From dcd5ffeb7ad0216b5d9a90f59c23c54208e2bdde Mon Sep 17 00:00:00 2001
From: hrydgard <hrydgard@gmail.com>
Date: Sat, 16 Aug 2008 10:49:08 +0000
Subject: [PATCH] More accurate audio interrupts (preparation for homebrew
 audio support). some minor cleanup in gl plugin.

git-svn-id: https://dolphin-emu.googlecode.com/svn/trunk@226 8ced0084-cf51-0410-be5f-012b33b47a6e
---
 Source/Core/Core/Src/HW/AudioInterface.cpp    | 29 ++++--
 Source/Core/Core/Src/HW/AudioInterface.h      |  3 +
 Source/Core/Core/Src/HW/DSP.cpp               | 88 ++++++++++++++-----
 Source/Core/Core/Src/HW/DSP.h                 |  4 +-
 Source/Core/Core/Src/HW/GPFifo.cpp            |  4 +
 Source/Core/Core/Src/HW/GPFifo.h              |  2 +
 Source/Core/Core/Src/HW/SystemTimers.cpp      | 65 ++++++++------
 .../Interpreter_SystemRegisters.cpp           | 18 ++--
 .../Src/PowerPC/Jit64/Jit_SystemRegisters.cpp |  4 +-
 .../Plugin_VideoOGL/Plugin_VideoOGL.vcproj    |  4 +
 .../Plugins/Plugin_VideoOGL/Src/Globals.cpp   |  7 ++
 Source/Plugins/Plugin_VideoOGL/Src/Globals.h  | 23 +----
 .../Plugins/Plugin_VideoOGL/Src/OS/Win32.cpp  |  9 ++
 .../Src/PixelShaderManager.cpp                |  9 +-
 .../Plugin_VideoOGL/Src/TextureMngr.cpp       |  7 +-
 .../Plugin_VideoOGL/Src/VertexLoader.cpp      |  2 +-
 16 files changed, 177 insertions(+), 101 deletions(-)

diff --git a/Source/Core/Core/Src/HW/AudioInterface.cpp b/Source/Core/Core/Src/HW/AudioInterface.cpp
index 89a826cc99..8f18bdd8d6 100644
--- a/Source/Core/Core/Src/HW/AudioInterface.cpp
+++ b/Source/Core/Core/Src/HW/AudioInterface.cpp
@@ -15,6 +15,13 @@
 // Official SVN repository and contact information can be found at
 // http://code.google.com/p/dolphin-emu/
 
+// This file is ONLY about disc streaming. It's a bit unfortunately named.
+// For the rest of the audio stuff, including the "real" AI, see DSP.cpp/h.
+
+// AI disc streaming is handled completely separately from the rest of the
+// audio processing. In short, it simply streams audio directly from disc
+// out through the speakers.
+
 #include "Common.h"
 
 #include "StreamADPCM.H"
@@ -125,13 +132,15 @@ void Read32(u32& _rReturnValue, const u32 _Address)
 		return;
 
 	case AI_INTERRUPT_TIMING:
+		// When sample counter reaches the value of this register, the interrupt AIINT should
+		// fire.
         LOG(AUDIO_INTERFACE, "AudioInterface(R) 0x%08x", _Address);
 		_rReturnValue = g_AudioRegister.m_InterruptTiming;
 		return;
 
 	default:
         LOG(AUDIO_INTERFACE, "AudioInterface(R) 0x%08x", _Address);
-		_dbg_assert_msg_(AUDIO_INTERFACE,0,"AudioInterface - Read from ???");
+		_dbg_assert_msg_(AUDIO_INTERFACE, 0, "AudioInterface - Read from ???");
 		_rReturnValue = 0;
 		return;
 	}
@@ -196,7 +205,8 @@ void Write32(const u32 _Value, const u32 _Address)
 		break;
 
 	case AI_SAMPLE_COUNTER:
-		_dbg_assert_msg_(AUDIO_INTERFACE,0,"AudioInterface - m_SampleCounter is Read only");
+		// _dbg_assert_msg_(AUDIO_INTERFACE, 0, "AudioInterface - m_SampleCounter is Read only");
+		g_AudioRegister.m_SampleCounter = _Value;
 		break;
 
 	case AI_INTERRUPT_TIMING:		
@@ -229,7 +239,7 @@ void GenerateAudioInterrupt()
 	UpdateInterrupts();
 }
 
-// Callback for the DSP streaming
+// Callback for the disc streaming
 // WARNING - called from audio thread
 unsigned __int32 Callback_GetStreaming(short* _pDestBuffer, unsigned __int32 _numSamples)
 {
@@ -240,9 +250,9 @@ unsigned __int32 Callback_GetStreaming(short* _pDestBuffer, unsigned __int32 _nu
 		const int lvolume = g_AudioRegister.m_Volume.leftVolume;
 		const int rvolume = g_AudioRegister.m_Volume.rightVolume;
 
-		for (unsigned int i=0; i<_numSamples; i++)
+		for (unsigned int i = 0; i < _numSamples; i++)
 		{
-			if (pos==0)
+			if (pos == 0)
 			{
 				ReadStreamBlock(pcm);
 			}
@@ -253,13 +263,13 @@ unsigned __int32 Callback_GetStreaming(short* _pDestBuffer, unsigned __int32 _nu
 			pos++;
 			if (pos == 28) 
 			{
-				pos=0;
+				pos = 0;
 			}
 		}
 	}
 	else
 	{
-		for (unsigned int i=0; i<_numSamples*2; i++)
+		for (unsigned int i = 0; i < _numSamples * 2; i++)
 		{
 			_pDestBuffer[i] = 0; //silence!
 		}
@@ -269,7 +279,7 @@ unsigned __int32 Callback_GetStreaming(short* _pDestBuffer, unsigned __int32 _nu
 }
 
 // WARNING - called from audio thread
-void ReadStreamBlock(short* _pPCM)
+void ReadStreamBlock(short *_pPCM)
 {
 	char tempADPCM[32];
 	if (DVDInterface::DVDReadADPCM((u8*)tempADPCM, 32))
@@ -289,6 +299,7 @@ void ReadStreamBlock(short* _pPCM)
     // our whole streaming code is "faked" ... so it shouldn't increase the sample counter
     // streaming will never work correctly this way, but at least the program will think all is alright.
 
+	// This call must not be done wihout going through CoreTiming's threadsafe option.
 	// IncreaseSampleCount(28); 
 }
 
@@ -298,7 +309,7 @@ void IncreaseSampleCount(const u32 _iAmount)
 	{
 		g_AudioRegister.m_SampleCounter += _iAmount;
 		if (g_AudioRegister.m_Control.AIINTVLD && 
-            (g_AudioRegister.m_SampleCounter > g_AudioRegister.m_InterruptTiming))
+            (g_AudioRegister.m_SampleCounter >= g_AudioRegister.m_InterruptTiming))
 		{			
 			GenerateAudioInterrupt();
 		}
diff --git a/Source/Core/Core/Src/HW/AudioInterface.h b/Source/Core/Core/Src/HW/AudioInterface.h
index 2e8da4b33c..4b5e759550 100644
--- a/Source/Core/Core/Src/HW/AudioInterface.h
+++ b/Source/Core/Core/Src/HW/AudioInterface.h
@@ -14,6 +14,9 @@
 
 // Official SVN repository and contact information can be found at
 // http://code.google.com/p/dolphin-emu/
+
+// See CPP file for comments.
+
 #ifndef _AUDIOINTERFACE_H
 #define _AUDIOINTERFACE_H
 
diff --git a/Source/Core/Core/Src/HW/DSP.cpp b/Source/Core/Core/Src/HW/DSP.cpp
index 6126128db3..c941f07507 100644
--- a/Source/Core/Core/Src/HW/DSP.cpp
+++ b/Source/Core/Core/Src/HW/DSP.cpp
@@ -15,6 +15,22 @@
 // Official SVN repository and contact information can be found at
 // http://code.google.com/p/dolphin-emu/
 
+// AID / AUDIO_DMA controls pushing audio out to the SRC and then the speakers.
+// The audio DMA pushes audio through a small FIFO 32 bytes at a time, as needed.
+// Since the SRC behind the fifo eats stereo 16-bit data at a sample rate of 32khz,
+// that is, 4 bytes at 32 khz, which is 32 bytes at 4 khz. We should thus schedule an
+// event that runs at 4khz, that eats audio from the fifo, and all the rest will follow.
+// Then we will have homebrew audio.
+
+// The AID interrupt is set when the fifo STARTS a transfer. It latches address and count
+// into internal registers and starts copying. This means that the interrupt handler can simply
+// set the registers to where the next buffer is, and start filling it. When the DMA is complete,
+// it will automatically relatch and fire a new interrupt.
+
+// Then there's the DSP... what likely happens is that the fifo-latched-interrupt handler
+// kicks off the DSP, requesting it to fill up the just used buffer through the AXList (or 
+// whatever it might be called in Nintendo games).
+
 #include "DSP.h"
 
 #include "../CoreTiming.h"
@@ -80,8 +96,8 @@ union UDSPControl
 		unsigned DSPAssertInt	: 1;
 		unsigned DSPHalt		: 1;
 
-		unsigned AI				: 1;
-		unsigned AI_mask   		: 1;
+		unsigned AID			: 1;
+		unsigned AID_mask   	: 1;
 		unsigned ARAM			: 1;
 		unsigned ARAM_mask		: 1;
 		unsigned DSP			: 1;
@@ -107,14 +123,14 @@ struct DSPState
 	}
 };	
 
-// UDSPControl
+// Blocks are 32 bytes.
 union UAudioDMAControl
 {
     u16 Hex;
     struct  
     {        
-        unsigned NumSamples	    : 15;
-        unsigned Enabled		: 1;
+        unsigned NumBlocks  : 15;
+		unsigned Enabled    : 1;
     };
 
     UAudioDMAControl(u16 _Hex = 0) : Hex(_Hex)
@@ -125,8 +141,9 @@ union UAudioDMAControl
 struct AudioDMA
 {
 	u32 SourceAddress;
+	u32 ReadAddress;
 	UAudioDMAControl AudioDMAControl;
-	u32 SamplesLeft;
+	int BlocksLeft;
 };
 
 // ARDMA
@@ -162,6 +179,11 @@ void WriteARAM(u8 _iValue, u32 _iAddress);
 bool Update_DSP_ReadRegister();
 void Update_DSP_WriteRegister();
 
+int GetDSPSampleRate()
+{
+	return 32000;  // TODO - can also be 48000
+}
+
 void Init()
 {
 	g_ARAM = (u8 *)AllocateMemoryPages(ARAM_SIZE);
@@ -236,7 +258,8 @@ void Read16(u16& _uReturnValue, const u32 _iAddress)
 		// DMA_REGS 0x5030+
 		// ==================================================================================
 		case AUDIO_DMA_BYTES_LEFT:
-			_uReturnValue = g_audioDMA.SamplesLeft;
+			// Hmm. Would be stupid to ask for bytes left. Assume it wants blocks left.
+			_uReturnValue = g_audioDMA.BlocksLeft;
 			return;
 
 		case AUDIO_DMA_START_LO:
@@ -305,12 +328,12 @@ void Write16(const u16 _Value, const u32 _Address)
 			g_dspState.DSPControl.DSPInit       = tmpControl.DSPInit;
 
 			// Interrupt (mask)
-			g_dspState.DSPControl.AI_mask	= tmpControl.AI_mask;
+			g_dspState.DSPControl.AID_mask	= tmpControl.AID_mask;
 			g_dspState.DSPControl.ARAM_mask	= tmpControl.ARAM_mask;
 			g_dspState.DSPControl.DSP_mask	= tmpControl.DSP_mask;
 
 			// Interrupt
-			if (tmpControl.AI)		g_dspState.DSPControl.AI		= 0;
+			if (tmpControl.AID)		g_dspState.DSPControl.AID		= 0;
 			if (tmpControl.ARAM)	g_dspState.DSPControl.ARAM	    = 0;
 			if (tmpControl.DSP)		g_dspState.DSPControl.DSP		= 0;
 
@@ -332,6 +355,7 @@ void Write16(const u16 _Value, const u32 _Address)
 
 	// ==================================================================================
 	// AR_REGS 0x501x+
+	// DMA back and forth between ARAM and RAM
 	// ==================================================================================
 
 	case 0x5012:
@@ -369,6 +393,7 @@ void Write16(const u16 _Value, const u32 _Address)
 
 	// ==================================================================================
 	// Audio DMA_REGS 0x5030+
+	// This is the DMA that goes straight out the speaker. 
 	// ==================================================================================
 	case AUDIO_DMA_START_HI:
 		g_audioDMA.SourceAddress = (g_audioDMA.SourceAddress & 0xFFFF) | (_Value<<16);
@@ -379,16 +404,20 @@ void Write16(const u16 _Value, const u32 _Address)
 		break;
 
 	case AUDIO_DMA_CONTROL_LEN:			// called by AIStartDMA()
-		g_audioDMA.AudioDMAControl.Hex = _Value;
-		g_audioDMA.SamplesLeft = g_audioDMA.AudioDMAControl.NumSamples;
-
-		if (g_audioDMA.AudioDMAControl.Enabled)
 		{
-			PluginDSP::DSP_SendAIBuffer(g_audioDMA.SourceAddress, g_audioDMA.AudioDMAControl.NumSamples);
-			g_audioDMA.SamplesLeft = 0;
+		UAudioDMAControl old_control = g_audioDMA.AudioDMAControl;
+		g_audioDMA.AudioDMAControl.Hex = _Value;
+
+		if (!old_control.Enabled && g_audioDMA.AudioDMAControl.Enabled)
+		{
+			// Enabled bit was flipped to true, let's latch address & length and call the interrupt.
+			g_audioDMA.BlocksLeft = g_audioDMA.AudioDMAControl.NumBlocks;
+			g_audioDMA.ReadAddress = g_audioDMA.SourceAddress;
+			GenerateDSPInterrupt(DSP::INT_AID);
+			LOG(DSP, "AID DMA started - source address %08x, length %i blocks", g_audioDMA.SourceAddress, g_audioDMA.AudioDMAControl.NumBlocks);
 		}
 		break;
-
+		}
 	case AUDIO_DMA_BYTES_LEFT:
 		_dbg_assert_(DSPINTERFACE,0);
 		break;
@@ -399,6 +428,25 @@ void Write16(const u16 _Value, const u32 _Address)
 	}
 }
 
+// This happens at 4 khz, since 32 bytes at 4khz = 4 bytes at 32 khz (16bit stereo pcm)
+void UpdateAudioDMA()
+{
+	if (g_audioDMA.AudioDMAControl.Enabled && g_audioDMA.BlocksLeft) {
+		// TODO : Read audio at g_audioDMA.ReadAddress in RAM and push onto an external audio fifo in the emulator,
+		// to be mixed with the disc streaming output. If that audio queue fills up, we have to delay the emulator.
+		g_audioDMA.ReadAddress += 32;
+		g_audioDMA.BlocksLeft--;
+		if (!g_audioDMA.BlocksLeft) {
+			// No need to turn off the DMA - we can only get here if we had blocks left when we
+			// entered this function, and no longer have any.
+			// Latch new parameters
+			g_audioDMA.BlocksLeft = g_audioDMA.AudioDMAControl.NumBlocks;
+			g_audioDMA.ReadAddress = g_audioDMA.SourceAddress;
+			GenerateDSPInterrupt(DSP::INT_AID);
+		}
+	}
+}
+
 void Read32(u32& _uReturnValue, const u32 _iAddress)
 {
 	LOG(DSPINTERFACE, "DSPInterface(r) 0x%08x", _iAddress);
@@ -450,7 +498,7 @@ void Write32(const u32 _iValue, const u32 _iAddress)
 //	
 void UpdateInterrupts()
 {
-	if ((g_dspState.DSPControl.AI   & g_dspState.DSPControl.AI_mask) ||
+	if ((g_dspState.DSPControl.AID  & g_dspState.DSPControl.AID_mask) ||
 		(g_dspState.DSPControl.ARAM & g_dspState.DSPControl.ARAM_mask) ||
 		(g_dspState.DSPControl.DSP  & g_dspState.DSPControl.DSP_mask))
 	{
@@ -468,7 +516,7 @@ void GenerateDSPInterrupt(DSPInterruptType type, bool _bSet)
 	{
 	case INT_DSP:	g_dspState.DSPControl.DSP		= _bSet ? 1 : 0; break;
 	case INT_ARAM:	g_dspState.DSPControl.ARAM	    = _bSet ? 1 : 0; break;
-	case INT_AI:	g_dspState.DSPControl.AI		= _bSet ? 1 : 0; break;
+	case INT_AID:	g_dspState.DSPControl.AID		= _bSet ? 1 : 0; break;
 	}
 
 	UpdateInterrupts();
@@ -504,7 +552,7 @@ void Update_ARAM_DMA()
 		u32 iARAMAddress = g_arDMA.ARAddr;
 		
 		// TODO(??): sanity check instead of writing bogus data?
-		for (u32 i=0; i<g_arDMA.Cnt.count; i++)
+		for (u32 i = 0; i < g_arDMA.Cnt.count; i++)
 		{
 			u32 tmp = (iARAMAddress < ARAM_SIZE) ? g_ARAM[iARAMAddress] : 0x05050505;
 			Memory::Write_U8(tmp, iMemAddress);				
@@ -520,7 +568,7 @@ void Update_ARAM_DMA()
 
 		//write to g_ARAM
 		LOG(DSPINTERFACE,"g_ARAM DMA write %08x bytes from %08x to Aram: %08x",g_arDMA.Cnt.count, g_arDMA.MMAddr, g_arDMA.ARAddr);
-		for (u32 i=0; i<g_arDMA.Cnt.count; i++)
+		for (u32 i = 0; i < g_arDMA.Cnt.count; i++)
 		{
 			if (iARAMAddress < ARAM_SIZE)
 				g_ARAM[iARAMAddress] = Memory::Read_U8(iMemAddress);
diff --git a/Source/Core/Core/Src/HW/DSP.h b/Source/Core/Core/Src/HW/DSP.h
index 662bf59f02..c20cb9f12d 100644
--- a/Source/Core/Core/Src/HW/DSP.h
+++ b/Source/Core/Core/Src/HW/DSP.h
@@ -27,7 +27,7 @@ enum DSPInterruptType
 {
 	INT_DSP		= 0,
 	INT_ARAM	= 1,
-	INT_AI		= 2
+	INT_AID		= 2
 };
 
 void Init();
@@ -49,8 +49,8 @@ u8 ReadARAM(const u32 _uAddress);
 // Debugger Helper
 u8* GetARAMPtr();
 
-
 void UpdateAudioDMA();
+int GetDSPSampleRate();
 
 }// end of namespace DSP
 
diff --git a/Source/Core/Core/Src/HW/GPFifo.cpp b/Source/Core/Core/Src/HW/GPFifo.cpp
index 19ddb43484..02b462e292 100644
--- a/Source/Core/Core/Src/HW/GPFifo.cpp
+++ b/Source/Core/Core/Src/HW/GPFifo.cpp
@@ -47,6 +47,10 @@ void Init()
 	ResetGatherPipe();
 }
 
+bool IsEmpty() {
+	return m_gatherPipeCount == 0;
+}
+
 void ResetGatherPipe()
 {
 	m_gatherPipeCount = 0;
diff --git a/Source/Core/Core/Src/HW/GPFifo.h b/Source/Core/Core/Src/HW/GPFifo.h
index a91767edce..b3377aeeed 100644
--- a/Source/Core/Core/Src/HW/GPFifo.h
+++ b/Source/Core/Core/Src/HW/GPFifo.h
@@ -40,6 +40,8 @@ void Init();
 void ResetGatherPipe();
 void CheckGatherPipe();
 
+bool IsEmpty();
+
 // Write
 void HWCALL Write8(const u8 _iValue, const u32 _iAddress);
 void HWCALL Write16(const u16 _iValue, const u32 _iAddress);
diff --git a/Source/Core/Core/Src/HW/SystemTimers.cpp b/Source/Core/Core/Src/HW/SystemTimers.cpp
index e37736b1e1..b079957b46 100644
--- a/Source/Core/Core/Src/HW/SystemTimers.cpp
+++ b/Source/Core/Core/Src/HW/SystemTimers.cpp
@@ -54,13 +54,22 @@ int
 	// VideoInterface::Update is stupid!
 	VI_PERIOD = GetTicksPerSecond() / (60*120), 
 
+	// TODO: The SI interfact actually has a register that determines the polling frequency.
+	// We should obey that instead of arbitrarly checking at 60fps.
 	SI_PERIOD = GetTicksPerSecond() / 60, //once a frame is good for controllers
 	
 	// These are the big question marks IMHO :)
+	// This one should simply be determined by the increasing counter in AI.
 	AI_PERIOD = GetTicksPerSecond() / 80,
+
+	// These shouldn't be period controlled either, most likely.
 	DSP_PERIOD = GetTicksPerSecond() / 250,
-	DSPINT_PERIOD = GetTicksPerSecond() / 230,
+
+	// This is completely arbitrary. If we find that we need lower latency, we can just
+	// increase this number.
     HLE_IPC_PERIOD = GetTicksPerSecond() / 250,
+
+	// This one is also fairly arbitrary. Every N cycles, run the GPU until it starves (in single core mode only).
 	GPU_PERIOD = 10000;
 
 u32 GetTicksPerSecond()
@@ -75,10 +84,28 @@ u32 ConvertMillisecondsToTicks(u32 _Milliseconds)
 
 void AICallback(u64 userdata, int cyclesLate)
 {
+	// Update disk streaming. All that code really needs a revamp, including replacing the codec with the one
+	// from in_cube.
 	AudioInterface::Update();
 	CoreTiming::ScheduleEvent(AI_PERIOD-cyclesLate, &AICallback, "AICallback");
 }
 
+void DSPCallback(u64 userdata, int cyclesLate)
+{
+	// Poke the DSP, make it do stuff. This should eventually be replaced with dsp->RunCycles(1000) or whatever,
+	// ~1/6th as many cycles as the period PPC-side.
+	PluginDSP::DSP_Update();
+	CoreTiming::ScheduleEvent(DSP_PERIOD-cyclesLate, &DSPCallback, "DSPCallback");
+}
+
+void AudioFifoCallback(u64 userdata, int cyclesLate)
+{	
+	int period = CPU_CORE_CLOCK / (DSP::GetDSPSampleRate() * 4 / 32);
+	DSP::UpdateAudioDMA();  // Push audio to speakers.
+
+	CoreTiming::ScheduleEvent(period - cyclesLate, &AudioFifoCallback, "AudioFifoCallback");
+}
+
 void IPC_HLE_UpdateCallback(u64 userdata, int cyclesLate)
 {	
     WII_IPC_HLE_Interface::Update();
@@ -100,18 +127,6 @@ void SICallback(u64 userdata, int cyclesLate)
 	CoreTiming::ScheduleEvent(SI_PERIOD-cyclesLate, &SICallback, "SICallback");
 }
 
-void DSPCallback(u64 userdata, int cyclesLate)
-{
-	PluginDSP::DSP_Update();
-	CoreTiming::ScheduleEvent(DSP_PERIOD-cyclesLate, &DSPCallback, "DSPCallback");
-}
-
-void DSPInterruptCallback(u64 userdata, int cyclesLate)
-{	
-	DSP::GenerateDSPInterrupt(DSP::INT_AI); 
-	CoreTiming::ScheduleEvent(DSPINT_PERIOD-cyclesLate, &DSPInterruptCallback, "DSPInterruptCallback");
-}
-
 void DecrementerCallback(u64 userdata, int cyclesLate)
 {
 	//Why is fakeDec too big here?
@@ -122,7 +137,6 @@ void DecrementerCallback(u64 userdata, int cyclesLate)
 
 void DecrementerSet()
 {
-	// MessageBox(0, "dec set",0,0);
 	u32 decValue = PowerPC::ppcState.spr[SPR_DEC];
 	fakeDec = decValue*TIMER_RATIO;
 	CoreTiming::RemoveEvent(DecrementerCallback);
@@ -131,10 +145,7 @@ void DecrementerSet()
 
 void AdvanceCallback(int cyclesExecuted)
 {
-	//int oldFakeDec = fakeDec;
 	fakeDec -= cyclesExecuted;
-	//if (fakeDec < 0 && oldFakeDec > 0)
-	//	fakeDec = TIMER_RATIO;
 	u64 timebase_ticks = CoreTiming::GetTicks() / TIMER_RATIO; //works since we are little endian and TL comes first :)
 	*(u64*)&TL = timebase_ticks;
  	if (fakeDec >= 0)
@@ -185,26 +196,24 @@ void Init()
 	if (Core::GetStartupParameter().bWii)
 	{
 		CPU_CORE_CLOCK = 721000000;
-		VI_PERIOD = GetTicksPerSecond() / (60*120), 
-		SI_PERIOD = GetTicksPerSecond() / 60, //once a frame is good for controllers
+		VI_PERIOD = GetTicksPerSecond() / (60*120); 
+		SI_PERIOD = GetTicksPerSecond() / 60; // once a frame is good for controllers
 		
 		// These are the big question marks IMHO :)
-		AI_PERIOD = GetTicksPerSecond() / 80,
-		DSP_PERIOD = (int)(GetTicksPerSecond() * 0.003f),
-		DSPINT_PERIOD = (int)(GetTicksPerSecond() * 0.003f);
+		AI_PERIOD = GetTicksPerSecond() / 80;
+		DSP_PERIOD = (int)(GetTicksPerSecond() * 0.003f);
 
         HLE_IPC_PERIOD = (int)(GetTicksPerSecond() * 0.003f);
 	}
 	else
 	{
 		CPU_CORE_CLOCK = 486000000;
-		VI_PERIOD = GetTicksPerSecond() / (60*120), 
-		SI_PERIOD = GetTicksPerSecond() / 60, //once a frame is good for controllers
+		VI_PERIOD = GetTicksPerSecond() / (60*120);
+		SI_PERIOD = GetTicksPerSecond() / 60; // once a frame is good for controllers
 		
 		// These are the big question marks IMHO :)
-		AI_PERIOD = GetTicksPerSecond() / 80,
-		DSP_PERIOD = (int)(GetTicksPerSecond() * 0.005f),
-		DSPINT_PERIOD = (int)(GetTicksPerSecond() *0.005f);
+		AI_PERIOD = GetTicksPerSecond() / 80;
+		DSP_PERIOD = (int)(GetTicksPerSecond() * 0.005f);
 	}
 	Common::Timer::IncreaseResolution();
 	memset(timeHistory, 0, sizeof(timeHistory));
@@ -216,7 +225,7 @@ void Init()
 	CoreTiming::ScheduleEvent(VI_PERIOD, &VICallback, "VICallback");
 	CoreTiming::ScheduleEvent(DSP_PERIOD, &DSPCallback, "DSPCallback");
 	CoreTiming::ScheduleEvent(SI_PERIOD, &SICallback, "SICallback");
-	CoreTiming::ScheduleEvent(DSPINT_PERIOD, &DSPInterruptCallback, "DSPInterruptCallback");
+	CoreTiming::ScheduleEvent(CPU_CORE_CLOCK / (32000 * 4 / 32), &AudioFifoCallback, "AudioFifoCallback");
 
 	if (!Core::GetStartupParameter().bUseDualCore)
 		CoreTiming::ScheduleEvent(GPU_PERIOD, &RunGPUCallback, "RunGPUCallback");
diff --git a/Source/Core/Core/Src/PowerPC/Interpreter/Interpreter_SystemRegisters.cpp b/Source/Core/Core/Src/PowerPC/Interpreter/Interpreter_SystemRegisters.cpp
index 7aa4f09f33..73d23191eb 100644
--- a/Source/Core/Core/Src/PowerPC/Interpreter/Interpreter_SystemRegisters.cpp
+++ b/Source/Core/Core/Src/PowerPC/Interpreter/Interpreter_SystemRegisters.cpp
@@ -234,11 +234,11 @@ void CInterpreter::mftb(UGeckoInstruction _inst)
 
 void CInterpreter::mfspr(UGeckoInstruction _inst)
 {
-	u32 iIndex = ((_inst.SPR & 0x1F) << 5) + ((_inst.SPR >> 5)&0x1F);
+	u32 iIndex = ((_inst.SPR & 0x1F) << 5) + ((_inst.SPR >> 5) & 0x1F);
 
 	//TODO - check processor privilege level - many of these require privilege
 	//XER LR CTR are the only ones available in user mode, time base can be read too.
-	//Not that gamecube games usually run user mode, but hey....
+	//Gamecube games always run in superuser mode, but hey....
 
 	switch (iIndex) 
 	{
@@ -247,11 +247,8 @@ void CInterpreter::mfspr(UGeckoInstruction _inst)
 	//	break;
 	case SPR_WPAR:
 		{
-			//There's supposed to be a bit here that indicates whether there is any data in the wp
-			//(or if it's full, not sure)
-			//MessageBox(NULL, "Read from SPR_WPAR", "????", MB_OK);
-			//Paper Mario reads here, this should be investigated ... TODO(ector)
-			bool wpar_empty = true;
+			// If wpar_empty ever is false, Paper Mario hangs. Strange.			
+			bool wpar_empty = true; //GPFifo::IsEmpty();
 			if (!wpar_empty)
 				rSPR(iIndex) |= 1;  // BNE = buffer not empty
 			else
@@ -268,9 +265,9 @@ void CInterpreter::mtspr(UGeckoInstruction _inst)
 	u32 oldValue = rSPR(iIndex);
 	rSPR(iIndex) = m_GPR[_inst.RD];
 
-	//TODO(ector) - check processor privilege level - many of these require privilege
-	//XER LR CTR are the only ones available in user mode
-	//Not that gamecube games usually run user mode, but hey....
+	//TODO - check processor privilege level - many of these require privilege
+	//XER LR CTR are the only ones available in user mode, time base can be read too.
+	//Gamecube games always run in superuser mode, but hey....
 
 	//Our DMA emulation is highly inaccurate - instead of properly emulating the queue
 	//and so on, we simply make all DMA:s complete instantaneously.
@@ -306,7 +303,6 @@ void CInterpreter::mtspr(UGeckoInstruction _inst)
 			//TODO(ector): Protect LC memory if LCE is false.
 			//TODO(ector): Honor PSE.
 
-			//
 			//_assert_msg_(GEKKO, WriteGatherPipeEnable, "Write gather pipe not enabled!");
 			//if ((HID2.PSE == 0))
 			//	MessageBox(NULL, "PSE in HID2 is set", "Warning", MB_OK);
diff --git a/Source/Core/Core/Src/PowerPC/Jit64/Jit_SystemRegisters.cpp b/Source/Core/Core/Src/PowerPC/Jit64/Jit_SystemRegisters.cpp
index 4ce970def6..8b3bab2b1a 100644
--- a/Source/Core/Core/Src/PowerPC/Jit64/Jit_SystemRegisters.cpp
+++ b/Source/Core/Core/Src/PowerPC/Jit64/Jit_SystemRegisters.cpp
@@ -77,7 +77,9 @@ namespace Jit64
 		int d = inst.RD;
 		switch (iIndex)
 		{
-
+		case SPR_WPAR:
+			Default(inst);
+			return;
 //		case SPR_DEC:
 			//MessageBox(NULL, "Read from DEC", "????", MB_OK);
 			//break;
diff --git a/Source/Plugins/Plugin_VideoOGL/Plugin_VideoOGL.vcproj b/Source/Plugins/Plugin_VideoOGL/Plugin_VideoOGL.vcproj
index a9a31c11e4..fae04ccc98 100644
--- a/Source/Plugins/Plugin_VideoOGL/Plugin_VideoOGL.vcproj
+++ b/Source/Plugins/Plugin_VideoOGL/Plugin_VideoOGL.vcproj
@@ -1048,6 +1048,10 @@
 			RelativePath=".\Src\Render.h"
 			>
 		</File>
+		<File
+			RelativePath=".\Src\SConscript"
+			>
+		</File>
 		<File
 			RelativePath=".\Src\XFB.cpp"
 			>
diff --git a/Source/Plugins/Plugin_VideoOGL/Src/Globals.cpp b/Source/Plugins/Plugin_VideoOGL/Src/Globals.cpp
index 82c75bf7bd..9e160eebb6 100644
--- a/Source/Plugins/Plugin_VideoOGL/Src/Globals.cpp
+++ b/Source/Plugins/Plugin_VideoOGL/Src/Globals.cpp
@@ -15,6 +15,13 @@
 // Official SVN repository and contact information can be found at
 // http://code.google.com/p/dolphin-emu/
 
+
+#include <wx/wx.h>
+#include <wx/filepicker.h>
+#include <wx/notebook.h>
+#include <wx/dialog.h>
+#include <wx/aboutdlg.h>
+
 #include "Globals.h"
 
 #include "pluginspecs_video.h"
diff --git a/Source/Plugins/Plugin_VideoOGL/Src/Globals.h b/Source/Plugins/Plugin_VideoOGL/Src/Globals.h
index 98d9d89a74..11a38bb936 100644
--- a/Source/Plugins/Plugin_VideoOGL/Src/Globals.h
+++ b/Source/Plugins/Plugin_VideoOGL/Src/Globals.h
@@ -19,12 +19,6 @@
 #ifndef _GLOBALS_H
 #define _GLOBALS_H
 
-#include <wx/wx.h>
-#include <wx/filepicker.h>
-#include <wx/notebook.h>
-#include <wx/dialog.h>
-#include <wx/aboutdlg.h>
-
 #define LOGGING
 
 #include "Common.h"
@@ -32,13 +26,6 @@
 
 #ifdef _WIN32
 
-#include <windows.h>
-#include <windowsx.h>
-#include <mmsystem.h>
-
-void OpenConsole();
-void CloseConsole();
-
 #define GLEW_STATIC
 
 #include "GLew/glew.h"
@@ -97,16 +84,8 @@ struct RECT
 #define GL_DEPTH24_STENCIL8_EXT 0x88F0
 #define GL_TEXTURE_STENCIL_SIZE_EXT 0x88F1
 #endif
-extern float MValue;
-// several macros
-// PLEASE DO NOT USE THE FOLLOWING SAFE*
-// They only encourage sloppy coding, and hide bugs. 
-#define SAFE_RELEASE_CGPROG(x) { if( (x) != NULL ) { cgDestroyProgram(x); x = NULL; } }
-#define SAFE_RELEASE_PROG(x) { if( (x) != 0 ) { glDeleteProgramsARB(1, &(x)); x = 0; } }
-#define SAFE_RELEASE_TEX(x) { if( (x) != 0 ) { glDeleteTextures(1, &(x)); x = 0; } }
-#define SAFE_RELEASE_BUF(x) { if( (x) != 0 ) { glDeleteBuffers(1, &(x)); x = 0; } }
 
-#define FORIT(it, v) for(it = (v).begin(); it != (v).end(); ++(it))
+extern float MValue;
 
 #define ERROR_LOG __Log
 
diff --git a/Source/Plugins/Plugin_VideoOGL/Src/OS/Win32.cpp b/Source/Plugins/Plugin_VideoOGL/Src/OS/Win32.cpp
index f4c3a45e68..54e3f9e993 100644
--- a/Source/Plugins/Plugin_VideoOGL/Src/OS/Win32.cpp
+++ b/Source/Plugins/Plugin_VideoOGL/Src/OS/Win32.cpp
@@ -18,11 +18,20 @@
 
 #include <windows.h>
 
+#include <wx/wx.h>
+#include <wx/filepicker.h>
+#include <wx/notebook.h>
+#include <wx/dialog.h>
+#include <wx/aboutdlg.h>
+
 #include "../Globals.h"
 
 #include "../../Core/Src/Core.h"
 #include "Win32.h"
 
+void OpenConsole();
+void CloseConsole();
+
 HINSTANCE g_hInstance;
 
 class wxDLLApp : public wxApp
diff --git a/Source/Plugins/Plugin_VideoOGL/Src/PixelShaderManager.cpp b/Source/Plugins/Plugin_VideoOGL/Src/PixelShaderManager.cpp
index 63018f6fdc..2f4b43738a 100644
--- a/Source/Plugins/Plugin_VideoOGL/Src/PixelShaderManager.cpp
+++ b/Source/Plugins/Plugin_VideoOGL/Src/PixelShaderManager.cpp
@@ -94,16 +94,17 @@ void PixelShaderMngr::Init()
     GL_REPORT_ERROR();
     if( err != GL_NO_ERROR ) {
         ERROR_LOG("Failed to create color matrix fragment program\n");
-        
-        SAFE_RELEASE_PROG(s_ColorMatrixProgram);
+        glDeleteProgramsARB(1, &s_ColorMatrixProgram);
+		s_ColorMatrixProgram = 0;
     }
 }
 
 void PixelShaderMngr::Shutdown()
 {
-    SAFE_RELEASE_PROG(s_ColorMatrixProgram);
+    glDeleteProgramsARB(1, &s_ColorMatrixProgram);
+	s_ColorMatrixProgram = 0;
     PSCache::iterator iter = pshaders.begin();
-    for (;iter!=pshaders.end();iter++)
+    for (; iter != pshaders.end(); iter++)
         iter->second.Destroy();
     pshaders.clear();
 }
diff --git a/Source/Plugins/Plugin_VideoOGL/Src/TextureMngr.cpp b/Source/Plugins/Plugin_VideoOGL/Src/TextureMngr.cpp
index 37b868f615..ae4b672e3c 100644
--- a/Source/Plugins/Plugin_VideoOGL/Src/TextureMngr.cpp
+++ b/Source/Plugins/Plugin_VideoOGL/Src/TextureMngr.cpp
@@ -80,14 +80,15 @@ void TextureMngr::TCacheEntry::SetTextureParameters(TexMode0& newmode)
 
 void TextureMngr::TCacheEntry::Destroy()
 {
-    SAFE_RELEASE_TEX(texture);
+    glDeleteTextures(1, &texture);
+	texture = 0;
 }
 
 void TextureMngr::Init()
 {
     temp = (u8*)AllocateMemoryPages(TEMP_SIZE);
     nTex2DEnabled = nTexRECTEnabled = 0;
-	TexDecoder_SetTexFmtOverlayOptions(g_Config.bTexFmtOverlayEnable,g_Config.bTexFmtOverlayCenter);
+	TexDecoder_SetTexFmtOverlayOptions(g_Config.bTexFmtOverlayEnable, g_Config.bTexFmtOverlayCenter);
 }
 
 void TextureMngr::Invalidate()
@@ -96,7 +97,7 @@ void TextureMngr::Invalidate()
     for (;iter!=textures.end();iter++)
         iter->second.Destroy();
     textures.clear();
-	TexDecoder_SetTexFmtOverlayOptions(g_Config.bTexFmtOverlayEnable,g_Config.bTexFmtOverlayCenter);
+	TexDecoder_SetTexFmtOverlayOptions(g_Config.bTexFmtOverlayEnable, g_Config.bTexFmtOverlayCenter);
 }
 
 void TextureMngr::Shutdown()
diff --git a/Source/Plugins/Plugin_VideoOGL/Src/VertexLoader.cpp b/Source/Plugins/Plugin_VideoOGL/Src/VertexLoader.cpp
index eb89f5a642..1cf5224d1d 100644
--- a/Source/Plugins/Plugin_VideoOGL/Src/VertexLoader.cpp
+++ b/Source/Plugins/Plugin_VideoOGL/Src/VertexLoader.cpp
@@ -1087,7 +1087,7 @@ void VertexManager::Flush()
 
     int offset = 0;
     vector< pair<int, int> >::iterator it;
-    FORIT (it, s_vStoredPrimitives) {
+    for (it = s_vStoredPrimitives.begin(); it != s_vStoredPrimitives.end(); ++it) {
         glDrawArrays(it->first, offset, it->second);
         offset += it->second;
     }