1. ffmpeg
  2. Commit 2d8a0815116cbe3f3720f80b5709d87774bc92b8
Browse code

RV40 MC functions

Originally committed as revision 15967 to svn://svn.ffmpeg.org/ffmpeg/trunk

Kostya Shishkov authored on 2008/12/01 15:36:22
Showing 3 changed files
libavcodec/dsputil.c
History View file @ 2d8a081
... ... 
@@ -2743,6 +2743,23 @@ void ff_intrax8dsp_init(DSPContext* c, AVCodecContext *avctx);
2743 2743 
 /* H264 specific */
2744 2744 
 void ff_h264dspenc_init(DSPContext* c, AVCodecContext *avctx);
2745 2745
2746 
+#if defined(CONFIG_RV40_DECODER)
2747 
+static void put_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){
2748 
+    put_pixels16_xy2_c(dst, src, stride, 16);
2749 
+}
2750 
+static void avg_rv40_qpel16_mc33_c(uint8_t *dst, uint8_t *src, int stride){
2751 
+    avg_pixels16_xy2_c(dst, src, stride, 16);
2752 
+}
2753 
+static void put_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){
2754 
+    put_pixels8_xy2_c(dst, src, stride, 8);
2755 
+}
2756 
+static void avg_rv40_qpel8_mc33_c(uint8_t *dst, uint8_t *src, int stride){
2757 
+    avg_pixels8_xy2_c(dst, src, stride, 8);
2758 
+}
2759 
+
2760 
+void ff_rv40dsp_init(DSPContext* c, AVCodecContext *avctx);
2761 
+#endif /* CONFIG_RV40_DECODER */
2762 
+
2746 2763 
 static void wmv2_mspel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int w){
2747 2764 
     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
2748 2765 
     int i;
... ... 
@@ -4411,6 +4428,13 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
4411 4411 
 #if defined(CONFIG_H264_ENCODER)
4412 4412 
     ff_h264dspenc_init(c,avctx);
4413 4413 
 #endif
4414 
+#if defined(CONFIG_RV40_DECODER)
4415 
+    ff_rv40dsp_init(c,avctx);
4416 
+    c->put_rv40_qpel_pixels_tab[0][15] = put_rv40_qpel16_mc33_c;
4417 
+    c->avg_rv40_qpel_pixels_tab[0][15] = avg_rv40_qpel16_mc33_c;
4418 
+    c->put_rv40_qpel_pixels_tab[1][15] = put_rv40_qpel8_mc33_c;
4419 
+    c->avg_rv40_qpel_pixels_tab[1][15] = avg_rv40_qpel8_mc33_c;
4420 
+#endif
4414 4421
4415 4422 
     c->put_mspel_pixels_tab[0]= put_mspel8_mc00_c;
4416 4423 
     c->put_mspel_pixels_tab[1]= put_mspel8_mc10_c;
libavcodec/dsputil.h
History View file @ 2d8a081
... ... 
@@ -484,6 +484,12 @@ typedef struct DSPContext {
484 484 
      * @param shift number of bits to discard from product
485 485 
      */
486 486 
     int32_t (*scalarproduct_int16)(int16_t *v1, int16_t *v2/*align 16*/, int len, int shift);
487 
+
488 
+    /* rv40 functions */
489 
+    qpel_mc_func put_rv40_qpel_pixels_tab[4][16];
490 
+    qpel_mc_func avg_rv40_qpel_pixels_tab[4][16];
491 
+    h264_chroma_mc_func put_rv40_chroma_pixels_tab[3];
492 
+    h264_chroma_mc_func avg_rv40_chroma_pixels_tab[3];
487 493 
 } DSPContext;
488 494
489 495 
 void dsputil_static_init(void);
libavcodec/rv40dsp.c
History View file @ 2d8a081
490 496 
new file mode 100644
... ... 
@@ -0,0 +1,353 @@
0 
+/*
1 
+ * RV40 decoder motion compensation functions
2 
+ * Copyright (c) 2008 Konstantin Shishkov
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 
+/**
22 
+ * @file rv40dsp.c
23 
+ * RV40 decoder motion compensation functions
24 
+ */
25 
+
26 
+#include "avcodec.h"
27 
+#include "dsputil.h"
28 
+
29 
+#define RV40_LOWPASS(OPNAME, OP) \
30 
+static av_unused void OPNAME ## rv40_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,\
31 
+                                                     const int h, const int C1, const int C2, const int SHIFT){\
32 
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
33 
+    int i;\
34 
+    for(i=0; i<h; i++)\
35 
+    {\
36 
+        OP(dst[0], (src[-2] + src[ 3] - 5*(src[-1]+src[2]) + src[0]*C1 + src[1]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
37 
+        OP(dst[1], (src[-1] + src[ 4] - 5*(src[ 0]+src[3]) + src[1]*C1 + src[2]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
38 
+        OP(dst[2], (src[ 0] + src[ 5] - 5*(src[ 1]+src[4]) + src[2]*C1 + src[3]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
39 
+        OP(dst[3], (src[ 1] + src[ 6] - 5*(src[ 2]+src[5]) + src[3]*C1 + src[4]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
40 
+        OP(dst[4], (src[ 2] + src[ 7] - 5*(src[ 3]+src[6]) + src[4]*C1 + src[5]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
41 
+        OP(dst[5], (src[ 3] + src[ 8] - 5*(src[ 4]+src[7]) + src[5]*C1 + src[6]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
42 
+        OP(dst[6], (src[ 4] + src[ 9] - 5*(src[ 5]+src[8]) + src[6]*C1 + src[7]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
43 
+        OP(dst[7], (src[ 5] + src[10] - 5*(src[ 6]+src[9]) + src[7]*C1 + src[8]*C2 + (1<<(SHIFT-1))) >> SHIFT);\
44 
+        dst+=dstStride;\
45 
+        src+=srcStride;\
46 
+    }\
47 
+}\
48 
+\
49 
+static void OPNAME ## rv40_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,\
50 
+                                           const int w, const int C1, const int C2, const int SHIFT){\
51 
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
52 
+    int i;\
53 
+    for(i=0; i<w; i++)\
54 
+    {\
55 
+        const int srcB = src[-2*srcStride];\
56 
+        const int srcA = src[-1*srcStride];\
57 
+        const int src0 = src[0 *srcStride];\
58 
+        const int src1 = src[1 *srcStride];\
59 
+        const int src2 = src[2 *srcStride];\
60 
+        const int src3 = src[3 *srcStride];\
61 
+        const int src4 = src[4 *srcStride];\
62 
+        const int src5 = src[5 *srcStride];\
63 
+        const int src6 = src[6 *srcStride];\
64 
+        const int src7 = src[7 *srcStride];\
65 
+        const int src8 = src[8 *srcStride];\
66 
+        const int src9 = src[9 *srcStride];\
67 
+        const int src10= src[10*srcStride];\
68 
+        OP(dst[0*dstStride], (srcB + src3  - 5*(srcA+src2) + src0*C1 + src1*C2 + (1<<(SHIFT-1))) >> SHIFT);\
69 
+        OP(dst[1*dstStride], (srcA + src4  - 5*(src0+src3) + src1*C1 + src2*C2 + (1<<(SHIFT-1))) >> SHIFT);\
70 
+        OP(dst[2*dstStride], (src0 + src5  - 5*(src1+src4) + src2*C1 + src3*C2 + (1<<(SHIFT-1))) >> SHIFT);\
71 
+        OP(dst[3*dstStride], (src1 + src6  - 5*(src2+src5) + src3*C1 + src4*C2 + (1<<(SHIFT-1))) >> SHIFT);\
72 
+        OP(dst[4*dstStride], (src2 + src7  - 5*(src3+src6) + src4*C1 + src5*C2 + (1<<(SHIFT-1))) >> SHIFT);\
73 
+        OP(dst[5*dstStride], (src3 + src8  - 5*(src4+src7) + src5*C1 + src6*C2 + (1<<(SHIFT-1))) >> SHIFT);\
74 
+        OP(dst[6*dstStride], (src4 + src9  - 5*(src5+src8) + src6*C1 + src7*C2 + (1<<(SHIFT-1))) >> SHIFT);\
75 
+        OP(dst[7*dstStride], (src5 + src10 - 5*(src6+src9) + src7*C1 + src8*C2 + (1<<(SHIFT-1))) >> SHIFT);\
76 
+        dst++;\
77 
+        src++;\
78 
+    }\
79 
+}\
80 
+\
81 
+static void OPNAME ## rv40_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,\
82 
+                                            const int w, const int C1, const int C2, const int SHIFT){\
83 
+    OPNAME ## rv40_qpel8_v_lowpass(dst  , src  , dstStride, srcStride, 8, C1, C2, SHIFT);\
84 
+    OPNAME ## rv40_qpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, 8, C1, C2, SHIFT);\
85 
+    src += 8*srcStride;\
86 
+    dst += 8*dstStride;\
87 
+    OPNAME ## rv40_qpel8_v_lowpass(dst  , src  , dstStride, srcStride, w-8, C1, C2, SHIFT);\
88 
+    OPNAME ## rv40_qpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, w-8, C1, C2, SHIFT);\
89 
+}\
90 
+\
91 
+static void OPNAME ## rv40_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride,\
92 
+                                            const int h, const int C1, const int C2, const int SHIFT){\
93 
+    OPNAME ## rv40_qpel8_h_lowpass(dst  , src  , dstStride, srcStride, 8, C1, C2, SHIFT);\
94 
+    OPNAME ## rv40_qpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, 8, C1, C2, SHIFT);\
95 
+    src += 8*srcStride;\
96 
+    dst += 8*dstStride;\
97 
+    OPNAME ## rv40_qpel8_h_lowpass(dst  , src  , dstStride, srcStride, h-8, C1, C2, SHIFT);\
98 
+    OPNAME ## rv40_qpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, h-8, C1, C2, SHIFT);\
99 
+}\
100 
+\
101 
+
102 
+#define RV40_MC(OPNAME, SIZE) \
103 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc10_c(uint8_t *dst, uint8_t *src, int stride){\
104 
+    OPNAME ## rv40_qpel ## SIZE ## _h_lowpass(dst, src, stride, stride, SIZE, 52, 20, 6);\
105 
+}\
106 
+\
107 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc20_c(uint8_t *dst, uint8_t *src, int stride){\
108 
+    OPNAME ## rv40_qpel ## SIZE ## _h_lowpass(dst, src, stride, stride, SIZE, 20, 20, 5);\
109 
+}\
110 
+\
111 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc30_c(uint8_t *dst, uint8_t *src, int stride){\
112 
+    OPNAME ## rv40_qpel ## SIZE ## _h_lowpass(dst, src, stride, stride, SIZE, 20, 52, 6);\
113 
+}\
114 
+\
115 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc01_c(uint8_t *dst, uint8_t *src, int stride){\
116 
+    OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, src, stride, stride, SIZE, 52, 20, 6);\
117 
+}\
118 
+\
119 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc11_c(uint8_t *dst, uint8_t *src, int stride){\
120 
+    uint8_t full[SIZE*(SIZE+5)];\
121 
+    uint8_t * const full_mid= full + SIZE*2;\
122 
+    put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 52, 20, 6);\
123 
+    OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 52, 20, 6);\
124 
+}\
125 
+\
126 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc21_c(uint8_t *dst, uint8_t *src, int stride){\
127 
+    uint8_t full[SIZE*(SIZE+5)];\
128 
+    uint8_t * const full_mid= full + SIZE*2;\
129 
+    put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 20, 20, 5);\
130 
+    OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 52, 20, 6);\
131 
+}\
132 
+\
133 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc31_c(uint8_t *dst, uint8_t *src, int stride){\
134 
+    uint8_t full[SIZE*(SIZE+5)];\
135 
+    uint8_t * const full_mid= full + SIZE*2;\
136 
+    put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 20, 52, 6);\
137 
+    OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 52, 20, 6);\
138 
+}\
139 
+\
140 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc02_c(uint8_t *dst, uint8_t *src, int stride){\
141 
+    OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, src, stride, stride, SIZE, 20, 20, 5);\
142 
+}\
143 
+\
144 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc12_c(uint8_t *dst, uint8_t *src, int stride){\
145 
+    uint8_t full[SIZE*(SIZE+5)];\
146 
+    uint8_t * const full_mid= full + SIZE*2;\
147 
+    put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 52, 20, 6);\
148 
+    OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 20, 20, 5);\
149 
+}\
150 
+\
151 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc22_c(uint8_t *dst, uint8_t *src, int stride){\
152 
+    uint8_t full[SIZE*(SIZE+5)];\
153 
+    uint8_t * const full_mid= full + SIZE*2;\
154 
+    put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 20, 20, 5);\
155 
+    OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 20, 20, 5);\
156 
+}\
157 
+\
158 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc32_c(uint8_t *dst, uint8_t *src, int stride){\
159 
+    uint8_t full[SIZE*(SIZE+5)];\
160 
+    uint8_t * const full_mid= full + SIZE*2;\
161 
+    put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 20, 52, 6);\
162 
+    OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 20, 20, 5);\
163 
+}\
164 
+\
165 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc03_c(uint8_t *dst, uint8_t *src, int stride){\
166 
+    OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, src, stride, stride, SIZE, 20, 52, 6);\
167 
+}\
168 
+\
169 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc13_c(uint8_t *dst, uint8_t *src, int stride){\
170 
+    uint8_t full[SIZE*(SIZE+5)];\
171 
+    uint8_t * const full_mid= full + SIZE*2;\
172 
+    put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 52, 20, 6);\
173 
+    OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 20, 52, 6);\
174 
+}\
175 
+\
176 
+static void OPNAME ## rv40_qpel ## SIZE ## _mc23_c(uint8_t *dst, uint8_t *src, int stride){\
177 
+    uint8_t full[SIZE*(SIZE+5)];\
178 
+    uint8_t * const full_mid= full + SIZE*2;\
179 
+    put_rv40_qpel ## SIZE ## _h_lowpass(full, src - 2*stride, SIZE, stride, SIZE+5, 20, 20, 5);\
180 
+    OPNAME ## rv40_qpel ## SIZE ## _v_lowpass(dst, full_mid, stride, SIZE, SIZE, 20, 52, 6);\
181 
+}\
182 
+\
183 
+
184 
+#define op_avg(a, b)  a = (((a)+cm[b]+1)>>1)
185 
+#define op_put(a, b)  a = cm[b]
186 
+
187 
+RV40_LOWPASS(put_       , op_put)
188 
+RV40_LOWPASS(avg_       , op_avg)
189 
+
190 
+#undef op_avg
191 
+#undef op_put
192 
+
193 
+RV40_MC(put_, 8)
194 
+RV40_MC(put_, 16)
195 
+RV40_MC(avg_, 8)
196 
+RV40_MC(avg_, 16)
197 
+
198 
+static const int rv40_bias[4][4] = {
199 
+    {  0, 16, 32, 16 },
200 
+    { 32, 28, 32, 28 },
201 
+    {  0, 32, 16, 32 },
202 
+    { 32, 28, 32, 28 }
203 
+};
204 
+
205 
+#define RV40_CHROMA_MC(OPNAME, OP)\
206 
+static void OPNAME ## rv40_chroma_mc4_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){\
207 
+    const int A=(8-x)*(8-y);\
208 
+    const int B=(  x)*(8-y);\
209 
+    const int C=(8-x)*(  y);\
210 
+    const int D=(  x)*(  y);\
211 
+    int i;\
212 
+    int bias = rv40_bias[y>>1][x>>1];\
213 
+    \
214 
+    assert(x<8 && y<8 && x>=0 && y>=0);\
215 
+\
216 
+    if(D){\
217 
+        for(i=0; i<h; i++){\
218 
+            OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + bias));\
219 
+            OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + bias));\
220 
+            OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + bias));\
221 
+            OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + bias));\
222 
+            dst+= stride;\
223 
+            src+= stride;\
224 
+        }\
225 
+    }else{\
226 
+        const int E= B+C;\
227 
+        const int step= C ? stride : 1;\
228 
+        for(i=0; i<h; i++){\
229 
+            OP(dst[0], (A*src[0] + E*src[step+0] + bias));\
230 
+            OP(dst[1], (A*src[1] + E*src[step+1] + bias));\
231 
+            OP(dst[2], (A*src[2] + E*src[step+2] + bias));\
232 
+            OP(dst[3], (A*src[3] + E*src[step+3] + bias));\
233 
+            dst+= stride;\
234 
+            src+= stride;\
235 
+        }\
236 
+    }\
237 
+}\
238 
+\
239 
+static void OPNAME ## rv40_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){\
240 
+    const int A=(8-x)*(8-y);\
241 
+    const int B=(  x)*(8-y);\
242 
+    const int C=(8-x)*(  y);\
243 
+    const int D=(  x)*(  y);\
244 
+    int i;\
245 
+    int bias = rv40_bias[y>>1][x>>1];\
246 
+    \
247 
+    assert(x<8 && y<8 && x>=0 && y>=0);\
248 
+\
249 
+    if(D){\
250 
+        for(i=0; i<h; i++){\
251 
+            OP(dst[0], (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + bias));\
252 
+            OP(dst[1], (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + bias));\
253 
+            OP(dst[2], (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + bias));\
254 
+            OP(dst[3], (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + bias));\
255 
+            OP(dst[4], (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + bias));\
256 
+            OP(dst[5], (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + bias));\
257 
+            OP(dst[6], (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + bias));\
258 
+            OP(dst[7], (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + bias));\
259 
+            dst+= stride;\
260 
+            src+= stride;\
261 
+        }\
262 
+    }else{\
263 
+        const int E= B+C;\
264 
+        const int step= C ? stride : 1;\
265 
+        for(i=0; i<h; i++){\
266 
+            OP(dst[0], (A*src[0] + E*src[step+0] + bias));\
267 
+            OP(dst[1], (A*src[1] + E*src[step+1] + bias));\
268 
+            OP(dst[2], (A*src[2] + E*src[step+2] + bias));\
269 
+            OP(dst[3], (A*src[3] + E*src[step+3] + bias));\
270 
+            OP(dst[4], (A*src[4] + E*src[step+4] + bias));\
271 
+            OP(dst[5], (A*src[5] + E*src[step+5] + bias));\
272 
+            OP(dst[6], (A*src[6] + E*src[step+6] + bias));\
273 
+            OP(dst[7], (A*src[7] + E*src[step+7] + bias));\
274 
+            dst+= stride;\
275 
+            src+= stride;\
276 
+        }\
277 
+    }\
278 
+}
279 
+
280 
+#define op_avg(a, b) a = (((a)+((b)>>6)+1)>>1)
281 
+#define op_put(a, b) a = ((b)>>6)
282 
+
283 
+RV40_CHROMA_MC(put_, op_put)
284 
+RV40_CHROMA_MC(avg_, op_avg)
285 
+
286 
+void ff_rv40dsp_init(DSPContext* c, AVCodecContext *avctx) {
287 
+    c->put_rv40_qpel_pixels_tab[0][ 0] = c->put_h264_qpel_pixels_tab[0][0];
288 
+    c->put_rv40_qpel_pixels_tab[0][ 1] = put_rv40_qpel16_mc10_c;
289 
+    c->put_rv40_qpel_pixels_tab[0][ 2] = put_rv40_qpel16_mc20_c;
290 
+    c->put_rv40_qpel_pixels_tab[0][ 3] = put_rv40_qpel16_mc30_c;
291 
+    c->put_rv40_qpel_pixels_tab[0][ 4] = put_rv40_qpel16_mc01_c;
292 
+    c->put_rv40_qpel_pixels_tab[0][ 5] = put_rv40_qpel16_mc11_c;
293 
+    c->put_rv40_qpel_pixels_tab[0][ 6] = put_rv40_qpel16_mc21_c;
294 
+    c->put_rv40_qpel_pixels_tab[0][ 7] = put_rv40_qpel16_mc31_c;
295 
+    c->put_rv40_qpel_pixels_tab[0][ 8] = put_rv40_qpel16_mc02_c;
296 
+    c->put_rv40_qpel_pixels_tab[0][ 9] = put_rv40_qpel16_mc12_c;
297 
+    c->put_rv40_qpel_pixels_tab[0][10] = put_rv40_qpel16_mc22_c;
298 
+    c->put_rv40_qpel_pixels_tab[0][11] = put_rv40_qpel16_mc32_c;
299 
+    c->put_rv40_qpel_pixels_tab[0][12] = put_rv40_qpel16_mc03_c;
300 
+    c->put_rv40_qpel_pixels_tab[0][13] = put_rv40_qpel16_mc13_c;
301 
+    c->put_rv40_qpel_pixels_tab[0][14] = put_rv40_qpel16_mc23_c;
302 
+    c->avg_rv40_qpel_pixels_tab[0][ 0] = c->avg_h264_qpel_pixels_tab[0][0];
303 
+    c->avg_rv40_qpel_pixels_tab[0][ 1] = avg_rv40_qpel16_mc10_c;
304 
+    c->avg_rv40_qpel_pixels_tab[0][ 2] = avg_rv40_qpel16_mc20_c;
305 
+    c->avg_rv40_qpel_pixels_tab[0][ 3] = avg_rv40_qpel16_mc30_c;
306 
+    c->avg_rv40_qpel_pixels_tab[0][ 4] = avg_rv40_qpel16_mc01_c;
307 
+    c->avg_rv40_qpel_pixels_tab[0][ 5] = avg_rv40_qpel16_mc11_c;
308 
+    c->avg_rv40_qpel_pixels_tab[0][ 6] = avg_rv40_qpel16_mc21_c;
309 
+    c->avg_rv40_qpel_pixels_tab[0][ 7] = avg_rv40_qpel16_mc31_c;
310 
+    c->avg_rv40_qpel_pixels_tab[0][ 8] = avg_rv40_qpel16_mc02_c;
311 
+    c->avg_rv40_qpel_pixels_tab[0][ 9] = avg_rv40_qpel16_mc12_c;
312 
+    c->avg_rv40_qpel_pixels_tab[0][10] = avg_rv40_qpel16_mc22_c;
313 
+    c->avg_rv40_qpel_pixels_tab[0][11] = avg_rv40_qpel16_mc32_c;
314 
+    c->avg_rv40_qpel_pixels_tab[0][12] = avg_rv40_qpel16_mc03_c;
315 
+    c->avg_rv40_qpel_pixels_tab[0][13] = avg_rv40_qpel16_mc13_c;
316 
+    c->avg_rv40_qpel_pixels_tab[0][14] = avg_rv40_qpel16_mc23_c;
317 
+    c->put_rv40_qpel_pixels_tab[1][ 0] = c->put_h264_qpel_pixels_tab[1][0];
318 
+    c->put_rv40_qpel_pixels_tab[1][ 1] = put_rv40_qpel8_mc10_c;
319 
+    c->put_rv40_qpel_pixels_tab[1][ 2] = put_rv40_qpel8_mc20_c;
320 
+    c->put_rv40_qpel_pixels_tab[1][ 3] = put_rv40_qpel8_mc30_c;
321 
+    c->put_rv40_qpel_pixels_tab[1][ 4] = put_rv40_qpel8_mc01_c;
322 
+    c->put_rv40_qpel_pixels_tab[1][ 5] = put_rv40_qpel8_mc11_c;
323 
+    c->put_rv40_qpel_pixels_tab[1][ 6] = put_rv40_qpel8_mc21_c;
324 
+    c->put_rv40_qpel_pixels_tab[1][ 7] = put_rv40_qpel8_mc31_c;
325 
+    c->put_rv40_qpel_pixels_tab[1][ 8] = put_rv40_qpel8_mc02_c;
326 
+    c->put_rv40_qpel_pixels_tab[1][ 9] = put_rv40_qpel8_mc12_c;
327 
+    c->put_rv40_qpel_pixels_tab[1][10] = put_rv40_qpel8_mc22_c;
328 
+    c->put_rv40_qpel_pixels_tab[1][11] = put_rv40_qpel8_mc32_c;
329 
+    c->put_rv40_qpel_pixels_tab[1][12] = put_rv40_qpel8_mc03_c;
330 
+    c->put_rv40_qpel_pixels_tab[1][13] = put_rv40_qpel8_mc13_c;
331 
+    c->put_rv40_qpel_pixels_tab[1][14] = put_rv40_qpel8_mc23_c;
332 
+    c->avg_rv40_qpel_pixels_tab[1][ 0] = c->avg_h264_qpel_pixels_tab[1][0];
333 
+    c->avg_rv40_qpel_pixels_tab[1][ 1] = avg_rv40_qpel8_mc10_c;
334 
+    c->avg_rv40_qpel_pixels_tab[1][ 2] = avg_rv40_qpel8_mc20_c;
335 
+    c->avg_rv40_qpel_pixels_tab[1][ 3] = avg_rv40_qpel8_mc30_c;
336 
+    c->avg_rv40_qpel_pixels_tab[1][ 4] = avg_rv40_qpel8_mc01_c;
337 
+    c->avg_rv40_qpel_pixels_tab[1][ 5] = avg_rv40_qpel8_mc11_c;
338 
+    c->avg_rv40_qpel_pixels_tab[1][ 6] = avg_rv40_qpel8_mc21_c;
339 
+    c->avg_rv40_qpel_pixels_tab[1][ 7] = avg_rv40_qpel8_mc31_c;
340 
+    c->avg_rv40_qpel_pixels_tab[1][ 8] = avg_rv40_qpel8_mc02_c;
341 
+    c->avg_rv40_qpel_pixels_tab[1][ 9] = avg_rv40_qpel8_mc12_c;
342 
+    c->avg_rv40_qpel_pixels_tab[1][10] = avg_rv40_qpel8_mc22_c;
343 
+    c->avg_rv40_qpel_pixels_tab[1][11] = avg_rv40_qpel8_mc32_c;
344 
+    c->avg_rv40_qpel_pixels_tab[1][12] = avg_rv40_qpel8_mc03_c;
345 
+    c->avg_rv40_qpel_pixels_tab[1][13] = avg_rv40_qpel8_mc13_c;
346 
+    c->avg_rv40_qpel_pixels_tab[1][14] = avg_rv40_qpel8_mc23_c;
347 
+
348 
+    c->put_rv40_chroma_pixels_tab[0]= put_rv40_chroma_mc8_c;
349 
+    c->put_rv40_chroma_pixels_tab[1]= put_rv40_chroma_mc4_c;
350 
+    c->avg_rv40_chroma_pixels_tab[0]= avg_rv40_chroma_mc8_c;
351 
+    c->avg_rv40_chroma_pixels_tab[1]= avg_rv40_chroma_mc4_c;
352 
+}