• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /*
2  * Altivec optimized MP3 decoding functions
3  * Copyright (c) 2010 Vitor Sessak
4  *
5  * This file is part of FFmpeg.
6  *
7  * FFmpeg is free software; you can redistribute it and/or
8  * modify it under the terms of the GNU Lesser General Public
9  * License as published by the Free Software Foundation; either
10  * version 2.1 of the License, or (at your option) any later version.
11  *
12  * FFmpeg is distributed in the hope that it will be useful,
13  * but WITHOUT ANY WARRANTY; without even the implied warranty of
14  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
15  * Lesser General Public License for more details.
16  *
17  * You should have received a copy of the GNU Lesser General Public
18  * License along with FFmpeg; if not, write to the Free Software
19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20  */
21 
22 #include "config.h"
23 #include "libavutil/attributes.h"
24 #include "libavutil/cpu.h"
25 #include "libavutil/internal.h"
26 #include "libavutil/mem_internal.h"
27 #include "libavutil/ppc/cpu.h"
28 #include "libavutil/ppc/util_altivec.h"
29 #include "libavcodec/mpegaudiodsp.h"
30 
31 #if HAVE_ALTIVEC
32 
33 #define MACS(rt, ra, rb) rt+=(ra)*(rb)
34 #define MLSS(rt, ra, rb) rt-=(ra)*(rb)
35 
36 #define SUM8(op, sum, w, p)               \
37 {                                         \
38     op(sum, (w)[0 * 64], (p)[0 * 64]);    \
39     op(sum, (w)[1 * 64], (p)[1 * 64]);    \
40     op(sum, (w)[2 * 64], (p)[2 * 64]);    \
41     op(sum, (w)[3 * 64], (p)[3 * 64]);    \
42     op(sum, (w)[4 * 64], (p)[4 * 64]);    \
43     op(sum, (w)[5 * 64], (p)[5 * 64]);    \
44     op(sum, (w)[6 * 64], (p)[6 * 64]);    \
45     op(sum, (w)[7 * 64], (p)[7 * 64]);    \
46 }
47 
apply_window(const float * buf,const float * win1,const float * win2,float * sum1,float * sum2,int len)48 static void apply_window(const float *buf, const float *win1,
49                          const float *win2, float *sum1, float *sum2, int len)
50 {
51     const vector float *win1a = (const vector float *) win1;
52     const vector float *win2a = (const vector float *) win2;
53     const vector float *bufa  = (const vector float *) buf;
54     vector float *sum1a = (vector float *) sum1;
55     vector float *sum2a = (vector float *) sum2;
56     vector float av_uninit(v0), av_uninit(v4);
57     vector float v1, v2, v3;
58 
59     len = len >> 2;
60 
61 #define MULT(a, b)                         \
62     {                                      \
63         v1 = vec_ld(a, win1a);             \
64         v2 = vec_ld(b, win2a);             \
65         v3 = vec_ld(a, bufa);              \
66         v0 = vec_madd(v3, v1, v0);         \
67         v4 = vec_madd(v2, v3, v4);         \
68     }
69 
70     while (len--) {
71         v0 = vec_xor(v0, v0);
72         v4 = vec_xor(v4, v4);
73 
74         MULT(   0,   0);
75         MULT( 256,  64);
76         MULT( 512, 128);
77         MULT( 768, 192);
78         MULT(1024, 256);
79         MULT(1280, 320);
80         MULT(1536, 384);
81         MULT(1792, 448);
82 
83         vec_st(v0, 0, sum1a);
84         vec_st(v4, 0, sum2a);
85         sum1a++;
86         sum2a++;
87         win1a++;
88         win2a++;
89         bufa++;
90     }
91 }
92 
apply_window_mp3(float * in,float * win,int * unused,float * out,ptrdiff_t incr)93 static void apply_window_mp3(float *in, float *win, int *unused, float *out,
94                              ptrdiff_t incr)
95 {
96     LOCAL_ALIGNED_16(float, suma, [17]);
97     LOCAL_ALIGNED_16(float, sumb, [17]);
98     LOCAL_ALIGNED_16(float, sumc, [17]);
99     LOCAL_ALIGNED_16(float, sumd, [17]);
100 
101     float sum;
102     int j;
103     float *out2 = out + 32 * incr;
104 
105     /* copy to avoid wrap */
106     memcpy(in + 512, in, 32 * sizeof(*in));
107 
108     apply_window(in + 16, win     , win + 512, suma, sumc, 16);
109     apply_window(in + 32, win + 48, win + 640, sumb, sumd, 16);
110 
111     SUM8(MLSS, suma[0], win + 32, in + 48);
112 
113     sumc[ 0] = 0;
114     sumb[16] = 0;
115     sumd[16] = 0;
116 
117     out[0  ]  = suma[   0];
118     out += incr;
119     out2 -= incr;
120     for(j=1;j<16;j++) {
121         *out  =  suma[   j] - sumd[16-j];
122         *out2 = -sumb[16-j] - sumc[   j];
123         out  += incr;
124         out2 -= incr;
125     }
126 
127     sum = 0;
128     SUM8(MLSS, sum, win + 16 + 32, in + 32);
129     *out = sum;
130 }
131 
132 #endif /* HAVE_ALTIVEC */
133 
ff_mpadsp_init_ppc(MPADSPContext * s)134 av_cold void ff_mpadsp_init_ppc(MPADSPContext *s)
135 {
136 #if HAVE_ALTIVEC
137     if (!PPC_ALTIVEC(av_get_cpu_flags()))
138         return;
139 
140     s->apply_window_float = apply_window_mp3;
141 #endif /* HAVE_ALTIVEC */
142 }
143