Optimizations: Improved emulation performance (5-15% speed boost depending on game)

Core/BaseApuChannel.h

@@ -90,7 +90,7 @@ public:
 		return 0;
 	}
 
-	virtual void AddOutput(int8_t output)
+	void AddOutput(int8_t output)
 	{
 		if(output != _lastOutput) {
 			_mixer->AddDelta(_channel, _previousCycle, output - _lastOutput);

Core/MMC5.h

@@ -171,42 +171,40 @@ private:
 
 	void ProcessCpuClock() override
 	{
-		if(!PPU::GetControlFlags().BackgroundEnabled && !PPU::GetControlFlags().SpritesEnabled) {
-			_ppuInFrame = false;
-		}
-
 		_audio.Clock();
 	}
 
 	virtual void NotifyVRAMAddressChange(uint16_t addr) override
 	{
-		if(_spriteFetch != IsSpriteFetch() || _largeSprites != PPU::GetControlFlags().LargeSprites) {
-			if(PPU::GetControlFlags().BackgroundEnabled || PPU::GetControlFlags().SpritesEnabled) {
+		if(PPU::GetControlFlags().BackgroundEnabled || PPU::GetControlFlags().SpritesEnabled) {
+			if(_spriteFetch != IsSpriteFetch() || _largeSprites != PPU::GetControlFlags().LargeSprites) {
 				UpdateChrBanks();
 			}
-		}
 
-		int16_t currentScanline = PPU::GetCurrentScanline();
-		if(currentScanline != _previousScanline) {
-			if(currentScanline >= 239 || currentScanline < 0) {
-				_ppuInFrame = false;
-			} else {
-				if(!_ppuInFrame) {
-					_ppuInFrame = true;
-					_irqCounter = 0;
-					_irqPending = false;
-					CPU::ClearIRQSource(IRQSource::External);
+			int16_t currentScanline = PPU::GetCurrentScanline();
+			if(currentScanline != _previousScanline) {
+				if(currentScanline >= 239 || currentScanline < 0) {
+					_ppuInFrame = false;
 				} else {
-					_irqCounter++;
-					if(_irqCounter == _irqCounterTarget) {
-						_irqPending = true;
-						if(_irqEnabled) {
-							CPU::SetIRQSource(IRQSource::External);
+					if(!_ppuInFrame) {
+						_ppuInFrame = true;
+						_irqCounter = 0;
+						_irqPending = false;
+						CPU::ClearIRQSource(IRQSource::External);
+					} else {
+						_irqCounter++;
+						if(_irqCounter == _irqCounterTarget) {
+							_irqPending = true;
+							if(_irqEnabled) {
+								CPU::SetIRQSource(IRQSource::External);
+							}
 						}
 					}
 				}
+				_previousScanline = currentScanline;
 			}
-			_previousScanline = currentScanline;
+		} else {
+			_ppuInFrame = false;
 		}
 	}
 

Core/MMC5Audio.h

@@ -15,20 +15,11 @@ private:
 		//"$5001 has no effect. The MMC5 pulse channels will not sweep, as they have no sweep unit."
 	}
 
-	bool IsMuted() override
-	{
-		//"Frequency values less than 8 do not silence the MMC5 pulse channels; they can output ultrasonic frequencies."
-		return false;
-	}
 public:
 	MMC5Square() : SquareChannel(AudioChannel::MMC5, nullptr, false)
 	{
 		_currentOutput = 0;
-	}
-	
-	virtual void AddOutput(int8_t output) override
-	{
-		_currentOutput = output;
+		_isMmc5Square = true;
 	}
 
 	int8_t GetOutput()
@@ -38,8 +29,14 @@ public:
 
 	void RunChannel()
 	{
-		Run(1);
-		EndFrame();
+		if(_timer == 0) {
+			_dutyPos = (_dutyPos - 1) & 0x07;
+			//"Frequency values less than 8 do not silence the MMC5 pulse channels; they can output ultrasonic frequencies."
+			_currentOutput = _dutySequences[_duty][_dutyPos] * GetVolume();
+			_timer = _period;
+		} else {
+			_timer--;
+		}
 	}
 };
 
@@ -80,8 +77,10 @@ protected:
 		}
 
 		int16_t summedOutput = (_square1.GetOutput() + _square2.GetOutput()) * 4 + _pcmOutput;
-		APU::AddExpansionAudioDelta(AudioChannel::MMC5, summedOutput - _lastOutput);
-		_lastOutput = summedOutput;
+		if(summedOutput != _lastOutput) {
+			APU::AddExpansionAudioDelta(AudioChannel::MMC5, summedOutput - _lastOutput);
+			_lastOutput = summedOutput;
+		}
 
 		_square1.ReloadCounter();
 		_square2.ReloadCounter();

Core/PPU.cpp

@@ -577,6 +577,10 @@ void PPU::LoadSprite(uint8_t spriteY, uint8_t tileIndex, uint8_t attributes, uin
 		info.TileAddr = tileAddr;
 		info.OffsetY = lineOffset;
 		info.SpriteX = spriteX;
+		
+		for(int i = 0; i < 8 && spriteX + i + 1 < 257; i++) {
+			_hasSprite[spriteX + i + 1] = true;
+		}
 	} 
 	
 	if(fetchLastSprite) {
@@ -650,10 +654,10 @@ uint32_t PPU::GetPixelColor()
 		}
 	}
 
-	if(_cycle > _minimumDrawSpriteCycle) {
+	if(_hasSprite[_cycle] && _cycle > _minimumDrawSpriteCycle) {
 		//SpriteMask = true: Hide sprites in leftmost 8 pixels of screen
 		for(uint8_t i = 0; i < _spriteCount; i++) {
-			int32_t shift = -((int32_t)_spriteTiles[i].SpriteX - (int32_t)_cycle + 1);
+			int32_t shift = (int32_t)_cycle - _spriteTiles[i].SpriteX - 1;
 			if(shift >= 0 && shift < 8) {
 				_lastSprite = &_spriteTiles[i];
 				uint32_t spriteColor;
@@ -699,10 +703,7 @@ void PPU::DrawPixel()
 
 void PPU::ProcessScanline()
 {
-	if(_scanline == -1 && _cycle == 0) {
-		_statusFlags.SpriteOverflow = false;
-		_statusFlags.Sprite0Hit = false;
-	} else if(_cycle > 0 && _cycle <= 256) {
+	if(_cycle > 0 && _cycle <= 256) {
 		LoadTileInfo();
 
 		if(_prevRenderingEnabled && (_cycle & 0x07) == 0) {
@@ -723,6 +724,7 @@ void PPU::ProcessScanline()
 	} else if(_cycle >= 257 && _cycle <= 320) {
 		if(_cycle == 257) {
 			_spriteIndex = 0;
+			memset(_hasSprite, 0, sizeof(_hasSprite));
 			if(_prevRenderingEnabled) {
 				//copy horizontal scrolling value from t
 				_state.VideoRamAddr = (_state.VideoRamAddr & ~0x041F) | (_state.TmpVideoRamAddr & 0x041F);
@@ -773,6 +775,9 @@ void PPU::ProcessScanline()
 				_cycle = 340;
 			}
 		}
+	} else if(_scanline == -1 && _cycle == 0) {
+		_statusFlags.SpriteOverflow = false;
+		_statusFlags.Sprite0Hit = false;
 	}
 }
 
@@ -1047,5 +1052,9 @@ void PPU::StreamState(bool saving)
 
 		SetNesModel(_nesModel);
 		UpdateMinimumDrawCycles();
+
+		for(int i = 0; i < 257; i++) {
+			_hasSprite[i] = true;
+		}
 	}
 }

Core/PPU.h

@@ -41,6 +41,7 @@ class PPU : public IMemoryHandler, public Snapshotable
 
 		uint8_t _spriteRAM[0x100];
 		uint8_t _secondarySpriteRAM[0x20];
+		bool _hasSprite[257];
 
 		uint16_t *_currentOutputBuffer;
 		uint16_t *_outputBuffers[2];

Core/SoundMixer.cpp

@@ -84,7 +84,13 @@ void SoundMixer::PlayAudioBuffer(uint32_t time)
 	EndFrame(time);
 
 	size_t sampleCount = blip_read_samples(_blipBufLeft, _outputBuffer, SoundMixer::MaxSamplesPerFrame, 1);
-	blip_read_samples(_blipBufRight, _outputBuffer + 1, SoundMixer::MaxSamplesPerFrame, 1);
+	if(_hasPanning) {
+		blip_read_samples(_blipBufRight, _outputBuffer + 1, SoundMixer::MaxSamplesPerFrame, 1);
+	} else {
+		for(int i = 0; i < sampleCount * 2; i+=2) {
+			_outputBuffer[i + 1] = _outputBuffer[i];
+		}
+	}
 
 	//Apply low pass filter/volume reduction when in background (based on options)
 	if(!VideoDecoder::GetInstance()->IsRecording() && !_waveRecorder && !EmulationSettings::CheckFlag(EmulationFlags::NsfPlayerEnabled) && EmulationSettings::CheckFlag(EmulationFlags::InBackground)) {
@@ -129,8 +135,9 @@ void SoundMixer::PlayAudioBuffer(uint32_t time)
 		//Update sample rate for next frame if setting changed
 		_sampleRate = EmulationSettings::GetSampleRate();
 		UpdateRates(true);
+	} else {
+		UpdateRates(false);
 	}
-	UpdateRates(false);
 }
 
 void SoundMixer::SetNesModel(NesModel model)
@@ -151,6 +158,20 @@ void SoundMixer::UpdateRates(bool forceUpdate)
 		blip_set_rates(_blipBufLeft, _clockRate, _sampleRate);
 		blip_set_rates(_blipBufRight, _clockRate, _sampleRate);
 	}
+
+	bool hasPanning = false;
+	for(uint32_t i = 0; i < MaxChannelCount; i++) {
+		_volumes[i] = EmulationSettings::GetChannelVolume((AudioChannel)i);
+		_panning[i] = EmulationSettings::GetChannelPanning((AudioChannel)i);
+		if(_panning[i] != 1.0) {
+			if(_hasPanning) {
+				blip_clear(_blipBufLeft);
+				blip_clear(_blipBufRight);
+			}
+			_hasPanning = true;
+		}
+	}
+	_hasPanning = hasPanning;
 }
 
 double SoundMixer::GetChannelOutput(AudioChannel channel, bool forRightChannel)
@@ -167,8 +188,8 @@ int16_t SoundMixer::GetOutputVolume(bool forRightChannel)
 	double squareOutput = GetChannelOutput(AudioChannel::Square1, forRightChannel) + GetChannelOutput(AudioChannel::Square2, forRightChannel);
 	double tndOutput = 3 * GetChannelOutput(AudioChannel::Triangle, forRightChannel) + 2 * GetChannelOutput(AudioChannel::Noise, forRightChannel) + GetChannelOutput(AudioChannel::DMC, forRightChannel);
 
-	uint16_t squareVolume = (uint16_t)(95.52 / (8128.0 / squareOutput + 100.0) * 5000);
-	uint16_t tndVolume = (uint16_t)(163.67 / (24329.0 / tndOutput + 100.0) * 5000);
+	uint16_t squareVolume = (uint16_t)(477600 / (8128.0 / squareOutput + 100.0));
+	uint16_t tndVolume = (uint16_t)(818350 / (24329.0 / tndOutput + 100.0));
 	
 	return (int16_t)(squareVolume + tndVolume +
 		GetChannelOutput(AudioChannel::FDS, forRightChannel) * 20 +
@@ -189,7 +210,7 @@ void SoundMixer::AddDelta(AudioChannel channel, uint32_t time, int16_t delta)
 
 void SoundMixer::EndFrame(uint32_t time)
 {
-	double masterVolume = EmulationSettings::GetMasterVolume();
+	double masterVolume = EmulationSettings::GetMasterVolume() * _fadeRatio;
 	sort(_timestamps.begin(), _timestamps.end());
 	_timestamps.erase(std::unique(_timestamps.begin(), _timestamps.end()), _timestamps.end());
 
@@ -206,16 +227,20 @@ void SoundMixer::EndFrame(uint32_t time)
 		}
 
 		int16_t currentOutput = GetOutputVolume(false);
-		blip_add_delta(_blipBufLeft, stamp, (int)((currentOutput - _previousOutputLeft) * masterVolume * _fadeRatio));
+		blip_add_delta(_blipBufLeft, stamp, (int)((currentOutput - _previousOutputLeft) * masterVolume));
 		_previousOutputLeft = currentOutput;
 
-		currentOutput = GetOutputVolume(true);
-		blip_add_delta(_blipBufRight, stamp, (int)((currentOutput - _previousOutputRight) * masterVolume * _fadeRatio));
-		_previousOutputRight = currentOutput;
+		if(_hasPanning) {
+			currentOutput = GetOutputVolume(true);
+			blip_add_delta(_blipBufRight, stamp, (int)((currentOutput - _previousOutputRight) * masterVolume));
+			_previousOutputRight = currentOutput;
+		}
 	}
 
 	blip_end_frame(_blipBufLeft, time);
-	blip_end_frame(_blipBufRight, time);
+	if(_hasPanning) {
+		blip_end_frame(_blipBufRight, time);
+	}
 
 	if(muteFrame) {
 		_muteFrameCount++;
@@ -224,11 +249,6 @@ void SoundMixer::EndFrame(uint32_t time)
 	}
 
 	//Reset everything
-	for(uint32_t i = 0; i < MaxChannelCount; i++) {
-		_volumes[i] = EmulationSettings::GetChannelVolume((AudioChannel)i);
-		_panning[i] = EmulationSettings::GetChannelPanning((AudioChannel)i);
-	}
-
 	_timestamps.clear();
 	memset(_channelOutput, 0, sizeof(_channelOutput));
 }

Core/SoundMixer.h

@@ -52,6 +52,8 @@ private:
 	uint32_t _sampleRate;
 	uint32_t _clockRate;
 
+	bool _hasPanning;
+
 	double GetChannelOutput(AudioChannel channel, bool forRightChannel);
 	int16_t GetOutputVolume(bool forRightChannel);
 	void EndFrame(uint32_t time);

Core/SquareChannel.h

@@ -6,7 +6,7 @@
 
 class SquareChannel : public ApuEnvelope
 {
-private:
+protected:
 	const uint8_t _dutySequences[4][8] = {
 		{ 0, 0, 0, 0, 0, 0, 0, 1 },
 		{ 0, 0, 0, 0, 0, 0, 1, 1 },
@@ -15,6 +15,7 @@ private:
 	};
 
 	bool _isChannel1 = false;
+	bool _isMmc5Square = false;
 
 	uint8_t _duty = 0;
 	uint8_t _dutyPos = 0;
@@ -28,7 +29,7 @@ private:
 	uint32_t _sweepTargetPeriod = 0;
 	uint16_t _realPeriod = 0;
 	
-	virtual bool IsMuted()
+	bool IsMuted()
 	{
 		//A period of t < 8, either set explicitly or via a sweep period update, silences the corresponding pulse channel.
 		return _realPeriod < 8 || (!_sweepNegate && _sweepTargetPeriod > 0x7FF);
@@ -161,8 +162,10 @@ public:
 				ResetEnvelope();
 				break;
 		}
-
-		UpdateOutput();
+		
+		if(!_isMmc5Square) {
+			UpdateOutput();
+		}
 	}
 
 	void TickSweep()