| 1 | /* |
| 2 | * Copyright (C) 2012 Michael Niedermayer (michaelni@gmx.at) |
| 3 | * |
| 4 | * This file is part of libswresample |
| 5 | * |
| 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. |
| 10 | * |
| 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. |
| 15 | * |
| 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 |
| 19 | */ |
| 20 | |
| 21 | #include "libavutil/x86/cpu.h" |
| 22 | #include "libswresample/swresample_internal.h" |
| 23 | #include "libswresample/audioconvert.h" |
| 24 | |
| 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) |
| 29 | PROTO4(_) |
| 30 | PROTO4(_pack_2ch_) |
| 31 | PROTO4(_pack_6ch_) |
| 32 | PROTO4(_unpack_2ch_) |
| 33 | |
| 34 | av_cold void swri_audio_convert_init_x86(struct AudioConvert *ac, |
| 35 | enum AVSampleFormat out_fmt, |
| 36 | enum AVSampleFormat in_fmt, |
| 37 | int channels){ |
| 38 | int mm_flags = av_get_cpu_flags(); |
| 39 | |
| 40 | ac->simd_f= NULL; |
| 41 | |
| 42 | //FIXME add memcpy case |
| 43 | |
| 44 | #define MULTI_CAPS_FUNC(flag, cap) \ |
| 45 | if (EXTERNAL_##flag(mm_flags)) {\ |
| 46 | 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)\ |
| 47 | ac->simd_f = ff_int16_to_int32_a_ ## cap;\ |
| 48 | 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)\ |
| 49 | ac->simd_f = ff_int32_to_int16_a_ ## cap;\ |
| 50 | } |
| 51 | |
| 52 | MULTI_CAPS_FUNC(MMX, mmx) |
| 53 | MULTI_CAPS_FUNC(SSE2, sse2) |
| 54 | |
| 55 | if(EXTERNAL_MMX(mm_flags)) { |
| 56 | if(channels == 6) { |
| 57 | 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) |
| 58 | ac->simd_f = ff_pack_6ch_float_to_float_a_mmx; |
| 59 | } |
| 60 | } |
| 61 | if(EXTERNAL_SSE(mm_flags)) { |
| 62 | if(channels == 6) { |
| 63 | 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) |
| 64 | ac->simd_f = ff_pack_6ch_float_to_float_a_sse; |
| 65 | } |
| 66 | } |
| 67 | if(EXTERNAL_SSE2(mm_flags)) { |
| 68 | 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) |
| 69 | ac->simd_f = ff_int32_to_float_a_sse2; |
| 70 | 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) |
| 71 | ac->simd_f = ff_int16_to_float_a_sse2; |
| 72 | 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) |
| 73 | ac->simd_f = ff_float_to_int32_a_sse2; |
| 74 | 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) |
| 75 | ac->simd_f = ff_float_to_int16_a_sse2; |
| 76 | |
| 77 | if(channels == 2) { |
| 78 | 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) |
| 79 | ac->simd_f = ff_pack_2ch_int32_to_int32_a_sse2; |
| 80 | if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S16P) |
| 81 | ac->simd_f = ff_pack_2ch_int16_to_int16_a_sse2; |
| 82 | if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_S16P) |
| 83 | ac->simd_f = ff_pack_2ch_int16_to_int32_a_sse2; |
| 84 | if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_S32P) |
| 85 | ac->simd_f = ff_pack_2ch_int32_to_int16_a_sse2; |
| 86 | |
| 87 | 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) |
| 88 | ac->simd_f = ff_unpack_2ch_int32_to_int32_a_sse2; |
| 89 | if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S16) |
| 90 | ac->simd_f = ff_unpack_2ch_int16_to_int16_a_sse2; |
| 91 | if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16) |
| 92 | ac->simd_f = ff_unpack_2ch_int16_to_int32_a_sse2; |
| 93 | if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S32) |
| 94 | ac->simd_f = ff_unpack_2ch_int32_to_int16_a_sse2; |
| 95 | |
| 96 | if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P) |
| 97 | ac->simd_f = ff_pack_2ch_int32_to_float_a_sse2; |
| 98 | if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP) |
| 99 | ac->simd_f = ff_pack_2ch_float_to_int32_a_sse2; |
| 100 | if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S16P) |
| 101 | ac->simd_f = ff_pack_2ch_int16_to_float_a_sse2; |
| 102 | if( out_fmt == AV_SAMPLE_FMT_S16 && in_fmt == AV_SAMPLE_FMT_FLTP) |
| 103 | ac->simd_f = ff_pack_2ch_float_to_int16_a_sse2; |
| 104 | if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S32) |
| 105 | ac->simd_f = ff_unpack_2ch_int32_to_float_a_sse2; |
| 106 | if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_FLT) |
| 107 | ac->simd_f = ff_unpack_2ch_float_to_int32_a_sse2; |
| 108 | if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16) |
| 109 | ac->simd_f = ff_unpack_2ch_int16_to_float_a_sse2; |
| 110 | if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_FLT) |
| 111 | ac->simd_f = ff_unpack_2ch_float_to_int16_a_sse2; |
| 112 | } |
| 113 | if(channels == 6) { |
| 114 | if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P) |
| 115 | ac->simd_f = ff_pack_6ch_int32_to_float_a_sse2; |
| 116 | if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP) |
| 117 | ac->simd_f = ff_pack_6ch_float_to_int32_a_sse2; |
| 118 | } |
| 119 | } |
| 120 | if(EXTERNAL_SSSE3(mm_flags)) { |
| 121 | if(channels == 2) { |
| 122 | if( out_fmt == AV_SAMPLE_FMT_S16P && in_fmt == AV_SAMPLE_FMT_S16) |
| 123 | ac->simd_f = ff_unpack_2ch_int16_to_int16_a_ssse3; |
| 124 | if( out_fmt == AV_SAMPLE_FMT_S32P && in_fmt == AV_SAMPLE_FMT_S16) |
| 125 | ac->simd_f = ff_unpack_2ch_int16_to_int32_a_ssse3; |
| 126 | if( out_fmt == AV_SAMPLE_FMT_FLTP && in_fmt == AV_SAMPLE_FMT_S16) |
| 127 | ac->simd_f = ff_unpack_2ch_int16_to_float_a_ssse3; |
| 128 | } |
| 129 | } |
| 130 | if(EXTERNAL_AVX(mm_flags)) { |
| 131 | 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) |
| 132 | ac->simd_f = ff_int32_to_float_a_avx; |
| 133 | if(channels == 6) { |
| 134 | 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) |
| 135 | ac->simd_f = ff_pack_6ch_float_to_float_a_avx; |
| 136 | if( out_fmt == AV_SAMPLE_FMT_FLT && in_fmt == AV_SAMPLE_FMT_S32P) |
| 137 | ac->simd_f = ff_pack_6ch_int32_to_float_a_avx; |
| 138 | if( out_fmt == AV_SAMPLE_FMT_S32 && in_fmt == AV_SAMPLE_FMT_FLTP) |
| 139 | ac->simd_f = ff_pack_6ch_float_to_int32_a_avx; |
| 140 | } |
| 141 | } |
| 142 | if(EXTERNAL_AVX2(mm_flags)) { |
| 143 | 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) |
| 144 | ac->simd_f = ff_float_to_int32_a_avx2; |
| 145 | } |
| 146 | } |