| Line | Branch | Exec | Source |
|---|---|---|---|
| 1 | /* | ||
| 2 | * Copyright (c) 2002 Michael Niedermayer <michaelni@gmx.at> | ||
| 3 | * Copyright (c) 2013 Paul B Mahol | ||
| 4 | * | ||
| 5 | * This file is part of FFmpeg. | ||
| 6 | * | ||
| 7 | * FFmpeg is free software; you can redistribute it and/or | ||
| 8 | * modify it under the terms of the GNU Lesser General Public | ||
| 9 | * License as published by the Free Software Foundation; either | ||
| 10 | * version 2.1 of the License, or (at your option) any later version. | ||
| 11 | * | ||
| 12 | * FFmpeg is distributed in the hope that it will be useful, | ||
| 13 | * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
| 14 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU | ||
| 15 | * Lesser General Public License for more details. | ||
| 16 | * | ||
| 17 | * You should have received a copy of the GNU Lesser General Public | ||
| 18 | * License along with FFmpeg; if not, write to the Free Software | ||
| 19 | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA | ||
| 20 | */ | ||
| 21 | |||
| 22 | #include "libavutil/attributes.h" | ||
| 23 | #include "libavutil/x86/cpu.h" | ||
| 24 | #include "libavutil/x86/asm.h" | ||
| 25 | #include "libavfilter/vf_noise.h" | ||
| 26 | |||
| 27 | #if HAVE_INLINE_ASM | ||
| 28 | #if HAVE_6REGS | ||
| 29 | ✗ | static void line_noise_avg_mmx(uint8_t *dst, const uint8_t *src, | |
| 30 | int len, const int8_t * const *shift) | ||
| 31 | { | ||
| 32 | ✗ | x86_reg mmx_len = len & (~7); | |
| 33 | |||
| 34 | ✗ | __asm__ volatile( | |
| 35 | "mov %5, %%"FF_REG_a" \n\t" | ||
| 36 | ".p2align 4 \n\t" | ||
| 37 | "1: \n\t" | ||
| 38 | "movq (%1, %%"FF_REG_a"), %%mm1 \n\t" | ||
| 39 | "movq (%0, %%"FF_REG_a"), %%mm0 \n\t" | ||
| 40 | "paddb (%2, %%"FF_REG_a"), %%mm1\n\t" | ||
| 41 | "paddb (%3, %%"FF_REG_a"), %%mm1\n\t" | ||
| 42 | "movq %%mm0, %%mm2 \n\t" | ||
| 43 | "movq %%mm1, %%mm3 \n\t" | ||
| 44 | "punpcklbw %%mm0, %%mm0 \n\t" | ||
| 45 | "punpckhbw %%mm2, %%mm2 \n\t" | ||
| 46 | "punpcklbw %%mm1, %%mm1 \n\t" | ||
| 47 | "punpckhbw %%mm3, %%mm3 \n\t" | ||
| 48 | "pmulhw %%mm0, %%mm1 \n\t" | ||
| 49 | "pmulhw %%mm2, %%mm3 \n\t" | ||
| 50 | "paddw %%mm1, %%mm1 \n\t" | ||
| 51 | "paddw %%mm3, %%mm3 \n\t" | ||
| 52 | "paddw %%mm0, %%mm1 \n\t" | ||
| 53 | "paddw %%mm2, %%mm3 \n\t" | ||
| 54 | "psrlw $8, %%mm1 \n\t" | ||
| 55 | "psrlw $8, %%mm3 \n\t" | ||
| 56 | "packuswb %%mm3, %%mm1 \n\t" | ||
| 57 | "movq %%mm1, (%4, %%"FF_REG_a") \n\t" | ||
| 58 | "add $8, %%"FF_REG_a" \n\t" | ||
| 59 | " js 1b \n\t" | ||
| 60 | ✗ | :: "r" (src+mmx_len), "r" (shift[0]+mmx_len), "r" (shift[1]+mmx_len), "r" (shift[2]+mmx_len), | |
| 61 | ✗ | "r" (dst+mmx_len), "g" (-mmx_len) | |
| 62 | : "%"FF_REG_a | ||
| 63 | ); | ||
| 64 | |||
| 65 | ✗ | if (mmx_len != len){ | |
| 66 | ✗ | const int8_t *shift2[3] = { shift[0]+mmx_len, shift[1]+mmx_len, shift[2]+mmx_len }; | |
| 67 | ✗ | ff_line_noise_avg_c(dst+mmx_len, src+mmx_len, len-mmx_len, shift2); | |
| 68 | } | ||
| 69 | ✗ | } | |
| 70 | #endif /* HAVE_6REGS */ | ||
| 71 | |||
| 72 | ✗ | static void line_noise_mmxext(uint8_t *dst, const uint8_t *src, | |
| 73 | const int8_t *noise, int len, int shift) | ||
| 74 | { | ||
| 75 | ✗ | x86_reg mmx_len = len & (~7); | |
| 76 | ✗ | noise += shift; | |
| 77 | |||
| 78 | ✗ | __asm__ volatile( | |
| 79 | "mov %3, %%"FF_REG_a" \n\t" | ||
| 80 | "pcmpeqb %%mm7, %%mm7 \n\t" | ||
| 81 | "psllw $15, %%mm7 \n\t" | ||
| 82 | "packsswb %%mm7, %%mm7 \n\t" | ||
| 83 | ".p2align 4 \n\t" | ||
| 84 | "1: \n\t" | ||
| 85 | "movq (%0, %%"FF_REG_a"), %%mm0 \n\t" | ||
| 86 | "movq (%1, %%"FF_REG_a"), %%mm1 \n\t" | ||
| 87 | "pxor %%mm7, %%mm0 \n\t" | ||
| 88 | "paddsb %%mm1, %%mm0 \n\t" | ||
| 89 | "pxor %%mm7, %%mm0 \n\t" | ||
| 90 | "movntq %%mm0, (%2, %%"FF_REG_a") \n\t" | ||
| 91 | "add $8, %%"FF_REG_a" \n\t" | ||
| 92 | " js 1b \n\t" | ||
| 93 | ✗ | :: "r" (src+mmx_len), "r" (noise+mmx_len), "r" (dst+mmx_len), "g" (-mmx_len) | |
| 94 | : "%"FF_REG_a | ||
| 95 | ); | ||
| 96 | ✗ | if (mmx_len != len) | |
| 97 | ✗ | ff_line_noise_c(dst+mmx_len, src+mmx_len, noise+mmx_len, len-mmx_len, 0); | |
| 98 | ✗ | } | |
| 99 | #endif /* HAVE_INLINE_ASM */ | ||
| 100 | |||
| 101 | ✗ | av_cold void ff_noise_init_x86(NoiseContext *n) | |
| 102 | { | ||
| 103 | #if HAVE_INLINE_ASM | ||
| 104 | ✗ | int cpu_flags = av_get_cpu_flags(); | |
| 105 | |||
| 106 | ✗ | if (INLINE_MMX(cpu_flags)) { | |
| 107 | #if HAVE_6REGS | ||
| 108 | ✗ | n->line_noise_avg = line_noise_avg_mmx; | |
| 109 | #endif | ||
| 110 | } | ||
| 111 | ✗ | if (INLINE_MMXEXT(cpu_flags)) { | |
| 112 | ✗ | n->line_noise = line_noise_mmxext; | |
| 113 | } | ||
| 114 | #endif | ||
| 115 | ✗ | } | |
| 116 |