23 #ifndef AVCODEC_X86_VP9DSP_INIT_H
24 #define AVCODEC_X86_VP9DSP_INIT_H
29 #define cat(a, bpp, b) a##bpp##b
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)
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])
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)
47 #define decl_ipred_fn(type, sz, bpp, opt) \
48 void ff_vp9_ipred_##type##_##sz##x##sz##_##bpp##_##opt(uint8_t *dst, \
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)
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, \
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)
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]) \
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); \
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)
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) \
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]); \
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)
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)
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)
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) \
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, \
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, \
125 ff_filters_##f_opt[f][my - 1]); \
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)
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)
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
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
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)
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)
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)
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)
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)
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)
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)
void ff_vp9dsp_init_10bpp_x86(VP9DSPContext *dsp, int bitexact)
void ff_vp9dsp_init_16bpp_x86(VP9DSPContext *dsp)
void ff_vp9dsp_init_12bpp_x86(VP9DSPContext *dsp, int bitexact)