1 /*
2  * Copyright (c) 2015 Shivraj Patil (Shivraj.Patil@imgtec.com)
3  *
4  * This file is part of FFmpeg.
5  *
6  * FFmpeg is free software; you can redistribute it and/or
7  * modify it under the terms of the GNU Lesser General Public
8  * License as published by the Free Software Foundation; either
9  * version 2.1 of the License, or (at your option) any later version.
10  *
11  * FFmpeg is distributed in the hope that it will be useful,
12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
14  * Lesser General Public License for more details.
15  *
16  * You should have received a copy of the GNU Lesser General Public
17  * License along with FFmpeg; if not, write to the Free Software
18  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19  */
20 
21 #include "libavutil/mips/cpu.h"
22 #include "config.h"
23 #include "libavutil/common.h"
24 #include "libavcodec/vp9dsp.h"
25 #include "vp9dsp_mips.h"
26 
27 #if HAVE_MSA
vp9dsp_intrapred_init_msa(VP9DSPContext * dsp,int bpp)28 static av_cold void vp9dsp_intrapred_init_msa(VP9DSPContext *dsp, int bpp)
29 {
30     if (bpp == 8) {
31 #define init_intra_pred_msa(tx, sz)                             \
32     dsp->intra_pred[tx][VERT_PRED]    = ff_vert_##sz##_msa;     \
33     dsp->intra_pred[tx][HOR_PRED]     = ff_hor_##sz##_msa;      \
34     dsp->intra_pred[tx][DC_PRED]      = ff_dc_##sz##_msa;       \
35     dsp->intra_pred[tx][LEFT_DC_PRED] = ff_dc_left_##sz##_msa;  \
36     dsp->intra_pred[tx][TOP_DC_PRED]  = ff_dc_top_##sz##_msa;   \
37     dsp->intra_pred[tx][DC_128_PRED]  = ff_dc_128_##sz##_msa;   \
38     dsp->intra_pred[tx][DC_127_PRED]  = ff_dc_127_##sz##_msa;   \
39     dsp->intra_pred[tx][DC_129_PRED]  = ff_dc_129_##sz##_msa;   \
40     dsp->intra_pred[tx][TM_VP8_PRED]  = ff_tm_##sz##_msa;       \
41 
42     init_intra_pred_msa(TX_16X16, 16x16);
43     init_intra_pred_msa(TX_32X32, 32x32);
44 #undef init_intra_pred_msa
45 
46 #define init_intra_pred_msa(tx, sz)                             \
47     dsp->intra_pred[tx][DC_PRED]      = ff_dc_##sz##_msa;       \
48     dsp->intra_pred[tx][LEFT_DC_PRED] = ff_dc_left_##sz##_msa;  \
49     dsp->intra_pred[tx][TOP_DC_PRED]  = ff_dc_top_##sz##_msa;   \
50     dsp->intra_pred[tx][TM_VP8_PRED]  = ff_tm_##sz##_msa;       \
51 
52     init_intra_pred_msa(TX_4X4, 4x4);
53     init_intra_pred_msa(TX_8X8, 8x8);
54 #undef init_intra_pred_msa
55     }
56 }
57 
vp9dsp_itxfm_init_msa(VP9DSPContext * dsp,int bpp)58 static av_cold void vp9dsp_itxfm_init_msa(VP9DSPContext *dsp, int bpp)
59 {
60     if (bpp == 8) {
61 #define init_itxfm(tx, sz)                                         \
62     dsp->itxfm_add[tx][DCT_DCT]   = ff_idct_idct_##sz##_add_msa;   \
63     dsp->itxfm_add[tx][DCT_ADST]  = ff_iadst_idct_##sz##_add_msa;  \
64     dsp->itxfm_add[tx][ADST_DCT]  = ff_idct_iadst_##sz##_add_msa;  \
65     dsp->itxfm_add[tx][ADST_ADST] = ff_iadst_iadst_##sz##_add_msa  \
66 
67 #define init_idct(tx, nm)                        \
68     dsp->itxfm_add[tx][DCT_DCT]   =              \
69     dsp->itxfm_add[tx][ADST_DCT]  =              \
70     dsp->itxfm_add[tx][DCT_ADST]  =              \
71     dsp->itxfm_add[tx][ADST_ADST] = nm##_add_msa
72 
73     init_itxfm(TX_4X4, 4x4);
74     init_itxfm(TX_8X8, 8x8);
75     init_itxfm(TX_16X16, 16x16);
76     init_idct(TX_32X32, ff_idct_idct_32x32);
77 #undef init_itxfm
78 #undef init_idct
79     }
80 }
81 
vp9dsp_mc_init_msa(VP9DSPContext * dsp,int bpp)82 static av_cold void vp9dsp_mc_init_msa(VP9DSPContext *dsp, int bpp)
83 {
84     if (bpp == 8) {
85 #define init_fpel(idx1, idx2, sz, type)                                    \
86     dsp->mc[idx1][FILTER_8TAP_SMOOTH ][idx2][0][0] = ff_##type##sz##_msa;  \
87     dsp->mc[idx1][FILTER_8TAP_REGULAR][idx2][0][0] = ff_##type##sz##_msa;  \
88     dsp->mc[idx1][FILTER_8TAP_SHARP  ][idx2][0][0] = ff_##type##sz##_msa;  \
89     dsp->mc[idx1][FILTER_BILINEAR    ][idx2][0][0] = ff_##type##sz##_msa
90 
91 #define init_copy_avg(idx, sz)    \
92     init_fpel(idx, 0, sz, copy);  \
93     init_fpel(idx, 1, sz, avg)
94 
95 #define init_avg(idx, sz)  \
96     init_fpel(idx, 1, sz, avg)
97 
98     init_copy_avg(0, 64);
99     init_copy_avg(1, 32);
100     init_copy_avg(2, 16);
101     init_copy_avg(3,  8);
102     init_avg(4,  4);
103 
104 #undef init_copy_avg
105 #undef init_avg
106 #undef init_fpel
107 
108 #define init_subpel1(idx1, idx2, idxh, idxv, sz, dir, type)  \
109     dsp->mc[idx1][FILTER_BILINEAR    ][idx2][idxh][idxv] =   \
110         ff_##type##_bilin_##sz##dir##_msa;                   \
111     dsp->mc[idx1][FILTER_8TAP_SMOOTH ][idx2][idxh][idxv] =   \
112         ff_##type##_8tap_smooth_##sz##dir##_msa;             \
113     dsp->mc[idx1][FILTER_8TAP_REGULAR][idx2][idxh][idxv] =   \
114         ff_##type##_8tap_regular_##sz##dir##_msa;            \
115     dsp->mc[idx1][FILTER_8TAP_SHARP  ][idx2][idxh][idxv] =   \
116         ff_##type##_8tap_sharp_##sz##dir##_msa;
117 
118 #define init_subpel2(idx, idxh, idxv, dir, type)      \
119     init_subpel1(0, idx, idxh, idxv, 64, dir, type);  \
120     init_subpel1(1, idx, idxh, idxv, 32, dir, type);  \
121     init_subpel1(2, idx, idxh, idxv, 16, dir, type);  \
122     init_subpel1(3, idx, idxh, idxv,  8, dir, type);  \
123     init_subpel1(4, idx, idxh, idxv,  4, dir, type)
124 
125 #define init_subpel3(idx, type)         \
126     init_subpel2(idx, 1, 1, hv, type);  \
127     init_subpel2(idx, 0, 1, v, type);   \
128     init_subpel2(idx, 1, 0, h, type)
129 
130     init_subpel3(0, put);
131     init_subpel3(1, avg);
132 
133 #undef init_subpel1
134 #undef init_subpel2
135 #undef init_subpel3
136     }
137 }
138 
vp9dsp_loopfilter_init_msa(VP9DSPContext * dsp,int bpp)139 static av_cold void vp9dsp_loopfilter_init_msa(VP9DSPContext *dsp, int bpp)
140 {
141     if (bpp == 8) {
142         dsp->loop_filter_8[0][0] = ff_loop_filter_h_4_8_msa;
143         dsp->loop_filter_8[0][1] = ff_loop_filter_v_4_8_msa;
144         dsp->loop_filter_8[1][0] = ff_loop_filter_h_8_8_msa;
145         dsp->loop_filter_8[1][1] = ff_loop_filter_v_8_8_msa;
146         dsp->loop_filter_8[2][0] = ff_loop_filter_h_16_8_msa;
147         dsp->loop_filter_8[2][1] = ff_loop_filter_v_16_8_msa;
148 
149         dsp->loop_filter_16[0] = ff_loop_filter_h_16_16_msa;
150         dsp->loop_filter_16[1] = ff_loop_filter_v_16_16_msa;
151 
152         dsp->loop_filter_mix2[0][0][0] = ff_loop_filter_h_44_16_msa;
153         dsp->loop_filter_mix2[0][0][1] = ff_loop_filter_v_44_16_msa;
154         dsp->loop_filter_mix2[0][1][0] = ff_loop_filter_h_48_16_msa;
155         dsp->loop_filter_mix2[0][1][1] = ff_loop_filter_v_48_16_msa;
156         dsp->loop_filter_mix2[1][0][0] = ff_loop_filter_h_84_16_msa;
157         dsp->loop_filter_mix2[1][0][1] = ff_loop_filter_v_84_16_msa;
158         dsp->loop_filter_mix2[1][1][0] = ff_loop_filter_h_88_16_msa;
159         dsp->loop_filter_mix2[1][1][1] = ff_loop_filter_v_88_16_msa;
160     }
161 }
162 
vp9dsp_init_msa(VP9DSPContext * dsp,int bpp)163 static av_cold void vp9dsp_init_msa(VP9DSPContext *dsp, int bpp)
164 {
165     vp9dsp_intrapred_init_msa(dsp, bpp);
166     vp9dsp_itxfm_init_msa(dsp, bpp);
167     vp9dsp_mc_init_msa(dsp, bpp);
168     vp9dsp_loopfilter_init_msa(dsp, bpp);
169 }
170 #endif  // #if HAVE_MSA
171 
172 #if HAVE_MMI
vp9dsp_mc_init_mmi(VP9DSPContext * dsp)173 static av_cold void vp9dsp_mc_init_mmi(VP9DSPContext *dsp)
174 {
175 #define init_subpel1(idx1, idx2, idxh, idxv, sz, dir, type)  \
176     dsp->mc[idx1][FILTER_8TAP_SMOOTH ][idx2][idxh][idxv] =   \
177         ff_##type##_8tap_smooth_##sz##dir##_mmi;             \
178     dsp->mc[idx1][FILTER_8TAP_REGULAR][idx2][idxh][idxv] =   \
179         ff_##type##_8tap_regular_##sz##dir##_mmi;            \
180     dsp->mc[idx1][FILTER_8TAP_SHARP  ][idx2][idxh][idxv] =   \
181         ff_##type##_8tap_sharp_##sz##dir##_mmi;
182 
183 #define init_subpel2(idx, idxh, idxv, dir, type)      \
184     init_subpel1(0, idx, idxh, idxv, 64, dir, type);  \
185     init_subpel1(1, idx, idxh, idxv, 32, dir, type);  \
186     init_subpel1(2, idx, idxh, idxv, 16, dir, type);  \
187     init_subpel1(3, idx, idxh, idxv,  8, dir, type);  \
188     init_subpel1(4, idx, idxh, idxv,  4, dir, type)
189 
190 #define init_subpel3(idx, type)         \
191     init_subpel2(idx, 1, 1, hv, type);  \
192     init_subpel2(idx, 0, 1, v, type);   \
193     init_subpel2(idx, 1, 0, h, type)
194 
195     init_subpel3(0, put);
196     init_subpel3(1, avg);
197 
198 #undef init_subpel1
199 #undef init_subpel2
200 #undef init_subpel3
201 }
202 
vp9dsp_init_mmi(VP9DSPContext * dsp,int bpp)203 static av_cold void vp9dsp_init_mmi(VP9DSPContext *dsp, int bpp)
204 {
205     if (bpp == 8) {
206         vp9dsp_mc_init_mmi(dsp);
207     }
208 }
209 #endif  // #if HAVE_MMI
210 
ff_vp9dsp_init_mips(VP9DSPContext * dsp,int bpp)211 av_cold void ff_vp9dsp_init_mips(VP9DSPContext *dsp, int bpp)
212 {
213 #if HAVE_MSA || HAVE_MMI
214     int cpu_flags = av_get_cpu_flags();
215 #endif
216 
217 #if HAVE_MMI
218     if (have_mmi(cpu_flags))
219         vp9dsp_init_mmi(dsp, bpp);
220 #endif
221 
222 #if HAVE_MSA
223     if (have_msa(cpu_flags))
224         vp9dsp_init_msa(dsp, bpp);
225 #endif
226 }
227