2 * FFT/MDCT transform with SSE optimizations
3 * Copyright (c) 2002 Fabrice Bellard.
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.
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.
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
19 #include "../dsputil.h"
22 #ifdef HAVE_BUILTIN_VECTOR
24 #include <xmmintrin.h>
26 static const float p1p1p1m1[4] __attribute__((aligned(16))) =
27 { 1.0, 1.0, 1.0, -1.0 };
29 static const float p1p1m1p1[4] __attribute__((aligned(16))) =
30 { 1.0, 1.0, -1.0, 1.0 };
32 static const float p1p1m1m1[4] __attribute__((aligned(16))) =
33 { 1.0, 1.0, -1.0, -1.0 };
36 static void print_v4sf(const char *str, __m128 a)
38 float *p = (float *)&a;
39 printf("%s: %f %f %f %f\n",
40 str, p[0], p[1], p[2], p[3]);
44 /* XXX: handle reverse case */
45 void ff_fft_calc_sse(FFTContext *s, FFTComplex *z)
50 register FFTComplex *p, *q;
51 FFTComplex *cptr, *cptr1;
57 __m128 *r, a, b, a1, c1, c2;
60 c1 = *(__m128 *)p1p1m1m1;
61 c2 = *(__m128 *)p1p1p1m1;
63 c2 = *(__m128 *)p1p1m1p1;
65 c2 = *(__m128 *)p1p1p1m1;
70 b = _mm_shuffle_ps(a, a, _MM_SHUFFLE(1, 0, 3, 2));
71 a = _mm_mul_ps(a, c1);
72 /* do the pass 0 butterfly */
76 b = _mm_shuffle_ps(a1, a1, _MM_SHUFFLE(1, 0, 3, 2));
77 a1 = _mm_mul_ps(a1, c1);
78 /* do the pass 0 butterfly */
79 b = _mm_add_ps(a1, b);
81 /* multiply third by -i */
82 b = _mm_shuffle_ps(b, b, _MM_SHUFFLE(2, 3, 1, 0));
83 b = _mm_mul_ps(b, c2);
85 /* do the pass 1 butterfly */
86 r[0] = _mm_add_ps(a, b);
87 r[1] = _mm_sub_ps(a, b);
106 __m128 a, b, c, t1, t2;
115 _mm_shuffle_ps(b, b, _MM_SHUFFLE(2, 2, 0, 0)));
116 c = *(__m128 *)(cptr + 2);
119 _mm_shuffle_ps(b, b, _MM_SHUFFLE(3, 3, 1, 1)));
120 b = _mm_add_ps(t1, t2);
123 *(__m128 *)p = _mm_add_ps(a, b);
124 *(__m128 *)q = _mm_sub_ps(a, b);
135 nblocks = nblocks >> 1;
136 nloops = nloops << 1;
137 } while (nblocks != 0);