| 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/mem.h" |
| 22 | #include "libavutil/x86/asm.h" |
| 23 | #include "libavutil/x86/cpu.h" |
| 24 | #include "libavfilter/vf_idet.h" |
| 25 | |
| 26 | #if HAVE_YASM |
| 27 | |
| 28 | /* declares main callable idet_filter_line_{mmx,mmxext,sse2}() */ |
| 29 | #define FUNC_MAIN_DECL(KIND, SPAN) \ |
| 30 | int ff_idet_filter_line_##KIND(const uint8_t *a, const uint8_t *b, \ |
| 31 | const uint8_t *c, int w); \ |
| 32 | static int idet_filter_line_##KIND(const uint8_t *a, const uint8_t *b, \ |
| 33 | const uint8_t *c, int w) { \ |
| 34 | int sum = 0; \ |
| 35 | const int left_over = w & (SPAN - 1); \ |
| 36 | w -= left_over; \ |
| 37 | if (w > 0) \ |
| 38 | sum += ff_idet_filter_line_##KIND(a, b, c, w); \ |
| 39 | if (left_over > 0) \ |
| 40 | sum += ff_idet_filter_line_c(a + w, b + w, c + w, left_over); \ |
| 41 | return sum; \ |
| 42 | } |
| 43 | |
| 44 | |
| 45 | #define FUNC_MAIN_DECL_16bit(KIND, SPAN) \ |
| 46 | int ff_idet_filter_line_16bit_##KIND(const uint16_t *a, const uint16_t *b, \ |
| 47 | const uint16_t *c, int w); \ |
| 48 | static int idet_filter_line_16bit_##KIND(const uint16_t *a, const uint16_t *b, \ |
| 49 | const uint16_t *c, int w) { \ |
| 50 | int sum = 0; \ |
| 51 | const int left_over = w & (SPAN - 1); \ |
| 52 | w -= left_over; \ |
| 53 | if (w > 0) \ |
| 54 | sum += ff_idet_filter_line_16bit_##KIND(a, b, c, w); \ |
| 55 | if (left_over > 0) \ |
| 56 | sum += ff_idet_filter_line_c_16bit(a + w, b + w, c + w, left_over); \ |
| 57 | return sum; \ |
| 58 | } |
| 59 | |
| 60 | FUNC_MAIN_DECL(sse2, 16) |
| 61 | FUNC_MAIN_DECL_16bit(sse2, 8) |
| 62 | #if ARCH_X86_32 |
| 63 | FUNC_MAIN_DECL(mmx, 8) |
| 64 | FUNC_MAIN_DECL(mmxext, 8) |
| 65 | FUNC_MAIN_DECL_16bit(mmx, 4) |
| 66 | #endif |
| 67 | |
| 68 | #endif |
| 69 | av_cold void ff_idet_init_x86(IDETContext *idet, int for_16b) |
| 70 | { |
| 71 | #if HAVE_YASM |
| 72 | const int cpu_flags = av_get_cpu_flags(); |
| 73 | |
| 74 | #if ARCH_X86_32 |
| 75 | if (EXTERNAL_MMX(cpu_flags)) { |
| 76 | idet->filter_line = for_16b ? (ff_idet_filter_func)idet_filter_line_16bit_mmx : idet_filter_line_mmx; |
| 77 | } |
| 78 | if (EXTERNAL_MMXEXT(cpu_flags)) { |
| 79 | idet->filter_line = for_16b ? (ff_idet_filter_func)idet_filter_line_16bit_mmx : idet_filter_line_mmxext; |
| 80 | } |
| 81 | #endif // ARCH_x86_32 |
| 82 | |
| 83 | if (EXTERNAL_SSE2(cpu_flags)) { |
| 84 | idet->filter_line = for_16b ? (ff_idet_filter_func)idet_filter_line_16bit_sse2 : idet_filter_line_sse2; |
| 85 | } |
| 86 | #endif // HAVE_YASM |
| 87 | } |