1 /*
2 * SIMD-optimized lossless video encoding functions
3 * Copyright (c) 2000, 2001 Fabrice Bellard
4 * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
5 *
6 * MMX optimization by Nick Kurshev <nickols_k@mail.ru>
7 *
8 * This file is part of FFmpeg.
9 *
10 * FFmpeg is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU Lesser General Public
12 * License as published by the Free Software Foundation; either
13 * version 2.1 of the License, or (at your option) any later version.
14 *
15 * FFmpeg is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 * Lesser General Public License for more details.
19 *
20 * You should have received a copy of the GNU Lesser General Public
21 * License along with FFmpeg; if not, write to the Free Software
22 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23 */
24
25 #include "libavutil/attributes.h"
26 #include "libavutil/cpu.h"
27 #include "libavutil/x86/asm.h"
28 #include "libavutil/x86/cpu.h"
29 #include "libavcodec/lossless_videoencdsp.h"
30 #include "libavcodec/mathops.h"
31
32 void ff_diff_bytes_mmx(uint8_t *dst, const uint8_t *src1, const uint8_t *src2,
33 intptr_t w);
34 void ff_diff_bytes_sse2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2,
35 intptr_t w);
36 void ff_diff_bytes_avx2(uint8_t *dst, const uint8_t *src1, const uint8_t *src2,
37 intptr_t w);
38
39 void ff_sub_left_predict_avx(uint8_t *dst, uint8_t *src,
40 ptrdiff_t stride, ptrdiff_t width, int height);
41
42 #if HAVE_INLINE_ASM
43
sub_median_pred_mmxext(uint8_t * dst,const uint8_t * src1,const uint8_t * src2,intptr_t w,int * left,int * left_top)44 static void sub_median_pred_mmxext(uint8_t *dst, const uint8_t *src1,
45 const uint8_t *src2, intptr_t w,
46 int *left, int *left_top)
47 {
48 x86_reg i = 0;
49 uint8_t l, lt;
50
51 __asm__ volatile (
52 "movq (%1, %0), %%mm0 \n\t" // LT
53 "psllq $8, %%mm0 \n\t"
54 "1: \n\t"
55 "movq (%1, %0), %%mm1 \n\t" // T
56 "movq -1(%2, %0), %%mm2 \n\t" // L
57 "movq (%2, %0), %%mm3 \n\t" // X
58 "movq %%mm2, %%mm4 \n\t" // L
59 "psubb %%mm0, %%mm2 \n\t"
60 "paddb %%mm1, %%mm2 \n\t" // L + T - LT
61 "movq %%mm4, %%mm5 \n\t" // L
62 "pmaxub %%mm1, %%mm4 \n\t" // max(T, L)
63 "pminub %%mm5, %%mm1 \n\t" // min(T, L)
64 "pminub %%mm2, %%mm4 \n\t"
65 "pmaxub %%mm1, %%mm4 \n\t"
66 "psubb %%mm4, %%mm3 \n\t" // dst - pred
67 "movq %%mm3, (%3, %0) \n\t"
68 "add $8, %0 \n\t"
69 "movq -1(%1, %0), %%mm0 \n\t" // LT
70 "cmp %4, %0 \n\t"
71 " jb 1b \n\t"
72 : "+r" (i)
73 : "r" (src1), "r" (src2), "r" (dst), "r" ((x86_reg) w));
74
75 l = *left;
76 lt = *left_top;
77
78 dst[0] = src2[0] - mid_pred(l, src1[0], (l + src1[0] - lt) & 0xFF);
79
80 *left_top = src1[w - 1];
81 *left = src2[w - 1];
82 }
83
84 #endif /* HAVE_INLINE_ASM */
85
ff_llvidencdsp_init_x86(LLVidEncDSPContext * c)86 av_cold void ff_llvidencdsp_init_x86(LLVidEncDSPContext *c)
87 {
88 av_unused int cpu_flags = av_get_cpu_flags();
89
90 if (ARCH_X86_32 && EXTERNAL_MMX(cpu_flags)) {
91 c->diff_bytes = ff_diff_bytes_mmx;
92 }
93
94 #if HAVE_INLINE_ASM
95 if (INLINE_MMXEXT(cpu_flags)) {
96 c->sub_median_pred = sub_median_pred_mmxext;
97 }
98 #endif /* HAVE_INLINE_ASM */
99
100 if (EXTERNAL_SSE2(cpu_flags)) {
101 c->diff_bytes = ff_diff_bytes_sse2;
102 }
103
104 if (EXTERNAL_AVX(cpu_flags)) {
105 c->sub_left_predict = ff_sub_left_predict_avx;
106 }
107
108 if (EXTERNAL_AVX2_FAST(cpu_flags)) {
109 c->diff_bytes = ff_diff_bytes_avx2;
110 }
111 }
112