhysicalMemoryInfo() { + // Account for resolution scaling in memory limits + const size_t normal_memory = 6_GiB; + const size_t scaler_memory = 1_GiB * Settings::values.resolution_info.ScaleUp(1); + + // Calculate limits using memory budget VkPhysicalDeviceMemoryBudgetPropertiesEXT budget{}; budget.sType = VK_STRUCTURE_TYPE_PHYSICAL_DEVICE_MEMORY_BUDGET_PROPERTIES_EXT; const auto mem_info = @@ -1031,11 +1036,12 @@ void Device::CollectPhysicalMemoryInfo() { if (!is_integrated) { const u64 reserve_memory = std::min<u64>(device_access_memory / 8, 1_GiB); device_access_memory -= reserve_memory; + device_access_memory = std::min<u64>(device_access_memory, normal_memory + scaler_memory); return; } const s64 available_memory = static_cast<s64>(device_access_memory - device_initial_usage); device_access_memory = static_cast<u64>(std::max<s64>( - std::min<s64>(available_memory - 8_GiB, 4_GiB), static_cast<s64>(local_memory))); + std::min<s64>(available_memory - 8_GiB, 4_GiB), std::min<s64>(local_memory, 4_GiB))); } void Device::CollectToolingInfo() { diff --git a/src/yuzu/configuration/config.cpp b/src/yuzu/configuration/config.cpp index 70737c54e..662651196 100644 --- a/src/yuzu/configuration/config.cpp +++ b/src/yuzu/configuration/config.cpp @@ -711,6 +711,7 @@ void Config::ReadRendererValues() { ReadGlobalSetting(Settings::values.nvdec_emulation); ReadGlobalSetting(Settings::values.accelerate_astc); ReadGlobalSetting(Settings::values.async_astc); + ReadGlobalSetting(Settings::values.astc_recompression); ReadGlobalSetting(Settings::values.use_reactive_flushing); ReadGlobalSetting(Settings::values.shader_backend); ReadGlobalSetting(Settings::values.use_asynchronous_shaders); @@ -1359,6 +1360,10 @@ void Config::SaveRendererValues() { Settings::values.nvdec_emulation.UsingGlobal()); WriteGlobalSetting(Settings::values.accelerate_astc); WriteGlobalSetting(Settings::values.async_astc); + WriteSetting(QString::fromStdString(Settings::values.astc_recompression.GetLabel()), + static_cast<u32>(Settings::values.astc_recompression.GetValue(global)), + static_cast<u32>(Settings::values.astc_recompression.GetDefault()), + Settings::values.astc_recompression.UsingGlobal()); WriteGlobalSetting(Settings::values.use_reactive_flushing); WriteSetting(QString::fromStdString(Settings::values.shader_backend.GetLabel()), static_cast<u32>(Settings::values.shader_backend.GetValue(global)), diff --git a/src/yuzu/configuration/config.h b/src/yuzu/configuration/config.h index 7d26e9ab6..9cb9db6cf 100644 --- a/src/yuzu/configuration/config.h +++ b/src/yuzu/configuration/config.h @@ -208,3 +208,4 @@ Q_DECLARE_METATYPE(Settings::ScalingFilter); Q_DECLARE_METATYPE(Settings::AntiAliasing); Q_DECLARE_METATYPE(Settings::RendererBackend); Q_DECLARE_METATYPE(Settings::ShaderBackend); +Q_DECLARE_METATYPE(Settings::AstcRecompression); diff --git a/src/yuzu/configuration/configure_graphics_advanced.cpp b/src/yuzu/configuration/configure_graphics_advanced.cpp index 1f3e489d0..896863f87 100644 --- a/src/yuzu/configuration/configure_graphics_advanced.cpp +++ b/src/yuzu/configuration/configure_graphics_advanced.cpp @@ -27,6 +27,7 @@ void ConfigureGraphicsAdvanced::SetConfiguration() { ui->async_present->setEnabled(runtime_lock); ui->renderer_force_max_clock->setEnabled(runtime_lock); ui->async_astc->setEnabled(runtime_lock); + ui->astc_recompression_combobox->setEnabled(runtime_lock); ui->use_asynchronous_shaders->setEnabled(runtime_lock); ui->anisotropic_filtering_combobox->setEnabled(runtime_lock); ui->enable_compute_pipelines_checkbox->setEnabled(runtime_lock); @@ -47,14 +48,20 @@ void ConfigureGraphicsAdvanced::SetConfiguration() { static_cast<int>(Settings::values.gpu_accuracy.GetValue())); ui->anisotropic_filtering_combobox->setCurrentIndex( Settings::values.max_anisotropy.GetValue()); + ui->astc_recompression_combobox->setCurrentIndex( + static_cast<int>(Settings::values.astc_recompression.GetValue())); } else { ConfigurationShared::SetPerGameSetting(ui->gpu_accuracy, &Settings::values.gpu_accuracy); ConfigurationShared::SetPerGameSetting(ui->anisotropic_filtering_combobox, &Settings::values.max_anisotropy); + ConfigurationShared::SetPerGameSetting(ui->astc_recompression_combobox, + &Settings::values.astc_recompression); ConfigurationShared::SetHighlight(ui->label_gpu_accuracy, !Settings::values.gpu_accuracy.UsingGlobal()); ConfigurationShared::SetHighlight(ui->af_label, !Settings::values.max_anisotropy.UsingGlobal()); + ConfigurationShared::SetHighlight(ui->label_astc_recompression, + !Settings::values.astc_recompression.UsingGlobal()); } } @@ -71,6 +78,8 @@ void ConfigureGraphicsAdvanced::ApplyConfiguration() { ui->use_reactive_flushing, use_reactive_flushing); ConfigurationShared::ApplyPerGameSetting(&Settings::values.async_astc, ui->async_astc, async_astc); + ConfigurationShared::ApplyPerGameSetting(&Settings::values.astc_recompression, + ui->astc_recompression_combobox); ConfigurationShared::ApplyPerGameSetting(&Settings::values.use_asynchronous_shaders, ui->use_asynchronous_shaders, use_asynchronous_shaders); @@ -105,6 +114,8 @@ void ConfigureGraphicsAdvanced::SetupPerGameUI() { Settings::values.renderer_force_max_clock.UsingGlobal()); ui->use_reactive_flushing->setEnabled(Settings::values.use_reactive_flushing.UsingGlobal()); ui->async_astc->setEnabled(Settings::values.async_astc.UsingGlobal()); + ui->astc_recompression_combobox->setEnabled( + Settings::values.astc_recompression.UsingGlobal()); ui->use_asynchronous_shaders->setEnabled( Settings::values.use_asynchronous_shaders.UsingGlobal()); ui->use_fast_gpu_time->setEnabled(Settings::values.use_fast_gpu_time.UsingGlobal()); @@ -144,6 +155,9 @@ void ConfigureGraphicsAdvanced::SetupPerGameUI() { ConfigurationShared::SetColoredComboBox( ui->anisotropic_filtering_combobox, ui->af_label, static_cast<int>(Settings::values.max_anisotropy.GetValue(true))); + ConfigurationShared::SetColoredComboBox( + ui->astc_recompression_combobox, ui->label_astc_recompression, + static_cast<int>(Settings::values.astc_recompression.GetValue(true))); } void ConfigureGraphicsAdvanced::ExposeComputeOption() { diff --git a/src/yuzu/configuration/configure_graphics_advanced.ui b/src/yuzu/configuration/configure_graphics_advanced.ui index 9ef7c8e8f..37757a918 100644 --- a/src/yuzu/configuration/configure_graphics_advanced.ui +++ b/src/yuzu/configuration/configure_graphics_advanced.ui @@ -70,6 +70,50 @@ </widget> </item> <item> + <widget class="QWidget" name="astc_recompression_layout" native="true"> + <layout class="QHBoxLayout" name="horizontalLayout_3"> + <property name="leftMargin"> + <number>0</number> + </property> + <property name="topMargin"> + <number>0</number> + </property> + <property name="rightMargin"> + <number>0</number> + </property> + <property name="bottomMargin"> + <number>0</number> + </property> + <item> + <widget class="QLabel" name="label_astc_recompression"> + <property name="text"> + <string>ASTC recompression:</string> + </property> + </widget> + </item> + <item> + <widget class="QComboBox" name="astc_recompression_combobox"> + <item> + <property name="text"> + <string>Uncompressed (Best quality)</string> + </property> + </item> + <item> + <property name="text"> + <string>BC1 (Low quality)</string> + </property> + </item> + <item> + <property name="text"> + <string>BC3 (Medium quality)</string> + </property> + </item> + </widget> + </item> + </layout> + </widget> + </item> + <item> <widget class="QCheckBox" name="async_present"> <property name="text"> <string>Enable asynchronous presentation (Vulkan only)</string> diff --git a/src/yuzu_cmd/config.cpp b/src/yuzu_cmd/config.cpp index dc9a3d68f..c5bc472ca 100644 --- a/src/yuzu_cmd/config.cpp +++ b/src/yuzu_cmd/config.cpp @@ -318,6 +318,7 @@ void Config::ReadValues() { ReadSetting("Renderer", Settings::values.nvdec_emulation); ReadSetting("Renderer", Settings::values.accelerate_astc); ReadSetting("Renderer", Settings::values.async_astc); + ReadSetting("Renderer", Settings::values.astc_recompression); ReadSetting("Renderer", Settings::values.use_fast_gpu_time); ReadSetting("Renderer", Settings::values.use_vulkan_driver_pipeline_cache); diff --git a/src/yuzu_cmd/default_ini.h b/src/yuzu_cmd/default_ini.h index 5e7c3ac04..644a30e59 100644 --- a/src/yuzu_cmd/default_ini.h +++ b/src/yuzu_cmd/default_ini.h @@ -360,6 +360,10 @@ accelerate_astc = # 0 (default): Off, 1: On async_astc = +# Recompress ASTC textures to a different format. +# 0 (default): Uncompressed, 1: BC1 (Low quality), 2: BC3: (Medium quality) +async_astc = + # Turns on the speed limiter, which will limit the emulation speed to the desired speed limit value # 0: Off, 1: On (default) use_speed_limit = diff --git a/src/yuzu_cmd/yuzu.cpp b/src/yuzu_cmd/yuzu.cpp index 5f39ece32..7b6d49c63 100644 --- a/src/yuzu_cmd/yuzu.cpp +++ b/src/yuzu_cmd/yuzu.cpp @@ -227,7 +227,7 @@ int main(int argc, char** argv) { }; while (optind < argc) { - int arg = getopt_long(argc, argv, "g:fhvp::c:", long_options, &option_index); + int arg = getopt_long(argc, argv, "g:fhvp::c:u:", long_options, &option_index); if (arg != -1) { switch (static_cast<char>(arg)) { case 'c': @@ -283,7 +283,7 @@ int main(int argc, char** argv) { break; case 'u': selected_user = atoi(optarg); - return 0; + break; case 'v': PrintVersion(); return 0; |