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/attributes.h"
22 #include "libavutil/x86/cpu.h"
23 #include "libswresample/swresample_internal.h"
24 #include "libswresample/audioconvert.h"
26 #define PROTO(pre, in, out, cap) void ff ## pre ## in## _to_ ##out## _a_ ##cap(uint8_t **dst, const uint8_t **src, int len);
27 #define PROTO2(pre, out, cap) PROTO(pre, int16, out, cap) PROTO(pre, int32, out, cap) PROTO(pre, float, out, cap)
28 #define PROTO3(pre, cap) PROTO2(pre, int16, cap) PROTO2(pre, int32, cap) PROTO2(pre, float, cap)
29 #define PROTO4(pre) PROTO3(pre, mmx) PROTO3(pre, sse) PROTO3(pre, sse2) PROTO3(pre, ssse3) PROTO3(pre, sse4) PROTO3(pre, avx) PROTO3(pre, avx2)
37 av_cold void swri_audio_convert_init_x86(struct AudioConvert *ac,
38 enum AVSampleFormat out_fmt,
39 enum AVSampleFormat in_fmt,
41 int mm_flags = av_get_cpu_flags();
45 //FIXME add memcpy case
47 #define MULTI_CAPS_FUNC(flag, cap) \
48 if (EXTERNAL_##flag(mm_flags)) {\
49 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)\
50 ac->simd_f = ff_int16_to_int32_a_ ## cap;\
51 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)\
52 ac->simd_f = ff_int32_to_int16_a_ ## cap;\
55 MULTI_CAPS_FUNC(MMX, mmx)
56 MULTI_CAPS_FUNC(SSE2, sse2)
58 if(EXTERNAL_MMX(mm_flags)) {
60 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)
61 ac->simd_f = ff_pack_6ch_float_to_float_a_mmx;
64 if(EXTERNAL_SSE(mm_flags)) {
66 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)
67 ac->simd_f = ff_pack_6ch_float_to_float_a_sse;
69 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)
70 ac->simd_f = ff_unpack_6ch_float_to_float_a_sse;
73 if(EXTERNAL_SSE2(mm_flags)) {
74 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)
75 ac->simd_f = ff_int32_to_float_a_sse2;
76 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)
77 ac->simd_f = ff_int16_to_float_a_sse2;
78 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)
79 ac->simd_f = ff_float_to_int32_a_sse2;
80 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)
81 ac->simd_f = ff_float_to_int16_a_sse2;
84 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)
85 ac->simd_f = ff_pack_2ch_int32_to_int32_a_sse2;
86 if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S16P)
87 ac->simd_f = ff_pack_2ch_int16_to_int16_a_sse2;
88 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S16P)
89 ac->simd_f = ff_pack_2ch_int16_to_int32_a_sse2;
90 if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S32P)
91 ac->simd_f = ff_pack_2ch_int32_to_int16_a_sse2;
93 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)
94 ac->simd_f = ff_unpack_2ch_int32_to_int32_a_sse2;
95 if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S16)
96 ac->simd_f = ff_unpack_2ch_int16_to_int16_a_sse2;
97 if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16)
98 ac->simd_f = ff_unpack_2ch_int16_to_int32_a_sse2;
99 if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S32)
100 ac->simd_f = ff_unpack_2ch_int32_to_int16_a_sse2;
102 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
103 ac->simd_f = ff_pack_2ch_int32_to_float_a_sse2;
104 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
105 ac->simd_f = ff_pack_2ch_float_to_int32_a_sse2;
106 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S16P)
107 ac->simd_f = ff_pack_2ch_int16_to_float_a_sse2;
108 if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_FLTP)
109 ac->simd_f = ff_pack_2ch_float_to_int16_a_sse2;
110 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32)
111 ac->simd_f = ff_unpack_2ch_int32_to_float_a_sse2;
112 if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLT)
113 ac->simd_f = ff_unpack_2ch_float_to_int32_a_sse2;
114 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16)
115 ac->simd_f = ff_unpack_2ch_int16_to_float_a_sse2;
116 if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_FLT)
117 ac->simd_f = ff_unpack_2ch_float_to_int16_a_sse2;
120 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
121 ac->simd_f = ff_pack_6ch_int32_to_float_a_sse2;
122 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
123 ac->simd_f = ff_pack_6ch_float_to_int32_a_sse2;
125 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32)
126 ac->simd_f = ff_unpack_6ch_int32_to_float_a_sse2;
127 if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLT)
128 ac->simd_f = ff_unpack_6ch_float_to_int32_a_sse2;
131 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)
132 ac->simd_f = ff_pack_8ch_float_to_float_a_sse2;
133 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
134 ac->simd_f = ff_pack_8ch_int32_to_float_a_sse2;
135 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
136 ac->simd_f = ff_pack_8ch_float_to_int32_a_sse2;
139 if(EXTERNAL_SSSE3(mm_flags)) {
141 if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S16)
142 ac->simd_f = ff_unpack_2ch_int16_to_int16_a_ssse3;
143 if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16)
144 ac->simd_f = ff_unpack_2ch_int16_to_int32_a_ssse3;
145 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16)
146 ac->simd_f = ff_unpack_2ch_int16_to_float_a_ssse3;
149 if(EXTERNAL_AVX_FAST(mm_flags)) {
150 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)
151 ac->simd_f = ff_int32_to_float_a_avx;
153 if(EXTERNAL_AVX(mm_flags)) {
155 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)
156 ac->simd_f = ff_pack_6ch_float_to_float_a_avx;
157 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
158 ac->simd_f = ff_pack_6ch_int32_to_float_a_avx;
159 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
160 ac->simd_f = ff_pack_6ch_float_to_int32_a_avx;
162 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)
163 ac->simd_f = ff_unpack_6ch_float_to_float_a_avx;
164 if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32)
165 ac->simd_f = ff_unpack_6ch_int32_to_float_a_avx;
166 if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLT)
167 ac->simd_f = ff_unpack_6ch_float_to_int32_a_avx;
170 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)
171 ac->simd_f = ff_pack_8ch_float_to_float_a_avx;
172 if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P)
173 ac->simd_f = ff_pack_8ch_int32_to_float_a_avx;
174 if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP)
175 ac->simd_f = ff_pack_8ch_float_to_int32_a_avx;
178 if(EXTERNAL_AVX2_FAST(mm_flags)) {
179 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)
180 ac->simd_f = ff_float_to_int32_a_avx2;