2 * Copyright (C) 2012 Michael Niedermayer (michaelni@gmx.at)
4 * This file is part of libswresample
6 * libswresample is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * libswresample is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with libswresample; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21 #include "libavutil/x86/cpu.h"
22 #include "libswresample/swresample_internal.h"
23 #include "libswresample/audioconvert.h"
25 #define PROTO(pre, in, out, cap) void ff ## pre ## in## _to_ ##out## _a_ ##cap(uint8_t **dst, const uint8_t **src, int len);
26 #define PROTO2(pre, out, cap) PROTO(pre, int16, out, cap) PROTO(pre, int32, out, cap) PROTO(pre, float, out, cap)
27 #define PROTO3(pre, cap) PROTO2(pre, int16, cap) PROTO2(pre, int32, cap) PROTO2(pre, float, cap)
28 #define PROTO4(pre) PROTO3(pre, mmx) PROTO3(pre, sse) PROTO3(pre, sse2) PROTO3(pre, ssse3) PROTO3(pre, sse4) PROTO3(pre, avx) PROTO3(pre, avx2)
36 av_cold void swri_audio_convert_init_x86(struct AudioConvert *ac,
37 enum AVSampleFormat out_fmt,
38 enum AVSampleFormat in_fmt,
40 int mm_flags = av_get_cpu_flags();
44 //FIXME add memcpy case
46 #define MULTI_CAPS_FUNC(flag, cap) \
47 if (EXTERNAL_##flag(mm_flags)) {\
48 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S16 || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16P)\
49 ac->simd_f = ff_int16_to_int32_a_ ## cap;\
50 if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S32 || out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S32P)\
51 ac->simd_f = ff_int32_to_int16_a_ ## cap;\
54 MULTI_CAPS_FUNC(MMX, mmx)
55 MULTI_CAPS_FUNC(SSE2, sse2)
57 if(EXTERNAL_MMX(mm_flags)) {
59 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
60 ac->simd_f = ff_pack_6ch_float_to_float_a_mmx;
63 if(EXTERNAL_SSE(mm_flags)) {
65 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
66 ac->simd_f = ff_pack_6ch_float_to_float_a_sse;
68 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S32)
69 ac->simd_f = ff_unpack_6ch_float_to_float_a_sse;
72 if(EXTERNAL_SSE2(mm_flags)) {
73 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32 || out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32P)
74 ac->simd_f = ff_int32_to_float_a_sse2;
75 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S16 || out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16P)
76 ac->simd_f = ff_int16_to_float_a_sse2;
77 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLTP)
78 ac->simd_f = ff_float_to_int32_a_sse2;
79 if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_FLTP)
80 ac->simd_f = ff_float_to_int16_a_sse2;
83 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
84 ac->simd_f = ff_pack_2ch_int32_to_int32_a_sse2;
85 if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S16P)
86 ac->simd_f = ff_pack_2ch_int16_to_int16_a_sse2;
87 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S16P)
88 ac->simd_f = ff_pack_2ch_int16_to_int32_a_sse2;
89 if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S32P)
90 ac->simd_f = ff_pack_2ch_int32_to_int16_a_sse2;
92 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S32)
93 ac->simd_f = ff_unpack_2ch_int32_to_int32_a_sse2;
94 if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S16)
95 ac->simd_f = ff_unpack_2ch_int16_to_int16_a_sse2;
96 if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16)
97 ac->simd_f = ff_unpack_2ch_int16_to_int32_a_sse2;
98 if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S32)
99 ac->simd_f = ff_unpack_2ch_int32_to_int16_a_sse2;
101 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
102 ac->simd_f = ff_pack_2ch_int32_to_float_a_sse2;
103 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
104 ac->simd_f = ff_pack_2ch_float_to_int32_a_sse2;
105 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S16P)
106 ac->simd_f = ff_pack_2ch_int16_to_float_a_sse2;
107 if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_FLTP)
108 ac->simd_f = ff_pack_2ch_float_to_int16_a_sse2;
109 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32)
110 ac->simd_f = ff_unpack_2ch_int32_to_float_a_sse2;
111 if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLT)
112 ac->simd_f = ff_unpack_2ch_float_to_int32_a_sse2;
113 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16)
114 ac->simd_f = ff_unpack_2ch_int16_to_float_a_sse2;
115 if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_FLT)
116 ac->simd_f = ff_unpack_2ch_float_to_int16_a_sse2;
119 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
120 ac->simd_f = ff_pack_6ch_int32_to_float_a_sse2;
121 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
122 ac->simd_f = ff_pack_6ch_float_to_int32_a_sse2;
124 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32)
125 ac->simd_f = ff_unpack_6ch_int32_to_float_a_sse2;
126 if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLT)
127 ac->simd_f = ff_unpack_6ch_float_to_int32_a_sse2;
130 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
131 ac->simd_f = ff_pack_8ch_float_to_float_a_sse2;
132 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
133 ac->simd_f = ff_pack_8ch_int32_to_float_a_sse2;
134 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
135 ac->simd_f = ff_pack_8ch_float_to_int32_a_sse2;
138 if(EXTERNAL_SSSE3(mm_flags)) {
140 if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S16)
141 ac->simd_f = ff_unpack_2ch_int16_to_int16_a_ssse3;
142 if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16)
143 ac->simd_f = ff_unpack_2ch_int16_to_int32_a_ssse3;
144 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16)
145 ac->simd_f = ff_unpack_2ch_int16_to_float_a_ssse3;
148 if(EXTERNAL_AVX_FAST(mm_flags)) {
149 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32 || out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32P)
150 ac->simd_f = ff_int32_to_float_a_avx;
152 if(EXTERNAL_AVX(mm_flags)) {
154 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
155 ac->simd_f = ff_pack_6ch_float_to_float_a_avx;
156 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
157 ac->simd_f = ff_pack_6ch_int32_to_float_a_avx;
158 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
159 ac->simd_f = ff_pack_6ch_float_to_int32_a_avx;
161 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S32)
162 ac->simd_f = ff_unpack_6ch_float_to_float_a_avx;
163 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32)
164 ac->simd_f = ff_unpack_6ch_int32_to_float_a_avx;
165 if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLT)
166 ac->simd_f = ff_unpack_6ch_float_to_int32_a_avx;
169 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_FLTP || out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S32P)
170 ac->simd_f = ff_pack_8ch_float_to_float_a_avx;
171 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
172 ac->simd_f = ff_pack_8ch_int32_to_float_a_avx;
173 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
174 ac->simd_f = ff_pack_8ch_float_to_int32_a_avx;
177 if(EXTERNAL_AVX2_FAST(mm_flags)) {
178 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLT || out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLTP)
179 ac->simd_f = ff_float_to_int32_a_avx2;