2 * x86-optimized AC-3 DSP utils
3 * Copyright (c) 2011 Justin Ruggles
5 * This file is part of Libav.
7 * Libav 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 * Libav 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 Libav; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
22 #include "libavutil/x86/asm.h"
23 #include "libavutil/x86/cpu.h"
24 #include "dsputil_mmx.h"
25 #include "libavcodec/ac3dsp.h"
27 extern void ff_ac3_exponent_min_mmx (uint8_t *exp, int num_reuse_blocks, int nb_coefs);
28 extern void ff_ac3_exponent_min_mmxext(uint8_t *exp, int num_reuse_blocks, int nb_coefs);
29 extern void ff_ac3_exponent_min_sse2 (uint8_t *exp, int num_reuse_blocks, int nb_coefs);
31 extern int ff_ac3_max_msb_abs_int16_mmx (const int16_t *src, int len);
32 extern int ff_ac3_max_msb_abs_int16_mmx2 (const int16_t *src, int len);
33 extern int ff_ac3_max_msb_abs_int16_sse2 (const int16_t *src, int len);
34 extern int ff_ac3_max_msb_abs_int16_ssse3(const int16_t *src, int len);
36 extern void ff_ac3_lshift_int16_mmx (int16_t *src, unsigned int len, unsigned int shift);
37 extern void ff_ac3_lshift_int16_sse2(int16_t *src, unsigned int len, unsigned int shift);
39 extern void ff_ac3_rshift_int32_mmx (int32_t *src, unsigned int len, unsigned int shift);
40 extern void ff_ac3_rshift_int32_sse2(int32_t *src, unsigned int len, unsigned int shift);
42 extern void ff_float_to_fixed24_3dnow(int32_t *dst, const float *src, unsigned int len);
43 extern void ff_float_to_fixed24_sse (int32_t *dst, const float *src, unsigned int len);
44 extern void ff_float_to_fixed24_sse2 (int32_t *dst, const float *src, unsigned int len);
46 extern int ff_ac3_compute_mantissa_size_sse2(uint16_t mant_cnt[6][16]);
48 extern void ff_ac3_extract_exponents_3dnow(uint8_t *exp, int32_t *coef, int nb_coefs);
49 extern void ff_ac3_extract_exponents_sse2 (uint8_t *exp, int32_t *coef, int nb_coefs);
50 extern void ff_ac3_extract_exponents_ssse3(uint8_t *exp, int32_t *coef, int nb_coefs);
52 av_cold void ff_ac3dsp_init_x86(AC3DSPContext *c, int bit_exact)
54 int mm_flags = av_get_cpu_flags();
56 if (EXTERNAL_MMX(mm_flags)) {
57 c->ac3_exponent_min = ff_ac3_exponent_min_mmx;
58 c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_mmx;
59 c->ac3_lshift_int16 = ff_ac3_lshift_int16_mmx;
60 c->ac3_rshift_int32 = ff_ac3_rshift_int32_mmx;
62 if (EXTERNAL_AMD3DNOW(mm_flags)) {
63 c->extract_exponents = ff_ac3_extract_exponents_3dnow;
65 c->float_to_fixed24 = ff_float_to_fixed24_3dnow;
68 if (EXTERNAL_MMXEXT(mm_flags)) {
69 c->ac3_exponent_min = ff_ac3_exponent_min_mmxext;
70 c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_mmx2;
72 if (EXTERNAL_SSE(mm_flags)) {
73 c->float_to_fixed24 = ff_float_to_fixed24_sse;
75 if (EXTERNAL_SSE2(mm_flags)) {
76 c->ac3_exponent_min = ff_ac3_exponent_min_sse2;
77 c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_sse2;
78 c->float_to_fixed24 = ff_float_to_fixed24_sse2;
79 c->compute_mantissa_size = ff_ac3_compute_mantissa_size_sse2;
80 c->extract_exponents = ff_ac3_extract_exponents_sse2;
81 if (!(mm_flags & AV_CPU_FLAG_SSE2SLOW)) {
82 c->ac3_lshift_int16 = ff_ac3_lshift_int16_sse2;
83 c->ac3_rshift_int32 = ff_ac3_rshift_int32_sse2;
86 if (EXTERNAL_SSSE3(mm_flags)) {
87 c->ac3_max_msb_abs_int16 = ff_ac3_max_msb_abs_int16_ssse3;
88 if (!(mm_flags & AV_CPU_FLAG_ATOM)) {
89 c->extract_exponents = ff_ac3_extract_exponents_ssse3;