FFmpeg
libavcodec
x86
vp9dsp_init.h
Go to the documentation of this file.
1
/*
2
* VP9 SIMD optimizations
3
*
4
* Copyright (c) 2013 Ronald S. Bultje <rsbultje gmail com>
5
*
6
* This file is part of FFmpeg.
7
*
8
* FFmpeg is free software; you can redistribute it and/or
9
* modify it under the terms of the GNU Lesser General Public
10
* License as published by the Free Software Foundation; either
11
* version 2.1 of the License, or (at your option) any later version.
12
*
13
* FFmpeg is distributed in the hope that it will be useful,
14
* but WITHOUT ANY WARRANTY; without even the implied warranty of
15
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
16
* Lesser General Public License for more details.
17
*
18
* You should have received a copy of the GNU Lesser General Public
19
* License along with FFmpeg; if not, write to the Free Software
20
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
21
*/
22
23
#ifndef AVCODEC_X86_VP9DSP_INIT_H
24
#define AVCODEC_X86_VP9DSP_INIT_H
25
26
#include "
libavcodec/vp9dsp.h
"
27
28
// hack to force-expand BPC
29
#define cat(a, bpp, b) a##bpp##b
30
31
#define decl_fpel_func(avg, sz, bpp, opt) \
32
void ff_vp9_##avg##sz##bpp##_##opt(uint8_t *dst, ptrdiff_t dst_stride, \
33
const uint8_t *src, ptrdiff_t src_stride, \
34
int h, int mx, int my)
35
36
#define decl_mc_func(avg, sz, dir, opt, type, f_sz, bpp) \
37
void ff_vp9_##avg##_8tap_1d_##dir##_##sz##_##bpp##_##opt(uint8_t *dst, ptrdiff_t dst_stride, \
38
const uint8_t *src, ptrdiff_t src_stride, \
39
int h, const type (*filter)[f_sz])
40
41
#define decl_mc_funcs(sz, opt, type, fsz, bpp) \
42
decl_mc_func(put, sz, h, opt, type, fsz, bpp); \
43
decl_mc_func(avg, sz, h, opt, type, fsz, bpp); \
44
decl_mc_func(put, sz, v, opt, type, fsz, bpp); \
45
decl_mc_func(avg, sz, v, opt, type, fsz, bpp)
46
47
#define decl_ipred_fn(type, sz, bpp, opt) \
48
void ff_vp9_ipred_##type##_##sz##x##sz##_##bpp##_##opt(uint8_t *dst, \
49
ptrdiff_t stride, \
50
const uint8_t *l, \
51
const uint8_t *a)
52
53
#define decl_ipred_fns(type, bpp, opt4, opt8_16_32) \
54
decl_ipred_fn(type, 4, bpp, opt4); \
55
decl_ipred_fn(type, 8, bpp, opt8_16_32); \
56
decl_ipred_fn(type, 16, bpp, opt8_16_32); \
57
decl_ipred_fn(type, 32, bpp, opt8_16_32)
58
59
#define decl_itxfm_func(typea, typeb, size, bpp, opt) \
60
void cat(ff_vp9_##typea##_##typeb##_##size##x##size##_add_, bpp, _##opt)(uint8_t *dst, \
61
ptrdiff_t stride, \
62
int16_t *block, \
63
int eob)
64
65
#define decl_itxfm_funcs(size, bpp, opt) \
66
decl_itxfm_func(idct, idct, size, bpp, opt); \
67
decl_itxfm_func(iadst, idct, size, bpp, opt); \
68
decl_itxfm_func(idct, iadst, size, bpp, opt); \
69
decl_itxfm_func(iadst, iadst, size, bpp, opt)
70
71
#define mc_rep_func(avg, sz, hsz, hszb, dir, opt, type, f_sz, bpp) \
72
static av_always_inline void \
73
ff_vp9_##avg##_8tap_1d_##dir##_##sz##_##bpp##_##opt(uint8_t *dst, ptrdiff_t dst_stride, \
74
const uint8_t *src, ptrdiff_t src_stride, \
75
int h, const type (*filter)[f_sz]) \
76
{ \
77
ff_vp9_##avg##_8tap_1d_##dir##_##hsz##_##bpp##_##opt(dst, dst_stride, src, \
78
src_stride, h, filter); \
79
ff_vp9_##avg##_8tap_1d_##dir##_##hsz##_##bpp##_##opt(dst + hszb, dst_stride, src + hszb, \
80
src_stride, h, filter); \
81
}
82
83
#define mc_rep_funcs(sz, hsz, hszb, opt, type, fsz, bpp) \
84
mc_rep_func(put, sz, hsz, hszb, h, opt, type, fsz, bpp) \
85
mc_rep_func(avg, sz, hsz, hszb, h, opt, type, fsz, bpp) \
86
mc_rep_func(put, sz, hsz, hszb, v, opt, type, fsz, bpp) \
87
mc_rep_func(avg, sz, hsz, hszb, v, opt, type, fsz, bpp)
88
89
#define filter_8tap_1d_fn(op, sz, f, f_opt, fname, dir, dvar, bpp, opt) \
90
static void op##_8tap_##fname##_##sz##dir##_##bpp##_##opt(uint8_t *dst, ptrdiff_t dst_stride, \
91
const uint8_t *src, ptrdiff_t src_stride, \
92
int h, int mx, int my) \
93
{ \
94
ff_vp9_##op##_8tap_1d_##dir##_##sz##_##bpp##_##opt(dst, dst_stride, src, src_stride, \
95
h, ff_filters_##f_opt[f][dvar - 1]); \
96
}
97
98
#define filters_8tap_1d_fn(op, sz, dir, dvar, bpp, opt, f_opt) \
99
filter_8tap_1d_fn(op, sz, FILTER_8TAP_REGULAR, f_opt, regular, dir, dvar, bpp, opt) \
100
filter_8tap_1d_fn(op, sz, FILTER_8TAP_SHARP, f_opt, sharp, dir, dvar, bpp, opt) \
101
filter_8tap_1d_fn(op, sz, FILTER_8TAP_SMOOTH, f_opt, smooth, dir, dvar, bpp, opt)
102
103
#define filters_8tap_1d_fn2(op, sz, bpp, opt, f_opt) \
104
filters_8tap_1d_fn(op, sz, h, mx, bpp, opt, f_opt) \
105
filters_8tap_1d_fn(op, sz, v, my, bpp, opt, f_opt)
106
107
#define filters_8tap_1d_fn3(op, bpp, opt4, opt8, f_opt) \
108
filters_8tap_1d_fn2(op, 64, bpp, opt8, f_opt) \
109
filters_8tap_1d_fn2(op, 32, bpp, opt8, f_opt) \
110
filters_8tap_1d_fn2(op, 16, bpp, opt8, f_opt) \
111
filters_8tap_1d_fn2(op, 8, bpp, opt8, f_opt) \
112
filters_8tap_1d_fn2(op, 4, bpp, opt4, f_opt)
113
114
#define filter_8tap_2d_fn(op, sz, f, f_opt, fname, align, bpp, bytes, opt) \
115
static void op##_8tap_##fname##_##sz##hv_##bpp##_##opt(uint8_t *dst, ptrdiff_t dst_stride, \
116
const uint8_t *src, ptrdiff_t src_stride, \
117
int h, int mx, int my) \
118
{ \
119
LOCAL_ALIGNED_##align(uint8_t, temp, [71 * 64 * bytes]); \
120
ff_vp9_put_8tap_1d_h_##sz##_##bpp##_##opt(temp, 64 * bytes, src - 3 * src_stride, \
121
src_stride, h + 7, \
122
ff_filters_##f_opt[f][mx - 1]); \
123
ff_vp9_##op##_8tap_1d_v_##sz##_##bpp##_##opt(dst, dst_stride, temp + 3 * bytes * 64, \
124
64 * bytes, h, \
125
ff_filters_##f_opt[f][my - 1]); \
126
}
127
128
#define filters_8tap_2d_fn(op, sz, align, bpp, bytes, opt, f_opt) \
129
filter_8tap_2d_fn(op, sz, FILTER_8TAP_REGULAR, f_opt, regular, align, bpp, bytes, opt) \
130
filter_8tap_2d_fn(op, sz, FILTER_8TAP_SHARP, f_opt, sharp, align, bpp, bytes, opt) \
131
filter_8tap_2d_fn(op, sz, FILTER_8TAP_SMOOTH, f_opt, smooth, align, bpp, bytes, opt)
132
133
#define filters_8tap_2d_fn2(op, align, bpp, bytes, opt4, opt8, f_opt) \
134
filters_8tap_2d_fn(op, 64, align, bpp, bytes, opt8, f_opt) \
135
filters_8tap_2d_fn(op, 32, align, bpp, bytes, opt8, f_opt) \
136
filters_8tap_2d_fn(op, 16, align, bpp, bytes, opt8, f_opt) \
137
filters_8tap_2d_fn(op, 8, align, bpp, bytes, opt8, f_opt) \
138
filters_8tap_2d_fn(op, 4, align, bpp, bytes, opt4, f_opt)
139
140
#define init_fpel_func(idx1, idx2, sz, type, bpp, opt) \
141
dsp->mc[idx1][FILTER_8TAP_SMOOTH ][idx2][0][0] = \
142
dsp->mc[idx1][FILTER_8TAP_REGULAR][idx2][0][0] = \
143
dsp->mc[idx1][FILTER_8TAP_SHARP ][idx2][0][0] = \
144
dsp->mc[idx1][FILTER_BILINEAR ][idx2][0][0] = ff_vp9_##type##sz##bpp##_##opt
145
146
#define init_subpel1(idx1, idx2, idxh, idxv, sz, dir, type, bpp, opt) \
147
dsp->mc[idx1][FILTER_8TAP_SMOOTH ][idx2][idxh][idxv] = \
148
type##_8tap_smooth_##sz##dir##_##bpp##_##opt; \
149
dsp->mc[idx1][FILTER_8TAP_REGULAR][idx2][idxh][idxv] = \
150
type##_8tap_regular_##sz##dir##_##bpp##_##opt; \
151
dsp->mc[idx1][FILTER_8TAP_SHARP ][idx2][idxh][idxv] = \
152
type##_8tap_sharp_##sz##dir##_##bpp##_##opt
153
154
#define init_subpel2(idx1, idx2, sz, type, bpp, opt) \
155
init_subpel1(idx1, idx2, 1, 1, sz, hv, type, bpp, opt); \
156
init_subpel1(idx1, idx2, 0, 1, sz, v, type, bpp, opt); \
157
init_subpel1(idx1, idx2, 1, 0, sz, h, type, bpp, opt)
158
159
#define init_subpel3_32_64(idx, type, bpp, opt) \
160
init_subpel2(0, idx, 64, type, bpp, opt); \
161
init_subpel2(1, idx, 32, type, bpp, opt)
162
163
#define init_subpel3_8to64(idx, type, bpp, opt) \
164
init_subpel3_32_64(idx, type, bpp, opt); \
165
init_subpel2(2, idx, 16, type, bpp, opt); \
166
init_subpel2(3, idx, 8, type, bpp, opt)
167
168
#define init_subpel3(idx, type, bpp, opt) \
169
init_subpel3_8to64(idx, type, bpp, opt); \
170
init_subpel2(4, idx, 4, type, bpp, opt)
171
172
#define init_ipred_func(type, enum, sz, bpp, opt) \
173
dsp->intra_pred[TX_##sz##X##sz][enum##_PRED] = \
174
cat(ff_vp9_ipred_##type##_##sz##x##sz##_, bpp, _##opt)
175
176
#define init_8_16_32_ipred_funcs(type, enum, bpp, opt) \
177
init_ipred_func(type, enum, 8, bpp, opt); \
178
init_ipred_func(type, enum, 16, bpp, opt); \
179
init_ipred_func(type, enum, 32, bpp, opt)
180
181
#define init_ipred_funcs(type, enum, bpp, opt) \
182
init_ipred_func(type, enum, 4, bpp, opt); \
183
init_8_16_32_ipred_funcs(type, enum, bpp, opt)
184
185
void
ff_vp9dsp_init_10bpp_x86
(
VP9DSPContext
*dsp,
int
bitexact);
186
void
ff_vp9dsp_init_12bpp_x86
(
VP9DSPContext
*dsp,
int
bitexact);
187
void
ff_vp9dsp_init_16bpp_x86
(
VP9DSPContext
*dsp);
188
189
#endif
/* AVCODEC_X86_VP9DSP_INIT_H */
VP9DSPContext
Definition:
vp9dsp.h:39
ff_vp9dsp_init_10bpp_x86
void ff_vp9dsp_init_10bpp_x86(VP9DSPContext *dsp, int bitexact)
ff_vp9dsp_init_12bpp_x86
void ff_vp9dsp_init_12bpp_x86(VP9DSPContext *dsp, int bitexact)
vp9dsp.h
ff_vp9dsp_init_16bpp_x86
void ff_vp9dsp_init_16bpp_x86(VP9DSPContext *dsp)
Definition:
vp9dsp_init_16bpp.c:72
Generated on Wed Aug 24 2022 21:31:06 for FFmpeg by
1.8.17