initial ci of sparse coefficients FIR filter

svn path=/trunk/externals/iemlib/; revision=14521

2 files changed, 311 insertions, 0 deletions

diff --git a/iemlib1/sparse_FIR~-help.pd b/iemlib1/sparse_FIR~-help.pd
new file mode 100644
index 0000000..3eb7ba2
--- /dev/null
+++ b/iemlib1/sparse_FIR~-help.pd
@@ -0,0 +1,88 @@
+#N canvas 156 22 842 608 10;
+#X floatatom 4 133 5 0 0 0 - - -;
+#X text 32 96 % cpu;
+#X obj 4 93 dsp;
+#X obj 4 73 tgl 15 0 empty empty dsp_ON/OFF 2 -8 0 10 -225271 -42246
+-90881 1 1;
+#X floatatom 183 494 7 0 0 0 - - -;
+#X obj 112 373 dac~;
+#X obj 112 347 gainvu~ 300;
+#X obj 174 369 vu 15 120 empty empty 8 -8 0 10 -66577 -1 1 0;
+#X obj 294 216 vsl 15 128 0 127 0 1 empty empty gain -7 -8 0 10 -225271
+-42246 -90881 7500 1;
+#X obj 113 95 noise~;
+#X floatatom 173 512 7 0 0 0 - - -;
+#X floatatom 17 113 5 0 0 0 - - -;
+#X text 170 567 IEM KUG;
+#X text 146 555 musil;
+#X text 182 555 @;
+#X text 190 555 iem.at;
+#X text 153 577 Graz \, Austria;
+#X floatatom 142 323 5 0 0 0 - - -;
+#X obj 113 179 sparse_FIR~ 256;
+#X msg 177 112 size 256;
+#X msg 367 535 0 1 1 1;
+#X msg 367 556 0 1 40 -1 80 1 120 -1 160 1;
+#X msg 369 312 0 1 10 1 20 1 30 1 40 1 50 1 60 1 70 1 80 1;
+#X obj 707 38 t b b;
+#X msg 767 101 set;
+#X msg 668 101 add2 \$1 \$2;
+#X obj 707 18 bng 15 250 50 0 empty empty empty 17 7 0 10 -262144 -1
+-1;
+#X msg 176 137 order 384;
+#X msg 366 342 matrix 9 2 0 1 10 1 20 1 30 1 40 1 50 1 60 1 70 1 80
+1;
+#X msg 363 371 matrix 2 9 0 1 10 1 20 1 30 1 40 1 50 1 60 1 70 1 80
+1;
+#X text 103 544 (c) Thomas Musil 2000 - 2010;
+#X text 17 16 sparse_FIR~;
+#X msg 370 581 0 0;
+#X msg 370 502 0 -0.1 1 -0.18 2 -0.1 3 0.15 4 0.4 5 0.7 6 0.4 7 0.15
+8 -0.1 9 -0.18 10 -0.1;
+#X msg 510 126 0 0.04 1 0.04 2 0.04 3 0.04 4 0.04 5 0.04 6 0.04 7 0.04
+8 0.04 9 0.04 10 0.04 11 0.04 12 0.04 13 0.04 14 0.04 15 0.04 16 0.04
+17 0.04 18 0.04 19 0.04 20 0.04 21 0.04 22 0.04 23 0.04 24 0.04 25
+0.04 26 0.04 27 0.04 28 0.04 29 0.04;
+#X obj 668 81 pack 0 0.04;
+#X obj 668 59 for++ 0 29;
+#X text 366 228 A list of alternating index- and value- doubles will
+change the coefficients of convolution. The number of pairs should
+be less than the order of FIR.;
+#X text 367 276 A matrix message will do the same.;
+#X obj 568 93 loadbang;
+#X text 134 16 convolve a signal with a coefficent list of non zero
+index/value pairs. All values between the determined values will be
+set to zero and won't be calculated (sparse).;
+#X text 227 182 1.arg: <float> maximum convolution-length or order
+;
+#X text 176 93 will change the max. order of FIR filter.;
+#X connect 2 0 0 0;
+#X connect 2 1 11 0;
+#X connect 3 0 2 0;
+#X connect 6 0 5 0;
+#X connect 6 0 5 1;
+#X connect 6 1 17 0;
+#X connect 6 2 7 0;
+#X connect 7 0 10 0;
+#X connect 7 1 4 0;
+#X connect 8 0 6 1;
+#X connect 9 0 18 0;
+#X connect 18 0 6 0;
+#X connect 19 0 18 0;
+#X connect 20 0 18 0;
+#X connect 21 0 18 0;
+#X connect 22 0 18 0;
+#X connect 23 0 36 0;
+#X connect 23 1 24 0;
+#X connect 24 0 34 0;
+#X connect 25 0 34 0;
+#X connect 26 0 23 0;
+#X connect 27 0 18 0;
+#X connect 28 0 18 0;
+#X connect 29 0 18 0;
+#X connect 32 0 18 0;
+#X connect 33 0 18 0;
+#X connect 34 0 18 0;
+#X connect 35 0 25 0;
+#X connect 36 0 35 0;
+#X connect 39 0 34 0;
diff --git a/iemlib1/src/sparse_FIR~.c b/iemlib1/src/sparse_FIR~.c
new file mode 100644
index 0000000..0a1043a
--- /dev/null
+++ b/iemlib1/src/sparse_FIR~.c
@@ -0,0 +1,223 @@
+/* For information on usage and redistribution, and for a DISCLAIMER OF ALL
+* WARRANTIES, see the file, "LICENSE.txt," in this distribution.
+
+iemlib1 written by Thomas Musil, Copyright (c) IEM KUG Graz Austria 2000 - 2010 */
+
+
+#include "m_pd.h"
+#include "iemlib.h"
+
+
+/* ---------- sparse_FIR~ - sparse_FIR-filter with coef-matrix-message ----------- */
+/* the list or matrix message should have an index and its value, index , value, aso.... */
+
+typedef struct _sparse_FIR_tilde
+{
+  t_object  x_obj;
+  t_float   *x_coef_beg;
+  int       *x_index_beg;
+  int       x_n_coef;
+  int       x_n_coef_malloc;
+  t_float   *x_history_beg;
+  int       x_rw_index;
+  int       x_sparse_FIR_order;
+  t_float   x_msi;
+} t_sparse_FIR_tilde;
+
+static t_class *sparse_FIR_tilde_class;
+
+static t_int *sparse_FIR_tilde_perform(t_int *w)
+{
+  t_float *in = (t_float *)(w[1]);
+  t_float *out = (t_float *)(w[2]);
+  t_sparse_FIR_tilde *x = (t_sparse_FIR_tilde *)(w[3]);
+  int n = (t_int)(w[4]);
+  int rw_index = x->x_rw_index;
+  int i, j, ix;
+  int order = x->x_sparse_FIR_order;
+  int n_coef = x->x_n_coef;
+  int n_coef8;
+  t_float sum=0.0f;
+  t_float *coef = x->x_coef_beg;
+  int *index = x->x_index_beg;
+  t_float *write_hist1=x->x_history_beg;
+  t_float *write_hist2;
+  t_float *read_hist;
+  t_float *coef_vec;
+  int *index_vec;
+  t_float *hist_vec;
+  
+  if((order < 1) || (n_coef < 1))
+    goto sparse_FIR_tilde_perf_zero;
+  
+  n_coef8 = n_coef / 8;
+  write_hist1 = x->x_history_beg;
+  write_hist2 = write_hist1 + order;
+  read_hist = write_hist2;
+  
+  for(i=0; i<n; i++)
+  {
+    write_hist1[rw_index] = in[i];
+    write_hist2[rw_index] = in[i];
+    sum = 0.0f;
+    coef_vec = coef;
+    index_vec = index;
+    hist_vec = &read_hist[rw_index];
+    for(j=0; j<n_coef8; j++)
+    {
+      ix = index_vec[0];
+      sum += coef_vec[0] * hist_vec[ix];
+      ix = index_vec[1];
+      sum += coef_vec[1] * hist_vec[ix];
+      ix = index_vec[2];
+      sum += coef_vec[2] * hist_vec[ix];
+      ix = index_vec[3];
+      sum += coef_vec[3] * hist_vec[ix];
+      ix = index_vec[4];
+      sum += coef_vec[4] * hist_vec[ix];
+      ix = index_vec[5];
+      sum += coef_vec[5] * hist_vec[ix];
+      ix = index_vec[6];
+      sum += coef_vec[6] * hist_vec[ix];
+      ix = index_vec[7];
+      sum += coef_vec[7] * hist_vec[ix];
+      coef_vec += 8;
+      index_vec += 8;
+    }
+    for(j=n_coef8*8; j<n_coef; j++)
+    {
+      ix = index[j];
+      sum += coef[j] * read_hist[rw_index+ix];
+    }
+    out[i] = sum;
+    rw_index++;
+    if(rw_index >= order)
+      rw_index -= order;
+  }
+  
+  x->x_rw_index = rw_index;
+  return(w+5);
+  
+sparse_FIR_tilde_perf_zero:
+  
+  while(n--)
+    *out++ = 0.0f;
+  return(w+5);
+}
+
+static void sparse_FIR_tilde_list(t_sparse_FIR_tilde *x, t_symbol *s, int argc, t_atom *argv)
+{
+  int order = x->x_sparse_FIR_order;
+  int n_arg2 = argc/2, index, i;
+  int n_coef = 0;
+  int *index_pointer = x->x_index_beg;
+  t_float *coef_pointer = x->x_coef_beg;
+  t_float coef;
+  
+  for(i=0; i<n_arg2; i++)
+  {
+    index = (int)atom_getfloat(argv++);
+    coef = (t_float)atom_getfloat(argv++);
+    if((index >= 0) && (index < order))
+    {
+      *index_pointer++ = -index;
+      *coef_pointer++ = coef;
+      n_coef++;
+    }
+  }
+  x->x_n_coef = n_coef;
+}
+
+static void sparse_FIR_tilde_matrix(t_sparse_FIR_tilde *x, t_symbol *s, int argc, t_atom *argv)
+{
+  int row, col;
+  
+  if(argc < 2)
+  {
+    post("sparse_FIR~ : corrupt matrix passed");
+    return;
+  }
+  row = (int)atom_getfloat(argv++);
+  col = (int)atom_getfloat(argv++);
+  if((row < 1)||(col < 1))
+  {
+    post("sparse_FIR~ : corrupt matrix passed");
+    return;
+  }
+  if((row*col) < (argc - 2))
+  {
+    post("sparse_FIR~ WARNING: row column product less than message content!");
+    sparse_FIR_tilde_list(x, &s_list, row*col, argv);
+  }
+  else if((row*col) > (argc-2))
+  {
+    post("sparse_FIR~ WARNING: row column product greater than message content!");
+    sparse_FIR_tilde_list(x, &s_list, argc-2, argv);
+  }
+  else
+    sparse_FIR_tilde_list(x, &s_list, argc-2, argv);
+}
+
+static void sparse_FIR_tilde_order(t_sparse_FIR_tilde *x, t_floatarg fn)
+{
+  int n = (int)fn;
+  
+  if(n > 0)
+  {
+    if(n > x->x_sparse_FIR_order)/* resize */
+    {
+      x->x_history_beg =  (t_float *)resizebytes(x->x_history_beg, 2*x->x_n_coef_malloc*sizeof(t_float), 2*n*sizeof(t_float));
+      x->x_index_beg =  (int *)resizebytes(x->x_index_beg, x->x_n_coef_malloc*sizeof(int), n*sizeof(int));
+      x->x_coef_beg =  (t_float *)resizebytes(x->x_coef_beg, x->x_n_coef_malloc*sizeof(t_float), n*sizeof(t_float));
+      x->x_n_coef_malloc = n;
+    }
+    x->x_sparse_FIR_order = n;
+    x->x_rw_index = 0;
+  }
+}
+
+static void sparse_FIR_tilde_dsp(t_sparse_FIR_tilde *x, t_signal **sp)
+{
+  dsp_add(sparse_FIR_tilde_perform, 4, sp[0]->s_vec, sp[1]->s_vec, x, sp[0]->s_n);
+}
+
+static void *sparse_FIR_tilde_new(t_floatarg fn)
+{
+  t_sparse_FIR_tilde *x = (t_sparse_FIR_tilde *)pd_new(sparse_FIR_tilde_class);
+  int n=(int)fn;
+  
+  outlet_new(&x->x_obj, &s_signal);
+  x->x_msi = 0;
+  x->x_n_coef = 0;
+  if(n < 1)
+    n = 1;
+  x->x_sparse_FIR_order = n;
+  x->x_n_coef_malloc = n;
+  x->x_history_beg = (t_float *)getbytes((2*x->x_n_coef_malloc)*sizeof(t_float));
+  x->x_index_beg = (int *)getbytes(x->x_n_coef_malloc*sizeof(int));
+  x->x_coef_beg = (t_float *)getbytes(x->x_n_coef_malloc*sizeof(t_float));
+  x->x_rw_index = 0;
+  return(x);
+}
+
+static void sparse_FIR_tilde_free(t_sparse_FIR_tilde *x)
+{
+  if(x->x_history_beg)
+    freebytes(x->x_history_beg, (2*x->x_n_coef_malloc)*sizeof(t_float));
+  if(x->x_index_beg)
+    freebytes(x->x_index_beg, x->x_n_coef_malloc*sizeof(int));
+  if(x->x_coef_beg)
+    freebytes(x->x_coef_beg, x->x_n_coef_malloc*sizeof(t_float));
+}
+
+void sparse_FIR_tilde_setup(void)
+{
+  sparse_FIR_tilde_class = class_new(gensym("sparse_FIR~"), (t_newmethod)sparse_FIR_tilde_new,
+    (t_method)sparse_FIR_tilde_free, sizeof(t_sparse_FIR_tilde), 0, A_DEFFLOAT, 0);
+  CLASS_MAINSIGNALIN(sparse_FIR_tilde_class, t_sparse_FIR_tilde, x_msi);
+  class_addmethod(sparse_FIR_tilde_class, (t_method)sparse_FIR_tilde_dsp, gensym("dsp"), 0);
+  class_addlist(sparse_FIR_tilde_class, (t_method)sparse_FIR_tilde_list);
+  class_addmethod(sparse_FIR_tilde_class, (t_method)sparse_FIR_tilde_matrix, gensym("matrix"), A_GIMME, 0);
+  class_addmethod(sparse_FIR_tilde_class, (t_method)sparse_FIR_tilde_order, gensym("order"), A_FLOAT, 0);
+  class_addmethod(sparse_FIR_tilde_class, (t_method)sparse_FIR_tilde_order, gensym("size"), A_FLOAT, 0);
+}