2 * MMX optimized MP3 decoding functions
3 * Copyright (c) 2010 Vitor Sessak
5 * This file is part of FFmpeg.
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.
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.
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
22 #include "libavutil/cpu.h"
23 #include "libavutil/x86_cpu.h"
24 #include "libavcodec/dsputil.h"
25 #include "libavcodec/mpegaudiodsp.h"
27 #define MACS(rt, ra, rb) rt+=(ra)*(rb)
28 #define MLSS(rt, ra, rb) rt-=(ra)*(rb)
30 #define SUM8(op, sum, w, p) \
32 op(sum, (w)[0 * 64], (p)[0 * 64]); \
33 op(sum, (w)[1 * 64], (p)[1 * 64]); \
34 op(sum, (w)[2 * 64], (p)[2 * 64]); \
35 op(sum, (w)[3 * 64], (p)[3 * 64]); \
36 op(sum, (w)[4 * 64], (p)[4 * 64]); \
37 op(sum, (w)[5 * 64], (p)[5 * 64]); \
38 op(sum, (w)[6 * 64], (p)[6 * 64]); \
39 op(sum, (w)[7 * 64], (p)[7 * 64]); \
42 static void apply_window(const float *buf, const float *win1,
43 const float *win2, float *sum1, float *sum2, int len)
45 x86_reg count = - 4*len;
46 const float *win1a = win1+len;
47 const float *win2a = win2+len;
48 const float *bufa = buf+len;
49 float *sum1a = sum1+len;
50 float *sum2a = sum2+len;
54 "movaps " #a "(%1,%0), %%xmm1 \n\t" \
55 "movaps " #a "(%3,%0), %%xmm2 \n\t" \
56 "mulps %%xmm2, %%xmm1 \n\t" \
57 "subps %%xmm1, %%xmm0 \n\t" \
58 "mulps " #b "(%2,%0), %%xmm2 \n\t" \
59 "subps %%xmm2, %%xmm4 \n\t" \
63 "xorps %%xmm0, %%xmm0 \n\t"
64 "xorps %%xmm4, %%xmm4 \n\t"
75 "movaps %%xmm0, (%4,%0) \n\t"
76 "movaps %%xmm4, (%5,%0) \n\t"
80 :"r"(win1a), "r"(win2a), "r"(bufa), "r"(sum1a), "r"(sum2a)
86 static void apply_window_mp3(float *in, float *win, int *unused, float *out,
89 LOCAL_ALIGNED_16(float, suma, [17]);
90 LOCAL_ALIGNED_16(float, sumb, [17]);
91 LOCAL_ALIGNED_16(float, sumc, [17]);
92 LOCAL_ALIGNED_16(float, sumd, [17]);
96 /* copy to avoid wrap */
97 memcpy(in + 512, in, 32 * sizeof(*in));
99 apply_window(in + 16, win , win + 512, suma, sumc, 16);
100 apply_window(in + 32, win + 48, win + 640, sumb, sumd, 16);
102 SUM8(MACS, suma[0], win + 32, in + 48);
108 #define SUMS(suma, sumb, sumc, sumd, out1, out2) \
109 "movups " #sumd "(%4), %%xmm0 \n\t" \
110 "shufps $0x1b, %%xmm0, %%xmm0 \n\t" \
111 "subps " #suma "(%1), %%xmm0 \n\t" \
112 "movaps %%xmm0," #out1 "(%0) \n\t" \
114 "movups " #sumc "(%3), %%xmm0 \n\t" \
115 "shufps $0x1b, %%xmm0, %%xmm0 \n\t" \
116 "addps " #sumb "(%2), %%xmm0 \n\t" \
117 "movaps %%xmm0," #out2 "(%0) \n\t"
121 SUMS( 0, 48, 4, 52, 0, 112)
122 SUMS(16, 32, 20, 36, 16, 96)
123 SUMS(32, 16, 36, 20, 32, 80)
124 SUMS(48, 0, 52, 4, 48, 64)
127 :"r"(&suma[0]), "r"(&sumb[0]), "r"(&sumc[0]), "r"(&sumd[0])
133 float *out2 = out + 32 * incr;
138 *out = -suma[ j] + sumd[16-j];
139 *out2 = sumb[16-j] + sumc[ j];
146 SUM8(MLSS, sum, win + 16 + 32, in + 32);
150 void ff_mpadsp_init_mmx(MPADSPContext *s)
152 int mm_flags = av_get_cpu_flags();
154 if (mm_flags & AV_CPU_FLAG_SSE2) {
155 s->apply_window_float = apply_window_mp3;