X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=libavcodec%2Fdsputil.c;h=6aa54538ec75e227373a5b4345df913b651999ea;hb=5a6b9443ed6136416363c64566f8374a7d084b17;hp=1bee9b969e9e3cbebd04e1db7923114fd7b2171f;hpb=27c61ac53df2876b80633ec5b69229d06f466131;p=ffmpeg

diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index 1bee9b969e9..6aa54538ec7 100644
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -3,19 +3,21 @@
  * Copyright (c) 2000, 2001 Fabrice Bellard.
  * Copyright (c) 2002-2004 Michael Niedermayer <michaelni@gmx.at>
  *
- * This library is free software; you can redistribute it and/or
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
  * modify it under the terms of the GNU Lesser General Public
  * License as published by the Free Software Foundation; either
- * version 2 of the License, or (at your option) any later version.
+ * version 2.1 of the License, or (at your option) any later version.
  *
- * This library is distributed in the hope that it will be useful,
+ * FFmpeg is distributed in the hope that it will be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  * Lesser General Public License for more details.
  *
  * You should have received a copy of the GNU Lesser General Public
- * License along with this library; if not, write to the Free Software
- * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  *
  * gmc & q-pel & 32/64 bit based MC by Michael Niedermayer <michaelni@gmx.at>
  */
@@ -30,12 +32,16 @@
 #include "mpegvideo.h"
 #include "simple_idct.h"
 #include "faandct.h"
+#include "snow.h"
 
 /* snow.c */
 void ff_spatial_dwt(int *buffer, int width, int height, int stride, int type, int decomposition_count);
 
-uint8_t cropTbl[256 + 2 * MAX_NEG_CROP] = {0, };
-uint32_t squareTbl[512] = {0, };
+/* vorbis.c */
+void vorbis_inverse_coupling(float *mag, float *ang, int blocksize);
+
+uint8_t ff_cropTbl[256 + 2 * MAX_NEG_CROP] = {0, };
+uint32_t ff_squareTbl[512] = {0, };
 
 const uint8_t ff_zigzag_direct[64] = {
     0,   1,  8, 16,  9,  2,  3, 10,
@@ -62,7 +68,7 @@ const uint8_t ff_zigzag248_direct[64] = {
 };
 
 /* not permutated inverse zigzag_direct + 1 for MMX quantizer */
-uint16_t __align8 inv_zigzag_direct16[64] = {0, };
+DECLARE_ALIGNED_8(uint16_t, inv_zigzag_direct16[64]) = {0, };
 
 const uint8_t ff_alternate_horizontal_scan[64] = {
     0,  1,   2,  3,  8,  9, 16, 17,
@@ -87,7 +93,7 @@ const uint8_t ff_alternate_vertical_scan[64] = {
 };
 
 /* a*inverse[b]>>32 == a/b for all 0<=a<=65536 && 2<=b<=255 */
-const uint32_t inverse[256]={
+const uint32_t ff_inverse[256]={
          0, 4294967295U,2147483648U,1431655766, 1073741824,  858993460,  715827883,  613566757,
  536870912,  477218589,  429496730,  390451573,  357913942,  330382100,  306783379,  286331154,
  268435456,  252645136,  238609295,  226050911,  214748365,  204522253,  195225787,  186737709,
@@ -159,7 +165,7 @@ static int pix_sum_c(uint8_t * pix, int line_size)
 static int pix_norm1_c(uint8_t * pix, int line_size)
 {
     int s, i, j;
-    uint32_t *sq = squareTbl + 256;
+    uint32_t *sq = ff_squareTbl + 256;
 
     s = 0;
     for (i = 0; i < 16; i++) {
@@ -225,7 +231,7 @@ static void bswap_buf(uint32_t *dst, uint32_t *src, int w){
 static int sse4_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h)
 {
     int s, i;
-    uint32_t *sq = squareTbl + 256;
+    uint32_t *sq = ff_squareTbl + 256;
 
     s = 0;
     for (i = 0; i < h; i++) {
@@ -242,7 +248,7 @@ static int sse4_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h)
 static int sse8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h)
 {
     int s, i;
-    uint32_t *sq = squareTbl + 256;
+    uint32_t *sq = ff_squareTbl + 256;
 
     s = 0;
     for (i = 0; i < h; i++) {
@@ -263,7 +269,7 @@ static int sse8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h)
 static int sse16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 {
     int s, i;
-    uint32_t *sq = squareTbl + 256;
+    uint32_t *sq = ff_squareTbl + 256;
 
     s = 0;
     for (i = 0; i < h; i++) {
@@ -291,35 +297,34 @@ static int sse16_c(void *v, uint8_t *pix1, uint8_t *pix2, int line_size, int h)
 }
 
 
+#ifdef CONFIG_SNOW_ENCODER //dwt is in snow.c
 static inline int w_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int w, int h, int type){
-#ifdef CONFIG_SNOW_ENCODER //idwt is in snow.c
     int s, i, j;
     const int dec_count= w==8 ? 3 : 4;
-    int tmp[16*16];
-#if 0
+    int tmp[32*32];
     int level, ori;
     static const int scale[2][2][4][4]={
       {
         {
-            //8x8 dec=3
+            // 9/7 8x8 dec=3
             {268, 239, 239, 213},
             {  0, 224, 224, 152},
             {  0, 135, 135, 110},
         },{
-            //16x16 dec=4
+            // 9/7 16x16 or 32x32 dec=4
             {344, 310, 310, 280},
             {  0, 320, 320, 228},
             {  0, 175, 175, 136},
             {  0, 129, 129, 102},
         }
       },{
-        {//FIXME 5/3
-            //8x8 dec=3
+        {
+            // 5/3 8x8 dec=3
             {275, 245, 245, 218},
             {  0, 230, 230, 156},
             {  0, 138, 138, 113},
         },{
-            //16x16 dec=4
+            // 5/3 16x16 or 32x32 dec=4
             {352, 317, 317, 286},
             {  0, 328, 328, 233},
             {  0, 180, 180, 140},
@@ -327,51 +332,39 @@ static inline int w_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, in
         }
       }
     };
-#endif
 
     for (i = 0; i < h; i++) {
         for (j = 0; j < w; j+=4) {
-            tmp[16*i+j+0] = (pix1[j+0] - pix2[j+0])<<4;
-            tmp[16*i+j+1] = (pix1[j+1] - pix2[j+1])<<4;
-            tmp[16*i+j+2] = (pix1[j+2] - pix2[j+2])<<4;
-            tmp[16*i+j+3] = (pix1[j+3] - pix2[j+3])<<4;
+            tmp[32*i+j+0] = (pix1[j+0] - pix2[j+0])<<4;
+            tmp[32*i+j+1] = (pix1[j+1] - pix2[j+1])<<4;
+            tmp[32*i+j+2] = (pix1[j+2] - pix2[j+2])<<4;
+            tmp[32*i+j+3] = (pix1[j+3] - pix2[j+3])<<4;
         }
         pix1 += line_size;
         pix2 += line_size;
     }
 
-    ff_spatial_dwt(tmp, w, h, 16, type, dec_count);
+    ff_spatial_dwt(tmp, w, h, 32, type, dec_count);
 
     s=0;
-#if 0
+    assert(w==h);
     for(level=0; level<dec_count; level++){
         for(ori= level ? 1 : 0; ori<4; ori++){
-            int sx= (ori&1) ? 1<<level: 0;
-            int stride= 16<<(dec_count-level);
+            int size= w>>(dec_count-level);
+            int sx= (ori&1) ? size : 0;
+            int stride= 32<<(dec_count-level);
             int sy= (ori&2) ? stride>>1 : 0;
-            int size= 1<<level;
 
             for(i=0; i<size; i++){
                 for(j=0; j<size; j++){
                     int v= tmp[sx + sy + i*stride + j] * scale[type][dec_count-3][level][ori];
-                    s += ABS(v);
+                    s += FFABS(v);
                 }
             }
         }
     }
-#endif
-    for (i = 0; i < h; i++) {
-        for (j = 0; j < w; j+=4) {
-            s+= ABS(tmp[16*i+j+0]);
-            s+= ABS(tmp[16*i+j+1]);
-            s+= ABS(tmp[16*i+j+2]);
-            s+= ABS(tmp[16*i+j+3]);
-        }
-    }
     assert(s>=0);
-
-    return s>>2;
-#endif
+    return s>>9;
 }
 
 static int w53_8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
@@ -390,6 +383,15 @@ static int w97_16_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int
     return w_c(v, pix1, pix2, line_size, 16, h, 0);
 }
 
+int w53_32_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
+    return w_c(v, pix1, pix2, line_size, 32, h, 1);
+}
+
+int w97_32_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size, int h){
+    return w_c(v, pix1, pix2, line_size, 32, h, 0);
+}
+#endif
+
 static void get_pixels_c(DCTELEM *restrict block, const uint8_t *pixels, int line_size)
 {
     int i;
@@ -434,7 +436,7 @@ static void put_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels,
                                  int line_size)
 {
     int i;
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<8;i++) {
@@ -456,7 +458,7 @@ static void put_pixels_clamped4_c(const DCTELEM *block, uint8_t *restrict pixels
                                  int line_size)
 {
     int i;
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<4;i++) {
@@ -474,7 +476,7 @@ static void put_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels
                                  int line_size)
 {
     int i;
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<2;i++) {
@@ -511,7 +513,7 @@ static void add_pixels_clamped_c(const DCTELEM *block, uint8_t *restrict pixels,
                           int line_size)
 {
     int i;
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<8;i++) {
@@ -532,7 +534,7 @@ static void add_pixels_clamped4_c(const DCTELEM *block, uint8_t *restrict pixels
                           int line_size)
 {
     int i;
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<4;i++) {
@@ -549,7 +551,7 @@ static void add_pixels_clamped2_c(const DCTELEM *block, uint8_t *restrict pixels
                           int line_size)
 {
     int i;
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     /* read the pixels */
     for(i=0;i<2;i++) {
@@ -1143,7 +1145,7 @@ static void gmc1_c(uint8_t *dst, uint8_t *src, int stride, int h, int x16, int y
     }
 }
 
-static void gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
+void ff_gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
                   int dxx, int dxy, int dyx, int dyy, int shift, int r, int width, int height)
 {
     int y, vx, vy;
@@ -1176,19 +1178,19 @@ static void gmc_c(uint8_t *dst, uint8_t *src, int stride, int h, int ox, int oy,
                                            + src[index+stride+1]*   frac_x )*   frac_y
                                         + r)>>(shift*2);
                 }else{
-                    index= src_x + clip(src_y, 0, height)*stride;
+                    index= src_x + av_clip(src_y, 0, height)*stride;
                     dst[y*stride + x]= ( (  src[index         ]*(s-frac_x)
                                           + src[index       +1]*   frac_x )*s
                                         + r)>>(shift*2);
                 }
             }else{
                 if((unsigned)src_y < height){
-                    index= clip(src_x, 0, width) + src_y*stride;
+                    index= av_clip(src_x, 0, width) + src_y*stride;
                     dst[y*stride + x]= (  (  src[index         ]*(s-frac_y)
                                            + src[index+stride  ]*   frac_y )*s
                                         + r)>>(shift*2);
                 }else{
-                    index= clip(src_x, 0, width) + clip(src_y, 0, height)*stride;
+                    index= av_clip(src_x, 0, width) + av_clip(src_y, 0, height)*stride;
                     dst[y*stride + x]=    src[index         ];
                 }
             }
@@ -1487,75 +1489,33 @@ H264_CHROMA_MC(avg_       , op_avg)
 #undef op_avg
 #undef op_put
 
-static inline void copy_block4(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
-{
+static void put_no_rnd_h264_chroma_mc8_c(uint8_t *dst/*align 8*/, uint8_t *src/*align 1*/, int stride, int h, int x, int y){
+    const int A=(8-x)*(8-y);
+    const int B=(  x)*(8-y);
+    const int C=(8-x)*(  y);
+    const int D=(  x)*(  y);
     int i;
-    for(i=0; i<h; i++)
-    {
-        ST32(dst   , LD32(src   ));
-        dst+=dstStride;
-        src+=srcStride;
-    }
-}
-
-static inline void copy_block8(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
-{
-    int i;
-    for(i=0; i<h; i++)
-    {
-        ST32(dst   , LD32(src   ));
-        ST32(dst+4 , LD32(src+4 ));
-        dst+=dstStride;
-        src+=srcStride;
-    }
-}
-
-static inline void copy_block16(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
-{
-    int i;
-    for(i=0; i<h; i++)
-    {
-        ST32(dst   , LD32(src   ));
-        ST32(dst+4 , LD32(src+4 ));
-        ST32(dst+8 , LD32(src+8 ));
-        ST32(dst+12, LD32(src+12));
-        dst+=dstStride;
-        src+=srcStride;
-    }
-}
 
-static inline void copy_block17(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
-{
-    int i;
-    for(i=0; i<h; i++)
-    {
-        ST32(dst   , LD32(src   ));
-        ST32(dst+4 , LD32(src+4 ));
-        ST32(dst+8 , LD32(src+8 ));
-        ST32(dst+12, LD32(src+12));
-        dst[16]= src[16];
-        dst+=dstStride;
-        src+=srcStride;
-    }
-}
+    assert(x<8 && y<8 && x>=0 && y>=0);
 
-static inline void copy_block9(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h)
-{
-    int i;
     for(i=0; i<h; i++)
     {
-        ST32(dst   , LD32(src   ));
-        ST32(dst+4 , LD32(src+4 ));
-        dst[8]= src[8];
-        dst+=dstStride;
-        src+=srcStride;
+        dst[0] = (A*src[0] + B*src[1] + C*src[stride+0] + D*src[stride+1] + 32 - 4) >> 6;
+        dst[1] = (A*src[1] + B*src[2] + C*src[stride+1] + D*src[stride+2] + 32 - 4) >> 6;
+        dst[2] = (A*src[2] + B*src[3] + C*src[stride+2] + D*src[stride+3] + 32 - 4) >> 6;
+        dst[3] = (A*src[3] + B*src[4] + C*src[stride+3] + D*src[stride+4] + 32 - 4) >> 6;
+        dst[4] = (A*src[4] + B*src[5] + C*src[stride+4] + D*src[stride+5] + 32 - 4) >> 6;
+        dst[5] = (A*src[5] + B*src[6] + C*src[stride+5] + D*src[stride+6] + 32 - 4) >> 6;
+        dst[6] = (A*src[6] + B*src[7] + C*src[stride+6] + D*src[stride+7] + 32 - 4) >> 6;
+        dst[7] = (A*src[7] + B*src[8] + C*src[stride+7] + D*src[stride+8] + 32 - 4) >> 6;
+        dst+= stride;
+        src+= stride;
     }
 }
 
-
 #define QPEL_MC(r, OPNAME, RND, OP) \
 static void OPNAME ## mpeg4_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     for(i=0; i<h; i++)\
     {\
@@ -1574,7 +1534,7 @@ static void OPNAME ## mpeg4_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstS
 \
 static void OPNAME ## mpeg4_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     const int w=8;\
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     for(i=0; i<w; i++)\
     {\
@@ -1601,7 +1561,7 @@ static void OPNAME ## mpeg4_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstS
 }\
 \
 static void OPNAME ## mpeg4_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){\
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     \
     for(i=0; i<h; i++)\
@@ -1628,7 +1588,7 @@ static void OPNAME ## mpeg4_qpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dst
 }\
 \
 static void OPNAME ## mpeg4_qpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     const int w=16;\
     for(i=0; i<w; i++)\
@@ -2052,9 +2012,71 @@ QPEL_MC(0, avg_       , _       , op_avg)
 
 #if 1
 #define H264_LOWPASS(OPNAME, OP, OP2) \
+static void OPNAME ## h264_qpel2_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+    const int h=2;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
+    int i;\
+    for(i=0; i<h; i++)\
+    {\
+        OP(dst[0], (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]));\
+        OP(dst[1], (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]));\
+        dst+=dstStride;\
+        src+=srcStride;\
+    }\
+}\
+\
+static void OPNAME ## h264_qpel2_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
+    const int w=2;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
+    int i;\
+    for(i=0; i<w; i++)\
+    {\
+        const int srcB= src[-2*srcStride];\
+        const int srcA= src[-1*srcStride];\
+        const int src0= src[0 *srcStride];\
+        const int src1= src[1 *srcStride];\
+        const int src2= src[2 *srcStride];\
+        const int src3= src[3 *srcStride];\
+        const int src4= src[4 *srcStride];\
+        OP(dst[0*dstStride], (src0+src1)*20 - (srcA+src2)*5 + (srcB+src3));\
+        OP(dst[1*dstStride], (src1+src2)*20 - (src0+src3)*5 + (srcA+src4));\
+        dst++;\
+        src++;\
+    }\
+}\
+\
+static void OPNAME ## h264_qpel2_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
+    const int h=2;\
+    const int w=2;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
+    int i;\
+    src -= 2*srcStride;\
+    for(i=0; i<h+5; i++)\
+    {\
+        tmp[0]= (src[0]+src[1])*20 - (src[-1]+src[2])*5 + (src[-2]+src[3]);\
+        tmp[1]= (src[1]+src[2])*20 - (src[0 ]+src[3])*5 + (src[-1]+src[4]);\
+        tmp+=tmpStride;\
+        src+=srcStride;\
+    }\
+    tmp -= tmpStride*(h+5-2);\
+    for(i=0; i<w; i++)\
+    {\
+        const int tmpB= tmp[-2*tmpStride];\
+        const int tmpA= tmp[-1*tmpStride];\
+        const int tmp0= tmp[0 *tmpStride];\
+        const int tmp1= tmp[1 *tmpStride];\
+        const int tmp2= tmp[2 *tmpStride];\
+        const int tmp3= tmp[3 *tmpStride];\
+        const int tmp4= tmp[4 *tmpStride];\
+        OP2(dst[0*dstStride], (tmp0+tmp1)*20 - (tmpA+tmp2)*5 + (tmpB+tmp3));\
+        OP2(dst[1*dstStride], (tmp1+tmp2)*20 - (tmp0+tmp3)*5 + (tmpA+tmp4));\
+        dst++;\
+        tmp++;\
+    }\
+}\
 static void OPNAME ## h264_qpel4_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     const int h=4;\
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     for(i=0; i<h; i++)\
     {\
@@ -2069,7 +2091,7 @@ static void OPNAME ## h264_qpel4_h_lowpass(uint8_t *dst, uint8_t *src, int dstSt
 \
 static void OPNAME ## h264_qpel4_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     const int w=4;\
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     for(i=0; i<w; i++)\
     {\
@@ -2094,7 +2116,7 @@ static void OPNAME ## h264_qpel4_v_lowpass(uint8_t *dst, uint8_t *src, int dstSt
 static void OPNAME ## h264_qpel4_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
     const int h=4;\
     const int w=4;\
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     src -= 2*srcStride;\
     for(i=0; i<h+5; i++)\
@@ -2129,7 +2151,7 @@ static void OPNAME ## h264_qpel4_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t
 \
 static void OPNAME ## h264_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     const int h=8;\
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     for(i=0; i<h; i++)\
     {\
@@ -2148,7 +2170,7 @@ static void OPNAME ## h264_qpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstSt
 \
 static void OPNAME ## h264_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
     const int w=8;\
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     for(i=0; i<w; i++)\
     {\
@@ -2181,7 +2203,7 @@ static void OPNAME ## h264_qpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstSt
 static void OPNAME ## h264_qpel8_hv_lowpass(uint8_t *dst, int16_t *tmp, uint8_t *src, int dstStride, int tmpStride, int srcStride){\
     const int h=8;\
     const int w=8;\
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;\
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
     int i;\
     src -= 2*srcStride;\
     for(i=0; i<h+5; i++)\
@@ -2398,6 +2420,7 @@ static void OPNAME ## h264_qpel ## SIZE ## _mc32_c(uint8_t *dst, uint8_t *src, i
 
 H264_LOWPASS(put_       , op_put, op2_put)
 H264_LOWPASS(avg_       , op_avg, op2_avg)
+H264_MC(put_, 2)
 H264_MC(put_, 4)
 H264_MC(put_, 8)
 H264_MC(put_, 16)
@@ -2411,11 +2434,11 @@ H264_MC(avg_, 16)
 #undef op2_put
 #endif
 
-#define op_scale1(x)  block[x] = clip_uint8( (block[x]*weight + offset) >> log2_denom )
-#define op_scale2(x)  dst[x] = clip_uint8( (src[x]*weights + dst[x]*weightd + offset) >> (log2_denom+1))
+#define op_scale1(x)  block[x] = av_clip_uint8( (block[x]*weight + offset) >> log2_denom )
+#define op_scale2(x)  dst[x] = av_clip_uint8( (src[x]*weights + dst[x]*weightd + offset) >> (log2_denom+1))
 #define H264_WEIGHT(W,H) \
 static void weight_h264_pixels ## W ## x ## H ## _c(uint8_t *block, int stride, int log2_denom, int weight, int offset){ \
-    int attribute_unused x, y; \
+    int y; \
     offset <<= log2_denom; \
     if(log2_denom) offset += 1<<(log2_denom-1); \
     for(y=0; y<H; y++, block += stride){ \
@@ -2440,10 +2463,9 @@ static void weight_h264_pixels ## W ## x ## H ## _c(uint8_t *block, int stride,
         op_scale1(15); \
     } \
 } \
-static void biweight_h264_pixels ## W ## x ## H ## _c(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offsetd, int offsets){ \
-    int attribute_unused x, y; \
-    int offset = (offsets + offsetd + 1) >> 1; \
-    offset = ((offset << 1) + 1) << log2_denom; \
+static void biweight_h264_pixels ## W ## x ## H ## _c(uint8_t *dst, uint8_t *src, int stride, int log2_denom, int weightd, int weights, int offset){ \
+    int y; \
+    offset = ((offset + 1) | 1) << log2_denom; \
     for(y=0; y<H; y++, dst += stride, src += stride){ \
         op_scale2(0); \
         op_scale2(1); \
@@ -2483,7 +2505,7 @@ H264_WEIGHT(2,2)
 #undef H264_WEIGHT
 
 static void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int h){
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
     int i;
 
     for(i=0; i<h; i++){
@@ -2500,8 +2522,40 @@ static void wmv2_mspel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int
     }
 }
 
+#ifdef CONFIG_CAVS_DECODER
+/* AVS specific */
+void ff_cavsdsp_init(DSPContext* c, AVCodecContext *avctx);
+
+void ff_put_cavs_qpel8_mc00_c(uint8_t *dst, uint8_t *src, int stride) {
+    put_pixels8_c(dst, src, stride, 8);
+}
+void ff_avg_cavs_qpel8_mc00_c(uint8_t *dst, uint8_t *src, int stride) {
+    avg_pixels8_c(dst, src, stride, 8);
+}
+void ff_put_cavs_qpel16_mc00_c(uint8_t *dst, uint8_t *src, int stride) {
+    put_pixels16_c(dst, src, stride, 16);
+}
+void ff_avg_cavs_qpel16_mc00_c(uint8_t *dst, uint8_t *src, int stride) {
+    avg_pixels16_c(dst, src, stride, 16);
+}
+#endif /* CONFIG_CAVS_DECODER */
+
+#if defined(CONFIG_VC1_DECODER) || defined(CONFIG_WMV3_DECODER)
+/* VC-1 specific */
+void ff_vc1dsp_init(DSPContext* c, AVCodecContext *avctx);
+
+void ff_put_vc1_mspel_mc00_c(uint8_t *dst, uint8_t *src, int stride, int rnd) {
+    put_pixels8_c(dst, src, stride, 8);
+}
+#endif /* CONFIG_VC1_DECODER||CONFIG_WMV3_DECODER */
+
+#if defined(CONFIG_H264_ENCODER)
+/* H264 specific */
+void ff_h264dsp_init(DSPContext* c, AVCodecContext *avctx);
+#endif /* CONFIG_H264_ENCODER */
+
 static void wmv2_mspel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, int w){
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
     int i;
 
     for(i=0; i<w; i++){
@@ -2603,9 +2657,9 @@ static void h263_v_loop_filter_c(uint8_t *src, int stride, int qscale){
         src[x-1*stride] = p1;
         src[x+0*stride] = p2;
 
-        ad1= ABS(d1)>>1;
+        ad1= FFABS(d1)>>1;
 
-        d2= clip((p0-p3)/4, -ad1, ad1);
+        d2= av_clip((p0-p3)/4, -ad1, ad1);
 
         src[x-2*stride] = p0 - d2;
         src[x+  stride] = p3 + d2;
@@ -2638,9 +2692,9 @@ static void h263_h_loop_filter_c(uint8_t *src, int stride, int qscale){
         src[y*stride-1] = p1;
         src[y*stride+0] = p2;
 
-        ad1= ABS(d1)>>1;
+        ad1= FFABS(d1)>>1;
 
-        d2= clip((p0-p3)/4, -ad1, ad1);
+        d2= av_clip((p0-p3)/4, -ad1, ad1);
 
         src[y*stride-2] = p0 - d2;
         src[y*stride+1] = p3 + d2;
@@ -2690,25 +2744,25 @@ static inline void h264_loop_filter_luma_c(uint8_t *pix, int xstride, int ystrid
             const int q1 = pix[1*xstride];
             const int q2 = pix[2*xstride];
 
-            if( ABS( p0 - q0 ) < alpha &&
-                ABS( p1 - p0 ) < beta &&
-                ABS( q1 - q0 ) < beta ) {
+            if( FFABS( p0 - q0 ) < alpha &&
+                FFABS( p1 - p0 ) < beta &&
+                FFABS( q1 - q0 ) < beta ) {
 
                 int tc = tc0[i];
                 int i_delta;
 
-                if( ABS( p2 - p0 ) < beta ) {
-                    pix[-2*xstride] = p1 + clip( (( p2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - p1, -tc0[i], tc0[i] );
+                if( FFABS( p2 - p0 ) < beta ) {
+                    pix[-2*xstride] = p1 + av_clip( (( p2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - p1, -tc0[i], tc0[i] );
                     tc++;
                 }
-                if( ABS( q2 - q0 ) < beta ) {
-                    pix[   xstride] = q1 + clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc0[i], tc0[i] );
+                if( FFABS( q2 - q0 ) < beta ) {
+                    pix[   xstride] = q1 + av_clip( (( q2 + ( ( p0 + q0 + 1 ) >> 1 ) ) >> 1) - q1, -tc0[i], tc0[i] );
                     tc++;
                 }
 
-                i_delta = clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
-                pix[-xstride] = clip_uint8( p0 + i_delta );    /* p0' */
-                pix[0]        = clip_uint8( q0 - i_delta );    /* q0' */
+                i_delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
+                pix[-xstride] = av_clip_uint8( p0 + i_delta );    /* p0' */
+                pix[0]        = av_clip_uint8( q0 - i_delta );    /* q0' */
             }
             pix += ystride;
         }
@@ -2738,14 +2792,14 @@ static inline void h264_loop_filter_chroma_c(uint8_t *pix, int xstride, int ystr
             const int q0 = pix[0];
             const int q1 = pix[1*xstride];
 
-            if( ABS( p0 - q0 ) < alpha &&
-                ABS( p1 - p0 ) < beta &&
-                ABS( q1 - q0 ) < beta ) {
+            if( FFABS( p0 - q0 ) < alpha &&
+                FFABS( p1 - p0 ) < beta &&
+                FFABS( q1 - q0 ) < beta ) {
 
-                int delta = clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
+                int delta = av_clip( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
 
-                pix[-xstride] = clip_uint8( p0 + delta );    /* p0' */
-                pix[0]        = clip_uint8( q0 - delta );    /* q0' */
+                pix[-xstride] = av_clip_uint8( p0 + delta );    /* p0' */
+                pix[0]        = av_clip_uint8( q0 - delta );    /* q0' */
             }
             pix += ystride;
         }
@@ -2769,9 +2823,9 @@ static inline void h264_loop_filter_chroma_intra_c(uint8_t *pix, int xstride, in
         const int q0 = pix[0];
         const int q1 = pix[1*xstride];
 
-        if( ABS( p0 - q0 ) < alpha &&
-            ABS( p1 - p0 ) < beta &&
-            ABS( q1 - q0 ) < beta ) {
+        if( FFABS( p0 - q0 ) < alpha &&
+            FFABS( p1 - p0 ) < beta &&
+            FFABS( q1 - q0 ) < beta ) {
 
             pix[-xstride] = ( 2*p1 + p0 + q1 + 2 ) >> 2;   /* p0' */
             pix[0]        = ( 2*q1 + q0 + p1 + 2 ) >> 2;   /* q0' */
@@ -3000,9 +3054,9 @@ static int nsse16_c(void *v, uint8_t *s1, uint8_t *s2, int stride, int h){
         }
         if(y+1<h){
             for(x=0; x<15; x++){
-                score2+= ABS(  s1[x  ] - s1[x  +stride]
+                score2+= FFABS(  s1[x  ] - s1[x  +stride]
                              - s1[x+1] + s1[x+1+stride])
-                        -ABS(  s2[x  ] - s2[x  +stride]
+                        -FFABS(  s2[x  ] - s2[x  +stride]
                              - s2[x+1] + s2[x+1+stride]);
             }
         }
@@ -3010,8 +3064,8 @@ static int nsse16_c(void *v, uint8_t *s1, uint8_t *s2, int stride, int h){
         s2+= stride;
     }
 
-    if(c) return score1 + ABS(score2)*c->avctx->nsse_weight;
-    else  return score1 + ABS(score2)*8;
+    if(c) return score1 + FFABS(score2)*c->avctx->nsse_weight;
+    else  return score1 + FFABS(score2)*8;
 }
 
 static int nsse8_c(void *v, uint8_t *s1, uint8_t *s2, int stride, int h){
@@ -3026,9 +3080,9 @@ static int nsse8_c(void *v, uint8_t *s1, uint8_t *s2, int stride, int h){
         }
         if(y+1<h){
             for(x=0; x<7; x++){
-                score2+= ABS(  s1[x  ] - s1[x  +stride]
+                score2+= FFABS(  s1[x  ] - s1[x  +stride]
                              - s1[x+1] + s1[x+1+stride])
-                        -ABS(  s2[x  ] - s2[x  +stride]
+                        -FFABS(  s2[x  ] - s2[x  +stride]
                              - s2[x+1] + s2[x+1+stride]);
             }
         }
@@ -3036,8 +3090,8 @@ static int nsse8_c(void *v, uint8_t *s1, uint8_t *s2, int stride, int h){
         s2+= stride;
     }
 
-    if(c) return score1 + ABS(score2)*c->avctx->nsse_weight;
-    else  return score1 + ABS(score2)*8;
+    if(c) return score1 + FFABS(score2)*c->avctx->nsse_weight;
+    else  return score1 + FFABS(score2)*8;
 }
 
 static int try_8x8basis_c(int16_t rem[64], int16_t weight[64], int16_t basis[64], int scale){
@@ -3142,12 +3196,14 @@ void ff_set_cmp(DSPContext* c, me_cmp_func *cmp, int type){
         case FF_CMP_NSSE:
             cmp[i]= c->nsse[i];
             break;
+#ifdef CONFIG_SNOW_ENCODER
         case FF_CMP_W53:
             cmp[i]= c->w53[i];
             break;
         case FF_CMP_W97:
             cmp[i]= c->w97[i];
             break;
+#endif
         default:
             av_log(NULL, AV_LOG_ERROR,"internal error in cmp function selection\n");
         }
@@ -3225,7 +3281,7 @@ o2= (i1)-(i2);
     y= a-b;\
 }
 
-#define BUTTERFLYA(x,y) (ABS((x)+(y)) + ABS((x)-(y)))
+#define BUTTERFLYA(x,y) (FFABS((x)+(y)) + FFABS((x)-(y)))
 
 static int hadamard8_diff8x8_c(/*MpegEncContext*/ void *s, uint8_t *dst, uint8_t *src, int stride, int h){
     int i;
@@ -3322,14 +3378,14 @@ static int hadamard8_intra8x8_c(/*MpegEncContext*/ void *s, uint8_t *src, uint8_
             +BUTTERFLYA(temp[8*3+i], temp[8*7+i]);
     }
 
-    sum -= ABS(temp[8*0] + temp[8*4]); // -mean
+    sum -= FFABS(temp[8*0] + temp[8*4]); // -mean
 
     return sum;
 }
 
 static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
-    uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
+    DECLARE_ALIGNED_8(uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]);
     DCTELEM * const temp= (DCTELEM*)aligned_temp;
     int sum=0, i;
 
@@ -3339,7 +3395,7 @@ static int dct_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2
     s->dsp.fdct(temp);
 
     for(i=0; i<64; i++)
-        sum+= ABS(temp[i]);
+        sum+= FFABS(temp[i]);
 
     return sum;
 }
@@ -3388,7 +3444,7 @@ static int dct264_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s
 #undef DST
 
 #define SRC(x) dct[x][i]
-#define DST(x,v) sum += ABS(v)
+#define DST(x,v) sum += FFABS(v)
     for( i = 0; i < 8; i++ )
         DCT8_1D
 #undef SRC
@@ -3399,7 +3455,7 @@ static int dct264_sad8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s
 
 static int dct_max8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
-    uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
+    DECLARE_ALIGNED_8(uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]);
     DCTELEM * const temp= (DCTELEM*)aligned_temp;
     int sum=0, i;
 
@@ -3409,16 +3465,14 @@ static int dct_max8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2
     s->dsp.fdct(temp);
 
     for(i=0; i<64; i++)
-        sum= FFMAX(sum, ABS(temp[i]));
+        sum= FFMAX(sum, FFABS(temp[i]));
 
     return sum;
 }
 
-void simple_idct(DCTELEM *block); //FIXME
-
 static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
-    uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64*2/8];
+    DECLARE_ALIGNED_8 (uint64_t, aligned_temp[sizeof(DCTELEM)*64*2/8]);
     DCTELEM * const temp= (DCTELEM*)aligned_temp;
     DCTELEM * const bak = ((DCTELEM*)aligned_temp)+64;
     int sum=0, i;
@@ -3443,8 +3497,8 @@ static int quant_psnr8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *s
 static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
     const uint8_t *scantable= s->intra_scantable.permutated;
-    uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
-    uint64_t __align8 aligned_bak[stride];
+    DECLARE_ALIGNED_8 (uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]);
+    DECLARE_ALIGNED_8 (uint64_t, aligned_bak[stride]);
     DCTELEM * const temp= (DCTELEM*)aligned_temp;
     uint8_t * const bak= (uint8_t*)aligned_bak;
     int i, last, run, bits, level, distoration, start_i;
@@ -3522,7 +3576,7 @@ static int rd8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int
 static int bit8x8_c(/*MpegEncContext*/ void *c, uint8_t *src1, uint8_t *src2, int stride, int h){
     MpegEncContext * const s= (MpegEncContext *)c;
     const uint8_t *scantable= s->intra_scantable.permutated;
-    uint64_t __align8 aligned_temp[sizeof(DCTELEM)*64/8];
+    DECLARE_ALIGNED_8 (uint64_t, aligned_temp[sizeof(DCTELEM)*64/8]);
     DCTELEM * const temp= (DCTELEM*)aligned_temp;
     int i, last, run, bits, level, start_i;
     const int esc_length= s->ac_esc_length;
@@ -3585,8 +3639,8 @@ static int vsad_intra16_c(/*MpegEncContext*/ void *c, uint8_t *s, uint8_t *dummy
 
     for(y=1; y<h; y++){
         for(x=0; x<16; x+=4){
-            score+= ABS(s[x  ] - s[x  +stride]) + ABS(s[x+1] - s[x+1+stride])
-                   +ABS(s[x+2] - s[x+2+stride]) + ABS(s[x+3] - s[x+3+stride]);
+            score+= FFABS(s[x  ] - s[x  +stride]) + FFABS(s[x+1] - s[x+1+stride])
+                   +FFABS(s[x+2] - s[x+2+stride]) + FFABS(s[x+3] - s[x+3+stride]);
         }
         s+= stride;
     }
@@ -3600,7 +3654,7 @@ static int vsad16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int st
 
     for(y=1; y<h; y++){
         for(x=0; x<16; x++){
-            score+= ABS(s1[x  ] - s2[x ] - s1[x  +stride] + s2[x +stride]);
+            score+= FFABS(s1[x  ] - s2[x ] - s1[x  +stride] + s2[x +stride]);
         }
         s1+= stride;
         s2+= stride;
@@ -3643,12 +3697,47 @@ static int vsse16_c(/*MpegEncContext*/ void *c, uint8_t *s1, uint8_t *s2, int st
 WARPER8_16_SQ(hadamard8_diff8x8_c, hadamard8_diff16_c)
 WARPER8_16_SQ(hadamard8_intra8x8_c, hadamard8_intra16_c)
 WARPER8_16_SQ(dct_sad8x8_c, dct_sad16_c)
+#ifdef CONFIG_GPL
 WARPER8_16_SQ(dct264_sad8x8_c, dct264_sad16_c)
+#endif
 WARPER8_16_SQ(dct_max8x8_c, dct_max16_c)
 WARPER8_16_SQ(quant_psnr8x8_c, quant_psnr16_c)
 WARPER8_16_SQ(rd8x8_c, rd16_c)
 WARPER8_16_SQ(bit8x8_c, bit16_c)
 
+static void vector_fmul_c(float *dst, const float *src, int len){
+    int i;
+    for(i=0; i<len; i++)
+        dst[i] *= src[i];
+}
+
+static void vector_fmul_reverse_c(float *dst, const float *src0, const float *src1, int len){
+    int i;
+    src1 += len-1;
+    for(i=0; i<len; i++)
+        dst[i] = src0[i] * src1[-i];
+}
+
+void ff_vector_fmul_add_add_c(float *dst, const float *src0, const float *src1, const float *src2, int src3, int len, int step){
+    int i;
+    for(i=0; i<len; i++)
+        dst[i*step] = src0[i] * src1[i] + src2[i] + src3;
+}
+
+void ff_float_to_int16_c(int16_t *dst, const float *src, int len){
+    int i;
+    for(i=0; i<len; i++) {
+        int_fast32_t tmp = ((int32_t*)src)[i];
+        if(tmp & 0xf0000){
+            tmp = (0x43c0ffff - tmp)>>31;
+            // is this faster on some gcc/cpu combinations?
+//          if(tmp > 0x43c0ffff) tmp = 0xFFFF;
+//          else                 tmp = 0;
+        }
+        dst[i] = tmp - 0x8000;
+    }
+}
+
 /* XXX: those functions should be suppressed ASAP when all IDCTs are
  converted */
 static void ff_jref_idct_put(uint8_t *dest, int line_size, DCTELEM *block)
@@ -3686,40 +3775,62 @@ static void ff_jref_idct2_add(uint8_t *dest, int line_size, DCTELEM *block)
 
 static void ff_jref_idct1_put(uint8_t *dest, int line_size, DCTELEM *block)
 {
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     dest[0] = cm[(block[0] + 4)>>3];
 }
 static void ff_jref_idct1_add(uint8_t *dest, int line_size, DCTELEM *block)
 {
-    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 
     dest[0] = cm[dest[0] + ((block[0] + 4)>>3)];
 }
 
+static void just_return() { return; }
+
 /* init static data */
 void dsputil_static_init(void)
 {
     int i;
 
-    for(i=0;i<256;i++) cropTbl[i + MAX_NEG_CROP] = i;
+    for(i=0;i<256;i++) ff_cropTbl[i + MAX_NEG_CROP] = i;
     for(i=0;i<MAX_NEG_CROP;i++) {
-        cropTbl[i] = 0;
-        cropTbl[i + MAX_NEG_CROP + 256] = 255;
+        ff_cropTbl[i] = 0;
+        ff_cropTbl[i + MAX_NEG_CROP + 256] = 255;
     }
 
     for(i=0;i<512;i++) {
-        squareTbl[i] = (i - 256) * (i - 256);
+        ff_squareTbl[i] = (i - 256) * (i - 256);
     }
 
     for(i=0; i<64; i++) inv_zigzag_direct16[ff_zigzag_direct[i]]= i+1;
 }
 
+int ff_check_alignment(void){
+    static int did_fail=0;
+    DECLARE_ALIGNED_16(int, aligned);
+
+    if((int)&aligned & 15){
+        if(!did_fail){
+#if defined(HAVE_MMX) || defined(HAVE_ALTIVEC)
+            av_log(NULL, AV_LOG_ERROR,
+                "Compiler did not align stack variables. Libavcodec has been miscompiled\n"
+                "and may be very slow or crash. This is not a bug in libavcodec,\n"
+                "but in the compiler. Do not report crashes to FFmpeg developers.\n");
+#endif
+            did_fail=1;
+        }
+        return -1;
+    }
+    return 0;
+}
 
 void dsputil_init(DSPContext* c, AVCodecContext *avctx)
 {
     int i;
 
+    ff_check_alignment();
+
 #ifdef CONFIG_ENCODERS
     if(avctx->dct_algo==FF_DCT_FASTINT) {
         c->fdct = fdct_ifast;
@@ -3776,6 +3887,8 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
 
     c->h264_idct_add= ff_h264_idct_add_c;
     c->h264_idct8_add= ff_h264_idct8_add_c;
+    c->h264_idct_dc_add= ff_h264_idct_dc_add_c;
+    c->h264_idct8_dc_add= ff_h264_idct8_dc_add_c;
 
     c->get_pixels = get_pixels_c;
     c->diff_pixels = diff_pixels_c;
@@ -3785,7 +3898,7 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->add_pixels8 = add_pixels8_c;
     c->add_pixels4 = add_pixels4_c;
     c->gmc1 = gmc1_c;
-    c->gmc = gmc_c;
+    c->gmc = ff_gmc_c;
     c->clear_blocks = clear_blocks_c;
     c->pix_sum = pix_sum_c;
     c->pix_norm1 = pix_norm1_c;
@@ -3877,6 +3990,7 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     dspfunc(put_h264_qpel, 0, 16);
     dspfunc(put_h264_qpel, 1, 8);
     dspfunc(put_h264_qpel, 2, 4);
+    dspfunc(put_h264_qpel, 3, 2);
     dspfunc(avg_h264_qpel, 0, 16);
     dspfunc(avg_h264_qpel, 1, 8);
     dspfunc(avg_h264_qpel, 2, 4);
@@ -3888,6 +4002,7 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->avg_h264_chroma_pixels_tab[0]= avg_h264_chroma_mc8_c;
     c->avg_h264_chroma_pixels_tab[1]= avg_h264_chroma_mc4_c;
     c->avg_h264_chroma_pixels_tab[2]= avg_h264_chroma_mc2_c;
+    c->put_no_rnd_h264_chroma_pixels_tab[0]= put_no_rnd_h264_chroma_mc8_c;
 
     c->weight_h264_pixels_tab[0]= weight_h264_pixels16x16_c;
     c->weight_h264_pixels_tab[1]= weight_h264_pixels16x8_c;
@@ -3910,6 +4025,16 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->biweight_h264_pixels_tab[8]= biweight_h264_pixels2x4_c;
     c->biweight_h264_pixels_tab[9]= biweight_h264_pixels2x2_c;
 
+#ifdef CONFIG_CAVS_DECODER
+    ff_cavsdsp_init(c,avctx);
+#endif
+#if defined(CONFIG_VC1_DECODER) || defined(CONFIG_WMV3_DECODER)
+    ff_vc1dsp_init(c,avctx);
+#endif
+#if defined(CONFIG_H264_ENCODER)
+    ff_h264dsp_init(c,avctx);
+#endif
+
     c->put_mspel_pixels_tab[0]= put_mspel8_mc00_c;
     c->put_mspel_pixels_tab[1]= put_mspel8_mc10_c;
     c->put_mspel_pixels_tab[2]= put_mspel8_mc20_c;
@@ -3927,7 +4052,9 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->hadamard8_diff[4]= hadamard8_intra16_c;
     SET_CMP_FUNC(dct_sad)
     SET_CMP_FUNC(dct_max)
+#ifdef CONFIG_GPL
     SET_CMP_FUNC(dct264_sad)
+#endif
     c->sad[0]= pix_abs16_c;
     c->sad[1]= pix_abs8_c;
     c->sse[0]= sse16_c;
@@ -3942,10 +4069,12 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->vsse[4]= vsse_intra16_c;
     c->nsse[0]= nsse16_c;
     c->nsse[1]= nsse8_c;
+#ifdef CONFIG_SNOW_ENCODER
     c->w53[0]= w53_16_c;
     c->w53[1]= w53_8_c;
     c->w97[0]= w97_16_c;
     c->w97[1]= w97_8_c;
+#endif
 
     c->add_bytes= add_bytes_c;
     c->diff_bytes= diff_bytes_c;
@@ -3958,6 +4087,7 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->h264_h_loop_filter_chroma= h264_h_loop_filter_chroma_c;
     c->h264_v_loop_filter_chroma_intra= h264_v_loop_filter_chroma_intra_c;
     c->h264_h_loop_filter_chroma_intra= h264_h_loop_filter_chroma_intra_c;
+    c->h264_loop_filter_strength= NULL;
 
     c->h263_h_loop_filter= h263_h_loop_filter_c;
     c->h263_v_loop_filter= h263_v_loop_filter_c;
@@ -3967,6 +4097,30 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->try_8x8basis= try_8x8basis_c;
     c->add_8x8basis= add_8x8basis_c;
 
+#ifdef CONFIG_SNOW_DECODER
+    c->vertical_compose97i = ff_snow_vertical_compose97i;
+    c->horizontal_compose97i = ff_snow_horizontal_compose97i;
+    c->inner_add_yblock = ff_snow_inner_add_yblock;
+#endif
+
+#ifdef CONFIG_VORBIS_DECODER
+    c->vorbis_inverse_coupling = vorbis_inverse_coupling;
+#endif
+    c->vector_fmul = vector_fmul_c;
+    c->vector_fmul_reverse = vector_fmul_reverse_c;
+    c->vector_fmul_add_add = ff_vector_fmul_add_add_c;
+    c->float_to_int16 = ff_float_to_int16_c;
+
+    c->shrink[0]= ff_img_copy_plane;
+    c->shrink[1]= ff_shrink22;
+    c->shrink[2]= ff_shrink44;
+    c->shrink[3]= ff_shrink88;
+
+    c->prefetch= just_return;
+
+    memset(c->put_2tap_qpel_pixels_tab, 0, sizeof(c->put_2tap_qpel_pixels_tab));
+    memset(c->avg_2tap_qpel_pixels_tab, 0, sizeof(c->avg_2tap_qpel_pixels_tab));
+
 #ifdef HAVE_MMX
     dsputil_init_mmx(c, avctx);
 #endif
@@ -3991,6 +4145,16 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
 #ifdef ARCH_SH4
     dsputil_init_sh4(c,avctx);
 #endif
+#ifdef ARCH_BFIN
+    dsputil_init_bfin(c,avctx);
+#endif
+
+    for(i=0; i<64; i++){
+        if(!c->put_2tap_qpel_pixels_tab[0][i])
+            c->put_2tap_qpel_pixels_tab[0][i]= c->put_h264_qpel_pixels_tab[0][i];
+        if(!c->avg_2tap_qpel_pixels_tab[0][i])
+            c->avg_2tap_qpel_pixels_tab[0][i]= c->avg_h264_qpel_pixels_tab[0][i];
+    }
 
     switch(c->idct_permutation_type){
     case FF_NO_IDCT_PERM: