further tomfoolery

Makefile

@@ -94,11 +94,11 @@ $(UTILS): %: $(BIN)/%
 
 $(BIN)/%.exe: $(BIN)/%
 	@echo '  OBJCOPY '$@
-	@$(OBJCOPY) $< $@
+	@$(OBJCOPY) -S $< $@
 
 $(BIN)/%.dll: $(BIN)/%.so
 	@echo '  OBJCOPY '$@
-	@$(OBJCOPY) $< $@
+	@$(OBJCOPY) -S $< $@
 
 $(BIN)/ladspa/%.so: $(BIN)/ladspa/%.o
 	@echo '  CXXLD   '$@

crap/eq.hpp

@@ -1,5 +1,3 @@
-#define BLOCK_SIZE 256
-
 #include <stdio.h>
 #include <string.h>
 

crap/eq_const.hpp

@@ -1,5 +1,3 @@
-#define BLOCK_SIZE 256
-
 #include <string.h>
 
 #include "util.hpp"

crap/eq_const_T420.hpp

@@ -1,5 +1,3 @@
-#define BLOCK_SIZE 256
-
 #include <string.h>
 
 #include "util.hpp"

crap/eq_const_T420_svf.hpp

@@ -1,5 +1,3 @@
-#define BLOCK_SIZE 256
-
 #include <string.h>
 #include <stdio.h>
 #include <assert.h>

crap/mugi4.hpp

@@ -1,7 +1,3 @@
-#define BLOCK_SIZE 256
-#define OVERSAMPLING 2
-#define FULL_SIZE (BLOCK_SIZE*OVERSAMPLING)
-
 /*
 an implementation of:
 S. D Angelo and V. Välimäki. Generalized Moog Ladder Filter: Part II
@@ -18,6 +14,9 @@ https://aaltodoc.aalto.fi/bitstream/handle/123456789/14420/article6.pdf
 #include "Param.hpp"
 #include "Crap.hpp"
 #include "os2piir.hpp"
+
+#define OVERSAMPLING 2
+#define FULL_SIZE (BLOCK_SIZE*OVERSAMPLING)
 #include "Buffer2OS2.hpp"
 
 #define VT 0.026

crap/tube.hpp

@@ -1,7 +1,3 @@
-#define OVERSAMPLING 2
-#define BLOCK_SIZE 256
-#define FULL_SIZE (BLOCK_SIZE*OVERSAMPLING)
-
 #include <string.h>
 #include <stdio.h>
 
@@ -9,6 +5,9 @@
 #include "Param.hpp"
 #include "Crap.hpp"
 #include "os2piir.hpp"
+
+#define OVERSAMPLING 2
+#define FULL_SIZE (BLOCK_SIZE*OVERSAMPLING)
 #include "Buffer2OS2.hpp"
 
 typedef struct {

include/Buffer2.hpp

@@ -16,15 +16,15 @@ struct Buffer2 : public Mixin {
 				rem = count - pos;
 
 			for (ulong i = 0; i < rem; i++) {
-				buf[i].v[0] = in_L[i];
-				buf[i].v[1] = in_R[i];
+				buf[i][0] = in_L[i];
+				buf[i][1] = in_R[i];
 			}
 
 			process2(buf, rem);
 
 			for (ulong i = 0; i < rem; i++) {
-				out_L[i] = buf[i].v[0];
-				out_R[i] = buf[i].v[1];
+				out_L[i] = buf[i][0];
+				out_R[i] = buf[i][1];
 			}
 
 			in_L += BLOCK_SIZE;

include/Buffer4.hpp

@@ -19,19 +19,19 @@ struct Buffer4 : public Mixin {
 				rem = count - pos;
 
 			for (ulong i = 0, j = 0; i < rem; i += 2, j++) {
-				buf_L[j].v[0] = in_L[i+0];
-				buf_L[j].v[1] = in_L[i+1];
-				buf_R[j].v[0] = in_R[i+0];
-				buf_R[j].v[1] = in_R[i+1];
+				buf_L[j][0] = in_L[i+0];
+				buf_L[j][1] = in_L[i+1];
+				buf_R[j][0] = in_R[i+0];
+				buf_R[j][1] = in_R[i+1];
 			}
 
 			process2(buf_L, buf_R, rem);
 
 			for (ulong i = 0, j = 0; i < rem; i += 2, j++) {
-				out_L[i+0] = buf_L[j].v[0];
-				out_L[i+1] = buf_L[j].v[1];
-				out_R[i+0] = buf_R[j].v[0];
-				out_R[i+1] = buf_R[j].v[1];
+				out_L[i+0] = buf_L[j][0];
+				out_L[i+1] = buf_L[j][1];
+				out_R[i+0] = buf_R[j][0];
+				out_R[i+1] = buf_R[j][1];
 			}
 
 			in_L += BLOCK_SIZE;

include/Dumber.hpp

@@ -83,4 +83,3 @@ struct Dumber : public DumberBase<T> {
 	Dumber(DumberBase<T> v2) : DumberBase<T>(v2)
 	{}
 };
-

include/svf.hpp

@@ -123,18 +123,18 @@ svf_run_block_mat(svf_matrix<T> *RESTRICT mat, T *RESTRICT buf, ulong count)
 	T t1, t2, t3, t4;
 	T memory = mat->memory;
 	for (ulong i = 0; i < count/2; i++) {
-		memory.v[0] = buf[i].v[0];
-		memory.v[1] = buf[i].v[1];
+		memory[0] = buf[i][0];
+		memory[1] = buf[i][1];
 		t1 = mat->a*memory;
 		t2 = mat->b*memory;
 		t3 = mat->c*memory;
 		t4 = mat->d*memory;
-		memory.v[0] = t1.v[0] + t1.v[1] + t1.v[2] + t1.v[3];
-		memory.v[1] = t2.v[0] + t2.v[1] + t2.v[2] + t2.v[3];
-		memory.v[2] = t3.v[0] + t3.v[1] + t3.v[2] + t3.v[3];
-		memory.v[3] = t4.v[0] + t4.v[1] + t4.v[2] + t4.v[3];
-		buf[i].v[0] = memory.v[0];
-		buf[i].v[1] = memory.v[1];
+		memory[0] = t1[0] + t1[1] + t1[2] + t1[3];
+		memory[1] = t2[0] + t2[1] + t2[2] + t2[3];
+		memory[2] = t3[0] + t3[1] + t3[2] + t3[3];
+		memory[3] = t4[0] + t4[1] + t4[2] + t4[3];
+		buf[i][0] = memory[0];
+		buf[i][1] = memory[1];
 	}
 	mat->memory = memory;
 }

include/util.hpp

@@ -14,6 +14,10 @@
 
 typedef unsigned long ulong; // __attribute((aligned(16)));
 
+#ifndef BLOCK_SIZE
+#define BLOCK_SIZE 256
+#endif
+
 #include "Dumber.hpp"
 #include "vectors.hpp"
 

include/vectors.hpp

@@ -73,6 +73,16 @@ struct Dumber<_v4sf> : public DumberBase<_v4sf> {
 	TEMPLATE inline
 	Dumber(T x)
 	{ v = (_v4sf){float(x), float(x), float(x), float(x)}; }
+
+	inline float &
+	operator[](int index) {
+		return ((float *)&v)[index];
+	}
+
+	inline const float &
+	operator[](int index) const {
+		return ((float *)&v)[index];
+	}
 };
 
 typedef Dumber<_v2df> v2df;

util/bench.cpp

@@ -9,7 +9,7 @@
 #include "ladspa.hpp"
 #include "util.hpp"
 
-#define BLOCK_SIZE 2048
+#define BENCH_BLOCK 2048
 
 void *plug = NULL;
 static float *audio_buffer;
@@ -109,12 +109,12 @@ main(int argc, char **argv)
 		if (LADSPA_IS_PORT_AUDIO(d->PortDescriptors[i]))
 			audio_count++;
 
-	audio_buffer = (decltype(audio_buffer)) calloc(audio_count*BLOCK_SIZE, sizeof(float));
+	audio_buffer = (decltype(audio_buffer)) calloc(audio_count*BENCH_BLOCK, sizeof(float));
 
 	int a = 0;
 	for (int i = 0; i < d->PortCount; i++) {
 		if (LADSPA_IS_PORT_AUDIO(d->PortDescriptors[i])) {
-			d->connect_port(h, i, audio_buffer + a++*BLOCK_SIZE);
+			d->connect_port(h, i, audio_buffer + a++*BENCH_BLOCK);
 		} else {
 			float *x;
 			x = (decltype(x)) alloca(sizeof(float));
@@ -124,12 +124,12 @@ main(int argc, char **argv)
 	}
 
 	unsigned int mirand = time(NULL);
-	for (int i = 0; i < audio_count*BLOCK_SIZE; i++)
+	for (int i = 0; i < audio_count*BENCH_BLOCK; i++)
 		audio_buffer[i] = whitenoise(mirand);
 
 	if (d->activate) d->activate(h);
 	for (int i = 0; i < 64*64*8; i++)
-		d->run(h, BLOCK_SIZE);
+		d->run(h, BENCH_BLOCK);
 	if (d->deactivate) d->deactivate(h);
 
 	d->cleanup(h);