| Line | Branch | Exec | Source |
|---|---|---|---|
| 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/attributes.h" | ||
| 22 | #include "libavutil/mem.h" | ||
| 23 | #include "libavutil/x86/cpu.h" | ||
| 24 | #include "libswresample/swresample_internal.h" | ||
| 25 | |||
| 26 | #define D(type, simd) \ | ||
| 27 | mix_1_1_func_type ff_mix_1_1_a_## type ## _ ## simd;\ | ||
| 28 | mix_2_1_func_type ff_mix_2_1_a_## type ## _ ## simd; | ||
| 29 | |||
| 30 | D(float, sse) | ||
| 31 | D(float, avx) | ||
| 32 | D(int16, sse2) | ||
| 33 | |||
| 34 | 1889 | av_cold int swri_rematrix_init_x86(struct SwrContext *s){ | |
| 35 | 1889 | int mm_flags = av_get_cpu_flags(); | |
| 36 | 1889 | int nb_in = s->used_ch_layout.nb_channels; | |
| 37 | 1889 | int nb_out = s->out.ch_count; | |
| 38 | 1889 | int num = nb_in * nb_out; | |
| 39 | int i,j; | ||
| 40 | |||
| 41 | 1889 | s->mix_1_1_simd = NULL; | |
| 42 | 1889 | s->mix_2_1_simd = NULL; | |
| 43 | |||
| 44 |
2/2✓ Branch 0 taken 280 times.
✓ Branch 1 taken 1609 times.
|
1889 | if (s->midbuf.fmt == AV_SAMPLE_FMT_S16P){ |
| 45 |
2/2✓ Branch 0 taken 260 times.
✓ Branch 1 taken 20 times.
|
280 | if(EXTERNAL_SSE2(mm_flags)) { |
| 46 | 260 | s->mix_1_1_simd = ff_mix_1_1_a_int16_sse2; | |
| 47 | 260 | s->mix_2_1_simd = ff_mix_2_1_a_int16_sse2; | |
| 48 | } | ||
| 49 | 280 | s->native_simd_matrix = av_calloc(num, 2 * sizeof(int16_t)); | |
| 50 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 280 times.
|
280 | if (!s->native_simd_matrix) |
| 51 | ✗ | return AVERROR(ENOMEM); | |
| 52 | |||
| 53 |
2/2✓ Branch 0 taken 1271 times.
✓ Branch 1 taken 280 times.
|
1551 | for(i=0; i<nb_out; i++){ |
| 54 | 1271 | int sh = 0; | |
| 55 |
2/2✓ Branch 0 taken 5704 times.
✓ Branch 1 taken 1271 times.
|
6975 | for(j=0; j<nb_in; j++) |
| 56 | 5704 | sh = FFMAX(sh, FFABS(((int*)s->native_matrix)[i * nb_in + j])); | |
| 57 | 1271 | sh = FFMAX(av_log2(sh) - 14, 0); | |
| 58 |
2/2✓ Branch 0 taken 5704 times.
✓ Branch 1 taken 1271 times.
|
6975 | for(j=0; j<nb_in; j++) { |
| 59 | 5704 | ((int16_t*)s->native_simd_matrix)[2*(i * nb_in + j)+1] = 15 - sh; | |
| 60 | 5704 | ((int16_t*)s->native_simd_matrix)[2*(i * nb_in + j)] = | |
| 61 | 5704 | ((((int*)s->native_matrix)[i * nb_in + j]) + (1<<sh>>1)) >> sh; | |
| 62 | } | ||
| 63 | } | ||
| 64 | 280 | s->native_simd_one.i16[1] = 14; | |
| 65 | 280 | s->native_simd_one.i16[0] = 16384; | |
| 66 |
2/2✓ Branch 0 taken 1250 times.
✓ Branch 1 taken 359 times.
|
1609 | } else if(s->midbuf.fmt == AV_SAMPLE_FMT_FLTP){ |
| 67 |
2/2✓ Branch 0 taken 1242 times.
✓ Branch 1 taken 8 times.
|
1250 | if(EXTERNAL_SSE(mm_flags)) { |
| 68 | 1242 | s->mix_1_1_simd = ff_mix_1_1_a_float_sse; | |
| 69 | 1242 | s->mix_2_1_simd = ff_mix_2_1_a_float_sse; | |
| 70 | } | ||
| 71 |
3/4✓ Branch 0 taken 1242 times.
✓ Branch 1 taken 8 times.
✓ Branch 2 taken 1242 times.
✗ Branch 3 not taken.
|
1250 | if(EXTERNAL_AVX_FAST(mm_flags)) { |
| 72 | 1242 | s->mix_1_1_simd = ff_mix_1_1_a_float_avx; | |
| 73 | 1242 | s->mix_2_1_simd = ff_mix_2_1_a_float_avx; | |
| 74 | } | ||
| 75 | 1250 | s->native_simd_matrix = av_calloc(num, sizeof(float)); | |
| 76 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 1250 times.
|
1250 | if (!s->native_simd_matrix) |
| 77 | ✗ | return AVERROR(ENOMEM); | |
| 78 | 1250 | memcpy(s->native_simd_matrix, s->native_matrix, num * sizeof(float)); | |
| 79 | 1250 | s->native_simd_one.f = s->native_one.f; | |
| 80 | } | ||
| 81 | |||
| 82 | 1889 | return 0; | |
| 83 | } | ||
| 84 |