added [absgn~] as external (written by tim blechmann) and abstraction

the main purpose of this external is speed

svn path=/trunk/externals/zexy/; revision=4364

6 files changed, 252 insertions, 52 deletions

diff --git a/abs/absgn~-help.pd b/abs/absgn~-help.pd
new file mode 100644
index 0000000..4f718fd
--- /dev/null
+++ b/abs/absgn~-help.pd
@@ -0,0 +1,30 @@
+#N canvas 201 257 575 388 10;
+#X obj 92 103 sig~;
+#X text 272 20 part of zexy;
+#X obj 93 176 snapshot~;
+#X floatatom 93 204 5 0 0 0 - - -;
+#X obj 146 153 metro 100;
+#X obj 146 131 tgl 15 0 empty empty empty 0 -6 0 8 -262144 -1 -1 1
+1;
+#X floatatom 92 81 5 0 0 0 - - -;
+#X text 16 243 note:;
+#X text 41 285 performance is worse than with the external version
+;
+#X text 45 315 if you want to use the external version of this object
+\, you have to load zexy;
+#X text 43 255 this is the abstraction version of this object \, which
+is using [expr~];
+#X obj 92 152 absgn~;
+#X text 349 61 updated for zexy-2.1;
+#X obj 163 176 snapshot~;
+#X floatatom 163 202 5 0 0 0 - - -;
+#X text 65 46 absolute value and signum of a signal;
+#X connect 0 0 11 0;
+#X connect 2 0 3 0;
+#X connect 4 0 2 0;
+#X connect 4 0 13 0;
+#X connect 5 0 4 0;
+#X connect 6 0 0 0;
+#X connect 11 0 2 0;
+#X connect 11 1 13 0;
+#X connect 13 0 14 0;
diff --git a/abs/absgn~.pd b/abs/absgn~.pd
new file mode 100644
index 0000000..873bbab
--- /dev/null
+++ b/abs/absgn~.pd
@@ -0,0 +1,17 @@
+#N canvas 0 0 482 310 10;
+#X obj 30 88 inlet~;
+#X obj 30 136 outlet~;
+#X text 319 21 part of zexy;
+#X obj 187 137 outlet~;
+#X obj 31 113 abs~;
+#X obj 187 113 sgn~;
+#X text 36 44 [absgn~];
+#X text 250 261 updated for zexy-2.1;
+#X text 41 183 [absgn~] \, [abs~] and [sgn~] are provided both as externals
+(fast!) and abstractions (slow!!) \; if you are using this abstraction
+of [absgn~] \, chances are high \, that [abs~] and [sgn~] are abstractions
+too.;
+#X connect 0 0 4 0;
+#X connect 0 0 5 0;
+#X connect 4 0 1 0;
+#X connect 5 0 3 0;
diff --git a/examples/sigbinops+.pd b/examples/sigbinops+.pd
index 1223ffc..93bbf8d 100644
--- a/examples/sigbinops+.pd
+++ b/examples/sigbinops+.pd
@@ -1,61 +1,67 @@
 #N canvas 470 149 594 490 10;
 #X obj 62 193 abs~;
 #X obj 120 194 sgn~;
-#X obj 262 191 <~;
-#X obj 296 190 ==~;
-#X obj 330 191 >~;
-#X obj 388 190 &&~;
-#X obj 423 190 ||~;
-#X floatatom 423 238 0 0 0 0 - - -;
-#X floatatom 388 238 0 0 0 0 - - -;
-#X floatatom 330 239 0 0 0 0 - - -;
-#X floatatom 296 239 0 0 0 0 - - -;
+#X obj 342 191 <~;
+#X obj 376 190 ==~;
+#X obj 410 191 >~;
+#X obj 468 190 &&~;
+#X obj 503 190 ||~;
+#X floatatom 503 238 0 0 0 0 - - -;
+#X floatatom 468 238 0 0 0 0 - - -;
+#X floatatom 410 239 0 0 0 0 - - -;
+#X floatatom 376 239 0 0 0 0 - - -;
 #X floatatom 120 242 0 0 0 0 - - -;
 #X floatatom 62 243 0 0 0 0 - - -;
 #X obj 62 221 avg~;
 #X obj 120 220 avg~;
-#X floatatom 262 239 0 0 0 0 - - -;
-#X obj 262 217 avg~;
-#X obj 296 216 avg~;
-#X obj 330 216 avg~;
-#X obj 388 216 avg~;
-#X obj 423 216 avg~;
+#X floatatom 342 239 0 0 0 0 - - -;
+#X obj 342 217 avg~;
+#X obj 376 216 avg~;
+#X obj 410 216 avg~;
+#X obj 468 216 avg~;
+#X obj 503 216 avg~;
 #X obj 62 163 sig~;
 #X floatatom 62 141 0 0 0 0 - - -;
-#X obj 330 126 sig~;
-#X floatatom 330 104 0 0 0 0 - - -;
-#X obj 404 126 sig~;
-#X floatatom 404 104 0 0 0 0 - - -;
-#X floatatom 440 448 0 0 0 0 - - -;
-#X floatatom 400 448 0 0 0 0 - - -;
-#X floatatom 349 448 0 0 0 0 - - -;
-#X floatatom 308 449 0 0 0 0 - - -;
-#X floatatom 274 449 0 0 0 0 - - -;
-#X obj 274 427 avg~;
-#X obj 308 426 avg~;
-#X obj 349 425 avg~;
-#X obj 400 426 avg~;
-#X obj 440 426 avg~;
-#X obj 349 337 sig~;
-#X floatatom 349 315 0 0 0 0 - - -;
-#X floatatom 430 316 0 0 0 0 - - -;
-#X obj 274 401 <~ 2;
-#X obj 308 400 ==~ 2;
-#X obj 349 400 >~ 2;
-#X obj 400 400 &&~ 2;
-#X obj 440 400 ||~ 2;
+#X obj 410 126 sig~;
+#X floatatom 410 104 0 0 0 0 - - -;
+#X obj 484 126 sig~;
+#X floatatom 484 104 0 0 0 0 - - -;
+#X floatatom 520 448 0 0 0 0 - - -;
+#X floatatom 480 448 0 0 0 0 - - -;
+#X floatatom 429 448 0 0 0 0 - - -;
+#X floatatom 388 449 0 0 0 0 - - -;
+#X floatatom 354 449 0 0 0 0 - - -;
+#X obj 354 427 avg~;
+#X obj 388 426 avg~;
+#X obj 429 425 avg~;
+#X obj 480 426 avg~;
+#X obj 520 426 avg~;
+#X obj 429 337 sig~;
+#X floatatom 429 315 0 0 0 0 - - -;
+#X floatatom 510 316 0 0 0 0 - - -;
+#X obj 354 401 <~ 2;
+#X obj 388 400 ==~ 2;
+#X obj 429 400 >~ 2;
+#X obj 480 400 &&~ 2;
+#X obj 520 400 ||~ 2;
 #X text 83 30 more math and binary operations for signals;
 #X text 28 263 absolute;
 #X text 50 278 value;
 #X text 112 263 signum;
-#X text 249 267 greater;
-#X text 330 268 less;
-#X text 296 284 equal;
-#X text 393 261 logical;
-#X text 386 274 AND;
-#X text 431 274 OR;
+#X text 329 267 greater;
+#X text 410 268 less;
+#X text 376 284 equal;
+#X text 473 261 logical;
+#X text 466 274 AND;
+#X text 511 274 OR;
 #X text 430 16 part of zexy;
-#X text 72 58 (this patch might be very CPU-consumptious \, because
+#X floatatom 230 242 0 0 0 0 - - -;
+#X floatatom 192 243 0 0 0 0 - - -;
+#X obj 192 221 avg~;
+#X obj 230 220 avg~;
+#X obj 192 193 absgn~;
+#X text 196 267 (both);
+#X text 72 58 (this help-patch might be very CPU-consumptious \, because
 of the float-atoms....);
 #X connect 0 0 13 0;
 #X connect 1 0 14 0;
@@ -73,6 +79,7 @@ of the float-atoms....);
 #X connect 20 0 7 0;
 #X connect 21 0 0 0;
 #X connect 21 0 1 0;
+#X connect 21 0 60 0;
 #X connect 22 0 21 0;
 #X connect 23 0 2 0;
 #X connect 23 0 3 0;
@@ -106,3 +113,7 @@ of the float-atoms....);
 #X connect 42 0 34 0;
 #X connect 43 0 35 0;
 #X connect 44 0 36 0;
+#X connect 58 0 57 0;
+#X connect 59 0 56 0;
+#X connect 60 0 58 0;
+#X connect 60 1 59 0;
diff --git a/src/absgn~.c b/src/absgn~.c
new file mode 100644
index 0000000..1d3d4bd
--- /dev/null
+++ b/src/absgn~.c
@@ -0,0 +1,140 @@
+/******************************************************
+ *
+ * zexy - implementation file
+ *
+ * (c) 2006 Tim Blechmann
+ *
+ *
+ ******************************************************
+ *
+ * license: GNU General Public License v.2
+ *
+ ******************************************************/
+
+#include "zexy.h"
+
+typedef struct _absgn
+{
+  t_object x_obj;
+  float x_f;
+} t_absgn;
+
+
+/* ------------------------ sigABSGN~ ----------------------------- */
+
+static t_class *sigABSGN_class;
+
+static t_int *sigABSGN_perform(t_int *w)
+{
+  t_float *in = (t_float *)(w[1]);
+  t_float *out = (t_float *)(w[2]);
+  t_float *out2 = (t_float *)(w[3]);
+  int n = (int)(w[4]);
+  
+  while (n--)
+  {
+      t_float val = *in++;
+      *out++ = fabsf(val);
+
+      if (val>0.) *out2++=1.;
+      else if (val<0.) *out2++=-1.;
+      else *out2++=0.;
+  }
+
+  
+  return (w+5);
+}
+
+#ifdef __SSE__
+static long l_bitmask[]   ={0x7fffffff, 0x7fffffff, 0x7fffffff, 0x7fffffff};
+static long l_sgnbitmask[]={0x80000000, 0x80000000, 0x80000000, 0x80000000};
+static t_int *sigABSGN_performSSE(t_int *w)
+{
+  __m128 *in = (__m128 *)(w[1]);
+  __m128 *out1 = (__m128 *)(w[2]);
+  __m128 *out2 = (__m128 *)(w[3]);
+  int n = (int)(w[4])>>3;
+  
+  const __m128 bitmask= _mm_loadu_ps((float*)l_bitmask);
+  const __m128 sgnmask= _mm_loadu_ps((float*)l_sgnbitmask);
+  const __m128 zero   = _mm_setzero_ps();
+  const __m128 one    = _mm_set1_ps(1.f);
+  
+  do {
+    __m128 val, val2, xmm0, xmm1, xmm2, xmm3;  
+    val=in[0];
+    xmm0   = _mm_cmpneq_ps(val, zero);   /* mask for non-zeros */
+    xmm1   = _mm_and_ps   (val, sgnmask);/* sign (without value) */
+    xmm0   = _mm_and_ps   (xmm0, one);   /* (abs) value: (val==0.f)?0.f:1.f */
+    out1[0]= _mm_and_ps   (val, bitmask);/* abs: set sign-bit to "+" */
+    out2[0]= _mm_or_ps    (xmm1, xmm0);  /* merge sign and value */
+
+    val2=in[1];
+    xmm2   = _mm_cmpneq_ps(val2, zero);   /* mask for non-zeros */
+    xmm3   = _mm_and_ps   (val2, sgnmask);/* sign (without value) */
+    xmm2   = _mm_and_ps   (xmm2, one);    /* (abs) value: (val==0.f)?0.f:1.f */
+    out1[1]= _mm_and_ps   (val2, bitmask);/* abs: set sign-bit to "+" */
+    out2[1]= _mm_or_ps    (xmm3, xmm2);   /* merge sign and value */
+
+    in +=2;
+    out+=2;
+    out2+=2;
+  }
+  while (--n);
+
+  return (w+5);
+}
+#endif /* __SSE__ */
+
+static void sigABSGN_dsp(t_absgn *x, t_signal **sp)
+{
+#ifdef __SSE__
+  if(
+     Z_SIMD_CHKBLOCKSIZE(sp[0]->s_n)&&
+     Z_SIMD_CHKALIGN(sp[0]->s_vec)&&
+     Z_SIMD_CHKALIGN(sp[1]->s_vec)&&
+     Z_SIMD_CHKALIGN(sp[2]->s_vec)
+     )
+    {
+      dsp_add(sigABSGN_performSSE, 4, sp[0]->s_vec, sp[1]->s_vec, sp[2]->s_vec, sp[0]->s_n);
+    } else
+#endif
+    {
+      dsp_add(sigABSGN_perform, 4, sp[0]->s_vec, sp[1]->s_vec, sp[2]->s_vec, sp[0]->s_n);
+    }
+}
+
+static void sigABSGN_helper(void)
+{
+  post("\n%c absgn~ \t\t:: absolute value and sign of a signal", HEARTSYMBOL);
+  post("         \t\t   copyright (c) Tim Blechmann 2006");
+}
+
+static void *sigABSGN_new(void)
+{
+  t_absgn *x = (t_absgn *)pd_new(sigABSGN_class);
+  x->x_f=0.f;
+
+  outlet_new(&x->x_obj, gensym("signal"));
+  outlet_new(&x->x_obj, gensym("signal"));
+
+  return (x);
+}
+
+void absgn_tilde_setup(void)
+{
+  sigABSGN_class = class_new(gensym("absgn~"), (t_newmethod)sigABSGN_new, 0,
+			   sizeof(t_absgn), 0, A_DEFFLOAT, 0);
+  CLASS_MAINSIGNALIN(sigABSGN_class, t_absgn, x_f);
+  class_addmethod(sigABSGN_class, (t_method)sigABSGN_dsp, gensym("dsp"), 0);
+  
+  class_addmethod(sigABSGN_class, (t_method)sigABSGN_helper, gensym("help"), 0);
+  class_sethelpsymbol(sigABSGN_class, gensym("zexy/sigbinops+"));
+
+  zexy_register("absgn~");
+}
+
+void z_absgn__setup(void)
+{
+  absgn_tilde_setup();
+}
diff --git a/src/z_zexy.c b/src/z_zexy.c
index f7c1dc3..e1693c2 100644
--- a/src/z_zexy.c
+++ b/src/z_zexy.c
@@ -8,14 +8,15 @@
 void z_zexy_setup(void)
 {
 	z_a2l_setup(); /* a2l.c */
+	z_absgn__setup(); /* absgn~.c */
 	z_abs__setup(); /* abs~.c */
 	z_atoi_setup(); /* atoi.c */
 	z_avg__setup(); /* avg~.c */
 	z_blockmirror__setup(); /* blockmirror~.c */
 	z_blockswap__setup(); /* blockswap~.c */
 	z_date_setup(); /* date.c */
-	z_demultiplex__setup(); /* demultiplex~.c */
 	z_demultiplex_setup(); /* demultiplex.c */
+	z_demultiplex__setup(); /* demultiplex~.c */
 	z_dfreq__setup(); /* dfreq~.c */
 	z_dirac__setup(); /* dirac~.c */
 	z_dot_setup(); /* dot.c */
@@ -36,14 +37,14 @@ void z_zexy_setup(void)
 	z_minmax_setup(); /* minmax.c */
 	z_msgfile_setup(); /* msgfile.c */
 	z_multiline__setup(); /* multiline~.c */
-	z_multiplex__setup(); /* multiplex~.c */
 	z_multiplex_setup(); /* multiplex.c */
+	z_multiplex__setup(); /* multiplex~.c */
 	z_niagara_setup(); /* niagara.c */
 	z_noish__setup(); /* noish~.c */
 	z_noisi__setup(); /* noisi~.c */
 	z_operating_system_setup(); /* operating_system.c */
-	z_pack__setup(); /* pack~.c */
 	z_packel_setup(); /* packel.c */
+	z_pack__setup(); /* pack~.c */
 	z_pdf__setup(); /* pdf~.c */
 	z_prime_setup(); /* prime.c */
 	z_quantize__setup(); /* quantize~.c */
@@ -70,7 +71,7 @@ void z_zexy_setup(void)
 	z_urn_setup(); /* urn.c */
 	z_winNT_portio_setup(); /* winNT_portio.c */
 	z_wrap_setup(); /* wrap.c */
-	z_z__setup(); /* z~.c */
 	z_z_sigbin_setup(); /* z_sigbin.c */
+	z_z__setup(); /* z~.c */
 }
 
diff --git a/src/z_zexy.h b/src/z_zexy.h
index 6e09612..db1419d 100644
--- a/src/z_zexy.h
+++ b/src/z_zexy.h
@@ -6,14 +6,15 @@
 #ifndef Z_ZEXY_H__
 #define Z_ZEXY_H__
 void z_a2l_setup(void); /* a2l.c */
+void z_absgn__setup(void); /* absgn~.c */
 void z_abs__setup(void); /* abs~.c */
 void z_atoi_setup(void); /* atoi.c */
 void z_avg__setup(void); /* avg~.c */
 void z_blockmirror__setup(void); /* blockmirror~.c */
 void z_blockswap__setup(void); /* blockswap~.c */
 void z_date_setup(void); /* date.c */
-void z_demultiplex__setup(void); /* demultiplex~.c */
 void z_demultiplex_setup(void); /* demultiplex.c */
+void z_demultiplex__setup(void); /* demultiplex~.c */
 void z_dfreq__setup(void); /* dfreq~.c */
 void z_dirac__setup(void); /* dirac~.c */
 void z_dot_setup(void); /* dot.c */
@@ -34,14 +35,14 @@ void z_mavg_setup(void); /* mavg.c */
 void z_minmax_setup(void); /* minmax.c */
 void z_msgfile_setup(void); /* msgfile.c */
 void z_multiline__setup(void); /* multiline~.c */
-void z_multiplex__setup(void); /* multiplex~.c */
 void z_multiplex_setup(void); /* multiplex.c */
+void z_multiplex__setup(void); /* multiplex~.c */
 void z_niagara_setup(void); /* niagara.c */
 void z_noish__setup(void); /* noish~.c */
 void z_noisi__setup(void); /* noisi~.c */
 void z_operating_system_setup(void); /* operating_system.c */
-void z_pack__setup(void); /* pack~.c */
 void z_packel_setup(void); /* packel.c */
+void z_pack__setup(void); /* pack~.c */
 void z_pdf__setup(void); /* pdf~.c */
 void z_prime_setup(void); /* prime.c */
 void z_quantize__setup(void); /* quantize~.c */
@@ -68,7 +69,7 @@ void z_unpack__setup(void); /* unpack~.c */
 void z_urn_setup(void); /* urn.c */
 void z_winNT_portio_setup(void); /* winNT_portio.c */
 void z_wrap_setup(void); /* wrap.c */
-void z_z__setup(void); /* z~.c */
 void z_z_sigbin_setup(void); /* z_sigbin.c */
+void z_z__setup(void); /* z~.c */
 #endif /* Z_ZEXY_H__ */