* preAlpha DV encoding support -- there's still a truckload

[ffmpeg] / libavcodec / dsputil.c

diff --git a/libavcodec/dsputil.c b/libavcodec/dsputil.c
index 1f9acf8cb7e05fd582959a9cbf0bc184f790605a..9e7cca734f99189f8b825b4516f85c014d431887 100644 (file)
--- a/libavcodec/dsputil.c
+++ b/libavcodec/dsputil.c
@@ -187,6 +187,23 @@ static int pix_norm1_c(uint8_t * pix, int line_size)
     return s;
 }
 
+static void bswap_buf(uint32_t *dst, uint32_t *src, int w){
+    int i;
+    
+    for(i=0; i+8<=w; i+=8){
+        dst[i+0]= bswap_32(src[i+0]);
+        dst[i+1]= bswap_32(src[i+1]);
+        dst[i+2]= bswap_32(src[i+2]);
+        dst[i+3]= bswap_32(src[i+3]);
+        dst[i+4]= bswap_32(src[i+4]);
+        dst[i+5]= bswap_32(src[i+5]);
+        dst[i+6]= bswap_32(src[i+6]);
+        dst[i+7]= bswap_32(src[i+7]);
+    }
+    for(;i<w; i++){
+        dst[i+0]= bswap_32(src[i+0]);
+    }
+}
 
 static int sse8_c(void *v, uint8_t * pix1, uint8_t * pix2, int line_size)
 {
@@ -982,7 +999,7 @@ static inline void put_tpel_pixels_mc12_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-       dst[j] = (2731*(3*src[j] + 4*src[j+1] + 2*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15;
+       dst[j] = (2731*(3*src[j] + 2*src[j+1] + 4*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15;
       }
       src += stride;
       dst += stride;
@@ -1004,7 +1021,7 @@ static inline void put_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src, int
     int i,j;
     for (i=0; i < height; i++) {
       for (j=0; j < width; j++) {
-       dst[j] = (2731*(3*src[j] + 2*src[j+1] + 4*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15;
+       dst[j] = (2731*(3*src[j] + 4*src[j+1] + 2*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15;
       }
       src += stride;
       dst += stride;
@@ -1021,6 +1038,103 @@ static inline void put_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src, int
       dst += stride;
     }
 }
+
+static inline void avg_tpel_pixels_mc00_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+    switch(width){
+    case 2: avg_pixels2_c (dst, src, stride, height); break;
+    case 4: avg_pixels4_c (dst, src, stride, height); break;
+    case 8: avg_pixels8_c (dst, src, stride, height); break;
+    case 16:avg_pixels16_c(dst, src, stride, height); break;
+    }
+}
+
+static inline void avg_tpel_pixels_mc10_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+    int i,j;
+    for (i=0; i < height; i++) {
+      for (j=0; j < width; j++) {
+       dst[j] = (dst[j] + ((683*(2*src[j] + src[j+1] + 1)) >> 11) + 1) >> 1;
+      }
+      src += stride;
+      dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc20_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+    int i,j;
+    for (i=0; i < height; i++) {
+      for (j=0; j < width; j++) {
+       dst[j] = (dst[j] + ((683*(src[j] + 2*src[j+1] + 1)) >> 11) + 1) >> 1;
+      }
+      src += stride;
+      dst += stride;
+    }
+}
+    
+static inline void avg_tpel_pixels_mc01_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+    int i,j;
+    for (i=0; i < height; i++) {
+      for (j=0; j < width; j++) {
+       dst[j] = (dst[j] + ((683*(2*src[j] + src[j+stride] + 1)) >> 11) + 1) >> 1;
+      }
+      src += stride;
+      dst += stride;
+    }
+}
+    
+static inline void avg_tpel_pixels_mc11_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+    int i,j;
+    for (i=0; i < height; i++) {
+      for (j=0; j < width; j++) {
+       dst[j] = (dst[j] + ((2731*(4*src[j] + 3*src[j+1] + 3*src[j+stride] + 2*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+      }
+      src += stride;
+      dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc12_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+    int i,j;
+    for (i=0; i < height; i++) {
+      for (j=0; j < width; j++) {
+       dst[j] = (dst[j] + ((2731*(3*src[j] + 2*src[j+1] + 4*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+      }
+      src += stride;
+      dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc02_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+    int i,j;
+    for (i=0; i < height; i++) {
+      for (j=0; j < width; j++) {
+       dst[j] = (dst[j] + ((683*(src[j] + 2*src[j+stride] + 1)) >> 11) + 1) >> 1;
+      }
+      src += stride;
+      dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc21_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+    int i,j;
+    for (i=0; i < height; i++) {
+      for (j=0; j < width; j++) {
+       dst[j] = (dst[j] + ((2731*(3*src[j] + 4*src[j+1] + 2*src[j+stride] + 3*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+      }
+      src += stride;
+      dst += stride;
+    }
+}
+
+static inline void avg_tpel_pixels_mc22_c(uint8_t *dst, const uint8_t *src, int stride, int width, int height){
+    int i,j;
+    for (i=0; i < height; i++) {
+      for (j=0; j < width; j++) {
+       dst[j] = (dst[j] + ((2731*(2*src[j] + 3*src[j+1] + 3*src[j+stride] + 4*src[j+stride+1] + 6)) >> 15) + 1) >> 1;
+      }
+      src += stride;
+      dst += stride;
+    }
+}
 #if 0
 #define TPEL_WIDTH(width)\
 static void put_tpel_pixels ## width ## _mc00_c(uint8_t *dst, const uint8_t *src, int stride, int height){\
@@ -2746,10 +2860,12 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     if(avctx->idct_algo==FF_IDCT_INT){
         c->idct_put= ff_jref_idct_put;
         c->idct_add= ff_jref_idct_add;
+        c->idct    = j_rev_dct;
         c->idct_permutation_type= FF_LIBMPEG2_IDCT_PERM;
     }else{ //accurate/default
         c->idct_put= simple_idct_put;
         c->idct_add= simple_idct_add;
+        c->idct    = simple_idct;
         c->idct_permutation_type= FF_NO_IDCT_PERM;
     }
 
@@ -2792,6 +2908,8 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     dspfunc(avg_no_rnd, 0, 16);
     dspfunc(avg, 1, 8);
     dspfunc(avg_no_rnd, 1, 8);
+    dspfunc(avg, 2, 4);
+    dspfunc(avg, 3, 2);
 #undef dspfunc
 
     c->put_tpel_pixels_tab[ 0] = put_tpel_pixels_mc00_c;
@@ -2804,6 +2922,16 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
     c->put_tpel_pixels_tab[ 9] = put_tpel_pixels_mc12_c;
     c->put_tpel_pixels_tab[10] = put_tpel_pixels_mc22_c;
 
+    c->avg_tpel_pixels_tab[ 0] = avg_tpel_pixels_mc00_c;
+    c->avg_tpel_pixels_tab[ 1] = avg_tpel_pixels_mc10_c;
+    c->avg_tpel_pixels_tab[ 2] = avg_tpel_pixels_mc20_c;
+    c->avg_tpel_pixels_tab[ 4] = avg_tpel_pixels_mc01_c;
+    c->avg_tpel_pixels_tab[ 5] = avg_tpel_pixels_mc11_c;
+    c->avg_tpel_pixels_tab[ 6] = avg_tpel_pixels_mc21_c;
+    c->avg_tpel_pixels_tab[ 8] = avg_tpel_pixels_mc02_c;
+    c->avg_tpel_pixels_tab[ 9] = avg_tpel_pixels_mc12_c;
+    c->avg_tpel_pixels_tab[10] = avg_tpel_pixels_mc22_c;
+
 #define dspfunc(PFX, IDX, NUM) \
     c->PFX ## _pixels_tab[IDX][ 0] = PFX ## NUM ## _mc00_c; \
     c->PFX ## _pixels_tab[IDX][ 1] = PFX ## NUM ## _mc10_c; \
@@ -2879,6 +3007,7 @@ void dsputil_init(DSPContext* c, AVCodecContext *avctx)
         
     c->add_bytes= add_bytes_c;
     c->diff_bytes= diff_bytes_c;
+    c->bswap_buf= bswap_buf;
 
 #ifdef HAVE_MMX
     dsputil_init_mmx(c, avctx);