1 /*
2 * Copyright © 2023, VideoLAN and dav1d authors
3 * Copyright © 2023, Loongson Technology Corporation Limited
4 * All rights reserved.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions are met:
8 *
9 * 1. Redistributions of source code must retain the above copyright notice, this
10 * list of conditions and the following disclaimer.
11 *
12 * 2. Redistributions in binary form must reproduce the above copyright notice,
13 * this list of conditions and the following disclaimer in the documentation
14 * and/or other materials provided with the distribution.
15 *
16 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
18 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
19 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR
20 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
21 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
22 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND
23 * ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
24 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
25 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
26 */
27
28 #ifndef DAV1D_SRC_LOONGARCH_MC_H
29 #define DAV1D_SRC_LOONGARCH_MC_H
30
31 #include "config.h"
32 #include "src/mc.h"
33 #include "src/cpu.h"
34
35 #define init_mc_fn(type, name, suffix) \
36 c->mc[type] = BF(dav1d_put_##name, suffix)
37 #define init_mct_fn(type, name, suffix) \
38 c->mct[type] = BF(dav1d_prep_##name, suffix)
39
40 decl_avg_fn(BF(dav1d_avg, lsx));
41 decl_w_avg_fn(BF(dav1d_w_avg, lsx));
42 decl_mask_fn(BF(dav1d_mask, lsx));
43 decl_warp8x8_fn(BF(dav1d_warp_affine_8x8, lsx));
44 decl_warp8x8t_fn(BF(dav1d_warp_affine_8x8t, lsx));
45 decl_w_mask_fn(BF(dav1d_w_mask_420, lsx));
46 decl_blend_fn(BF(dav1d_blend, lsx));
47 decl_blend_dir_fn(BF(dav1d_blend_v, lsx));
48 decl_blend_dir_fn(BF(dav1d_blend_h, lsx));
49 decl_emu_edge_fn(BF(dav1d_emu_edge, lsx));
50
51 decl_mc_fn(BF(dav1d_put_8tap_regular, lsx));
52 decl_mc_fn(BF(dav1d_put_8tap_regular_smooth, lsx));
53 decl_mc_fn(BF(dav1d_put_8tap_regular_sharp, lsx));
54 decl_mc_fn(BF(dav1d_put_8tap_smooth, lsx));
55 decl_mc_fn(BF(dav1d_put_8tap_smooth_regular, lsx));
56 decl_mc_fn(BF(dav1d_put_8tap_smooth_sharp, lsx));
57 decl_mc_fn(BF(dav1d_put_8tap_sharp, lsx));
58 decl_mc_fn(BF(dav1d_put_8tap_sharp_regular, lsx));
59 decl_mc_fn(BF(dav1d_put_8tap_sharp_smooth, lsx));
60
61 decl_avg_fn(BF(dav1d_avg, lasx));
62 decl_w_avg_fn(BF(dav1d_w_avg, lasx));
63 decl_mask_fn(BF(dav1d_mask, lasx));
64 decl_warp8x8_fn(BF(dav1d_warp_affine_8x8, lasx));
65 decl_warp8x8t_fn(BF(dav1d_warp_affine_8x8t, lasx));
66 decl_w_mask_fn(BF(dav1d_w_mask_420, lasx));
67 decl_blend_dir_fn(BF(dav1d_blend_h, lasx));
68
69 decl_mct_fn(BF(dav1d_prep_8tap_regular, lsx));
70 decl_mct_fn(BF(dav1d_prep_8tap_regular_smooth, lsx));
71 decl_mct_fn(BF(dav1d_prep_8tap_regular_sharp, lsx));
72 decl_mct_fn(BF(dav1d_prep_8tap_smooth, lsx));
73 decl_mct_fn(BF(dav1d_prep_8tap_smooth_regular, lsx));
74 decl_mct_fn(BF(dav1d_prep_8tap_smooth_sharp, lsx));
75 decl_mct_fn(BF(dav1d_prep_8tap_sharp, lsx));
76 decl_mct_fn(BF(dav1d_prep_8tap_sharp_regular, lsx));
77 decl_mct_fn(BF(dav1d_prep_8tap_sharp_smooth, lsx));
78
79 decl_mct_fn(BF(dav1d_prep_8tap_regular, lasx));
80 decl_mct_fn(BF(dav1d_prep_8tap_regular_smooth, lasx));
81 decl_mct_fn(BF(dav1d_prep_8tap_regular_sharp, lasx));
82 decl_mct_fn(BF(dav1d_prep_8tap_smooth, lasx));
83 decl_mct_fn(BF(dav1d_prep_8tap_smooth_regular, lasx));
84 decl_mct_fn(BF(dav1d_prep_8tap_smooth_sharp, lasx));
85 decl_mct_fn(BF(dav1d_prep_8tap_sharp, lasx));
86 decl_mct_fn(BF(dav1d_prep_8tap_sharp_regular, lasx));
87 decl_mct_fn(BF(dav1d_prep_8tap_sharp_smooth, lasx));
88
mc_dsp_init_loongarch(Dav1dMCDSPContext * const c)89 static ALWAYS_INLINE void mc_dsp_init_loongarch(Dav1dMCDSPContext *const c) {
90 #if BITDEPTH == 8
91 const unsigned flags = dav1d_get_cpu_flags();
92
93 if (!(flags & DAV1D_LOONGARCH_CPU_FLAG_LSX)) return;
94
95 c->avg = BF(dav1d_avg, lsx);
96 c->w_avg = BF(dav1d_w_avg, lsx);
97 c->mask = BF(dav1d_mask, lsx);
98 c->warp8x8 = BF(dav1d_warp_affine_8x8, lsx);
99 c->warp8x8t = BF(dav1d_warp_affine_8x8t, lsx);
100 c->w_mask[2] = BF(dav1d_w_mask_420, lsx);
101 c->blend = BF(dav1d_blend, lsx);
102 c->blend_v = BF(dav1d_blend_v, lsx);
103 c->blend_h = BF(dav1d_blend_h, lsx);
104 c->emu_edge = BF(dav1d_emu_edge, lsx);
105
106 init_mc_fn(FILTER_2D_8TAP_REGULAR, 8tap_regular, lsx);
107 init_mc_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_regular_smooth, lsx);
108 init_mc_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_regular_sharp, lsx);
109 init_mc_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_smooth_regular, lsx);
110 init_mc_fn(FILTER_2D_8TAP_SMOOTH, 8tap_smooth, lsx);
111 init_mc_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_smooth_sharp, lsx);
112 init_mc_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_sharp_regular, lsx);
113 init_mc_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_sharp_smooth, lsx);
114 init_mc_fn(FILTER_2D_8TAP_SHARP, 8tap_sharp, lsx);
115
116 init_mct_fn(FILTER_2D_8TAP_REGULAR, 8tap_regular, lsx);
117 init_mct_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_regular_smooth, lsx);
118 init_mct_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_regular_sharp, lsx);
119 init_mct_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_smooth_regular, lsx);
120 init_mct_fn(FILTER_2D_8TAP_SMOOTH, 8tap_smooth, lsx);
121 init_mct_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_smooth_sharp, lsx);
122 init_mct_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_sharp_regular, lsx);
123 init_mct_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_sharp_smooth, lsx);
124 init_mct_fn(FILTER_2D_8TAP_SHARP, 8tap_sharp, lsx);
125
126 if (!(flags & DAV1D_LOONGARCH_CPU_FLAG_LASX)) return;
127
128 c->avg = BF(dav1d_avg, lasx);
129 c->w_avg = BF(dav1d_w_avg, lasx);
130 c->mask = BF(dav1d_mask, lasx);
131 c->warp8x8 = BF(dav1d_warp_affine_8x8, lasx);
132 c->warp8x8t = BF(dav1d_warp_affine_8x8t, lasx);
133 c->w_mask[2] = BF(dav1d_w_mask_420, lasx);
134 c->blend_h = BF(dav1d_blend_h, lasx);
135
136 init_mct_fn(FILTER_2D_8TAP_REGULAR, 8tap_regular, lasx);
137 init_mct_fn(FILTER_2D_8TAP_REGULAR_SMOOTH, 8tap_regular_smooth, lasx);
138 init_mct_fn(FILTER_2D_8TAP_REGULAR_SHARP, 8tap_regular_sharp, lasx);
139 init_mct_fn(FILTER_2D_8TAP_SMOOTH_REGULAR, 8tap_smooth_regular, lasx);
140 init_mct_fn(FILTER_2D_8TAP_SMOOTH, 8tap_smooth, lasx);
141 init_mct_fn(FILTER_2D_8TAP_SMOOTH_SHARP, 8tap_smooth_sharp, lasx);
142 init_mct_fn(FILTER_2D_8TAP_SHARP_REGULAR, 8tap_sharp_regular, lasx);
143 init_mct_fn(FILTER_2D_8TAP_SHARP_SMOOTH, 8tap_sharp_smooth, lasx);
144 init_mct_fn(FILTER_2D_8TAP_SHARP, 8tap_sharp, lasx);
145 #endif
146 }
147
148 #endif /* DAV1D_SRC_LOONGARCH_MC_H */
149