631848265a27b9d7f5adf80fb7cf12383d653a36
[libav.git] / libavcodec / i386 / fft_sse.c
1 /*
2 * FFT/MDCT transform with SSE optimizations
3 * Copyright (c) 2002 Fabrice Bellard.
4 *
5 * This library is free software; you can redistribute it and/or
6 * modify it under the terms of the GNU Lesser General Public
7 * License as published by the Free Software Foundation; either
8 * version 2 of the License, or (at your option) any later version.
9 *
10 * This library 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 GNU
13 * Lesser General Public License for more details.
14 *
15 * You should have received a copy of the GNU Lesser General Public
16 * License along with this library; if not, write to the Free Software
17 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
18 */
19 #include "../dsputil.h"
20 #include <math.h>
21
22 #ifdef HAVE_BUILTIN_VECTOR
23
24 #include <xmmintrin.h>
25
26 static const int p1p1p1m1[4] __attribute__((aligned(16))) =
27 { 0, 0, 0, 1 << 31 };
28
29 static const int p1p1m1p1[4] __attribute__((aligned(16))) =
30 { 0, 0, 1 << 31, 0 };
31
32 static const int p1p1m1m1[4] __attribute__((aligned(16))) =
33 { 0, 0, 1 << 31, 1 << 31 };
34
35 #if 0
36 static void print_v4sf(const char *str, __m128 a)
37 {
38 float *p = (float *)&a;
39 printf("%s: %f %f %f %f\n",
40 str, p[0], p[1], p[2], p[3]);
41 }
42 #endif
43
44 /* XXX: handle reverse case */
45 void ff_fft_calc_sse(FFTContext *s, FFTComplex *z)
46 {
47 int ln = s->nbits;
48 int j, np, np2;
49 int nblocks, nloops;
50 register FFTComplex *p, *q;
51 FFTComplex *cptr, *cptr1;
52 int k;
53
54 np = 1 << ln;
55
56 {
57 __m128 *r, a, b, a1, c1, c2;
58
59 r = (__m128 *)&z[0];
60 c1 = *(__m128 *)p1p1m1m1;
61 if (s->inverse)
62 c2 = *(__m128 *)p1p1m1p1;
63 else
64 c2 = *(__m128 *)p1p1p1m1;
65
66 j = (np >> 2);
67 do {
68 a = r[0];
69 b = _mm_shuffle_ps(a, a, _MM_SHUFFLE(1, 0, 3, 2));
70 a = _mm_xor_ps(a, c1);
71 /* do the pass 0 butterfly */
72 a = _mm_add_ps(a, b);
73
74 a1 = r[1];
75 b = _mm_shuffle_ps(a1, a1, _MM_SHUFFLE(1, 0, 3, 2));
76 a1 = _mm_xor_ps(a1, c1);
77 /* do the pass 0 butterfly */
78 b = _mm_add_ps(a1, b);
79
80 /* multiply third by -i */
81 /* by toggling the sign bit */
82 b = _mm_shuffle_ps(b, b, _MM_SHUFFLE(2, 3, 1, 0));
83 b = _mm_xor_ps(b, c2);
84
85 /* do the pass 1 butterfly */
86 r[0] = _mm_add_ps(a, b);
87 r[1] = _mm_sub_ps(a, b);
88 r += 2;
89 } while (--j != 0);
90 }
91 /* pass 2 .. ln-1 */
92
93 nblocks = np >> 3;
94 nloops = 1 << 2;
95 np2 = np >> 1;
96
97 cptr1 = s->exptab1;
98 do {
99 p = z;
100 q = z + nloops;
101 j = nblocks;
102 do {
103 cptr = cptr1;
104 k = nloops >> 1;
105 do {
106 __m128 a, b, c, t1, t2;
107
108 a = *(__m128 *)p;
109 b = *(__m128 *)q;
110
111 /* complex mul */
112 c = *(__m128 *)cptr;
113 /* cre*re cim*re */
114 t1 = _mm_mul_ps(c,
115 _mm_shuffle_ps(b, b, _MM_SHUFFLE(2, 2, 0, 0)));
116 c = *(__m128 *)(cptr + 2);
117 /* -cim*im cre*im */
118 t2 = _mm_mul_ps(c,
119 _mm_shuffle_ps(b, b, _MM_SHUFFLE(3, 3, 1, 1)));
120 b = _mm_add_ps(t1, t2);
121
122 /* butterfly */
123 *(__m128 *)p = _mm_add_ps(a, b);
124 *(__m128 *)q = _mm_sub_ps(a, b);
125
126 p += 2;
127 q += 2;
128 cptr += 4;
129 } while (--k);
130
131 p += nloops;
132 q += nloops;
133 } while (--j);
134 cptr1 += nloops * 2;
135 nblocks = nblocks >> 1;
136 nloops = nloops << 1;
137 } while (nblocks != 0);
138 }
139
140 #endif