diff --git a/Data/User/GameConfig/GKBEAF.ini b/Data/User/GameConfig/GKBEAF.ini index 9934a2ba80..5e7ee300f9 100644 --- a/Data/User/GameConfig/GKBEAF.ini +++ b/Data/User/GameConfig/GKBEAF.ini @@ -1,5 +1,6 @@ # GKBEAF - Baten Kaitos [Core] Values set here will override the main dolphin settings. +SyncGPU = 1 [EmuState] The Emulation State. 1 is worst, 5 is best, 0 is not set. EmulationStateId = 4 EmulationIssues = diff --git a/Data/User/GameConfig/GKBPAF.ini b/Data/User/GameConfig/GKBPAF.ini index 936380ca29..afb3661d63 100644 --- a/Data/User/GameConfig/GKBPAF.ini +++ b/Data/User/GameConfig/GKBPAF.ini @@ -1,5 +1,6 @@ # GKBPAF - Baten Kaitos [Core] Values set here will override the main dolphin settings. +SyncGPU = 1 [EmuState] The Emulation State. 1 is worst, 5 is best, 0 is not set. EmulationStateId = 4 EmulationIssues = diff --git a/Data/User/GameConfig/GLSE64.ini b/Data/User/GameConfig/GLSE64.ini index f9b38b357c..70bea42316 100644 --- a/Data/User/GameConfig/GLSE64.ini +++ b/Data/User/GameConfig/GLSE64.ini @@ -1,8 +1,9 @@ # GLSE64 - LucasArts Gladius [Core] Values set here will override the main dolphin settings. TLBHack = 1 +SyncGPU = 1 [EmuState] The Emulation State. 1 is worst, 5 is best, 0 is not set. -EmulationStateId = 1 +EmulationStateId = 4 EmulationIssues = [OnFrame] Add memory patches to be applied every frame here. [ActionReplay] Add action replay cheats here. diff --git a/Source/Core/Core/Src/BootManager.cpp b/Source/Core/Core/Src/BootManager.cpp index 1e9e612fe0..88e0cebcb6 100644 --- a/Source/Core/Core/Src/BootManager.cpp +++ b/Source/Core/Core/Src/BootManager.cpp @@ -55,7 +55,7 @@ namespace BootManager struct ConfigCache { bool valid, bCPUThread, bSkipIdle, bEnableFPRF, bMMU, bDCBZOFF, - bVBeam, bFastDiscSpeed, bMergeBlocks, bDSPHLE, bHLE_BS2; + bVBeam, bSyncGPU, bFastDiscSpeed, bMergeBlocks, bDSPHLE, bHLE_BS2; int iTLBHack, iCPUCore; std::string strBackend; }; @@ -95,6 +95,7 @@ bool BootCore(const std::string& _rFilename) config_cache.bDCBZOFF = StartUp.bDCBZOFF; config_cache.iTLBHack = StartUp.iTLBHack; config_cache.bVBeam = StartUp.bVBeam; + config_cache.bSyncGPU = StartUp.bSyncGPU; config_cache.bFastDiscSpeed = StartUp.bFastDiscSpeed; config_cache.bMergeBlocks = StartUp.bMergeBlocks; config_cache.bDSPHLE = StartUp.bDSPHLE; @@ -109,6 +110,7 @@ bool BootCore(const std::string& _rFilename) game_ini.Get("Core", "TLBHack", &StartUp.iTLBHack, StartUp.iTLBHack); game_ini.Get("Core", "DCBZ", &StartUp.bDCBZOFF, StartUp.bDCBZOFF); game_ini.Get("Core", "VBeam", &StartUp.bVBeam, StartUp.bVBeam); + game_ini.Get("Core", "SyncGPU", &StartUp.bSyncGPU, StartUp.bSyncGPU); game_ini.Get("Core", "FastDiscSpeed", &StartUp.bFastDiscSpeed, StartUp.bFastDiscSpeed); game_ini.Get("Core", "BlockMerging", &StartUp.bMergeBlocks, StartUp.bMergeBlocks); game_ini.Get("Core", "DSPHLE", &StartUp.bDSPHLE, StartUp.bDSPHLE); @@ -168,6 +170,7 @@ void Stop() StartUp.bDCBZOFF = config_cache.bDCBZOFF; StartUp.iTLBHack = config_cache.iTLBHack; StartUp.bVBeam = config_cache.bVBeam; + StartUp.bSyncGPU = config_cache.bSyncGPU; StartUp.bFastDiscSpeed = config_cache.bFastDiscSpeed; StartUp.bMergeBlocks = config_cache.bMergeBlocks; StartUp.bDSPHLE = config_cache.bDSPHLE; diff --git a/Source/Core/Core/Src/ConfigManager.cpp b/Source/Core/Core/Src/ConfigManager.cpp index f6194c1bb7..5b7a94e2ef 100644 --- a/Source/Core/Core/Src/ConfigManager.cpp +++ b/Source/Core/Core/Src/ConfigManager.cpp @@ -399,6 +399,7 @@ void SConfig::LoadSettings() ini.Get("Core", "MMU", &m_LocalCoreStartupParameter.bMMU, false); ini.Get("Core", "TLBHack", &m_LocalCoreStartupParameter.iTLBHack, 0); ini.Get("Core", "VBeam", &m_LocalCoreStartupParameter.bVBeam, false); + ini.Get("Core", "SyncGPU", &m_LocalCoreStartupParameter.bSyncGPU, false); ini.Get("Core", "FastDiscSpeed", &m_LocalCoreStartupParameter.bFastDiscSpeed, false); ini.Get("Core", "DCBZ", &m_LocalCoreStartupParameter.bDCBZOFF, false); ini.Get("Core", "FrameLimit", &m_Framelimit, 1); // auto frame limit by default diff --git a/Source/Core/Core/Src/CoreParameter.cpp b/Source/Core/Core/Src/CoreParameter.cpp index a7a44be5ea..0a6bb795d5 100644 --- a/Source/Core/Core/Src/CoreParameter.cpp +++ b/Source/Core/Core/Src/CoreParameter.cpp @@ -50,7 +50,7 @@ SCoreStartupParameter::SCoreStartupParameter() bDPL2Decoder(false), iLatency(14), bRunCompareServer(false), bRunCompareClient(false), bMMU(false), bDCBZOFF(false), iTLBHack(0), bVBeam(false), - bFastDiscSpeed(false), + bSyncGPU(false), bFastDiscSpeed(false), SelectedLanguage(0), bWii(false), bConfirmStop(false), bHideCursor(false), bAutoHideCursor(false), bUsePanicHandlers(true), bOnScreenDisplayMessages(true), @@ -78,6 +78,7 @@ void SCoreStartupParameter::LoadDefaults() bDCBZOFF = false; iTLBHack = 0; bVBeam = false; + bSyncGPU = false; bFastDiscSpeed = false; bMergeBlocks = false; SelectedLanguage = 0; diff --git a/Source/Core/Core/Src/CoreParameter.h b/Source/Core/Core/Src/CoreParameter.h index 55d20b7b1b..896f111ab7 100644 --- a/Source/Core/Core/Src/CoreParameter.h +++ b/Source/Core/Core/Src/CoreParameter.h @@ -116,6 +116,7 @@ struct SCoreStartupParameter bool bDCBZOFF; int iTLBHack; bool bVBeam; + bool bSyncGPU; bool bFastDiscSpeed; int SelectedLanguage; diff --git a/Source/Core/Core/Src/HW/SystemTimers.cpp b/Source/Core/Core/Src/HW/SystemTimers.cpp index a041a3b0ad..10f5e3204e 100644 --- a/Source/Core/Core/Src/HW/SystemTimers.cpp +++ b/Source/Core/Core/Src/HW/SystemTimers.cpp @@ -74,6 +74,7 @@ IPC_HLE_PERIOD: For the Wiimote this is the call schedule: #include "Thread.h" #include "Timer.h" #include "VideoBackendBase.h" +#include "CommandProcessor.h" namespace SystemTimers @@ -110,6 +111,7 @@ int et_Dec; int et_VI; int et_SI; int et_AI; +int et_CP; int et_AudioDMA; int et_DSP; int et_IPC_HLE; @@ -127,6 +129,9 @@ int // This is a fixed value, don't change it AUDIO_DMA_PERIOD, + // Regulates the speed of the Command Processor + CP_PERIOD, + // This is completely arbitrary. If we find that we need lower latency, we can just // increase this number. IPC_HLE_PERIOD; @@ -187,6 +192,12 @@ void SICallback(u64 userdata, int cyclesLate) CoreTiming::ScheduleEvent(SerialInterface::GetTicksToNextSIPoll() - cyclesLate, et_SI); } +void CPCallback(u64 userdata, int cyclesLate) +{ + CommandProcessor::Update(); + CoreTiming::ScheduleEvent(CP_PERIOD - cyclesLate, et_CP); +} + void DecrementerCallback(u64 userdata, int cyclesLate) { PowerPC::ppcState.spr[SPR_DEC] = 0xFFFFFFFF; @@ -272,6 +283,9 @@ void Init() // System internal sample rate is fixed at 32KHz * 4 (16bit Stereo) / 32 bytes DMA AUDIO_DMA_PERIOD = CPU_CORE_CLOCK / (AudioInterface::GetAIDSampleRate() * 4 / 32); + // Emulated gekko <-> flipper bus speed ratio (cpu clock / flipper clock) + CP_PERIOD = GetTicksPerSecond() / 10000; + Common::Timer::IncreaseResolution(); // store and convert localtime at boot to timebase ticks CoreTiming::SetFakeTBStartValue((u64)(CPU_CORE_CLOCK / TIMER_RATIO) * (u64)CEXIIPL::GetGCTime()); @@ -284,6 +298,8 @@ void Init() et_AI = CoreTiming::RegisterEvent("AICallback", AICallback); et_VI = CoreTiming::RegisterEvent("VICallback", VICallback); et_SI = CoreTiming::RegisterEvent("SICallback", SICallback); + if (SConfig::GetInstance().m_LocalCoreStartupParameter.bSyncGPU) + et_CP = CoreTiming::RegisterEvent("CPCallback", CPCallback); et_DSP = CoreTiming::RegisterEvent("DSPCallback", DSPCallback); et_AudioDMA = CoreTiming::RegisterEvent("AudioDMACallback", AudioDMACallback); et_IPC_HLE = CoreTiming::RegisterEvent("IPC_HLE_UpdateCallback", IPC_HLE_UpdateCallback); @@ -294,6 +310,8 @@ void Init() CoreTiming::ScheduleEvent(DSP_PERIOD, et_DSP); CoreTiming::ScheduleEvent(VideoInterface::GetTicksPerFrame(), et_SI); CoreTiming::ScheduleEvent(AUDIO_DMA_PERIOD, et_AudioDMA); + if (SConfig::GetInstance().m_LocalCoreStartupParameter.bSyncGPU) + CoreTiming::ScheduleEvent(CP_PERIOD, et_CP); CoreTiming::ScheduleEvent(VideoInterface::GetTicksPerFrame(), et_PatchEngine); diff --git a/Source/Core/DolphinWX/Src/ISOProperties.cpp b/Source/Core/DolphinWX/Src/ISOProperties.cpp index c44aad5d5b..f7b35b8269 100644 --- a/Source/Core/DolphinWX/Src/ISOProperties.cpp +++ b/Source/Core/DolphinWX/Src/ISOProperties.cpp @@ -321,6 +321,8 @@ void CISOProperties::CreateGUIControls(bool IsWad) DCBZOFF->SetToolTip(_("Bypass the clearing of the data cache by the DCBZ instruction. Usually leave this option disabled.")); VBeam = new wxCheckBox(m_GameConfig, ID_VBEAM, _("Accurate VBeam emulation"), wxDefaultPosition, wxDefaultSize, wxCHK_3STATE|wxCHK_ALLOW_3RD_STATE_FOR_USER, wxDefaultValidator); VBeam->SetToolTip(_("If the FPS is erratic, this option may help. (ON = Compatible, OFF = Fast)")); + SyncGPU = new wxCheckBox(m_GameConfig, ID_SYNCGPU, _("Sychronise GPU thread"), wxDefaultPosition, wxDefaultSize, wxCHK_3STATE|wxCHK_ALLOW_3RD_STATE_FOR_USER, wxDefaultValidator); + SyncGPU->SetToolTip(_("Synchonises the GPU and CPU threads to help prevent random freezes in Dual Core mode. (ON = Compatible, OFF = Fast)")); FastDiscSpeed = new wxCheckBox(m_GameConfig, ID_DISCSPEED, _("Speed up Disc Transfer Rate"), wxDefaultPosition, wxDefaultSize, wxCHK_3STATE|wxCHK_ALLOW_3RD_STATE_FOR_USER, wxDefaultValidator); FastDiscSpeed->SetToolTip(_("Enable fast disc access. Needed for a few games. (ON = Fast, OFF = Compatible)")); BlockMerging = new wxCheckBox(m_GameConfig, ID_MERGEBLOCKS, _("Enable Block Merging"), wxDefaultPosition, wxDefaultSize, wxCHK_3STATE|wxCHK_ALLOW_3RD_STATE_FOR_USER, wxDefaultValidator); @@ -362,9 +364,10 @@ void CISOProperties::CreateGUIControls(bool IsWad) sbCoreOverrides->Add(CPUThread, 0, wxLEFT, 5); sbCoreOverrides->Add(SkipIdle, 0, wxLEFT, 5); sbCoreOverrides->Add(MMU, 0, wxLEFT, 5); - sbCoreOverrides->Add(DCBZOFF, 0, wxLEFT, 5); sbCoreOverrides->Add(TLBHack, 0, wxLEFT, 5); + sbCoreOverrides->Add(DCBZOFF, 0, wxLEFT, 5); sbCoreOverrides->Add(VBeam, 0, wxLEFT, 5); + sbCoreOverrides->Add(SyncGPU, 0, wxLEFT, 5); sbCoreOverrides->Add(FastDiscSpeed, 0, wxLEFT, 5); sbCoreOverrides->Add(BlockMerging, 0, wxLEFT, 5); sbCoreOverrides->Add(DSPHLE, 0, wxLEFT, 5); @@ -941,6 +944,11 @@ void CISOProperties::LoadGameConfig() else VBeam->Set3StateValue(wxCHK_UNDETERMINED); + if (GameIni.Get("Core", "SyncGPU", &bTemp)) + SyncGPU->Set3StateValue((wxCheckBoxState)bTemp); + else + SyncGPU->Set3StateValue(wxCHK_UNDETERMINED); + if (GameIni.Get("Core", "FastDiscSpeed", &bTemp)) FastDiscSpeed->Set3StateValue((wxCheckBoxState)bTemp); else @@ -1030,6 +1038,11 @@ bool CISOProperties::SaveGameConfig() else GameIni.Set("Core", "VBeam", VBeam->Get3StateValue()); + if (SyncGPU->Get3StateValue() == wxCHK_UNDETERMINED) + GameIni.DeleteKey("Core", "SyncGPU"); + else + GameIni.Set("Core", "SyncGPU", SyncGPU->Get3StateValue()); + if (FastDiscSpeed->Get3StateValue() == wxCHK_UNDETERMINED) GameIni.DeleteKey("Core", "FastDiscSpeed"); else diff --git a/Source/Core/DolphinWX/Src/ISOProperties.h b/Source/Core/DolphinWX/Src/ISOProperties.h index 499b242516..41e7d250a0 100644 --- a/Source/Core/DolphinWX/Src/ISOProperties.h +++ b/Source/Core/DolphinWX/Src/ISOProperties.h @@ -70,7 +70,7 @@ private: // Core wxCheckBox *CPUThread, *SkipIdle, *MMU, *DCBZOFF, *TLBHack; - wxCheckBox *VBeam, *FastDiscSpeed, *BlockMerging, *DSPHLE; + wxCheckBox *VBeam, *SyncGPU, *FastDiscSpeed, *BlockMerging, *DSPHLE; // Wii wxCheckBox *EnableWideScreen; // Video @@ -130,6 +130,7 @@ private: ID_DCBZOFF, ID_TLBHACK, ID_VBEAM, + ID_SYNCGPU, ID_DISCSPEED, ID_MERGEBLOCKS, ID_AUDIO_DSP_HLE, diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.cpp b/Source/Core/VideoCommon/Src/CommandProcessor.cpp index 7e976f256e..eb03730fc3 100644 --- a/Source/Core/VideoCommon/Src/CommandProcessor.cpp +++ b/Source/Core/VideoCommon/Src/CommandProcessor.cpp @@ -32,6 +32,8 @@ #include "HW/GPFifo.h" #include "HW/Memmap.h" #include "DLCache.h" +#include "HW/SystemTimers.h" +#include "Core.h" namespace CommandProcessor { @@ -64,6 +66,8 @@ volatile bool interruptTokenWaiting = false; volatile bool interruptFinishWaiting = false; volatile bool waitingForPEInterruptDisable = false; +volatile u32 VITicks = CommandProcessor::m_cpClockOrigin; + bool IsOnThread() { return SConfig::GetInstance().m_LocalCoreStartupParameter.bCPUThread; @@ -464,7 +468,7 @@ void STACKALIGN GatherPipeBursted() } if (IsOnThread()) - SetWatermarkFromGatherPipe(); + SetCpStatus(); // update the fifo-pointer if (fifo.CPWritePointer >= fifo.CPEnd) @@ -514,27 +518,6 @@ void AbortFrame() } -void SetWatermarkFromGatherPipe() -{ - fifo.bFF_HiWatermark = (fifo.CPReadWriteDistance > fifo.CPHiWatermark); - fifo.bFF_LoWatermark = (fifo.CPReadWriteDistance < fifo.CPLoWatermark); - isHiWatermarkActive = fifo.bFF_HiWatermark && fifo.bFF_HiWatermarkInt && m_CPCtrlReg.GPReadEnable; - isLoWatermarkActive = fifo.bFF_LoWatermark && fifo.bFF_LoWatermarkInt && m_CPCtrlReg.GPReadEnable; - - if (isHiWatermarkActive) - { - interruptSet = true; - INFO_LOG(COMMANDPROCESSOR,"Interrupt set"); - ProcessorInterface::SetInterrupt(INT_CAUSE_CP, true); - } - else if (isLoWatermarkActive) - { - interruptSet = true; - INFO_LOG(COMMANDPROCESSOR,"Interrupt set"); - ProcessorInterface::SetInterrupt(INT_CAUSE_CP, true); - } -} - void SetCpStatus() { // overflow & underflow check @@ -542,30 +525,33 @@ void SetCpStatus() fifo.bFF_LoWatermark = (fifo.CPReadWriteDistance < fifo.CPLoWatermark); // breakpoint - if (fifo.bFF_BPEnable) - { - if (fifo.CPBreakpoint == fifo.CPReadPointer) - { - if (!fifo.bFF_Breakpoint) + if (Core::IsGPUThread()) + { + if (fifo.bFF_BPEnable) + { + if (fifo.CPBreakpoint == fifo.CPReadPointer) { - INFO_LOG(COMMANDPROCESSOR, "Hit breakpoint at %i", fifo.CPReadPointer); - fifo.bFF_Breakpoint = true; - IncrementCheckContextId(); + if (!fifo.bFF_Breakpoint) + { + INFO_LOG(COMMANDPROCESSOR, "Hit breakpoint at %i", fifo.CPReadPointer); + fifo.bFF_Breakpoint = true; + IncrementCheckContextId(); + } } - } + else + { + if (fifo.bFF_Breakpoint) + INFO_LOG(COMMANDPROCESSOR, "Cleared breakpoint at %i", fifo.CPReadPointer); + fifo.bFF_Breakpoint = false; + } + } else { if (fifo.bFF_Breakpoint) INFO_LOG(COMMANDPROCESSOR, "Cleared breakpoint at %i", fifo.CPReadPointer); - fifo.bFF_Breakpoint = false; + fifo.bFF_Breakpoint = false; } - } - else - { - if (fifo.bFF_Breakpoint) - INFO_LOG(COMMANDPROCESSOR, "Cleared breakpoint at %i", fifo.CPReadPointer); - fifo.bFF_Breakpoint = false; - } + } bool bpInt = fifo.bFF_Breakpoint && fifo.bFF_BPInt; bool ovfInt = fifo.bFF_HiWatermark && fifo.bFF_HiWatermarkInt; @@ -581,10 +567,19 @@ void SetCpStatus() u64 userdata = interrupt?1:0; if (IsOnThread()) { - if(!interrupt || bpInt || undfInt) + if(!interrupt || bpInt || undfInt || ovfInt) { - interruptWaiting = true; - CommandProcessor::UpdateInterruptsFromVideoBackend(userdata); + if (Core::IsGPUThread()) + { + interruptWaiting = true; + CommandProcessor::UpdateInterruptsFromVideoBackend(userdata); + } + else if (Core::IsCPUThread()) + { + interruptSet = interrupt; + INFO_LOG(COMMANDPROCESSOR,"Interrupt set"); + ProcessorInterface::SetInterrupt(INT_CAUSE_CP, interrupt); + } } } else @@ -608,7 +603,7 @@ void ProcessFifoAllDistance() if (IsOnThread()) { while (!CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable && - fifo.CPReadWriteDistance && !AtBreakpoint() && !PixelEngine::WaitingForPEInterrupt()) + fifo.CPReadWriteDistance && !AtBreakpoint()) Common::YieldCPU(); } bProcessFifoAllDistance = false; @@ -629,8 +624,8 @@ void SetCpStatusRegister() { // Here always there is one fifo attached to the GPU m_CPStatusReg.Breakpoint = fifo.bFF_Breakpoint; - m_CPStatusReg.ReadIdle = !fifo.CPReadWriteDistance || (fifo.CPReadPointer == fifo.CPWritePointer) || (fifo.CPReadPointer == fifo.CPBreakpoint) ; - m_CPStatusReg.CommandIdle = !fifo.isGpuReadingData; + m_CPStatusReg.ReadIdle = !fifo.CPReadWriteDistance || AtBreakpoint() || (fifo.CPReadPointer == fifo.CPWritePointer); + m_CPStatusReg.CommandIdle = !fifo.CPReadWriteDistance || AtBreakpoint() || !fifo.bFF_GPReadEnable; m_CPStatusReg.UnderflowLoWatermark = fifo.bFF_LoWatermark; m_CPStatusReg.OverflowHiWatermark = fifo.bFF_HiWatermark; @@ -701,4 +696,12 @@ void SetCpClearRegister() // } } +void Update() +{ + while (VITicks > m_cpClockOrigin && fifo.isGpuReadingData && IsOnThread()) + Common::YieldCPU(); + + if (fifo.isGpuReadingData) + Common::AtomicAdd(VITicks, SystemTimers::GetTicksPerSecond() / 10000); +} } // end of namespace CommandProcessor diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.h b/Source/Core/VideoCommon/Src/CommandProcessor.h index 84b8fa7d66..6190c48e14 100644 --- a/Source/Core/VideoCommon/Src/CommandProcessor.h +++ b/Source/Core/VideoCommon/Src/CommandProcessor.h @@ -141,6 +141,9 @@ union UCPClearReg UCPClearReg(u16 _hex) {Hex = _hex; } }; +// Can be any number, low enough to not be below the number of clocks executed by the GPU per CP_PERIOD +const static u32 m_cpClockOrigin = 200000; + // Init void Init(); void Shutdown(); @@ -162,11 +165,14 @@ bool AllowIdleSkipping(); void SetCpClearRegister(); void SetCpControlRegister(); void SetCpStatusRegister(); -void SetWatermarkFromGatherPipe(); void ProcessFifoToLoWatermark(); void ProcessFifoAllDistance(); void ProcessFifoEvents(); void AbortFrame(); + +void Update(); +extern volatile u32 VITicks; + } // namespace CommandProcessor #endif // _COMMANDPROCESSOR_H diff --git a/Source/Core/VideoCommon/Src/Fifo.cpp b/Source/Core/VideoCommon/Src/Fifo.cpp index d4b4db3e42..4103f23bf2 100644 --- a/Source/Core/VideoCommon/Src/Fifo.cpp +++ b/Source/Core/VideoCommon/Src/Fifo.cpp @@ -26,6 +26,7 @@ #include "Fifo.h" #include "HW/Memmap.h" #include "Core.h" +#include "CoreTiming.h" volatile bool g_bSkipCurrentFrame = false; extern u8* g_pVideoData; @@ -72,6 +73,7 @@ void Fifo_Init() videoBuffer = (u8*)AllocateMemoryPages(FIFO_SIZE); size = 0; GpuRunningState = false; + Common::AtomicStore(CommandProcessor::VITicks, CommandProcessor::m_cpClockOrigin); } void Fifo_Shutdown() @@ -123,7 +125,7 @@ void ReadDataFromFifo(u8* _uData, u32 len) size -= pos; if (size + len > FIFO_SIZE) { - PanicAlert("FIFO out of bounds (sz = %i, at %08x)", size, pos); + PanicAlert("FIFO out of bounds (sz = %i, len = %i at %08x)", size, len, pos); } memmove(&videoBuffer[0], &videoBuffer[pos], size); g_pVideoData = videoBuffer; @@ -147,6 +149,7 @@ void RunGpuLoop() std::lock_guard lk(m_csHWVidOccupied); GpuRunningState = true; SCPFifoStruct &fifo = CommandProcessor::fifo; + u32 cyclesExecuted = 0; while (GpuRunningState) { @@ -156,6 +159,8 @@ void RunGpuLoop() CommandProcessor::SetCpStatus(); + Common::AtomicStore(CommandProcessor::VITicks, CommandProcessor::m_cpClockOrigin); + // check if we are able to run this buffer while (GpuRunningState && !CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable && fifo.CPReadWriteDistance && !AtBreakpoint()) { @@ -163,24 +168,30 @@ void RunGpuLoop() fifo.isGpuReadingData = true; CommandProcessor::isPossibleWaitingSetDrawDone = fifo.bFF_GPLinkEnable ? true : false; - - u32 readPtr = fifo.CPReadPointer; - u8 *uData = Memory::GetPointer(readPtr); - if (readPtr == fifo.CPEnd) readPtr = fifo.CPBase; + if (Common::AtomicLoad(CommandProcessor::VITicks) > CommandProcessor::m_cpClockOrigin || !Core::g_CoreStartupParameter.bSyncGPU) + { + u32 readPtr = fifo.CPReadPointer; + u8 *uData = Memory::GetPointer(readPtr); + + if (readPtr == fifo.CPEnd) readPtr = fifo.CPBase; else readPtr += 32; - - _assert_msg_(COMMANDPROCESSOR, (s32)fifo.CPReadWriteDistance - 32 >= 0 , - "Negative fifo.CPReadWriteDistance = %i in FIFO Loop !\nThat can produce inestabilty in the game. Please report it.", fifo.CPReadWriteDistance - 32); - - ReadDataFromFifo(uData, 32); - - OpcodeDecoder_Run(g_bSkipCurrentFrame); - Common::AtomicStore(fifo.CPReadPointer, readPtr); - Common::AtomicAdd(fifo.CPReadWriteDistance, -32); - if((GetVideoBufferEndPtr() - g_pVideoData) == 0) - Common::AtomicStore(fifo.SafeCPReadPointer, fifo.CPReadPointer); + _assert_msg_(COMMANDPROCESSOR, (s32)fifo.CPReadWriteDistance - 32 >= 0 , + "Negative fifo.CPReadWriteDistance = %i in FIFO Loop !\nThat can produce instabilty in the game. Please report it.", fifo.CPReadWriteDistance - 32); + + ReadDataFromFifo(uData, 32); + + cyclesExecuted = OpcodeDecoder_Run(g_bSkipCurrentFrame); + + if (Common::AtomicLoad(CommandProcessor::VITicks) > cyclesExecuted && Core::g_CoreStartupParameter.bSyncGPU) + Common::AtomicAdd(CommandProcessor::VITicks, -(s32)cyclesExecuted); + + Common::AtomicStore(fifo.CPReadPointer, readPtr); + Common::AtomicAdd(fifo.CPReadWriteDistance, -32); + if((GetVideoBufferEndPtr() - g_pVideoData) == 0) + Common::AtomicStore(fifo.SafeCPReadPointer, fifo.CPReadPointer); + } CommandProcessor::SetCpStatus(); @@ -190,7 +201,7 @@ void RunGpuLoop() VideoFifo_CheckAsyncRequest(); CommandProcessor::isPossibleWaitingSetDrawDone = false; } - + fifo.isGpuReadingData = false; @@ -219,23 +230,23 @@ bool AtBreakpoint() void RunGpu() { - SCPFifoStruct &fifo = CommandProcessor::fifo; - while (fifo.bFF_GPReadEnable && fifo.CPReadWriteDistance && !AtBreakpoint() ) - { - u8 *uData = Memory::GetPointer(fifo.CPReadPointer); + SCPFifoStruct &fifo = CommandProcessor::fifo; + while (fifo.bFF_GPReadEnable && fifo.CPReadWriteDistance && !AtBreakpoint() ) + { + u8 *uData = Memory::GetPointer(fifo.CPReadPointer); - SaveSSEState(); - LoadDefaultSSEState(); - ReadDataFromFifo(uData, 32); - OpcodeDecoder_Run(g_bSkipCurrentFrame); - LoadSSEState(); + SaveSSEState(); + LoadDefaultSSEState(); + ReadDataFromFifo(uData, 32); + u32 count = OpcodeDecoder_Run(g_bSkipCurrentFrame); + LoadSSEState(); - //DEBUG_LOG(COMMANDPROCESSOR, "Fifo wraps to base"); + //DEBUG_LOG(COMMANDPROCESSOR, "Fifo wraps to base"); - if (fifo.CPReadPointer == fifo.CPEnd) fifo.CPReadPointer = fifo.CPBase; - else fifo.CPReadPointer += 32; + if (fifo.CPReadPointer == fifo.CPEnd) fifo.CPReadPointer = fifo.CPBase; + else fifo.CPReadPointer += 32; - fifo.CPReadWriteDistance -= 32; - } - CommandProcessor::SetCpStatus(); + fifo.CPReadWriteDistance -= 32; + } + CommandProcessor::SetCpStatus(); } diff --git a/Source/Core/VideoCommon/Src/OpcodeDecoding.cpp b/Source/Core/VideoCommon/Src/OpcodeDecoding.cpp index 4f7f86d655..662dbc66a8 100644 --- a/Source/Core/VideoCommon/Src/OpcodeDecoding.cpp +++ b/Source/Core/VideoCommon/Src/OpcodeDecoding.cpp @@ -136,29 +136,38 @@ void ExecuteDisplayList(u32 address, u32 size) InterpretDisplayList(address, size); } -bool FifoCommandRunnable() +u32 FifoCommandRunnable(u32 &command_size) { + u32 cycleTime = 0; u32 buffer_size = (u32)(GetVideoBufferEndPtr() - g_pVideoData); if (buffer_size == 0) - return false; // can't peek + return 0; // can't peek u8 cmd_byte = DataPeek8(0); - u32 command_size = 0; switch (cmd_byte) { case GX_NOP: // Hm, this means that we scan over nop streams pretty slowly... + command_size = 1; + cycleTime = 6; + break; case GX_CMD_INVL_VC: // Invalidate Vertex Cache - no parameters + command_size = 1; + cycleTime = 6; + break; case GX_CMD_UNKNOWN_METRICS: // zelda 4 swords calls it and checks the metrics registers after that command_size = 1; + cycleTime = 6; break; case GX_LOAD_BP_REG: command_size = 5; + cycleTime = 12; break; case GX_LOAD_CP_REG: command_size = 6; + cycleTime = 12; break; case GX_LOAD_INDX_A: @@ -166,10 +175,39 @@ bool FifoCommandRunnable() case GX_LOAD_INDX_C: case GX_LOAD_INDX_D: command_size = 5; + cycleTime = 6; // TODO break; - case GX_CMD_CALL_DL: - command_size = 9; + case GX_CMD_CALL_DL: + { + // FIXME: Calculate the cycle time of the display list. + //u32 address = DataPeek32(1); + //u32 size = DataPeek32(5); + //u8* old_pVideoData = g_pVideoData; + //u8* startAddress = Memory::GetPointer(address); + + //// Avoid the crash if Memory::GetPointer failed .. + //if (startAddress != 0) + //{ + // g_pVideoData = startAddress; + // u8 *end = g_pVideoData + size; + // u32 step = 0; + // while (g_pVideoData < end) + // { + // cycleTime += FifoCommandRunnable(step); + // g_pVideoData += step; + // } + //} + //else + //{ + // cycleTime = 45; + //} + + //// reset to the old pointer + //g_pVideoData = old_pVideoData; + command_size = 9; + cycleTime = 45; // This is unverified + } break; case GX_LOAD_XF_REG: @@ -180,11 +218,12 @@ bool FifoCommandRunnable() command_size = 1 + 4; u32 Cmd2 = DataPeek32(1); int transfer_size = ((Cmd2 >> 16) & 15) + 1; - command_size += transfer_size * 4; + command_size += transfer_size * 4; + cycleTime = 18 + 6 * transfer_size; } else { - return false; + return 0; } } break; @@ -198,10 +237,11 @@ bool FifoCommandRunnable() command_size = 1 + 2; u16 numVertices = DataPeek16(1); command_size += numVertices * VertexLoaderManager::GetVertexSize(cmd_byte & GX_VAT_MASK); + cycleTime = 12 * numVertices; // This depends on the number of pixels rendered } else { - return false; + return 0; } } else @@ -248,11 +288,19 @@ bool FifoCommandRunnable() } if (command_size > buffer_size) - return false; + return 0; // INFO_LOG("OP detected: cmd_byte 0x%x size %i buffer %i",cmd_byte, command_size, buffer_size); + if (cycleTime == 0) + cycleTime = 6; - return true; + return cycleTime; +} + +u32 FifoCommandRunnable() +{ + u32 command_size = 0; + return FifoCommandRunnable(command_size); } static void Decode() @@ -461,16 +509,15 @@ void OpcodeDecoder_Shutdown() } } -void OpcodeDecoder_Run(bool skipped_frame) +u32 OpcodeDecoder_Run(bool skipped_frame) { - if (!skipped_frame) + u32 totalCycles = 0; + u32 cycles = FifoCommandRunnable(); + while (cycles > 0) { - while (FifoCommandRunnable()) - Decode(); - } - else - { - while (FifoCommandRunnable()) - DecodeSemiNop(); + skipped_frame ? DecodeSemiNop() : Decode(); + totalCycles += cycles; + cycles = FifoCommandRunnable(); } + return totalCycles; } diff --git a/Source/Core/VideoCommon/Src/OpcodeDecoding.h b/Source/Core/VideoCommon/Src/OpcodeDecoding.h index f2e9cd1321..71a15477a3 100644 --- a/Source/Core/VideoCommon/Src/OpcodeDecoding.h +++ b/Source/Core/VideoCommon/Src/OpcodeDecoding.h @@ -50,6 +50,6 @@ extern bool g_bRecordFifoData; void OpcodeDecoder_Init(); void OpcodeDecoder_Shutdown(); -void OpcodeDecoder_Run(bool skipped_frame); +u32 OpcodeDecoder_Run(bool skipped_frame); void ExecuteDisplayList(u32 address, u32 size); #endif // _OPCODE_DECODING_H diff --git a/Source/Core/VideoCommon/Src/PixelEngine.cpp b/Source/Core/VideoCommon/Src/PixelEngine.cpp index 33fedd1a81..24e8a02ab4 100644 --- a/Source/Core/VideoCommon/Src/PixelEngine.cpp +++ b/Source/Core/VideoCommon/Src/PixelEngine.cpp @@ -450,9 +450,4 @@ void ResetSetToken() } CommandProcessor::interruptTokenWaiting = false; } - -bool WaitingForPEInterrupt() -{ - return !CommandProcessor::waitingForPEInterruptDisable && (CommandProcessor::interruptFinishWaiting || CommandProcessor::interruptTokenWaiting || Common::AtomicLoad(interruptSetFinish) || Common::AtomicLoad(interruptSetToken)); -} } // end of namespace PixelEngine diff --git a/Source/Core/VideoCommon/Src/PixelEngine.h b/Source/Core/VideoCommon/Src/PixelEngine.h index 5e64300ef3..4da0f938d8 100644 --- a/Source/Core/VideoCommon/Src/PixelEngine.h +++ b/Source/Core/VideoCommon/Src/PixelEngine.h @@ -80,7 +80,6 @@ void SetToken(const u16 _token, const int _bSetTokenAcknowledge); void SetFinish(void); void ResetSetFinish(void); void ResetSetToken(void); -bool WaitingForPEInterrupt(); // Bounding box functionality. Paper Mario (both) are a couple of the few games that use it. extern u16 bbox[4];