diff options
Diffstat (limited to 'Source/Core/VideoCommon')
24 files changed, 404 insertions, 329 deletions
diff --git a/Source/Core/VideoCommon/SConscript b/Source/Core/VideoCommon/SConscript deleted file mode 100644 index 5499f422db..0000000000 --- a/Source/Core/VideoCommon/SConscript +++ /dev/null @@ -1,48 +0,0 @@ -# -*- python -*- - -Import('env') - -files = [ - 'Src/BPFunctions.cpp', - 'Src/BPMemory.cpp', - 'Src/BPStructs.cpp', - 'Src/CPMemory.cpp', - 'Src/CommandProcessor.cpp', - 'Src/DLCache.cpp', - 'Src/Debugger.cpp', - 'Src/Fifo.cpp', - 'Src/FramebufferManagerBase.cpp', - 'Src/HiresTextures.cpp', - 'Src/ImageWrite.cpp', - 'Src/IndexGenerator.cpp', - 'Src/LightingShaderGen.cpp', - 'Src/MainBase.cpp', - 'Src/OnScreenDisplay.cpp', - 'Src/OpcodeDecoding.cpp', - 'Src/OpenCL.cpp', - 'Src/OpenCL/OCLTextureDecoder.cpp', - 'Src/PixelEngine.cpp', - 'Src/PixelShaderGen.cpp', - 'Src/PixelShaderManager.cpp', - 'Src/RenderBase.cpp', - 'Src/Statistics.cpp', - 'Src/TextureCacheBase.cpp', - 'Src/TextureConversionShader.cpp', - 'Src/TextureDecoder.cpp', - 'Src/VertexLoader.cpp', - 'Src/VertexLoaderManager.cpp', - 'Src/VertexLoader_Color.cpp', - 'Src/VertexLoader_Normal.cpp', - 'Src/VertexLoader_Position.cpp', - 'Src/VertexLoader_TextCoord.cpp', - 'Src/VertexManagerBase.cpp', - 'Src/VertexShaderGen.cpp', - 'Src/VertexShaderManager.cpp', - 'Src/VideoConfig.cpp', - 'Src/VideoState.cpp', - 'Src/XFMemory.cpp', - 'Src/XFStructs.cpp', - 'Src/memcpy_amd.cpp', - ] - -env['LIBS'] += env.StaticLibrary('videocommon', files) diff --git a/Source/Core/VideoCommon/Src/AVIDump.cpp b/Source/Core/VideoCommon/Src/AVIDump.cpp index 670693690b..4c31798d9f 100644 --- a/Source/Core/VideoCommon/Src/AVIDump.cpp +++ b/Source/Core/VideoCommon/Src/AVIDump.cpp @@ -211,6 +211,7 @@ extern "C" { #include <libavcodec/avcodec.h> #include <libavformat/avformat.h> #include <libswscale/swscale.h> +#include <libavutil/mathematics.h> } AVFormatContext *s_FormatContext = NULL; diff --git a/Source/Core/VideoCommon/Src/BPMemory.cpp b/Source/Core/VideoCommon/Src/BPMemory.cpp index 5c4ed457e3..d5b762c1a4 100644 --- a/Source/Core/VideoCommon/Src/BPMemory.cpp +++ b/Source/Core/VideoCommon/Src/BPMemory.cpp @@ -66,3 +66,113 @@ void BPReload() } } } + + +void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size_t desc_size) +{ + const char* no_yes[2] = { "No", "Yes" }; + + u32 cmddata = Common::swap32(*(u32*)data) & 0xFFFFFF; + switch (data[0]) + { + // Macro to set the register name and make sure it was written correctly via compile time assertion + #define SetRegName(reg) \ + snprintf(name, name_size, #reg); \ + (void)(reg); + + case BPMEM_DISPLAYCOPYFILER: // 0x01 + // TODO: This is actually the sample pattern used for copies from an antialiased EFB + SetRegName(BPMEM_DISPLAYCOPYFILER); + // TODO: Description + break; + + case 0x02: // 0x02 + case 0x03: // 0x03 + case 0x04: // 0x04 + // TODO: same as BPMEM_DISPLAYCOPYFILER + break; + + case BPMEM_EFB_TL: // 0x49 + { + SetRegName(BPMEM_EFB_TL); + X10Y10 left_top; left_top.hex = cmddata; + snprintf(desc, desc_size, "Left: %d\nTop: %d", left_top.x, left_top.y); + } + break; + + case BPMEM_EFB_BR: // 0x4A + { + // TODO: Misleading name, should be BPMEM_EFB_WH instead + SetRegName(BPMEM_EFB_BR); + X10Y10 width_height; width_height.hex = cmddata; + snprintf(desc, desc_size, "Width: %d\nHeight: %d", width_height.x+1, width_height.y+1); + } + break; + + case BPMEM_EFB_ADDR: // 0x4B + SetRegName(BPMEM_EFB_ADDR); + snprintf(desc, desc_size, "Target address (32 byte aligned): 0x%06X", cmddata << 5); + break; + + case BPMEM_COPYYSCALE: // 0x4E + SetRegName(BPMEM_COPYYSCALE); + snprintf(desc, desc_size, "Scaling factor (XFB copy only): 0x%X (%f or inverted %f)", cmddata, (float)cmddata/256.f, 256.f/(float)cmddata); + break; + + case BPMEM_CLEAR_AR: // 0x4F + SetRegName(BPMEM_CLEAR_AR); + snprintf(desc, desc_size, "Alpha: 0x%02X\nRed: 0x%02X", (cmddata&0xFF00)>>8, cmddata&0xFF); + break; + + case BPMEM_CLEAR_GB: // 0x50 + SetRegName(BPMEM_CLEAR_GB); + snprintf(desc, desc_size, "Green: 0x%02X\nBlue: 0x%02X", (cmddata&0xFF00)>>8, cmddata&0xFF); + break; + + case BPMEM_CLEAR_Z: // 0x51 + SetRegName(BPMEM_CLEAR_Z); + snprintf(desc, desc_size, "Z value: 0x%06X", cmddata); + break; + + case BPMEM_TRIGGER_EFB_COPY: // 0x52 + { + SetRegName(BPMEM_TRIGGER_EFB_COPY); + UPE_Copy copy; copy.Hex = cmddata; + snprintf(desc, desc_size, "Clamping: %s\n" + "Converting from RGB to YUV: %s\n" + "Target pixel format: 0x%X\n" + "Gamma correction: %s\n" + "Mipmap filter: %s\n" + "Vertical scaling: %s\n" + "Clear: %s\n" + "Frame to field: 0x%01X\n" + "Copy to XFB: %s\n" + "Intensity format: %s\n" + "Automatic color conversion: %s", + (copy.clamp0 && copy.clamp1) ? "Top and Bottom" : (copy.clamp0) ? "Top only" : (copy.clamp1) ? "Bottom only" : "None", + no_yes[copy.yuv], + copy.tp_realFormat(), + (copy.gamma==0)?"1.0":(copy.gamma==1)?"1.7":(copy.gamma==2)?"2.2":"Invalid value 0x3?", + no_yes[copy.half_scale], + no_yes[copy.scale_invert], + no_yes[copy.clear], + copy.frame_to_field, + no_yes[copy.copy_to_xfb], + no_yes[copy.intensity_fmt], + no_yes[copy.auto_conv]); + } + break; + + case BPMEM_COPYFILTER0: // 0x53 + SetRegName(BPMEM_COPYFILTER0); + // TODO: Description + break; + + case BPMEM_COPYFILTER1: // 0x54 + SetRegName(BPMEM_COPYFILTER1); + // TODO: Description + break; + +#undef SET_REG_NAME + } +} diff --git a/Source/Core/VideoCommon/Src/BPMemory.h b/Source/Core/VideoCommon/Src/BPMemory.h index 0bb2bf9013..9284674aa1 100644 --- a/Source/Core/VideoCommon/Src/BPMemory.h +++ b/Source/Core/VideoCommon/Src/BPMemory.h @@ -62,7 +62,7 @@ #define BPMEM_COPYFILTER1 0x54 #define BPMEM_CLEARBBOX1 0x55 #define BPMEM_CLEARBBOX2 0x56 -#define BPMEM_UNKOWN_57 0x57 +#define BPMEM_UNKNOWN_57 0x57 #define BPMEM_REVBITS 0x58 #define BPMEM_SCISSOROFFSET 0x59 #define BPMEM_PRELOAD_ADDR 0x60 @@ -995,4 +995,6 @@ extern BPMemory bpmem; void LoadBPReg(u32 value0); +void GetBPRegInfo(const u8* data, char* name, size_t name_size, char* desc, size_t desc_size); + #endif // _BPMEMORY_H diff --git a/Source/Core/VideoCommon/Src/BPStructs.cpp b/Source/Core/VideoCommon/Src/BPStructs.cpp index 6cb102555b..d59523e117 100644 --- a/Source/Core/VideoCommon/Src/BPStructs.cpp +++ b/Source/Core/VideoCommon/Src/BPStructs.cpp @@ -463,8 +463,8 @@ void BPWritten(const BPCmd& bp) case BPMEM_REVBITS: // Always set to 0x0F when GX_InitRevBits() is called. break; - case BPMEM_UNKOWN_57: // Sunshine alternates this register between values 0x000 and 0xAAA - DEBUG_LOG(VIDEO, "Uknown BP Reg 0x57: %08x", bp.newvalue); + case BPMEM_UNKNOWN_57: // Sunshine alternates this register between values 0x000 and 0xAAA + DEBUG_LOG(VIDEO, "Unknown BP Reg 0x57: %08x", bp.newvalue); break; case BPMEM_PRELOAD_ADDR: @@ -481,6 +481,13 @@ void BPWritten(const BPCmd& bp) u8* ram_ptr = Memory::GetPointer(tmem_cfg.preload_addr << 5); u32 tmem_addr = tmem_cfg.preload_tmem_even * TMEM_LINE_SIZE; u32 size = tmem_cfg.preload_tile_info.count * 32; + + // Check if the game has overflowed TMEM, and copy up to the limit. + // Paper Mario does this when entering the Great Boogly Tree (Chap 2) + // TODO: Does this wrap? + if ((tmem_addr + size) > TMEM_SIZE) + size = TMEM_SIZE - tmem_addr; + memcpy(texMem + tmem_addr, ram_ptr, size); } break; diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.cpp b/Source/Core/VideoCommon/Src/CommandProcessor.cpp index fc85b8e39d..a4dddf537f 100644 --- a/Source/Core/VideoCommon/Src/CommandProcessor.cpp +++ b/Source/Core/VideoCommon/Src/CommandProcessor.cpp @@ -56,11 +56,12 @@ static bool bProcessFifoToLoWatermark = false; static bool bProcessFifoAllDistance = false; volatile bool isPossibleWaitingSetDrawDone = false; +volatile bool isHiWatermarkActive = false; volatile bool interruptSet= false; volatile bool interruptWaiting= false; volatile bool interruptTokenWaiting = false; volatile bool interruptFinishWaiting = false; -volatile bool OnOverflow = false; +volatile bool waitingForPEInterruptDisable = false; bool IsOnThread() { @@ -86,13 +87,12 @@ void DoState(PointerWrap &p) p.Do(bProcessFifoToLoWatermark); p.Do(bProcessFifoAllDistance); - + p.Do(isHiWatermarkActive); p.Do(isPossibleWaitingSetDrawDone); p.Do(interruptSet); p.Do(interruptWaiting); p.Do(interruptTokenWaiting); p.Do(interruptFinishWaiting); - p.Do(OnOverflow); } inline void WriteLow (volatile u32& _reg, u16 lowbits) {Common::AtomicStore(_reg,(_reg & 0xFFFF0000) | lowbits);} @@ -135,16 +135,14 @@ void Init() bProcessFifoToLoWatermark = false; bProcessFifoAllDistance = false; isPossibleWaitingSetDrawDone = false; - OnOverflow = false; + isHiWatermarkActive = false; et_UpdateInterrupts = CoreTiming::RegisterEvent("UpdateInterrupts", UpdateInterrupts_Wrapper); } void Read16(u16& _rReturnValue, const u32 _Address) { - INFO_LOG(COMMANDPROCESSOR, "(r): 0x%08x", _Address); - ProcessFifoEvents(); switch (_Address & 0xFFF) { case STATUS_REGISTER: @@ -173,11 +171,23 @@ void Read16(u16& _rReturnValue, const u32 _Address) case FIFO_LO_WATERMARK_HI: _rReturnValue = ReadHigh(fifo.CPLoWatermark); return; case FIFO_RW_DISTANCE_LO: - _rReturnValue = ReadLow (fifo.CPReadWriteDistance); + if (IsOnThread()) + if(fifo.CPWritePointer >= fifo.SafeCPReadPointer) + _rReturnValue = ReadLow (fifo.CPWritePointer - fifo.SafeCPReadPointer); + else + _rReturnValue = ReadLow (fifo.CPEnd - fifo.SafeCPReadPointer + fifo.CPWritePointer - fifo.CPBase + 32); + else + _rReturnValue = ReadLow (fifo.CPReadWriteDistance); DEBUG_LOG(COMMANDPROCESSOR, "read FIFO_RW_DISTANCE_LO : %04x", _rReturnValue); return; case FIFO_RW_DISTANCE_HI: - _rReturnValue = ReadHigh(fifo.CPReadWriteDistance); + if (IsOnThread()) + if(fifo.CPWritePointer >= fifo.SafeCPReadPointer) + _rReturnValue = ReadHigh (fifo.CPWritePointer - fifo.SafeCPReadPointer); + else + _rReturnValue = ReadHigh (fifo.CPEnd - fifo.SafeCPReadPointer + fifo.CPWritePointer - fifo.CPBase + 32); + else + _rReturnValue = ReadHigh(fifo.CPReadWriteDistance); DEBUG_LOG(COMMANDPROCESSOR, "read FIFO_RW_DISTANCE_HI : %04x", _rReturnValue); return; case FIFO_WRITE_POINTER_LO: @@ -358,6 +368,7 @@ void Write16(const u16 _Value, const u32 _Address) break; case FIFO_READ_POINTER_HI: WriteHigh((u32 &)fifo.CPReadPointer, _Value); + fifo.SafeCPReadPointer = fifo.CPReadPointer; DEBUG_LOG(COMMANDPROCESSOR,"\t write to FIFO_READ_POINTER_HI : %04x", _Value); break; @@ -390,10 +401,6 @@ void Write16(const u16 _Value, const u32 _Address) case FIFO_RW_DISTANCE_HI: WriteHigh((u32 &)fifo.CPReadWriteDistance, _Value); - DEBUG_LOG(COMMANDPROCESSOR,"try to write to FIFO_RW_DISTANCE_HI : %04x", _Value); - break; - case FIFO_RW_DISTANCE_LO: - WriteLow((u32 &)fifo.CPReadWriteDistance, _Value & 0xFFE0); if (fifo.CPReadWriteDistance == 0) { GPFifo::ResetGatherPipe(); @@ -403,6 +410,10 @@ void Write16(const u16 _Value, const u32 _Address) ResetVideoBuffer(); } IncrementCheckContextId(); + DEBUG_LOG(COMMANDPROCESSOR,"try to write to FIFO_RW_DISTANCE_HI : %04x", _Value); + break; + case FIFO_RW_DISTANCE_LO: + WriteLow((u32 &)fifo.CPReadWriteDistance, _Value & 0xFFE0); DEBUG_LOG(COMMANDPROCESSOR,"try to write to FIFO_RW_DISTANCE_LO : %04x", _Value); break; @@ -412,7 +423,6 @@ void Write16(const u16 _Value, const u32 _Address) if (!IsOnThread()) RunGpu(); - ProcessFifoEvents(); } void Read32(u32& _rReturnValue, const u32 _Address) @@ -434,6 +444,19 @@ void STACKALIGN GatherPipeBursted() { if (!IsOnThread()) RunGpu(); + else + { + // In multibuffer mode is not allowed write in the same fifo attached to the GPU. + // Fix Pokemon XD in DC mode. + if((ProcessorInterface::Fifo_CPUEnd == fifo.CPEnd) && (ProcessorInterface::Fifo_CPUBase == fifo.CPBase) + && fifo.CPReadWriteDistance > 0) + { + waitingForPEInterruptDisable = true; + ProcessFifoAllDistance(); + waitingForPEInterruptDisable = false; + } + + } return; } @@ -449,26 +472,7 @@ void STACKALIGN GatherPipeBursted() Common::AtomicAdd(fifo.CPReadWriteDistance, GATHER_PIPE_SIZE); if (!IsOnThread()) - { RunGpu(); - } - else - { - if(fifo.CPReadWriteDistance == fifo.CPEnd - fifo.CPBase - 32) - { - if(!OnOverflow) - NOTICE_LOG(COMMANDPROCESSOR,"FIFO is almost in overflown, BreakPoint: %i", fifo.bFF_Breakpoint); - OnOverflow = true; - while (!CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable && - fifo.CPReadWriteDistance > fifo.CPEnd - fifo.CPBase - 64) - Common::YieldCPU(); - } - else - { - OnOverflow = false; - } - } - _assert_msg_(COMMANDPROCESSOR, fifo.CPReadWriteDistance <= fifo.CPEnd - fifo.CPBase, "FIFO is overflown by GatherPipe !\nCPU thread is too fast!"); @@ -509,17 +513,15 @@ void AbortFrame() void SetOverflowStatusFromGatherPipe() { - if (!fifo.bFF_HiWatermarkInt) return; - fifo.bFF_HiWatermark = (fifo.CPReadWriteDistance > fifo.CPHiWatermark); - fifo.bFF_LoWatermark = (fifo.CPReadWriteDistance < fifo.CPLoWatermark); - - bool interrupt = fifo.bFF_HiWatermark && fifo.bFF_HiWatermarkInt && - m_CPCtrlReg.GPLinkEnable && m_CPCtrlReg.GPReadEnable; + isHiWatermarkActive = fifo.bFF_HiWatermark && fifo.bFF_HiWatermarkInt && m_CPCtrlReg.GPReadEnable; - if (interrupt != interruptSet && interrupt) - CommandProcessor::UpdateInterrupts(true); - + if (isHiWatermarkActive) + { + interruptSet = true; + INFO_LOG(COMMANDPROCESSOR,"Interrupt set"); + ProcessorInterface::SetInterrupt(INT_CAUSE_CP, true); + } } void SetCpStatus() @@ -527,14 +529,12 @@ void SetCpStatus() // overflow & underflow check fifo.bFF_HiWatermark = (fifo.CPReadWriteDistance > fifo.CPHiWatermark); fifo.bFF_LoWatermark = (fifo.CPReadWriteDistance < fifo.CPLoWatermark); - - // breakpoint + // breakpoint if (fifo.bFF_BPEnable) { if (fifo.CPBreakpoint == fifo.CPReadPointer) - { - + { if (!fifo.bFF_Breakpoint) { INFO_LOG(COMMANDPROCESSOR, "Hit breakpoint at %i", fifo.CPReadPointer); @@ -562,13 +562,18 @@ void SetCpStatus() bool interrupt = (bpInt || ovfInt || undfInt) && m_CPCtrlReg.GPReadEnable; + isHiWatermarkActive = ovfInt && m_CPCtrlReg.GPReadEnable; + if (interrupt != interruptSet && !interruptWaiting) { u64 userdata = interrupt?1:0; if (IsOnThread()) { - interruptWaiting = true; - CommandProcessor::UpdateInterruptsFromVideoBackend(userdata); + if(!interrupt || bpInt || undfInt) + { + interruptWaiting = true; + CommandProcessor::UpdateInterruptsFromVideoBackend(userdata); + } } else CommandProcessor::UpdateInterrupts(userdata); @@ -591,7 +596,7 @@ void ProcessFifoAllDistance() if (IsOnThread()) { while (!CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable && - fifo.CPReadWriteDistance && !AtBreakpoint()) + fifo.CPReadWriteDistance && !AtBreakpoint() && !PixelEngine::WaitingForPEInterrupt()) Common::YieldCPU(); } bProcessFifoAllDistance = false; @@ -611,13 +616,16 @@ void Shutdown() void SetCpStatusRegister() { // Here always there is one fifo attached to the GPU - m_CPStatusReg.Breakpoint = fifo.bFF_Breakpoint; - m_CPStatusReg.ReadIdle = (fifo.CPReadPointer == fifo.CPWritePointer) || (fifo.CPReadPointer == fifo.CPBreakpoint); + m_CPStatusReg.ReadIdle = !fifo.CPReadWriteDistance || (fifo.CPReadPointer == fifo.CPWritePointer) || (fifo.CPReadPointer == fifo.CPBreakpoint) ; m_CPStatusReg.CommandIdle = !fifo.CPReadWriteDistance; m_CPStatusReg.UnderflowLoWatermark = fifo.bFF_LoWatermark; m_CPStatusReg.OverflowHiWatermark = fifo.bFF_HiWatermark; + // HACK to compensate for slow response to PE interrupts in Time Splitters: Future Perfect + if (IsOnThread()) + PixelEngine::ResumeWaitingForPEInterrupt(); + INFO_LOG(COMMANDPROCESSOR,"\t Read from STATUS_REGISTER : %04x", m_CPStatusReg.Hex); DEBUG_LOG(COMMANDPROCESSOR, "(r) status: iBP %s | fReadIdle %s | fCmdIdle %s | iOvF %s | iUndF %s" , m_CPStatusReg.Breakpoint ? "ON" : "OFF" @@ -630,14 +638,14 @@ void SetCpStatusRegister() void SetCpControlRegister() { - // If the new fifo is being attached We make sure there wont be SetFinish event pending. // This protection fix eternal darkness booting, because the second SetFinish event when it is booting // seems invalid or has a bug and hang the game. if (!fifo.bFF_GPReadEnable && m_CPCtrlReg.GPReadEnable && !m_CPCtrlReg.BPEnable) { - PixelEngine::ResetSetFinish(); + ProcessFifoEvents(); + PixelEngine::ResetSetFinish(); } fifo.bFF_BPInt = m_CPCtrlReg.BPInt; @@ -652,9 +660,6 @@ void SetCpControlRegister() ProcessorInterface::Fifo_CPUBase = fifo.CPBase; ProcessorInterface::Fifo_CPUEnd = fifo.CPEnd; } - // If overflown happens process the fifo to LoWatemark - if (bProcessFifoToLoWatermark) - ProcessFifoToLoWatermark(); if(fifo.bFF_GPReadEnable && !m_CPCtrlReg.GPReadEnable) { @@ -666,7 +671,6 @@ void SetCpControlRegister() fifo.bFF_GPReadEnable = m_CPCtrlReg.GPReadEnable; } - DEBUG_LOG(COMMANDPROCESSOR, "\t GPREAD %s | BP %s | Int %s | OvF %s | UndF %s | LINK %s" , fifo.bFF_GPReadEnable ? "ON" : "OFF" , fifo.bFF_BPEnable ? "ON" : "OFF" diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.h b/Source/Core/VideoCommon/Src/CommandProcessor.h index db6772d66d..5d31453537 100644 --- a/Source/Core/VideoCommon/Src/CommandProcessor.h +++ b/Source/Core/VideoCommon/Src/CommandProcessor.h @@ -25,18 +25,18 @@ class PointerWrap; extern bool MT; - namespace CommandProcessor { extern SCPFifoStruct fifo; //This one is shared between gfx thread and emulator thread. extern volatile bool isPossibleWaitingSetDrawDone; //This one is used for sync gfx thread and emulator thread. +extern volatile bool isHiWatermarkActive; extern volatile bool interruptSet; extern volatile bool interruptWaiting; extern volatile bool interruptTokenWaiting; extern volatile bool interruptFinishWaiting; -extern volatile bool OnOverflow; - +extern volatile bool waitingForPEInterruptDisable; + // internal hardware addresses enum { diff --git a/Source/Core/VideoCommon/Src/Fifo.cpp b/Source/Core/VideoCommon/Src/Fifo.cpp index 842ff49e78..2e60055c97 100644 --- a/Source/Core/VideoCommon/Src/Fifo.cpp +++ b/Source/Core/VideoCommon/Src/Fifo.cpp @@ -22,6 +22,7 @@ #include "Atomic.h" #include "OpcodeDecoding.h" #include "CommandProcessor.h" +#include "PixelEngine.h" #include "ChunkFile.h" #include "Fifo.h" #include "HW/Memmap.h" @@ -137,8 +138,7 @@ void RunGpuLoop() CommandProcessor::SetCpStatus(); // check if we are able to run this buffer - while (!CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable && - fifo.CPReadWriteDistance && (!AtBreakpoint() || CommandProcessor::OnOverflow)) + while (GpuRunningState && !CommandProcessor::interruptWaiting && fifo.bFF_GPReadEnable && fifo.CPReadWriteDistance && !AtBreakpoint() && !PixelEngine::WaitingForPEInterrupt()) { if (!GpuRunningState) break; diff --git a/Source/Core/VideoCommon/Src/MainBase.cpp b/Source/Core/VideoCommon/Src/MainBase.cpp index af21ebbb94..0b1258a662 100644 --- a/Source/Core/VideoCommon/Src/MainBase.cpp +++ b/Source/Core/VideoCommon/Src/MainBase.cpp @@ -250,6 +250,11 @@ bool VideoBackendHardware::Video_IsPossibleWaitingSetDrawDone() return CommandProcessor::isPossibleWaitingSetDrawDone; } +bool VideoBackendHardware::Video_IsHiWatermarkActive() +{ + return CommandProcessor::isHiWatermarkActive; +} + void VideoBackendHardware::Video_AbortFrame() { CommandProcessor::AbortFrame(); diff --git a/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp b/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp index 0f4809a6cc..440e69efd6 100644 --- a/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp +++ b/Source/Core/VideoCommon/Src/OpenCL/OCLTextureDecoder.cpp @@ -114,7 +114,7 @@ void TexDecoder_OpenCL_Initialize() else { binary_size = input.GetSize(); - header = new char[HEADER_SIZE]; // TODO: memleak possible + header = new char[HEADER_SIZE]; binary = new char[binary_size]; input.ReadBytes(header, HEADER_SIZE); input.ReadBytes(binary, binary_size); @@ -143,8 +143,8 @@ void TexDecoder_OpenCL_Initialize() } } } - delete header; - delete binary; + delete [] header; + delete [] binary; } // If an error occurred using the kernel binary, recompile the kernels diff --git a/Source/Core/VideoCommon/Src/PixelEngine.cpp b/Source/Core/VideoCommon/Src/PixelEngine.cpp index 456e0fb535..03a3a7547a 100644 --- a/Source/Core/VideoCommon/Src/PixelEngine.cpp +++ b/Source/Core/VideoCommon/Src/PixelEngine.cpp @@ -180,7 +180,6 @@ void Init() void Read16(u16& _uReturnValue, const u32 _iAddress) { DEBUG_LOG(PIXELENGINE, "(r16) 0x%08x", _iAddress); - CommandProcessor::ProcessFifoEvents(); switch (_iAddress & 0xFFF) { // CPU Direct Access EFB Raster State Config @@ -269,6 +268,10 @@ void Read16(u16& _uReturnValue, const u32 _iAddress) case PE_PERF_5L: case PE_PERF_5H: INFO_LOG(PIXELENGINE, "(r16) perf counter @ %08x", _iAddress); + // git r90a2096a24f4 (svn r3663) added the PE_PERF cases, without setting + // _uReturnValue to anything, this reverts to the previous behaviour which allows + // The timer in SMS:Scrubbing Serena Beach to countdown correctly + _uReturnValue = 1; break; default: @@ -323,7 +326,6 @@ void Write16(const u16 _iValue, const u32 _iAddress) break; case PE_TOKEN_REG: - //LOG(PIXELENGINE,"WEIRD: program wrote token: %i",_iValue); PanicAlert("(w16) WTF? PowerPC program wrote token: %i", _iValue); //only the gx pipeline is supposed to be able to write here //g_token = _iValue; @@ -334,7 +336,6 @@ void Write16(const u16 _iValue, const u32 _iAddress) break; } - CommandProcessor::ProcessFifoEvents(); } void Write32(const u32 _iValue, const u32 _iAddress) @@ -358,22 +359,16 @@ void UpdateInterrupts() void UpdateTokenInterrupt(bool active) { - if(interruptSetToken != active) - { ProcessorInterface::SetInterrupt(INT_CAUSE_PE_TOKEN, active); interruptSetToken = active; - } } void UpdateFinishInterrupt(bool active) { - if(interruptSetFinish != active) - { ProcessorInterface::SetInterrupt(INT_CAUSE_PE_FINISH, active); interruptSetFinish = active; if (active) State::ProcessRequestedStates(0); - } } // TODO(mb2): Refactor SetTokenINT_OnMainThread(u64 userdata, int cyclesLate). @@ -392,8 +387,6 @@ void SetToken_OnMainThread(u64 userdata, int cyclesLate) CommandProcessor::interruptTokenWaiting = false; IncrementCheckContextId(); //} - //else - // LOGV(PIXELENGINE, 1, "VIDEO Backend wrote token: %i", CommandProcessor::fifo.PEToken); } void SetFinish_OnMainThread(u64 userdata, int cyclesLate) @@ -470,4 +463,17 @@ void ResetSetToken() } CommandProcessor::interruptTokenWaiting = false; } + +bool WaitingForPEInterrupt() +{ + return !CommandProcessor::waitingForPEInterruptDisable && (CommandProcessor::interruptFinishWaiting || CommandProcessor::interruptTokenWaiting || interruptSetFinish || interruptSetToken); +} + +void ResumeWaitingForPEInterrupt() +{ + interruptSetFinish = false; + interruptSetToken = false; + CommandProcessor::interruptFinishWaiting = false; + CommandProcessor::interruptTokenWaiting = false; +} } // end of namespace PixelEngine diff --git a/Source/Core/VideoCommon/Src/PixelEngine.h b/Source/Core/VideoCommon/Src/PixelEngine.h index dd0304fbe1..64f959009f 100644 --- a/Source/Core/VideoCommon/Src/PixelEngine.h +++ b/Source/Core/VideoCommon/Src/PixelEngine.h @@ -80,7 +80,8 @@ void SetToken(const u16 _token, const int _bSetTokenAcknowledge); void SetFinish(void); void ResetSetFinish(void); void ResetSetToken(void); -bool AllowIdleSkipping(); +bool WaitingForPEInterrupt(); +void ResumeWaitingForPEInterrupt(); // Bounding box functionality. Paper Mario (both) are a couple of the few games that use it. extern u16 bbox[4]; diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp index fe5827de05..7500997fef 100644 --- a/Source/Core/VideoCommon/Src/PixelShaderGen.cpp +++ b/Source/Core/VideoCommon/Src/PixelShaderGen.cpp @@ -158,7 +158,7 @@ void GetPixelShaderId(PIXELSHADERUID *uid, DSTALPHA_MODE dstAlphaMode, u32 compo } u32* ptr = &uid->values[2]; - for (int i = 0; i < bpmem.genMode.numtevstages+1; ++i) + for (unsigned int i = 0; i < bpmem.genMode.numtevstages+1; ++i) { StageHash(i, ptr); ptr += 4; // max: ptr = &uid->values[66] @@ -299,7 +299,7 @@ void ValidatePixelShaderIDs(API_TYPE api, PIXELSHADERUIDSAFE old_id, const std:: static void WriteStage(char *&p, int n, API_TYPE ApiType); static void SampleTexture(char *&p, const char *destination, const char *texcoords, const char *texswap, int texmap, API_TYPE ApiType); // static void WriteAlphaCompare(char *&p, int num, int comp); -static bool WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode); +static void WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode); static void WriteFog(char *&p); static const char *tevKSelTableC[] = // KCSEL @@ -556,26 +556,22 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType WRITE(p, "\n"); - WRITE(p, "uniform float4 "I_COLORS"[4] : register(c%d);\n", C_COLORS); - WRITE(p, "uniform float4 "I_KCOLORS"[4] : register(c%d);\n", C_KCOLORS); - WRITE(p, "uniform float4 "I_ALPHA"[1] : register(c%d);\n", C_ALPHA); - WRITE(p, "uniform float4 "I_TEXDIMS"[8] : register(c%d);\n", C_TEXDIMS); - if (ApiType & API_D3D9) - { - WRITE(p, "uniform float4 "I_VTEXSCALE"[4] : register(c%d);\n", C_VTEXSCALE); - } - WRITE(p, "uniform float4 "I_ZBIAS"[2] : register(c%d);\n", C_ZBIAS); - WRITE(p, "uniform float4 "I_INDTEXSCALE"[2] : register(c%d);\n", C_INDTEXSCALE); - WRITE(p, "uniform float4 "I_INDTEXMTX"[6] : register(c%d);\n", C_INDTEXMTX); - WRITE(p, "uniform float4 "I_FOG"[3] : register(c%d);\n", C_FOG); + WRITE(p, "uniform float4 " I_COLORS"[4] : register(c%d);\n", C_COLORS); + WRITE(p, "uniform float4 " I_KCOLORS"[4] : register(c%d);\n", C_KCOLORS); + WRITE(p, "uniform float4 " I_ALPHA"[1] : register(c%d);\n", C_ALPHA); + WRITE(p, "uniform float4 " I_TEXDIMS"[8] : register(c%d);\n", C_TEXDIMS); + WRITE(p, "uniform float4 " I_ZBIAS"[2] : register(c%d);\n", C_ZBIAS); + WRITE(p, "uniform float4 " I_INDTEXSCALE"[2] : register(c%d);\n", C_INDTEXSCALE); + WRITE(p, "uniform float4 " I_INDTEXMTX"[6] : register(c%d);\n", C_INDTEXMTX); + WRITE(p, "uniform float4 " I_FOG"[3] : register(c%d);\n", C_FOG); if(g_ActiveConfig.bEnablePixelLighting && g_ActiveConfig.backend_info.bSupportsPixelLighting) { WRITE(p,"typedef struct { float4 col; float4 cosatt; float4 distatt; float4 pos; float4 dir; } Light;\n"); - WRITE(p,"typedef struct { Light lights[8]; } s_"I_PLIGHTS";\n"); - WRITE(p, "uniform s_"I_PLIGHTS" "I_PLIGHTS" : register(c%d);\n", C_PLIGHTS); - WRITE(p, "typedef struct { float4 C0, C1, C2, C3; } s_"I_PMATERIALS";\n"); - WRITE(p, "uniform s_"I_PMATERIALS" "I_PMATERIALS" : register(c%d);\n", C_PMATERIALS); + WRITE(p,"typedef struct { Light lights[8]; } s_" I_PLIGHTS";\n"); + WRITE(p, "uniform s_" I_PLIGHTS" " I_PLIGHTS" : register(c%d);\n", C_PLIGHTS); + WRITE(p, "typedef struct { float4 C0, C1, C2, C3; } s_" I_PMATERIALS";\n"); + WRITE(p, "uniform s_" I_PMATERIALS" " I_PMATERIALS" : register(c%d);\n", C_PMATERIALS); } WRITE(p, "void main(\n"); @@ -623,25 +619,32 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType char* pmainstart = p; int Pretest = AlphaPreTest(); - if (dstAlphaMode == DSTALPHA_ALPHA_PASS && !DepthTextureEnable && Pretest >= 0) + if(Pretest >= 0 && !DepthTextureEnable) { if (!Pretest) { - // alpha test will always fail, so restart the shader and just make it an empty function + // alpha test will always fail, so restart the shader and just make it an empty function WRITE(p, "ocol0 = 0;\n"); + if(DepthTextureEnable) + WRITE(p, "depth = 1.f;\n"); + if(dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND) + WRITE(p, "ocol1 = 0;\n"); WRITE(p, "discard;\n"); if(ApiType != API_D3D11) WRITE(p, "return;\n"); } - else + else if (dstAlphaMode == DSTALPHA_ALPHA_PASS) { - WRITE(p, " ocol0 = "I_ALPHA"[0].aaaa;\n"); + WRITE(p, " ocol0 = " I_ALPHA"[0].aaaa;\n"); + } + if(!Pretest || dstAlphaMode == DSTALPHA_ALPHA_PASS) + { + WRITE(p, "}\n"); + return text; } - WRITE(p, "}\n"); - return text; } - WRITE(p, " float4 c0 = "I_COLORS"[1], c1 = "I_COLORS"[2], c2 = "I_COLORS"[3], prev = float4(0.0f, 0.0f, 0.0f, 0.0f), textemp = float4(0.0f, 0.0f, 0.0f, 0.0f), rastemp = float4(0.0f, 0.0f, 0.0f, 0.0f), konsttemp = float4(0.0f, 0.0f, 0.0f, 0.0f);\n" + WRITE(p, " float4 c0 = " I_COLORS"[1], c1 = " I_COLORS"[2], c2 = " I_COLORS"[3], prev = float4(0.0f, 0.0f, 0.0f, 0.0f), textemp = float4(0.0f, 0.0f, 0.0f, 0.0f), rastemp = float4(0.0f, 0.0f, 0.0f, 0.0f), konsttemp = float4(0.0f, 0.0f, 0.0f, 0.0f);\n" " float3 comp16 = float3(1.0f, 255.0f, 0.0f), comp24 = float3(1.0f, 255.0f, 255.0f*255.0f);\n" " float4 alphabump=float4(0.0f,0.0f,0.0f,0.0f);\n" " float3 tevcoord=float3(0.0f, 0.0f, 0.0f);\n" @@ -692,7 +695,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType WRITE(p, " uv%d.xy = uv%d.xy / uv%d.z;\n", i, i, i); } - WRITE(p, "uv%d.xy = uv%d.xy * "I_TEXDIMS"[%d].zw;\n", i, i, i); + WRITE(p, "uv%d.xy = uv%d.xy * " I_TEXDIMS"[%d].zw;\n", i, i, i); } } @@ -704,7 +707,7 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType int texcoord = bpmem.tevindref.getTexCoord(i); if (texcoord < numTexgen) - WRITE(p, "tempcoord = uv%d.xy * "I_INDTEXSCALE"[%d].%s;\n", texcoord, i/2, (i&1)?"zw":"xy"); + WRITE(p, "tempcoord = uv%d.xy * " I_INDTEXSCALE"[%d].%s;\n", texcoord, i/2, (i&1)?"zw":"xy"); else WRITE(p, "tempcoord = float2(0.0f, 0.0f);\n"); @@ -727,65 +730,53 @@ const char *GeneratePixelShaderCode(DSTALPHA_MODE dstAlphaMode, API_TYPE ApiType // emulation of unsigned 8 overflow when casting WRITE(p, "prev = frac(4.0f + prev * (255.0f/256.0f)) * (256.0f/255.0f);\n"); - // TODO: Why are we doing a second alpha pretest here? - if (!WriteAlphaTest(p, ApiType, dstAlphaMode)) + if(Pretest == -1) { - // alpha test will always fail, so restart the shader and just make it an empty function - p = pmainstart; - WRITE(p, "ocol0 = 0;\n"); - if(DepthTextureEnable) - WRITE(p, "depth = 1.f;\n"); - if(dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND) - WRITE(p, "ocol1 = 0;\n"); - WRITE(p, "discard;\n"); - if(ApiType != API_D3D11) - WRITE(p, "return;\n"); + WriteAlphaTest(p, ApiType, dstAlphaMode); } - else - { - if((bpmem.fog.c_proj_fsel.fsel != 0) || DepthTextureEnable) - { - // the screen space depth value = far z + (clip z / clip w) * z range - WRITE(p, "float zCoord = "I_ZBIAS"[1].x + (clipPos.z / clipPos.w) * "I_ZBIAS"[1].y;\n"); - } + if((bpmem.fog.c_proj_fsel.fsel != 0) || DepthTextureEnable) + { + // the screen space depth value = far z + (clip z / clip w) * z range + WRITE(p, "float zCoord = " I_ZBIAS"[1].x + (clipPos.z / clipPos.w) * " I_ZBIAS"[1].y;\n"); + } - if (DepthTextureEnable) + if (DepthTextureEnable) + { + // use the texture input of the last texture stage (textemp), hopefully this has been read and is in correct format... + if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable) { - // use the texture input of the last texture stage (textemp), hopefully this has been read and is in correct format... - if (bpmem.ztex2.op != ZTEXTURE_DISABLE && !bpmem.zcontrol.zcomploc && bpmem.zmode.testenable && bpmem.zmode.updateenable) - { - if (bpmem.ztex2.op == ZTEXTURE_ADD) - WRITE(p, "zCoord = dot("I_ZBIAS"[0].xyzw, textemp.xyzw) + "I_ZBIAS"[1].w + zCoord;\n"); - else - WRITE(p, "zCoord = dot("I_ZBIAS"[0].xyzw, textemp.xyzw) + "I_ZBIAS"[1].w;\n"); - - // scale to make result from frac correct - WRITE(p, "zCoord = zCoord * (16777215.0f/16777216.0f);\n"); - WRITE(p, "zCoord = frac(zCoord);\n"); - WRITE(p, "zCoord = zCoord * (16777216.0f/16777215.0f);\n"); - } - WRITE(p, "depth = zCoord;\n"); - } + if (bpmem.ztex2.op == ZTEXTURE_ADD) + WRITE(p, "zCoord = dot(" I_ZBIAS"[0].xyzw, textemp.xyzw) + " I_ZBIAS"[1].w + zCoord;\n"); + else + WRITE(p, "zCoord = dot(" I_ZBIAS"[0].xyzw, textemp.xyzw) + " I_ZBIAS"[1].w;\n"); - if (dstAlphaMode == DSTALPHA_ALPHA_PASS) - WRITE(p, " ocol0 = float4(prev.rgb, "I_ALPHA"[0].a);\n"); - else - { - WriteFog(p); - WRITE(p, " ocol0 = prev;\n"); + // scale to make result from frac correct + WRITE(p, "zCoord = zCoord * (16777215.0f/16777216.0f);\n"); + WRITE(p, "zCoord = frac(zCoord);\n"); + WRITE(p, "zCoord = zCoord * (16777216.0f/16777215.0f);\n"); } + WRITE(p, "depth = zCoord;\n"); + } - // On D3D11, use dual-source color blending to perform dst alpha in a - // single pass - if (dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND) - { - // Colors will be blended against the alpha from ocol1... - WRITE(p, " ocol1 = ocol0;\n"); - // ...and the alpha from ocol0 will be written to the framebuffer. - WRITE(p, " ocol0.a = "I_ALPHA"[0].a;\n"); - } + if (dstAlphaMode == DSTALPHA_ALPHA_PASS) + WRITE(p, " ocol0 = float4(prev.rgb, " I_ALPHA"[0].a);\n"); + else + { + WriteFog(p); + WRITE(p, " ocol0 = prev;\n"); } + + // On D3D11, use dual-source color blending to perform dst alpha in a + // single pass + if (dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND) + { + // Colors will be blended against the alpha from ocol1... + WRITE(p, " ocol1 = ocol0;\n"); + // ...and the alpha from ocol0 will be written to the framebuffer. + WRITE(p, " ocol0.a = " I_ALPHA"[0].a;\n"); + } + WRITE(p, "}\n"); if (text[sizeof(text) - 1] != 0x7C) PanicAlert("PixelShader generator - buffer too small, canary has been eaten!"); @@ -876,20 +867,20 @@ static void WriteStage(char *&p, int n, API_TYPE ApiType) if (bpmem.tevind[n].mid <= 3) { int mtxidx = 2*(bpmem.tevind[n].mid-1); - WRITE(p, "float2 indtevtrans%d = float2(dot("I_INDTEXMTX"[%d].xyz, indtevcrd%d), dot("I_INDTEXMTX"[%d].xyz, indtevcrd%d));\n", + WRITE(p, "float2 indtevtrans%d = float2(dot(" I_INDTEXMTX"[%d].xyz, indtevcrd%d), dot(" I_INDTEXMTX"[%d].xyz, indtevcrd%d));\n", n, mtxidx, n, mtxidx+1, n); } else if (bpmem.tevind[n].mid <= 7 && bHasTexCoord) { // s matrix _assert_(bpmem.tevind[n].mid >= 5); int mtxidx = 2*(bpmem.tevind[n].mid-5); - WRITE(p, "float2 indtevtrans%d = "I_INDTEXMTX"[%d].ww * uv%d.xy * indtevcrd%d.xx;\n", n, mtxidx, texcoord, n); + WRITE(p, "float2 indtevtrans%d = " I_INDTEXMTX"[%d].ww * uv%d.xy * indtevcrd%d.xx;\n", n, mtxidx, texcoord, n); } else if (bpmem.tevind[n].mid <= 11 && bHasTexCoord) { // t matrix _assert_(bpmem.tevind[n].mid >= 9); int mtxidx = 2*(bpmem.tevind[n].mid-9); - WRITE(p, "float2 indtevtrans%d = "I_INDTEXMTX"[%d].ww * uv%d.xy * indtevcrd%d.yy;\n", n, mtxidx, texcoord, n); + WRITE(p, "float2 indtevtrans%d = " I_INDTEXMTX"[%d].ww * uv%d.xy * indtevcrd%d.yy;\n", n, mtxidx, texcoord, n); } else WRITE(p, "float2 indtevtrans%d = 0;\n", n); @@ -1103,15 +1094,9 @@ static void WriteStage(char *&p, int n, API_TYPE ApiType) void SampleTexture(char *&p, const char *destination, const char *texcoords, const char *texswap, int texmap, API_TYPE ApiType) { if (ApiType == API_D3D11) - WRITE(p, "%s=Tex%d.Sample(samp%d, %s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap,texmap, texcoords, texmap, texswap); - else if (ApiType & API_D3D9) - { - // D3D9 uses different pixel to texel mapping, so we need to offset our sampling address by half a pixel (assuming native and virtual texture dimensions match each other, otherwise some math is involved). - // Read the MSDN article "Directly Mapping Texels to Pixels (Direct3D 9)" for further info. - WRITE(p, "%s=tex2D(samp%d, (%s.xy + 0.5f*"I_VTEXSCALE"[%d].%s) * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap/2, (texmap&1)?"zw":"xy", texmap, texswap); - } + WRITE(p, "%s=Tex%d.Sample(samp%d,%s.xy * " I_TEXDIMS"[%d].xy).%s;\n", destination, texmap,texmap, texcoords, texmap, texswap); else - WRITE(p, "%s=tex2D(samp%d, %s.xy * "I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap, texswap); + WRITE(p, "%s=tex2D(samp%d,%s.xy * " I_TEXDIMS"[%d].xy).%s;\n", destination, texmap, texcoords, texmap, texswap); } static const char *tevAlphaFuncsTable[] = @@ -1167,19 +1152,13 @@ static int AlphaPreTest() } -static bool WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode) +static void WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode) { static const char *alphaRef[2] = { I_ALPHA"[0].r", I_ALPHA"[0].g" - }; - - int Pretest = AlphaPreTest(); - if(Pretest >= 0) - { - return Pretest != 0; - } + }; // using discard then return works the same in cg and dx9 but not in dx11 WRITE(p, "if(!( "); @@ -1191,11 +1170,35 @@ static bool WriteAlphaTest(char *&p, API_TYPE ApiType,DSTALPHA_MODE dstAlphaMode compindex = bpmem.alphaFunc.comp1 % 8; WRITE(p, tevAlphaFuncsTable[compindex],alphaRef[1]);//lookup the second component from the alpha function table - WRITE(p, ")){ocol0 = 0;%s%s discard;%s}\n", - dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND ? "ocol1 = 0;" : "", - DepthTextureEnable ? "depth = 1.f;" : "", - (ApiType != API_D3D11) ? "return;" : ""); - return true; + WRITE(p, ")) {\n"); + + WRITE(p, "ocol0 = 0;\n"); + if (dstAlphaMode == DSTALPHA_DUAL_SOURCE_BLEND) + WRITE(p, "ocol1 = 0;\n"); + if (DepthTextureEnable) + WRITE(p, "depth = 1.f;\n"); + + // HAXX: zcomploc is a way to control whether depth test is done before + // or after texturing and alpha test. PC GPU does depth test before texturing ONLY if depth value is + // not updated during shader execution. + // We implement "depth test before texturing" by discarding the fragment + // when the alpha test fail. This is not a correct implementation because + // even if the depth test fails the fragment could be alpha blended. + // this implemnetation is a trick to keep speed. + // the correct, but slow, way to implement a correct zComploc is : + // 1 - if zcomplock is enebled make a first pass, with color channel write disabled updating only + // depth channel. + // 2 - in the next pass disable depth chanel update, but proccess the color data normally + // this way is the only CORRECT way to emulate perfectly the zcomplock behaviour + if (!(bpmem.zcontrol.zcomploc && bpmem.zmode.updateenable)) + { + WRITE(p, "discard;\n"); + if (ApiType != API_D3D11) + WRITE(p, "return;\n"); + } + + WRITE(p, "}\n"); + } static const char *tevFogFuncsTable[] = @@ -1218,13 +1221,13 @@ static void WriteFog(char *&p) { // perspective // ze = A/(B - (Zs >> B_SHF) - WRITE (p, " float ze = "I_FOG"[1].x / ("I_FOG"[1].y - (zCoord / "I_FOG"[1].w));\n"); + WRITE (p, " float ze = " I_FOG"[1].x / (" I_FOG"[1].y - (zCoord / " I_FOG"[1].w));\n"); } else { // orthographic // ze = a*Zs (here, no B_SHF) - WRITE (p, " float ze = "I_FOG"[1].x * zCoord;\n"); + WRITE (p, " float ze = " I_FOG"[1].x * zCoord;\n"); } // x_adjust = sqrt((x-center)^2 + k^2)/k @@ -1232,12 +1235,12 @@ static void WriteFog(char *&p) //this is complitly teorical as the real hard seems to use a table intead of calculate the values. if(bpmem.fogRange.Base.Enabled) { - WRITE (p, " float x_adjust = (2.0f * (clipPos.x / "I_FOG"[2].y)) - 1.0f - "I_FOG"[2].x;\n"); - WRITE (p, " x_adjust = sqrt(x_adjust * x_adjust + "I_FOG"[2].z * "I_FOG"[2].z) / "I_FOG"[2].z;\n"); + WRITE (p, " float x_adjust = (2.0f * (clipPos.x / " I_FOG"[2].y)) - 1.0f - " I_FOG"[2].x;\n"); + WRITE (p, " x_adjust = sqrt(x_adjust * x_adjust + " I_FOG"[2].z * " I_FOG"[2].z) / " I_FOG"[2].z;\n"); WRITE (p, " ze *= x_adjust;\n"); } - WRITE (p, " float fog = saturate(ze - "I_FOG"[1].z);\n"); + WRITE (p, " float fog = saturate(ze - " I_FOG"[1].z);\n"); if(bpmem.fog.c_proj_fsel.fsel > 3) { @@ -1249,7 +1252,7 @@ static void WriteFog(char *&p) WARN_LOG(VIDEO, "Unknown Fog Type! %08x", bpmem.fog.c_proj_fsel.fsel); } - WRITE(p, " prev.rgb = lerp(prev.rgb,"I_FOG"[0].rgb,fog);\n"); + WRITE(p, " prev.rgb = lerp(prev.rgb," I_FOG"[0].rgb,fog);\n"); -}
\ No newline at end of file +} diff --git a/Source/Core/VideoCommon/Src/PixelShaderGen.h b/Source/Core/VideoCommon/Src/PixelShaderGen.h index 7374251c96..31242a916e 100644 --- a/Source/Core/VideoCommon/Src/PixelShaderGen.h +++ b/Source/Core/VideoCommon/Src/PixelShaderGen.h @@ -24,27 +24,26 @@ #define I_KCOLORS "k" #define I_ALPHA "alphaRef" #define I_TEXDIMS "texdim" -#define I_VTEXSCALE "vtexscale" #define I_ZBIAS "czbias" #define I_INDTEXSCALE "cindscale" #define I_INDTEXMTX "cindmtx" #define I_FOG "cfog" #define I_PLIGHTS "cLights" -#define I_PMATERIALS "cmtrl" +#define I_PMATERIALS "cmtrl" #define C_COLORMATRIX 0 // 0 #define C_COLORS 0 // 0 #define C_KCOLORS (C_COLORS + 4) // 4 #define C_ALPHA (C_KCOLORS + 4) // 8 #define C_TEXDIMS (C_ALPHA + 1) // 9 -#define C_VTEXSCALE (C_TEXDIMS + 8) //17 - virtual texture scaling factor (e.g. custom textures, scaled EFB copies) -#define C_ZBIAS (C_VTEXSCALE + 4) //21 -#define C_INDTEXSCALE (C_ZBIAS + 2) //23 -#define C_INDTEXMTX (C_INDTEXSCALE + 2) //25 -#define C_FOG (C_INDTEXMTX + 6) //31 -#define C_PLIGHTS (C_FOG + 3) //34 -#define C_PMATERIALS (C_PLIGHTS + 40) //74 -#define C_PENVCONST_END (C_PMATERIALS + 4) //78 +#define C_ZBIAS (C_TEXDIMS + 8) //17 +#define C_INDTEXSCALE (C_ZBIAS + 2) //19 +#define C_INDTEXMTX (C_INDTEXSCALE + 2) //21 +#define C_FOG (C_INDTEXMTX + 6) //27 + +#define C_PLIGHTS (C_FOG + 3) +#define C_PMATERIALS (C_PLIGHTS + 40) +#define C_PENVCONST_END (C_PMATERIALS + 4) #define PIXELSHADERUID_MAX_VALUES 70 #define PIXELSHADERUID_MAX_VALUES_SAFE 120 diff --git a/Source/Core/VideoCommon/Src/PixelShaderManager.cpp b/Source/Core/VideoCommon/Src/PixelShaderManager.cpp index 3e0f6d73c4..2521f80500 100644 --- a/Source/Core/VideoCommon/Src/PixelShaderManager.cpp +++ b/Source/Core/VideoCommon/Src/PixelShaderManager.cpp @@ -36,11 +36,9 @@ static bool s_bFogRangeAdjustChanged; static int nLightsChanged[2]; // min,max static float lastRGBAfull[2][4][4]; static u8 s_nTexDimsChanged; -static u8 s_nVirtualTexScalesChanged; static u8 s_nIndTexScaleChanged; static u32 lastAlpha; static u32 lastTexDims[8]; // width | height << 16 | wrap_s << 28 | wrap_t << 30 -static float lastVirtualTexScales[16]; // even fields: width ratio; odd fields: height ratio static u32 lastZBias; static int nMaterialsChanged; @@ -63,7 +61,6 @@ void PixelShaderManager::Init() { lastAlpha = 0; memset(lastTexDims, 0, sizeof(lastTexDims)); - memset(lastVirtualTexScales, 0, sizeof(lastVirtualTexScales)); lastZBias = 0; memset(lastRGBAfull, 0, sizeof(lastRGBAfull)); Dirty(); @@ -73,7 +70,6 @@ void PixelShaderManager::Dirty() { s_nColorsChanged[0] = s_nColorsChanged[1] = 15; s_nTexDimsChanged = 0xFF; - s_nVirtualTexScalesChanged = 0xFF; s_nIndTexScaleChanged = 0xFF; s_nIndTexMtxChanged = 15; s_bAlphaChanged = s_bZBiasChanged = s_bZTextureTypeChanged = s_bDepthRangeChanged = true; @@ -87,7 +83,7 @@ void PixelShaderManager::Shutdown() } -void PixelShaderManager::SetConstants(API_TYPE api_type) +void PixelShaderManager::SetConstants() { for (int i = 0; i < 2; ++i) { @@ -113,16 +109,6 @@ void PixelShaderManager::SetConstants(API_TYPE api_type) s_nTexDimsChanged = 0; } - if ((api_type & API_D3D9) && s_nVirtualTexScalesChanged) - { - for (int i = 0; i < 8; i += 2) - { - if (s_nVirtualTexScalesChanged & (3<<i)) - SetPSVirtualTexScalePair(i/2); - } - s_nVirtualTexScalesChanged = 0; - } - if (s_bAlphaChanged) { SetPSConstant4f(C_ALPHA, (lastAlpha&0xff)/255.0f, ((lastAlpha>>8)&0xff)/255.0f, 0, ((lastAlpha>>16)&0xff)/255.0f); @@ -352,13 +338,6 @@ void PixelShaderManager::SetPSTextureDims(int texid) SetPSConstant4fv(C_TEXDIMS + texid, fdims); } -void PixelShaderManager::SetPSVirtualTexScalePair(int texpairid) -{ - PRIM_LOG("vtexscale%d: %f %f %f %f\n", texpairid, lastVirtualTexScales[texpairid*4], lastVirtualTexScales[texpairid*4+1], - lastVirtualTexScales[texpairid*4+2], lastVirtualTexScales[texpairid*4+3]); - SetPSConstant4fv(C_VTEXSCALE + texpairid, &lastVirtualTexScales[texpairid*4]); -} - // This one is high in profiles (0.5%). TODO: Move conversion out, only store the raw color value // and update it when the shader constant is set, only. void PixelShaderManager::SetColorChanged(int type, int num, bool high) @@ -397,25 +376,14 @@ void PixelShaderManager::SetDestAlpha(const ConstantAlpha& alpha) } } -void PixelShaderManager::SetTexDims(int texmapid, u32 width, u32 height, u32 virtual_width, u32 virtual_height, u32 wraps, u32 wrapt, API_TYPE api_type) +void PixelShaderManager::SetTexDims(int texmapid, u32 width, u32 height, u32 wraps, u32 wrapt) { u32 wh = width | (height << 16) | (wraps << 28) | (wrapt << 30); - - bool refresh = lastTexDims[texmapid] != wh; - if (api_type & API_D3D9) + if (lastTexDims[texmapid] != wh) { - refresh |= (lastVirtualTexScales[texmapid*2] != (float)width / (float)virtual_width); - refresh |= (lastVirtualTexScales[texmapid*2+1] != (float)height / (float)virtual_height); - } - - if (refresh) - { - lastTexDims[texmapid] = wh; - lastVirtualTexScales[texmapid*2] = (float)width / (float)virtual_width; - lastVirtualTexScales[texmapid*2+1] = (float)height / (float)virtual_height; + lastTexDims[texmapid] = wh; s_nTexDimsChanged |= 1 << texmapid; - s_nVirtualTexScalesChanged |= 1 << texmapid; - } + } } void PixelShaderManager::SetZTextureBias(u32 bias) diff --git a/Source/Core/VideoCommon/Src/PixelShaderManager.h b/Source/Core/VideoCommon/Src/PixelShaderManager.h index 5cbe86cae4..2d1c01cad6 100644 --- a/Source/Core/VideoCommon/Src/PixelShaderManager.h +++ b/Source/Core/VideoCommon/Src/PixelShaderManager.h @@ -26,20 +26,18 @@ class PixelShaderManager { static void SetPSTextureDims(int texid); - static void SetPSVirtualTexScalePair(int texpairid); - public: static void Init(); static void Dirty(); static void Shutdown(); - static void SetConstants(API_TYPE api_type); // sets pixel shader constants + static void SetConstants(); // sets pixel shader constants // constant management, should be called after memory is committed static void SetColorChanged(int type, int index, bool high); static void SetAlpha(const AlphaFunc& alpha); static void SetDestAlpha(const ConstantAlpha& alpha); - static void SetTexDims(int texmapid, u32 width, u32 height, u32 virtual_width, u32 virtual_height, u32 wraps, u32 wrapt, API_TYPE api_type); + static void SetTexDims(int texmapid, u32 width, u32 height, u32 wraps, u32 wrapt); static void SetZTextureBias(u32 bias); static void SetViewportChanged(); static void SetIndMatrixChanged(int matrixidx); diff --git a/Source/Core/VideoCommon/Src/RenderBase.cpp b/Source/Core/VideoCommon/Src/RenderBase.cpp index 3b06bbb1c0..ef0fff57f7 100644 --- a/Source/Core/VideoCommon/Src/RenderBase.cpp +++ b/Source/Core/VideoCommon/Src/RenderBase.cpp @@ -43,6 +43,7 @@ #include "XFMemory.h" #include "FifoPlayer/FifoRecorder.h" #include "AVIDump.h" +#include "VertexShaderManager.h" #include <cmath> #include <string> @@ -193,6 +194,7 @@ bool Renderer::CalculateTargetSize(int multiplier) { s_target_width = newEFBWidth; s_target_height = newEFBHeight; + VertexShaderManager::SetViewportChanged(); return true; } return false; diff --git a/Source/Core/VideoCommon/Src/TextureCacheBase.cpp b/Source/Core/VideoCommon/Src/TextureCacheBase.cpp index cc14499330..00ff9610be 100644 --- a/Source/Core/VideoCommon/Src/TextureCacheBase.cpp +++ b/Source/Core/VideoCommon/Src/TextureCacheBase.cpp @@ -172,8 +172,10 @@ void TextureCache::ClearRenderTargets() TexCache::iterator iter = textures.begin(), tcend = textures.end(); + for (; iter!=tcend; ++iter) - iter->second->type = TCET_NORMAL; + if (iter->second->type != TCET_EC_DYNAMIC) + iter->second->type = TCET_NORMAL; } TextureCache::TCacheEntryBase* TextureCache::Load(unsigned int stage, @@ -238,6 +240,9 @@ TextureCache::TCacheEntryBase* TextureCache::Load(unsigned int stage, // 2. a) For EFB copies, only the hash and the texture address need to match if (entry->IsEfbCopy() && tex_hash == entry->hash && address == entry->addr) { + if (entry->type != TCET_EC_VRAM) + entry->type = TCET_NORMAL; + // TODO: Print a warning if the format changes! In this case, we could reinterpret the internal texture object data to the new pixel format (similiar to what is already being done in Renderer::ReinterpretPixelFormat()) goto return_entry; } @@ -318,8 +323,8 @@ TextureCache::TCacheEntryBase* TextureCache::Load(unsigned int stage, entry->SetGeneralParameters(address, texture_size, full_format, entry->num_mipmaps); entry->SetDimensions(nativeW, nativeH, width, height); entry->hash = tex_hash; - if (g_ActiveConfig.bCopyEFBToTexture) entry->type = TCET_NORMAL; - else if (entry->IsEfbCopy()) entry->type = TCET_EC_DYNAMIC; + if (entry->IsEfbCopy() && !g_ActiveConfig.bCopyEFBToTexture) entry->type = TCET_EC_DYNAMIC; + else entry->type = TCET_NORMAL; // load texture entry->Load(width, height, expandedWidth, 0, (texLevels == 0)); @@ -647,8 +652,11 @@ void TextureCache::CopyRenderTargetToTexture(u32 dstAddr, unsigned int dstFormat if ((entry->type == TCET_EC_VRAM && entry->virtual_width == scaled_tex_w && entry->virtual_height == scaled_tex_h) || (entry->type == TCET_EC_DYNAMIC && entry->native_width == tex_w && entry->native_height == tex_h)) { - scaled_tex_w = tex_w; - scaled_tex_h = tex_h; + if (entry->type == TCET_EC_DYNAMIC) + { + scaled_tex_w = tex_w; + scaled_tex_h = tex_h; + } } else { diff --git a/Source/Core/VideoCommon/Src/TextureConversionShader.cpp b/Source/Core/VideoCommon/Src/TextureConversionShader.cpp index c921a1a9fd..2132a25258 100644 --- a/Source/Core/VideoCommon/Src/TextureConversionShader.cpp +++ b/Source/Core/VideoCommon/Src/TextureConversionShader.cpp @@ -259,7 +259,7 @@ void WriteIncrementSampleX(char*& p,API_TYPE ApiType) void WriteToBitDepth(char*& p, u8 depth, const char* src, const char* dest) { - float result = pow(2.0f, depth) - 1.0f; + float result = 255 / pow(2.0f, (8 - depth)); WRITE(p, " %s = floor(%s * %ff);\n", dest, src, result); } diff --git a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp index 5076f9f044..07ace04b97 100644 --- a/Source/Core/VideoCommon/Src/VertexShaderGen.cpp +++ b/Source/Core/VideoCommon/Src/VertexShaderGen.cpp @@ -178,31 +178,31 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type) char *p = text; WRITE(p, "//Vertex Shader: comp:%x, \n", components); - WRITE(p, "typedef struct { float4 T0, T1, T2; float4 N0, N1, N2; } s_"I_POSNORMALMATRIX";\n" + WRITE(p, "typedef struct { float4 T0, T1, T2; float4 N0, N1, N2; } s_" I_POSNORMALMATRIX";\n" "typedef struct { float4 t; } FLT4;\n" - "typedef struct { FLT4 T[24]; } s_"I_TEXMATRICES";\n" - "typedef struct { FLT4 T[64]; } s_"I_TRANSFORMMATRICES";\n" - "typedef struct { FLT4 T[32]; } s_"I_NORMALMATRICES";\n" - "typedef struct { FLT4 T[64]; } s_"I_POSTTRANSFORMMATRICES";\n" + "typedef struct { FLT4 T[24]; } s_" I_TEXMATRICES";\n" + "typedef struct { FLT4 T[64]; } s_" I_TRANSFORMMATRICES";\n" + "typedef struct { FLT4 T[32]; } s_" I_NORMALMATRICES";\n" + "typedef struct { FLT4 T[64]; } s_" I_POSTTRANSFORMMATRICES";\n" "typedef struct { float4 col; float4 cosatt; float4 distatt; float4 pos; float4 dir; } Light;\n" - "typedef struct { Light lights[8]; } s_"I_LIGHTS";\n" - "typedef struct { float4 C0, C1, C2, C3; } s_"I_MATERIALS";\n" - "typedef struct { float4 T0, T1, T2, T3; } s_"I_PROJECTION";\n" + "typedef struct { Light lights[8]; } s_" I_LIGHTS";\n" + "typedef struct { float4 C0, C1, C2, C3; } s_" I_MATERIALS";\n" + "typedef struct { float4 T0, T1, T2, T3; } s_" I_PROJECTION";\n" ); p = GenerateVSOutputStruct(p, components, api_type); // uniforms - WRITE(p, "uniform s_"I_TRANSFORMMATRICES" "I_TRANSFORMMATRICES" : register(c%d);\n", C_TRANSFORMMATRICES); - WRITE(p, "uniform s_"I_TEXMATRICES" "I_TEXMATRICES" : register(c%d);\n", C_TEXMATRICES); // also using tex matrices - WRITE(p, "uniform s_"I_NORMALMATRICES" "I_NORMALMATRICES" : register(c%d);\n", C_NORMALMATRICES); - WRITE(p, "uniform s_"I_POSNORMALMATRIX" "I_POSNORMALMATRIX" : register(c%d);\n", C_POSNORMALMATRIX); - WRITE(p, "uniform s_"I_POSTTRANSFORMMATRICES" "I_POSTTRANSFORMMATRICES" : register(c%d);\n", C_POSTTRANSFORMMATRICES); - WRITE(p, "uniform s_"I_LIGHTS" "I_LIGHTS" : register(c%d);\n", C_LIGHTS); - WRITE(p, "uniform s_"I_MATERIALS" "I_MATERIALS" : register(c%d);\n", C_MATERIALS); - WRITE(p, "uniform s_"I_PROJECTION" "I_PROJECTION" : register(c%d);\n", C_PROJECTION); - WRITE(p, "uniform float4 "I_DEPTHPARAMS" : register(c%d);\n", C_DEPTHPARAMS); + WRITE(p, "uniform s_" I_TRANSFORMMATRICES" " I_TRANSFORMMATRICES" : register(c%d);\n", C_TRANSFORMMATRICES); + WRITE(p, "uniform s_" I_TEXMATRICES" " I_TEXMATRICES" : register(c%d);\n", C_TEXMATRICES); // also using tex matrices + WRITE(p, "uniform s_" I_NORMALMATRICES" " I_NORMALMATRICES" : register(c%d);\n", C_NORMALMATRICES); + WRITE(p, "uniform s_" I_POSNORMALMATRIX" " I_POSNORMALMATRIX" : register(c%d);\n", C_POSNORMALMATRIX); + WRITE(p, "uniform s_" I_POSTTRANSFORMMATRICES" " I_POSTTRANSFORMMATRICES" : register(c%d);\n", C_POSTTRANSFORMMATRICES); + WRITE(p, "uniform s_" I_LIGHTS" " I_LIGHTS" : register(c%d);\n", C_LIGHTS); + WRITE(p, "uniform s_" I_MATERIALS" " I_MATERIALS" : register(c%d);\n", C_MATERIALS); + WRITE(p, "uniform s_" I_PROJECTION" " I_PROJECTION" : register(c%d);\n", C_PROJECTION); + WRITE(p, "uniform float4 " I_DEPTHPARAMS" : register(c%d);\n", C_DEPTHPARAMS); WRITE(p, "VS_OUTPUT main(\n"); @@ -258,11 +258,11 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type) WRITE(p, "int posmtx = fposmtx;\n"); } - WRITE(p, "float4 pos = float4(dot("I_TRANSFORMMATRICES".T[posmtx].t, rawpos), dot("I_TRANSFORMMATRICES".T[posmtx+1].t, rawpos), dot("I_TRANSFORMMATRICES".T[posmtx+2].t, rawpos), 1);\n"); + WRITE(p, "float4 pos = float4(dot(" I_TRANSFORMMATRICES".T[posmtx].t, rawpos), dot(" I_TRANSFORMMATRICES".T[posmtx+1].t, rawpos), dot(" I_TRANSFORMMATRICES".T[posmtx+2].t, rawpos), 1);\n"); if (components & VB_HAS_NRMALL) { WRITE(p, "int normidx = posmtx >= 32 ? (posmtx-32) : posmtx;\n"); - WRITE(p, "float3 N0 = "I_NORMALMATRICES".T[normidx].t.xyz, N1 = "I_NORMALMATRICES".T[normidx+1].t.xyz, N2 = "I_NORMALMATRICES".T[normidx+2].t.xyz;\n"); + WRITE(p, "float3 N0 = " I_NORMALMATRICES".T[normidx].t.xyz, N1 = " I_NORMALMATRICES".T[normidx+1].t.xyz, N2 = " I_NORMALMATRICES".T[normidx+2].t.xyz;\n"); } if (components & VB_HAS_NRM0) @@ -274,13 +274,13 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type) } else { - WRITE(p, "float4 pos = float4(dot("I_POSNORMALMATRIX".T0, rawpos), dot("I_POSNORMALMATRIX".T1, rawpos), dot("I_POSNORMALMATRIX".T2, rawpos), 1.0f);\n"); + WRITE(p, "float4 pos = float4(dot(" I_POSNORMALMATRIX".T0, rawpos), dot(" I_POSNORMALMATRIX".T1, rawpos), dot(" I_POSNORMALMATRIX".T2, rawpos), 1.0f);\n"); if (components & VB_HAS_NRM0) - WRITE(p, "float3 _norm0 = normalize(float3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm0), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm0), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm0)));\n"); + WRITE(p, "float3 _norm0 = normalize(float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm0), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm0), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm0)));\n"); if (components & VB_HAS_NRM1) - WRITE(p, "float3 _norm1 = float3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm1), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm1), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm1));\n"); + WRITE(p, "float3 _norm1 = float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm1), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm1), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm1));\n"); if (components & VB_HAS_NRM2) - WRITE(p, "float3 _norm2 = float3(dot("I_POSNORMALMATRIX".N0.xyz, rawnorm2), dot("I_POSNORMALMATRIX".N1.xyz, rawnorm2), dot("I_POSNORMALMATRIX".N2.xyz, rawnorm2));\n"); + WRITE(p, "float3 _norm2 = float3(dot(" I_POSNORMALMATRIX".N0.xyz, rawnorm2), dot(" I_POSNORMALMATRIX".N1.xyz, rawnorm2), dot(" I_POSNORMALMATRIX".N2.xyz, rawnorm2));\n"); } if (!(components & VB_HAS_NRM0)) @@ -288,7 +288,7 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type) - WRITE(p, "o.pos = float4(dot("I_PROJECTION".T0, pos), dot("I_PROJECTION".T1, pos), dot("I_PROJECTION".T2, pos), dot("I_PROJECTION".T3, pos));\n"); + WRITE(p, "o.pos = float4(dot(" I_PROJECTION".T0, pos), dot(" I_PROJECTION".T1, pos), dot(" I_PROJECTION".T2, pos), dot(" I_PROJECTION".T3, pos));\n"); WRITE(p, "float4 mat, lacc;\n" "float3 ldir, h;\n" @@ -367,7 +367,7 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type) if (components & (VB_HAS_NRM1|VB_HAS_NRM2)) { // transform the light dir into tangent space - WRITE(p, "ldir = normalize("I_LIGHTS".lights[%d].pos.xyz - pos.xyz);\n", texinfo.embosslightshift); + WRITE(p, "ldir = normalize(" I_LIGHTS".lights[%d].pos.xyz - pos.xyz);\n", texinfo.embosslightshift); WRITE(p, "o.tex%d.xyz = o.tex%d.xyz + float3(dot(ldir, _norm1), dot(ldir, _norm2), 0.0f);\n", i, texinfo.embosssourceshift); } else @@ -389,16 +389,16 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type) default: if (components & (VB_HAS_TEXMTXIDX0<<i)) { if (texinfo.projection == XF_TEXPROJ_STQ) - WRITE(p, "o.tex%d.xyz = float3(dot(coord, "I_TRANSFORMMATRICES".T[tex%d.z].t), dot(coord, "I_TRANSFORMMATRICES".T[tex%d.z+1].t), dot(coord, "I_TRANSFORMMATRICES".T[tex%d.z+2].t));\n", i, i, i, i); + WRITE(p, "o.tex%d.xyz = float3(dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z].t), dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z+1].t), dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z+2].t));\n", i, i, i, i); else { - WRITE(p, "o.tex%d.xyz = float3(dot(coord, "I_TRANSFORMMATRICES".T[tex%d.z].t), dot(coord, "I_TRANSFORMMATRICES".T[tex%d.z+1].t), 1);\n", i, i, i); + WRITE(p, "o.tex%d.xyz = float3(dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z].t), dot(coord, " I_TRANSFORMMATRICES".T[tex%d.z+1].t), 1);\n", i, i, i); } } else { if (texinfo.projection == XF_TEXPROJ_STQ) - WRITE(p, "o.tex%d.xyz = float3(dot(coord, "I_TEXMATRICES".T[%d].t), dot(coord, "I_TEXMATRICES".T[%d].t), dot(coord, "I_TEXMATRICES".T[%d].t));\n", i, 3*i, 3*i+1, 3*i+2); + WRITE(p, "o.tex%d.xyz = float3(dot(coord, " I_TEXMATRICES".T[%d].t), dot(coord, " I_TEXMATRICES".T[%d].t), dot(coord, " I_TEXMATRICES".T[%d].t));\n", i, 3*i, 3*i+1, 3*i+2); else - WRITE(p, "o.tex%d.xyz = float3(dot(coord, "I_TEXMATRICES".T[%d].t), dot(coord, "I_TEXMATRICES".T[%d].t), 1);\n", i, 3*i, 3*i+1); + WRITE(p, "o.tex%d.xyz = float3(dot(coord, " I_TEXMATRICES".T[%d].t), dot(coord, " I_TEXMATRICES".T[%d].t), 1);\n", i, 3*i, 3*i+1); } break; } @@ -407,9 +407,9 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type) const PostMtxInfo& postInfo = xfregs.postMtxInfo[i]; int postidx = postInfo.index; - WRITE(p, "float4 P0 = "I_POSTTRANSFORMMATRICES".T[%d].t;\n" - "float4 P1 = "I_POSTTRANSFORMMATRICES".T[%d].t;\n" - "float4 P2 = "I_POSTTRANSFORMMATRICES".T[%d].t;\n", + WRITE(p, "float4 P0 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n" + "float4 P1 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n" + "float4 P2 = " I_POSTTRANSFORMMATRICES".T[%d].t;\n", postidx&0x3f, (postidx+1)&0x3f, (postidx+2)&0x3f); if (texGenSpecialCase) { @@ -467,7 +467,7 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type) //if not early z culling will improve speed if (is_d3d) { - WRITE(p, "o.pos.z = "I_DEPTHPARAMS".x * o.pos.w + o.pos.z * "I_DEPTHPARAMS".y;\n"); + WRITE(p, "o.pos.z = " I_DEPTHPARAMS".x * o.pos.w + o.pos.z * " I_DEPTHPARAMS".y;\n"); } else { @@ -493,6 +493,13 @@ const char *GenerateVertexShaderCode(u32 components, API_TYPE api_type) //seems to get rather complicated } + if (api_type & API_D3D9) + { + // D3D9 is addressing pixel centers instead of pixel boundaries in clip space. + // Thus we need to offset the final position by half a pixel + WRITE(p, "o.pos = o.pos + float4(" I_DEPTHPARAMS".z, " I_DEPTHPARAMS".w, 0.f, 0.f);\n"); + } + WRITE(p, "return o;\n}\n"); diff --git a/Source/Core/VideoCommon/Src/VertexShaderGen.h b/Source/Core/VideoCommon/Src/VertexShaderGen.h index 0c522286b1..cb253a9b6c 100644 --- a/Source/Core/VideoCommon/Src/VertexShaderGen.h +++ b/Source/Core/VideoCommon/Src/VertexShaderGen.h @@ -35,7 +35,7 @@ #define I_TRANSFORMMATRICES "ctrmtx" #define I_NORMALMATRICES "cnmtx" #define I_POSTTRANSFORMMATRICES "cpostmtx" -#define I_DEPTHPARAMS "cDepth" +#define I_DEPTHPARAMS "cDepth" // farZ, zRange, scaled viewport width, scaled viewport height #define C_POSNORMALMATRIX 0 #define C_PROJECTION (C_POSNORMALMATRIX + 6) diff --git a/Source/Core/VideoCommon/Src/VertexShaderManager.cpp b/Source/Core/VideoCommon/Src/VertexShaderManager.cpp index d7d25c5c85..b6afbc61c6 100644 --- a/Source/Core/VideoCommon/Src/VertexShaderManager.cpp +++ b/Source/Core/VideoCommon/Src/VertexShaderManager.cpp @@ -306,7 +306,11 @@ void VertexShaderManager::SetConstants() if (bViewportChanged) { bViewportChanged = false; - SetVSConstant4f(C_DEPTHPARAMS,xfregs.viewport.farZ / 16777216.0f,xfregs.viewport.zRange / 16777216.0f,0.0f,0.0f); + SetVSConstant4f(C_DEPTHPARAMS, + xfregs.viewport.farZ / 16777216.0f, + xfregs.viewport.zRange / 16777216.0f, + -1.f / (float)g_renderer->EFBToScaledX((int)ceil(2.0f * xfregs.viewport.wd)), + 1.f / (float)g_renderer->EFBToScaledY((int)ceil(-2.0f * xfregs.viewport.ht))); // This is so implementation-dependent that we can't have it here. UpdateViewport(s_viewportCorrection); bProjectionChanged = true; diff --git a/Source/Core/VideoCommon/VideoCommon.vcxproj b/Source/Core/VideoCommon/VideoCommon.vcxproj index dd29d29ccb..f53c18cb37 100644 --- a/Source/Core/VideoCommon/VideoCommon.vcxproj +++ b/Source/Core/VideoCommon/VideoCommon.vcxproj @@ -266,7 +266,6 @@ <ItemGroup> <None Include="..\..\..\Data\User\OpenCL\TextureDecoder.cl" /> <None Include="CMakeLists.txt" /> - <None Include="Src\SConscript" /> </ItemGroup> <ItemGroup> <ProjectReference Include="..\..\..\Externals\CLRun\clrun\CLRun.vcxproj"> diff --git a/Source/Core/VideoCommon/VideoCommon.vcxproj.filters b/Source/Core/VideoCommon/VideoCommon.vcxproj.filters index 5ce157a78e..c933fbc939 100644 --- a/Source/Core/VideoCommon/VideoCommon.vcxproj.filters +++ b/Source/Core/VideoCommon/VideoCommon.vcxproj.filters @@ -248,7 +248,6 @@ </ClInclude> </ItemGroup> <ItemGroup> - <None Include="Src\SConscript" /> <None Include="CMakeLists.txt" /> <None Include="..\..\..\Data\User\OpenCL\TextureDecoder.cl"> <Filter>Decoding\OpenCL</Filter> |