| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | /* | ||
| 2 | * This file is part of FFmpeg. | ||
| 3 | * | ||
| 4 | * FFmpeg is free software; you can redistribute it and/or | ||
| 5 | * modify it under the terms of the GNU Lesser General Public | ||
| 6 | * License as published by the Free Software Foundation; either | ||
| 7 | * version 2.1 of the License, or (at your option) any later version. | ||
| 8 | * | ||
| 9 | * FFmpeg is distributed in the hope that it will be useful, | ||
| 10 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
| 11 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | ||
| 12 | * Lesser General Public License for more details. | ||
| 13 | * | ||
| 14 | * You should have received a copy of the GNU Lesser General Public | ||
| 15 | * License along with FFmpeg; if not, write to the Free Software | ||
| 16 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | ||
| 17 | */ | ||
| 18 | |||
| 19 | #include "libavutil/attributes.h" | ||
| 20 | #include "libavutil/cpu.h" | ||
| 21 | #include "libavutil/x86/cpu.h" | ||
| 22 | #include "libavfilter/vf_idetdsp.h" | ||
| 23 | |||
| 24 | /* declares main callable idet_filter_line_sse2() */ | ||
| 25 | #define FUNC_MAIN_DECL(KIND, SPAN) \ | ||
| 26 | int ff_idet_filter_line_##KIND(const uint8_t *a, const uint8_t *b, \ | ||
| 27 | const uint8_t *c, int w); \ | ||
| 28 | static int idet_filter_line_##KIND(const uint8_t *a, const uint8_t *b, \ | ||
| 29 | const uint8_t *c, int w) { \ | ||
| 30 | int sum = 0; \ | ||
| 31 | const int left_over = w & (SPAN - 1); \ | ||
| 32 | w -= left_over; \ | ||
| 33 | if (w > 0) \ | ||
| 34 | sum += ff_idet_filter_line_##KIND(a, b, c, w); \ | ||
| 35 | if (left_over > 0) \ | ||
| 36 | sum += ff_idet_filter_line_c(a + w, b + w, c + w, left_over); \ | ||
| 37 | return sum; \ | ||
| 38 | } | ||
| 39 | |||
| 40 | |||
| 41 | #define FUNC_MAIN_DECL_16bit(KIND, SPAN) \ | ||
| 42 | int ff_idet_filter_line_16bit_##KIND(const uint8_t *a, const uint8_t *b, \ | ||
| 43 | const uint8_t *c, int w); \ | ||
| 44 | static int idet_filter_line_16bit_##KIND(const uint8_t *a, const uint8_t *b, \ | ||
| 45 | const uint8_t *c, int w) { \ | ||
| 46 | int sum = 0; \ | ||
| 47 | const int left_over = w & (SPAN - 1); \ | ||
| 48 | const int w_main = w - left_over; \ | ||
| 49 | const int offset = w_main << 1; \ | ||
| 50 | if (w_main > 0) \ | ||
| 51 | sum += ff_idet_filter_line_16bit_##KIND(a, b, c, w_main); \ | ||
| 52 | if (left_over > 0) { \ | ||
| 53 | sum += ff_idet_filter_line_c_16bit(a + offset, b + offset, c + offset, \ | ||
| 54 | left_over); \ | ||
| 55 | } \ | ||
| 56 | return sum; \ | ||
| 57 | } | ||
| 58 | |||
| 59 |
2/4✓ Branch 0 taken 4 times.
✗ Branch 1 not taken.
✓ Branch 3 taken 4 times.
✗ Branch 4 not taken.
|
4 | FUNC_MAIN_DECL(sse2, 16) |
| 60 | ✗ | FUNC_MAIN_DECL_16bit(sse2, 8) | |
| 61 | |||
| 62 |
2/4✓ Branch 0 taken 2 times.
✗ Branch 1 not taken.
✓ Branch 3 taken 2 times.
✗ Branch 4 not taken.
|
2 | FUNC_MAIN_DECL(avx2, 32) |
| 63 | ✗ | FUNC_MAIN_DECL_16bit(avx2, 16) | |
| 64 | |||
| 65 | ✗ | FUNC_MAIN_DECL(avx512icl, 64) | |
| 66 | ✗ | FUNC_MAIN_DECL_16bit(avx512icl, 32) | |
| 67 | |||
| 68 | 31 | av_cold void ff_idet_dsp_init_x86(IDETDSPContext *dsp, int depth) | |
| 69 | { | ||
| 70 | 31 | const int cpu_flags = av_get_cpu_flags(); | |
| 71 | |||
| 72 |
2/2✓ Branch 0 taken 20 times.
✓ Branch 1 taken 11 times.
|
31 | if (EXTERNAL_SSE2(cpu_flags)) { |
| 73 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 20 times.
|
20 | dsp->filter_line = depth > 8 ? idet_filter_line_16bit_sse2 : idet_filter_line_sse2; |
| 74 | } | ||
| 75 |
2/2✓ Branch 0 taken 2 times.
✓ Branch 1 taken 29 times.
|
31 | if (EXTERNAL_AVX2(cpu_flags)) { |
| 76 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 2 times.
|
2 | dsp->filter_line = depth > 8 ? idet_filter_line_16bit_avx2 : idet_filter_line_avx2; |
| 77 | } | ||
| 78 |
1/2✗ Branch 0 not taken.
✓ Branch 1 taken 31 times.
|
31 | if (EXTERNAL_AVX512ICL(cpu_flags)) { |
| 79 | ✗ | dsp->filter_line = depth > 8 ? idet_filter_line_16bit_avx512icl : idet_filter_line_avx512icl; | |
| 80 | } | ||
| 81 | 31 | } | |
| 82 |