diff --git a/src/ABC.cpp b/src/ABC.cpp
index 774f8d3..2d02464 100644
--- a/src/ABC.cpp
+++ b/src/ABC.cpp
@@ -1,4 +1,5 @@
 #include "plugin.hpp"
+#include "simd_mask.hpp"
 
 #define MAX(a,b) (a>b)?a:b
 
@@ -65,7 +66,9 @@ struct ABC : Module {
 		NUM_LIGHTS
 	};
 
-	simd::float_4 mask[4];
+	// simd::float_4 mask[4];
+	
+	ChannelMask channelMask;
 
 
 	ABC() {
@@ -75,27 +78,19 @@ struct ABC : Module {
 		configParam(B2_LEVEL_PARAM, -1.0, 1.0, 0.0, "B2 Level");
 		configParam(C2_LEVEL_PARAM, -1.0, 1.0, 0.0, "C2 Level");
 
+/* 
 		__m128i tmp =  _mm_cmpeq_epi16(_mm_set_epi32(0,0,0,0),_mm_set_epi32(0,0,0,0));
-
+		
 		for(int i=0; i<4; i++) {
 			mask[3-i] = simd::float_4(_mm_castsi128_ps(tmp));
 			tmp = _mm_srli_si128(tmp, 4);
 		}
+*/
 
 	}
 
-	inline void load_input(Input &in, simd::float_4 *v, int numChannels) {
-		if(numChannels==1) {
-			for(int i=0; i<4; i++) v[i] = simd::float_4(in.getVoltage());
-		} else {
-			for(int c=0; c<numChannels; c+=4) v[c/4] = simd::float_4::load(in.getVoltages(c));
-		}
-	}
 
-	inline void crop_channels(simd::float_4 *vec, int numChannels) {
-		int c=numChannels/4;
-		vec[c] = simd::float_4(_mm_and_ps(vec[c].v, mask[numChannels-4*c].v));
-	}
+
 
 	void process(const ProcessArgs &args) override {
 
@@ -123,7 +118,7 @@ struct ABC : Module {
 		channels_1 = MAX(channels_1, channels_B1);
 		channels_1 = MAX(channels_1, channels_C1);
 
-		int channels_2 = channels_1;
+		int channels_2 = 1;
 		channels_2 = MAX(channels_2, channels_A2);
 		channels_2 = MAX(channels_2, channels_B2);
 		channels_2 = MAX(channels_2, channels_C2);
@@ -136,7 +131,7 @@ struct ABC : Module {
 
 
 		load_input(inputs[A1_INPUT], a1, channels_A1);
-		crop_channels(a1, channels_1);
+		channelMask.apply(a1, channels_1);
 
 		if(inputs[B1_INPUT].isConnected()) {
 			load_input(inputs[B1_INPUT], b1, channels_B1);
@@ -144,7 +139,7 @@ struct ABC : Module {
 		} else {
 			for(int c=0; c<channels_1; c+=4) b1[c/4] = simd::float_4(5.f*mult_B1);
 		}
-		crop_channels(b1, channels_1);
+		channelMask.apply(b1, channels_1);
 
 		if(inputs[C1_INPUT].isConnected()) {
 			load_input(inputs[C1_INPUT], c1, channels_C1);
@@ -152,14 +147,14 @@ struct ABC : Module {
 		} else {
 			for(int c=0; c<channels_1; c+=4) c1[c/4] = simd::float_4(10.f*mult_C1);
 		}
-		crop_channels(c1, channels_1);
+		channelMask.apply(c1, channels_1);
 
 		for(int c=0; c<channels_1; c+=4) out1[c/4] = clip4(a1[c/4] * b1[c/4] + c1[c/4]);
 
 
 
-		load_input(inputs[A1_INPUT], a1, channels_A1);
-		crop_channels(a1, channels_1);
+		load_input(inputs[A2_INPUT], a2, channels_A2);
+		channelMask.apply(a2, channels_2);
 
 		if(inputs[B2_INPUT].isConnected()) {
 			load_input(inputs[B2_INPUT], b2, channels_B2);
@@ -167,7 +162,7 @@ struct ABC : Module {
 		} else {
 			for(int c=0; c<channels_2; c+=4) b2[c/4] = simd::float_4(5.f*mult_B2);
 		}
-		crop_channels(b2, channels_2);
+		channelMask.apply(b2, channels_2);
 
 		if(inputs[C2_INPUT].isConnected()) {
 			load_input(inputs[C2_INPUT], c2, channels_C2);
@@ -175,7 +170,7 @@ struct ABC : Module {
 		} else {
 			for(int c=0; c<channels_2; c+=4) c2[c/4] = simd::float_4(10.f*mult_C2);
 		}
-		crop_channels(c2, channels_2);
+		channelMask.apply(c2, channels_2);
 
 		for(int c=0; c<channels_2; c+=4) out2[c/4] = clip4(a2[c/4] * b2[c/4] + c2[c/4]);
 
@@ -188,10 +183,12 @@ struct ABC : Module {
 		}
 		else {
 			for(int c=0; c<channels_1; c+=4) out2[c/4] += out1[c/4];
+			channels_2 = MAX(channels_1, channels_2);
+
 		}
 		if (outputs[OUT2_OUTPUT].isConnected()) {
-			outputs[OUT2_OUTPUT].setChannels(channels_1);
-			for(int c=0; c<channels_1; c+=4) out2[c/4].store(outputs[OUT2_OUTPUT].getVoltages(c));
+			outputs[OUT2_OUTPUT].setChannels(channels_2);
+			for(int c=0; c<channels_2; c+=4) out2[c/4].store(outputs[OUT2_OUTPUT].getVoltages(c));
 		}
 
 		// Lights
diff --git a/src/DualAtenuverter.cpp b/src/DualAtenuverter.cpp
index 4e81fe0..08ae9e6 100644
--- a/src/DualAtenuverter.cpp
+++ b/src/DualAtenuverter.cpp
@@ -1,4 +1,5 @@
 #include "plugin.hpp"
+#include "simd_mask.hpp"
 
 
 struct DualAtenuverter : Module {
@@ -49,7 +50,6 @@ struct DualAtenuverter : Module {
 		simd::float_4 offset1 = simd::float_4(params[OFFSET1_PARAM].getValue());
 		simd::float_4 offset2 = simd::float_4(params[OFFSET2_PARAM].getValue());
 
-
 		for (int c = 0; c < channels1; c += 4) out1[c / 4] = clamp(simd::float_4::load(inputs[IN1_INPUT].getVoltages(c)) * att1 + offset1, -10.f, 10.f);
 		for (int c = 0; c < channels2; c += 4) out2[c / 4] = clamp(simd::float_4::load(inputs[IN2_INPUT].getVoltages(c)) * att2 + offset2, -10.f, 10.f);
 		
diff --git a/src/EvenVCO.cpp b/src/EvenVCO.cpp
index 088c6a4..d3ac1a5 100644
--- a/src/EvenVCO.cpp
+++ b/src/EvenVCO.cpp
@@ -1,4 +1,5 @@
 #include "plugin.hpp"
+#include "simd_mask.hpp"
 
 #define MAX(a,b) (a>b)?a:b
 
@@ -26,38 +27,73 @@ struct EvenVCO : Module {
 		NUM_OUTPUTS
 	};
 
-	float phase = 0.0;
+	simd::float_4 phase[4];
+	simd::float_4 tri[4];
+
 	/** The value of the last sync input */
 	float sync = 0.0;
 	/** The outputs */
-	float tri = 0.0;
 	/** Whether we are past the pulse width already */
-	bool halfPhase = false;
+	bool halfPhase[PORT_MAX_CHANNELS];
 
-	dsp::MinBlepGenerator<16, 32> triSquareMinBlep;
-	dsp::MinBlepGenerator<16, 32> triMinBlep;
-	dsp::MinBlepGenerator<16, 32> sineMinBlep;
-	dsp::MinBlepGenerator<16, 32> doubleSawMinBlep;
-	dsp::MinBlepGenerator<16, 32> sawMinBlep;
-	dsp::MinBlepGenerator<16, 32> squareMinBlep;
+	dsp::MinBlepGenerator<16, 32> triSquareMinBlep[PORT_MAX_CHANNELS];
+	dsp::MinBlepGenerator<16, 32> triMinBlep[PORT_MAX_CHANNELS];
+	dsp::MinBlepGenerator<16, 32> sineMinBlep[PORT_MAX_CHANNELS];
+	dsp::MinBlepGenerator<16, 32> doubleSawMinBlep[PORT_MAX_CHANNELS];
+	dsp::MinBlepGenerator<16, 32> sawMinBlep[PORT_MAX_CHANNELS];
+	dsp::MinBlepGenerator<16, 32> squareMinBlep[PORT_MAX_CHANNELS];
 
 	dsp::RCFilter triFilter;
 
+	ChannelMask channelMask;
+
 	EvenVCO() {
 		config(NUM_PARAMS, NUM_INPUTS, NUM_OUTPUTS);
 		configParam(OCTAVE_PARAM, -5.0, 4.0, 0.0, "Octave", "'", 0.5);
 		configParam(TUNE_PARAM, -7.0, 7.0, 0.0, "Tune", " semitones");
 		configParam(PWM_PARAM, -1.0, 1.0, 0.0, "Pulse width");
+		
+		for(int i=0; i<4; i++) {
+			phase[i] = simd::float_4(0.0f);
+			tri[i] = simd::float_4(0.0f);
+		}
+		for(int c=0; c<PORT_MAX_CHANNELS; c++) halfPhase[c] = false;
+	}
 
+/* 
+	inline void load_input(Input &in, simd::float_4 *v, int numChannels) {
+		if(numChannels==1) {
+			for(int i=0; i<4; i++) v[i] = simd::float_4(in.getVoltage());
+		} else {
+			for(int c=0; c<numChannels; c+=4) v[c/4] = simd::float_4::load(in.getVoltages(c));
+		}
+	}
+*/
 
+	inline void add_input(Input &in, simd::float_4 *v, int numChannels) {
+		if(numChannels==1) {
+			for(int i=0; i<4; i++) v[i] = simd::float_4(in.getVoltage());
+		} else {
+			for(int c=0; c<numChannels; c+=4) v[c/4] = simd::float_4::load(in.getVoltages(c));
+		}
 	}
 
 	void process(const ProcessArgs &args) override {
-		// Compute frequency, pitch is 1V/oct
+
+        simd::float_4 pitch[4];
+		simd::float_4 pitch_1[4];
+		simd::float_4 pitch_2[4];
+		simd::float_4 pitch_fm[4];
+		simd::float_4 freq[4]; 
+		simd::float_4 pw[4];
+		simd::float_4 pwm[4];
+		simd::float_4 deltaPhase[4];
+		simd::float_4 oldPhase[4];
 
 		int channels_pitch1 = inputs[PITCH1_INPUT].getChannels();
 		int channels_pitch2 = inputs[PITCH2_INPUT].getChannels();
 		int channels_fm     = inputs[FM_INPUT].getChannels();
+		int channels_pwm	= inputs[PWM_INPUT].getChannels();
 
 		int channels = 1;
 		channels = MAX(channels, channels_pitch1);
@@ -66,68 +102,151 @@ struct EvenVCO : Module {
 
 		float pitch_0 = 1.f + std::round(params[OCTAVE_PARAM].getValue()) + params[TUNE_PARAM].getValue() / 12.f;
 
+		// Compute frequency, pitch is 1V/oct
+
+		for(int c=0; c<channels; c+=4) pitch[c/4] = simd::float_4(pitch_0);
+
+		if(inputs[PITCH1_INPUT].isConnected()) {
+			load_input(inputs[PITCH1_INPUT], pitch_1, channels_pitch1);
+			channelMask.apply(pitch_1, channels_pitch1);
+			for(int c=0; c<channels_pitch1; c+=4) pitch[c/4] += pitch_1[c/4];
+		}
+
+		if(inputs[PITCH2_INPUT].isConnected()) {
+			load_input(inputs[PITCH2_INPUT], pitch_2, channels_pitch2);
+			channelMask.apply(pitch_2, channels_pitch2);
+			for(int c=0; c<channels_pitch2; c+=4) pitch[c/4] += pitch_2[c/4];
+		}
+
+		if(inputs[FM_INPUT].isConnected()) {
+			load_input(inputs[FM_INPUT], pitch_fm, channels_fm);
+			channelMask.apply(pitch_fm, channels_fm);
+			for(int c=0; c<channels_fm; c+=4) pitch[c/4] += pitch_fm[c/4] / 4.f;
+		}
+
+		for(int c=0; c<channels; c+=4) {
+			freq[c/4] = dsp::FREQ_C4 * simd::pow(2.f, pitch[c/4]);
+			freq[c/4] = clamp(freq[c/4], 0.f, 20000.f);
+		}
 
-		pitch += inputs[PITCH1_INPUT].getVoltage() + inputs[PITCH2_INPUT].getVoltage();
-		pitch += inputs[FM_INPUT].getVoltage() / 4.f;
-		float freq = dsp::FREQ_C4 * std::pow(2.f, pitch);
-		freq = clamp(freq, 0.f, 20000.f);
 
 		// Pulse width
-		float pw = params[PWM_PARAM].getValue() + inputs[PWM_INPUT].getVoltage() / 5.f;
-		const float minPw = 0.05;
-		pw = rescale(clamp(pw, -1.f, 1.f), -1.f, 1.f, minPw, 1.f - minPw);
-
-		// Advance phase
-		float deltaPhase = clamp(freq * args.sampleTime, 1e-6f, 0.5f);
-		float oldPhase = phase;
-		phase += deltaPhase;
-
-		if (oldPhase < 0.5 && phase >= 0.5) {
-			float crossing = -(phase - 0.5) / deltaPhase;
-			triSquareMinBlep.insertDiscontinuity(crossing, 2.f);
-			doubleSawMinBlep.insertDiscontinuity(crossing, -2.f);
+
+		float pw_0 = params[PWM_PARAM].getValue();
+
+		for(int c=0; c<channels; c+=4) pw[c/4] = simd::float_4(pw_0);
+		
+		if(inputs[PWM_INPUT].isConnected()) {
+			load_input(inputs[PWM_INPUT], pwm, channels_pwm);
+			channelMask.apply(pwm, channels_pwm);
+			for(int c=0; c<channels_pwm; c+=4) pw[c/4] += pwm[c/4] / 5.f;
 		}
 
-		if (!halfPhase && phase >= pw) {
-			float crossing  = -(phase - pw) / deltaPhase;
-			squareMinBlep.insertDiscontinuity(crossing, 2.f);
-			halfPhase = true;
+		const simd::float_4 minPw_4 = simd::float_4(0.05f);
+		const simd::float_4 m_one_4 = simd::float_4(-1.0f);
+		const simd::float_4 one_4 = simd::float_4(1.0f);
+
+		for(int c=0; c<channels; c+=4) {
+			pw[c/4] = rescale(clamp(pw[c/4], m_one_4, one_4), m_one_4, one_4, minPw_4, one_4 - minPw_4);
+
+			// Advance phase
+			deltaPhase[c/4] = clamp(freq[c/4] * args.sampleTime, simd::float_4(1e-6f), simd::float_4(0.5f));
+			oldPhase[c/4] = phase[c/4];
+			phase[c/4] += deltaPhase[c/4];
+		}
+
+		// the next block can't be done with SIMD instructions:
+
+		for(int c=0; c<channels; c++) {
+
+			if (oldPhase[c/4].s[c%4] < 0.5 && phase[c/4].s[c%4] >= 0.5) {
+				float crossing = -(phase[c/4].s[c%4] - 0.5) / deltaPhase[c/4].s[c%4];
+				triSquareMinBlep[c].insertDiscontinuity(crossing, 2.f);
+				doubleSawMinBlep[c].insertDiscontinuity(crossing, -2.f);
+			}
+
+			if (!halfPhase[c] && phase[c/4].s[c%4] >= pw[c/4].s[c%4]) {
+				float crossing  = -(phase[c/4].s[c%4] - pw[c/4].s[c%4]) / deltaPhase[c/4].s[c%4];
+				squareMinBlep[c].insertDiscontinuity(crossing, 2.f);
+				halfPhase[c] = true;
+			}	
+
+			// Reset phase if at end of cycle
+			if (phase[c/4].s[c%4] >= 1.f) {
+				phase[c/4].s[c%4] -= 1.f;
+				float crossing = -phase[c/4].s[c%4] / deltaPhase[c/4].s[c%4];
+				triSquareMinBlep[c].insertDiscontinuity(crossing, -2.f);
+				doubleSawMinBlep[c].insertDiscontinuity(crossing, -2.f);
+				squareMinBlep[c].insertDiscontinuity(crossing, -2.f);
+				sawMinBlep[c].insertDiscontinuity(crossing, -2.f);
+				halfPhase[c] = false;
+			}
 		}
 
-		// Reset phase if at end of cycle
-		if (phase >= 1.f) {
-			phase -= 1.f;
-			float crossing = -phase / deltaPhase;
-			triSquareMinBlep.insertDiscontinuity(crossing, -2.f);
-			doubleSawMinBlep.insertDiscontinuity(crossing, -2.f);
-			squareMinBlep.insertDiscontinuity(crossing, -2.f);
-			sawMinBlep.insertDiscontinuity(crossing, -2.f);
-			halfPhase = false;
+		simd::float_4 triSquareMinBlepOut[4];
+		simd::float_4 doubleSawMinBlepOut[4];
+		simd::float_4 sawMinBlepOut[4];
+		simd::float_4 squareMinBlepOut[4];
+
+		simd::float_4 triSquare[4];
+		simd::float_4 sine[4];
+		simd::float_4 doubleSaw[4];
+
+		simd::float_4 even[4];
+		simd::float_4 saw[4];
+		simd::float_4 square[4];
+		simd::float_4 triOut[4];
+
+		for(int c=0; c<channels; c++) {
+			triSquareMinBlepOut[c/4].s[c%4] = triSquareMinBlep[c].process();
+			doubleSawMinBlepOut[c/4].s[c%4] = doubleSawMinBlep[c].process();
+			sawMinBlepOut[c/4].s[c%4] = sawMinBlep[c].process();
+			squareMinBlepOut[c/4].s[c%4] = squareMinBlep[c].process();
 		}
 
 		// Outputs
-		float triSquare = (phase < 0.5) ? -1.f : 1.f;
-		triSquare += triSquareMinBlep.process();
-
-		// Integrate square for triangle
-		tri += 4.f * triSquare * freq * args.sampleTime;
-		tri *= (1.f - 40.f * args.sampleTime);
-
-		float sine = -std::cos(2*M_PI * phase);
-		float doubleSaw = (phase < 0.5) ? (-1.f + 4.f*phase) : (-1.f + 4.f*(phase - 0.5));
-		doubleSaw += doubleSawMinBlep.process();
-		float even = 0.55 * (doubleSaw + 1.27 * sine);
-		float saw = -1.f + 2.f*phase;
-		saw += sawMinBlep.process();
-		float square = (phase < pw) ? -1.f : 1.f;
-		square += squareMinBlep.process();
-
-		// Set outputs
-		outputs[TRI_OUTPUT].setVoltage(5.f*tri);
-		outputs[SINE_OUTPUT].setVoltage(5.f*sine);
-		outputs[EVEN_OUTPUT].setVoltage(5.f*even);
-		outputs[SAW_OUTPUT].setVoltage(5.f*saw);
-		outputs[SQUARE_OUTPUT].setVoltage(5.f*square);
+
+		outputs[TRI_OUTPUT].setChannels(channels);
+		outputs[SINE_OUTPUT].setChannels(channels);
+		outputs[EVEN_OUTPUT].setChannels(channels);
+		outputs[SAW_OUTPUT].setChannels(channels);
+		outputs[SQUARE_OUTPUT].setChannels(channels);
+
+		for(int c=0; c<channels; c+=4) {
+			
+			triSquare[c/4] = simd::ifelse( (phase[c/4] < 0.5f*one_4), m_one_4, one_4);	
+			triSquare[c/4] += triSquareMinBlepOut[c/4];
+
+			// Integrate square for triangle
+	
+			tri[c/4] += (4.f * triSquare[c/4]) * (freq[c/4] * args.sampleTime);
+			tri[c/4] *= (1.f - 40.f * args.sampleTime);
+			triOut[c/4] = 5.f * tri[c/4];
+
+
+			sine[c/4] = 5.f * simd::cos(2*M_PI * phase[c/4]);
+		
+			doubleSaw[c/4] = simd::ifelse( (phase[c/4] < 0.5),  (-1.f + 4.f*phase[c/4]),  (-1.f + 4.f*(phase[c/4] - 0.5f)));
+			doubleSaw[c/4] += doubleSawMinBlepOut[c/4];
+			doubleSaw[c/4] *= 5.f;
+
+			even[c/4] = 0.55 * (doubleSaw[c/4] + 1.27 * sine[c/4]);
+			saw[c/4] = -1.f + 2.f*phase[c/4];
+			saw[c/4] += sawMinBlepOut[c/4];
+			saw[c/4] *= 5.f;
+
+			square[c/4] = simd::ifelse( (phase[c/4] < pw[c/4]),  m_one_4, one_4) ;
+			square[c/4] += squareMinBlepOut[c/4];
+			square[c/4] *= 5.f;
+
+			// Set outputs
+
+			triOut[c/4].store(outputs[TRI_OUTPUT].getVoltages(c));
+			sine[c/4].store(outputs[SINE_OUTPUT].getVoltages(c));
+			even[c/4].store(outputs[EVEN_OUTPUT].getVoltages(c));
+			saw[c/4].store(outputs[SAW_OUTPUT].getVoltages(c));
+			square[c/4].store(outputs[SQUARE_OUTPUT].getVoltages(c));
+		}
 	}
 };
 
diff --git a/src/Mixer.cpp b/src/Mixer.cpp
index 4e77bb6..5a6a229 100644
--- a/src/Mixer.cpp
+++ b/src/Mixer.cpp
@@ -1,4 +1,6 @@
 #include "plugin.hpp"
+#include "simd_mask.hpp"
+
 
 #define MAX(a,b) (a>b)?a:b
 #define MIN(a,b) (a<b)?a:b
@@ -30,9 +32,10 @@ struct Mixer : Module {
 		NUM_LIGHTS
 	};
 
-	simd::float_4 mask[4];
 	simd::float_4 minus_one;
 
+	ChannelMask channelMask;
+
 	Mixer() {
 		config(NUM_PARAMS, NUM_INPUTS, NUM_OUTPUTS, NUM_LIGHTS);
 		configParam(CH1_PARAM, 0.0, 1.0, 0.0, "Ch 1 level", "%", 0, 100);
@@ -42,12 +45,6 @@ struct Mixer : Module {
 
 		minus_one = simd::float_4(-1.0f);
 
-		__m128i tmp =  _mm_cmpeq_epi16(_mm_set_epi32(0,0,0,0),_mm_set_epi32(0,0,0,0));
-
-		for(int i=0; i<4; i++) {
-			mask[3-i] = simd::float_4(_mm_castsi128_ps(tmp));
-			tmp = _mm_srli_si128(tmp, 4);
-		}
 
 	}
 
@@ -90,28 +87,28 @@ struct Mixer : Module {
 			for(int c=0; c<channels1; c+=4) in1[c/4] = simd::float_4::load(inputs[IN1_INPUT].getVoltages(c)) * mult1;
 
 			for(i=0; i<channels1/4; i++) out[i] += in1[i];                          // add only "real" channels. 
-			out[i] += simd::float_4(_mm_and_ps(in1[i].v, mask[channels1-4*i].v));   // make sure we zero out spurious channels
+			out[i] += simd::float_4(_mm_and_ps(in1[i].v, channelMask[channels1-4*i].v));   // make sure we zero out spurious channels
 		}
 	
 		if(inputs[IN2_INPUT].isConnected()) {
 			for(int c=0; c<channels2; c+=4) in2[c/4] = simd::float_4::load(inputs[IN2_INPUT].getVoltages(c)) * mult2;
 
 			for(i=0; i<channels2/4; i++) out[i] += in2[i];
-			out[i] += simd::float_4(_mm_and_ps(in2[i].v, mask[channels2-4*i].v));
+			out[i] += simd::float_4(_mm_and_ps(in2[i].v, channelMask[channels2-4*i].v));
 		}
 
 		if(inputs[IN3_INPUT].isConnected()) {
 			for(int c=0; c<channels3; c+=4) in3[c/4] = simd::float_4::load(inputs[IN3_INPUT].getVoltages(c)) * mult3;
 
 			for(i=0; i<channels3/4; i++) out[i] += in3[i];
-			out[i] += simd::float_4(_mm_and_ps(in3[i].v, mask[channels3-4*i].v));
+			out[i] += simd::float_4(_mm_and_ps(in3[i].v, channelMask[channels3-4*i].v));
 		}
 
 		if(inputs[IN4_INPUT].isConnected()) {
 			for(int c=0; c<channels4; c+=4) in4[c/4] = simd::float_4::load(inputs[IN4_INPUT].getVoltages(c)) * mult4;
 
 			for(i=0; i<channels4/4; i++) out[i] += in4[i];
-			out[i] += simd::float_4(_mm_and_ps(in4[i].v, mask[channels4-4*i].v));
+			out[i] += simd::float_4(_mm_and_ps(in4[i].v, channelMask[channels4-4*i].v));
 		}
 
 
diff --git a/src/Rampage.cpp b/src/Rampage.cpp
index 0e99035..cd1192e 100644
--- a/src/Rampage.cpp
+++ b/src/Rampage.cpp
@@ -1,4 +1,5 @@
 #include "plugin.hpp"
+#include "simd_mask.hpp"
 
 
 static float shapeDelta(float delta, float tau, float shape) {
@@ -73,10 +74,18 @@ struct Rampage : Module {
 		NUM_LIGHTS
 	};
 
+	/*
 	float out[2] = {};
 	bool gate[2] = {};
-	dsp::SchmittTrigger trigger[2];
-	dsp::PulseGenerator endOfCyclePulse[2];
+	*/
+
+	simd::float_4 out[2][4];
+	simd::int32_4 gate[2][4]; // represent bool as integer: false = 0; true > o0
+
+	dsp::SchmittTrigger trigger[2, PORT_MAX_CHANNELS];
+	dsp::PulseGenerator endOfCyclePulse[2, PORT_MAX_CHANNELS];
+
+	ChannelMask channelMask; 
 
 	Rampage() {
 		config(NUM_PARAMS, NUM_INPUTS, NUM_OUTPUTS, NUM_LIGHTS);
@@ -93,12 +102,27 @@ struct Rampage : Module {
 		configParam(FALL_B_PARAM, 0.0, 1.0, 0.0, "Ch 2 fall time");
 		configParam(CYCLE_B_PARAM, 0.0, 1.0, 0.0, "Ch 2 cycle");
 		configParam(BALANCE_PARAM, 0.0, 1.0, 0.5, "Balance");
+
+		memset(out, 0, sizeof(out));
+		memset(gate, 0, sizeof(gate));
 	}
 
 	void process(const ProcessArgs &args) override {
-		for (int c = 0; c < 2; c++) {
-			float in = inputs[IN_A_INPUT + c].getVoltage();
-			if (trigger[c].process(params[TRIGG_A_PARAM + c].getValue() * 10.0 + inputs[TRIGG_A_INPUT + c].getVoltage() / 2.0)) {
+
+		int channels_in_A[2];
+		int channels_in_B[2];
+
+		for (int part=0; part<2; part++) {
+			channels_in_A[part] = inputs[IN_A_INPUT].getChannels()
+			channels_in_B[part] = inputs[IN_B_INPUT].getChannels()
+		}
+
+		for (int part = 0; part < 2; part++) {
+			simd::float_4 in[4];
+
+			load_input(inputs[IN_A_INPUT + part], in, channels_in_A[part]);
+
+			if (trigger[part].process(params[TRIGG_A_PARAM + part].getValue() * 10.0 + inputs[TRIGG_A_INPUT + part].getVoltage() / 2.0)) {
 				gate[c] = true;
 			}
 			if (gate[c]) {
diff --git a/src/simd_mask.hpp b/src/simd_mask.hpp
new file mode 100644
index 0000000..e9161a2
--- /dev/null
+++ b/src/simd_mask.hpp
@@ -0,0 +1,46 @@
+#pragma once
+
+#include "rack.hpp"
+
+
+struct ChannelMask {
+
+	rack::simd::float_4 mask[4];
+
+    ChannelMask() {
+        
+		__m128i tmp =  _mm_cmpeq_epi16(_mm_set_epi32(0,0,0,0),_mm_set_epi32(0,0,0,0));
+		
+		for(int i=0; i<4; i++) {
+			mask[3-i] = simd::float_4(_mm_castsi128_ps(tmp));
+			tmp = _mm_srli_si128(tmp, 4);
+		}
+        
+    };
+
+    ~ChannelMask() {};
+
+    inline const rack::simd::float_4 operator[](const int c) {return  mask[c];}
+
+	inline void apply(simd::float_4 *vec, int numChannels) {
+		int c=numChannels/4;
+		vec[c] = vec[c]&mask[numChannels-4*c];
+	}
+
+	inline void apply_all(simd::float_4 *vec, int numChannels) {
+		int c=numChannels/4;
+		vec[c] = vec[c]&mask[numChannels-4*c];
+		for(int i=c+1; i<4; i++) vec[i] = simd::float_4(0.f);
+	}
+
+
+};
+
+inline void load_input(Input &in, simd::float_4 *v, int numChannels) {
+	if(numChannels==1) {
+		for(int i=0; i<4; i++) v[i] = simd::float_4(in.getVoltage());
+	} else {
+		for(int c=0; c<numChannels; c+=4) v[c/4] = simd::float_4::load(in.getVoltages(c));
+	}
+}
+