← Back to branch summary

~siretart/ubuntu/utopic/blender/libav10

« back to all changes in this revision

Viewing changes to extern/fftw/dft/simd/codelets/t3bv_8.c

  • Committer: Bazaar Package Importer
  • Author(s): Kevin Roy
  • Date: 2011-02-08 22:20:54 UTC
  • mfrom: (1.4.2 upstream)
  • mto: (14.2.6 sid) (1.5.1)
  • mto: This revision was merged to the branch mainline in revision 27.
  • Revision ID: james.westby@ubuntu.com-20110208222054-kk0gwa4bu8h5lyq4
Tags: upstream-2.56.1-beta-svn34076
Import upstream version 2.56.1-beta-svn34076

Show diffs side-by-side

added added

removed removed

Lines of Context:
extern/fftw/dft/simd/codelets/t3bv_8.c
1
 
/*
2
 
 * Copyright (c) 2003, 2006 Matteo Frigo
3
 
 * Copyright (c) 2003, 2006 Massachusetts Institute of Technology
4
 
 *
5
 
 * This program is free software; you can redistribute it and/or modify
6
 
 * it under the terms of the GNU General Public License as published by
7
 
 * the Free Software Foundation; either version 2 of the License, or
8
 
 * (at your option) any later version.
9
 
 *
10
 
 * This program is distributed in the hope that it will be useful,
11
 
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12
 
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
13
 
 * GNU General Public License for more details.
14
 
 *
15
 
 * You should have received a copy of the GNU General Public License
16
 
 * along with this program; if not, write to the Free Software
17
 
 * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
18
 
 *
19
 
 */
20
 
 
21
 
/* This file was automatically generated --- DO NOT EDIT */
22
 
/* Generated on Sat Jul  1 22:33:15 EDT 2006 */
23
 
 
24
 
#include "codelet-dft.h"
25
 
 
26
 
#ifdef HAVE_FMA
27
 
 
28
 
/* Generated by: ../../../genfft/gen_twiddle_c -fma -reorder-insns -schedule-for-pipeline -simd -compact -variables 4 -pipeline-latency 8 -twiddle-log3 -precompute-twiddles -no-generate-bytw -n 8 -name t3bv_8 -include t3b.h -sign 1 */
29
 
 
30
 
/*
31
 
 * This function contains 37 FP additions, 32 FP multiplications,
32
 
 * (or, 27 additions, 22 multiplications, 10 fused multiply/add),
33
 
 * 43 stack variables, and 16 memory accesses
34
 
 */
35
 
/*
36
 
 * Generator Id's : 
37
 
 * $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $
38
 
 * $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
39
 
 * $Id: gen_twiddle_c.ml,v 1.14 2006-02-12 23:34:12 athena Exp $
40
 
 */
41
 
 
42
 
#include "t3b.h"
43
 
 
44
 
static const R *t3bv_8(R *ri, R *ii, const R *W, stride ios, INT m, INT dist)
45
 
{
46
 
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
47
 
     INT i;
48
 
     R *x;
49
 
     x = ii;
50
 
     for (i = m; i > 0; i = i - VL, x = x + (VL * dist), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(ios)) {
51
 
          V T2, T3, Tb, T1, T5, Tn, Tq, T8, Td, T4, Ta, Tp, Tg, Ti, T9;
52
 
          T2 = LDW(&(W[0]));
53
 
          T3 = LDW(&(W[TWVL * 2]));
54
 
          Tb = LDW(&(W[TWVL * 4]));
55
 
          T1 = LD(&(x[0]), dist, &(x[0]));
56
 
          T5 = LD(&(x[WS(ios, 4)]), dist, &(x[0]));
57
 
          Tn = LD(&(x[WS(ios, 2)]), dist, &(x[0]));
58
 
          Tq = LD(&(x[WS(ios, 6)]), dist, &(x[0]));
59
 
          T8 = LD(&(x[WS(ios, 1)]), dist, &(x[WS(ios, 1)]));
60
 
          Td = LD(&(x[WS(ios, 5)]), dist, &(x[WS(ios, 1)]));
61
 
          T4 = VZMUL(T2, T3);
62
 
          Ta = VZMULJ(T2, T3);
63
 
          Tp = VZMULJ(T2, Tb);
64
 
          Tg = LD(&(x[WS(ios, 7)]), dist, &(x[WS(ios, 1)]));
65
 
          Ti = LD(&(x[WS(ios, 3)]), dist, &(x[WS(ios, 1)]));
66
 
          T9 = VZMUL(T2, T8);
67
 
          {
68
 
               V T6, To, Tc, Tr, Th, Tj;
69
 
               T6 = VZMUL(T4, T5);
70
 
               To = VZMUL(Ta, Tn);
71
 
               Tc = VZMULJ(Ta, Tb);
72
 
               Tr = VZMUL(Tp, Tq);
73
 
               Th = VZMUL(Tb, Tg);
74
 
               Tj = VZMUL(T3, Ti);
75
 
               {
76
 
                    V Tx, T7, Te, Ts, Ty, Tk, TB;
77
 
                    Tx = VADD(T1, T6);
78
 
                    T7 = VSUB(T1, T6);
79
 
                    Te = VZMUL(Tc, Td);
80
 
                    Ts = VSUB(To, Tr);
81
 
                    Ty = VADD(To, Tr);
82
 
                    Tk = VSUB(Th, Tj);
83
 
                    TB = VADD(Th, Tj);
84
 
                    {
85
 
                         V Tf, TA, Tz, TD;
86
 
                         Tf = VSUB(T9, Te);
87
 
                         TA = VADD(T9, Te);
88
 
                         Tz = VSUB(Tx, Ty);
89
 
                         TD = VADD(Tx, Ty);
90
 
                         {
91
 
                              V TC, TE, Tl, Tt;
92
 
                              TC = VSUB(TA, TB);
93
 
                              TE = VADD(TA, TB);
94
 
                              Tl = VADD(Tf, Tk);
95
 
                              Tt = VSUB(Tf, Tk);
96
 
                              {
97
 
                                   V Tu, Tw, Tm, Tv;
98
 
                                   ST(&(x[0]), VADD(TD, TE), dist, &(x[0]));
99
 
                                   ST(&(x[WS(ios, 4)]), VSUB(TD, TE), dist, &(x[0]));
100
 
                                   ST(&(x[WS(ios, 2)]), VFMAI(TC, Tz), dist, &(x[0]));
101
 
                                   ST(&(x[WS(ios, 6)]), VFNMSI(TC, Tz), dist, &(x[0]));
102
 
                                   Tu = VFNMS(LDK(KP707106781), Tt, Ts);
103
 
                                   Tw = VFMA(LDK(KP707106781), Tt, Ts);
104
 
                                   Tm = VFNMS(LDK(KP707106781), Tl, T7);
105
 
                                   Tv = VFMA(LDK(KP707106781), Tl, T7);
106
 
                                   ST(&(x[WS(ios, 1)]), VFMAI(Tw, Tv), dist, &(x[WS(ios, 1)]));
107
 
                                   ST(&(x[WS(ios, 7)]), VFNMSI(Tw, Tv), dist, &(x[WS(ios, 1)]));
108
 
                                   ST(&(x[WS(ios, 5)]), VFMAI(Tu, Tm), dist, &(x[WS(ios, 1)]));
109
 
                                   ST(&(x[WS(ios, 3)]), VFNMSI(Tu, Tm), dist, &(x[WS(ios, 1)]));
110
 
                              }
111
 
                         }
112
 
                    }
113
 
               }
114
 
          }
115
 
     }
116
 
     return W;
117
 
}
118
 
 
119
 
static const tw_instr twinstr[] = {
120
 
     VTW(1),
121
 
     VTW(3),
122
 
     VTW(7),
123
 
     {TW_NEXT, VL, 0}
124
 
};
125
 
 
126
 
static const ct_desc desc = { 8, "t3bv_8", twinstr, &GENUS, {27, 22, 10, 0}, 0, 0, 0 };
127
 
 
128
 
void X(codelet_t3bv_8) (planner *p) {
129
 
     X(kdft_dit_register) (p, t3bv_8, &desc);
130
 
}
131
 
#else                           /* HAVE_FMA */
132
 
 
133
 
/* Generated by: ../../../genfft/gen_twiddle_c -simd -compact -variables 4 -pipeline-latency 8 -twiddle-log3 -precompute-twiddles -no-generate-bytw -n 8 -name t3bv_8 -include t3b.h -sign 1 */
134
 
 
135
 
/*
136
 
 * This function contains 37 FP additions, 24 FP multiplications,
137
 
 * (or, 37 additions, 24 multiplications, 0 fused multiply/add),
138
 
 * 31 stack variables, and 16 memory accesses
139
 
 */
140
 
/*
141
 
 * Generator Id's : 
142
 
 * $Id: algsimp.ml,v 1.9 2006-02-12 23:34:12 athena Exp $
143
 
 * $Id: fft.ml,v 1.4 2006-01-05 03:04:27 stevenj Exp $
144
 
 * $Id: gen_twiddle_c.ml,v 1.14 2006-02-12 23:34:12 athena Exp $
145
 
 */
146
 
 
147
 
#include "t3b.h"
148
 
 
149
 
static const R *t3bv_8(R *ri, R *ii, const R *W, stride ios, INT m, INT dist)
150
 
{
151
 
     DVK(KP707106781, +0.707106781186547524400844362104849039284835938);
152
 
     INT i;
153
 
     R *x;
154
 
     x = ii;
155
 
     for (i = m; i > 0; i = i - VL, x = x + (VL * dist), W = W + (TWVL * 6), MAKE_VOLATILE_STRIDE(ios)) {
156
 
          V T1, T4, T5, Tp, T6, T7, Tj;
157
 
          T1 = LDW(&(W[0]));
158
 
          T4 = LDW(&(W[TWVL * 2]));
159
 
          T5 = VZMULJ(T1, T4);
160
 
          Tp = VZMUL(T1, T4);
161
 
          T6 = LDW(&(W[TWVL * 4]));
162
 
          T7 = VZMULJ(T5, T6);
163
 
          Tj = VZMULJ(T1, T6);
164
 
          {
165
 
               V Ts, Tx, Tm, Ty, Ta, TA, Tf, TB, To, Tr, Tq;
166
 
               To = LD(&(x[0]), dist, &(x[0]));
167
 
               Tq = LD(&(x[WS(ios, 4)]), dist, &(x[0]));
168
 
               Tr = VZMUL(Tp, Tq);
169
 
               Ts = VSUB(To, Tr);
170
 
               Tx = VADD(To, Tr);
171
 
               {
172
 
                    V Ti, Tl, Th, Tk;
173
 
                    Th = LD(&(x[WS(ios, 2)]), dist, &(x[0]));
174
 
                    Ti = VZMUL(T5, Th);
175
 
                    Tk = LD(&(x[WS(ios, 6)]), dist, &(x[0]));
176
 
                    Tl = VZMUL(Tj, Tk);
177
 
                    Tm = VSUB(Ti, Tl);
178
 
                    Ty = VADD(Ti, Tl);
179
 
               }
180
 
               {
181
 
                    V T3, T9, T2, T8;
182
 
                    T2 = LD(&(x[WS(ios, 1)]), dist, &(x[WS(ios, 1)]));
183
 
                    T3 = VZMUL(T1, T2);
184
 
                    T8 = LD(&(x[WS(ios, 5)]), dist, &(x[WS(ios, 1)]));
185
 
                    T9 = VZMUL(T7, T8);
186
 
                    Ta = VSUB(T3, T9);
187
 
                    TA = VADD(T3, T9);
188
 
               }
189
 
               {
190
 
                    V Tc, Te, Tb, Td;
191
 
                    Tb = LD(&(x[WS(ios, 7)]), dist, &(x[WS(ios, 1)]));
192
 
                    Tc = VZMUL(T6, Tb);
193
 
                    Td = LD(&(x[WS(ios, 3)]), dist, &(x[WS(ios, 1)]));
194
 
                    Te = VZMUL(T4, Td);
195
 
                    Tf = VSUB(Tc, Te);
196
 
                    TB = VADD(Tc, Te);
197
 
               }
198
 
               {
199
 
                    V Tz, TC, TD, TE;
200
 
                    Tz = VSUB(Tx, Ty);
201
 
                    TC = VBYI(VSUB(TA, TB));
202
 
                    ST(&(x[WS(ios, 6)]), VSUB(Tz, TC), dist, &(x[0]));
203
 
                    ST(&(x[WS(ios, 2)]), VADD(Tz, TC), dist, &(x[0]));
204
 
                    TD = VADD(Tx, Ty);
205
 
                    TE = VADD(TA, TB);
206
 
                    ST(&(x[WS(ios, 4)]), VSUB(TD, TE), dist, &(x[0]));
207
 
                    ST(&(x[0]), VADD(TD, TE), dist, &(x[0]));
208
 
                    {
209
 
                         V Tn, Tv, Tu, Tw, Tg, Tt;
210
 
                         Tg = VMUL(LDK(KP707106781), VSUB(Ta, Tf));
211
 
                         Tn = VBYI(VSUB(Tg, Tm));
212
 
                         Tv = VBYI(VADD(Tm, Tg));
213
 
                         Tt = VMUL(LDK(KP707106781), VADD(Ta, Tf));
214
 
                         Tu = VSUB(Ts, Tt);
215
 
                         Tw = VADD(Ts, Tt);
216
 
                         ST(&(x[WS(ios, 3)]), VADD(Tn, Tu), dist, &(x[WS(ios, 1)]));
217
 
                         ST(&(x[WS(ios, 7)]), VSUB(Tw, Tv), dist, &(x[WS(ios, 1)]));
218
 
                         ST(&(x[WS(ios, 5)]), VSUB(Tu, Tn), dist, &(x[WS(ios, 1)]));
219
 
                         ST(&(x[WS(ios, 1)]), VADD(Tv, Tw), dist, &(x[WS(ios, 1)]));
220
 
                    }
221
 
               }
222
 
          }
223
 
     }
224
 
     return W;
225
 
}
226
 
 
227
 
static const tw_instr twinstr[] = {
228
 
     VTW(1),
229
 
     VTW(3),
230
 
     VTW(7),
231
 
     {TW_NEXT, VL, 0}
232
 
};
233
 
 
234
 
static const ct_desc desc = { 8, "t3bv_8", twinstr, &GENUS, {37, 24, 0, 0}, 0, 0, 0 };
235
 
 
236
 
void X(codelet_t3bv_8) (planner *p) {
237
 
     X(kdft_dit_register) (p, t3bv_8, &desc);
238
 
}
239
 
#endif                          /* HAVE_FMA */