h264: special case dc-only idct. ~1% faster overall

Originally committed as revision 4971 to svn://svn.ffmpeg.org/ffmpeg/trunk

1 files changed, 25 insertions, 0 deletions

diff --git a/libavcodec/h264idct.c b/libavcodec/h264idct.c
index a4ddf1d51d..3e44385d5e 100644
--- a/libavcodec/h264idct.c
+++ b/libavcodec/h264idct.c
@@ -139,3 +139,28 @@ void ff_h264_idct8_add_c(uint8_t *dst, DCTELEM *block, int stride){
         dst[i + 7*stride] = cm[ dst[i + 7*stride] + ((b0 - b7) >> 6) ];
     }
 }
+
+// assumes all AC coefs are 0
+void ff_h264_idct_dc_add_c(uint8_t *dst, DCTELEM *block, int stride){
+    int i, j;
+    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    int dc = (block[0] + 32) >> 6;
+    for( j = 0; j < 4; j++ )
+    {
+        for( i = 0; i < 4; i++ )
+            dst[i] = cm[ dst[i] + dc ];
+        dst += stride;
+    }
+}
+
+void ff_h264_idct8_dc_add_c(uint8_t *dst, DCTELEM *block, int stride){
+    int i, j;
+    uint8_t *cm = cropTbl + MAX_NEG_CROP;
+    int dc = (block[0] + 32) >> 6;
+    for( j = 0; j < 8; j++ )
+    {
+        for( i = 0; i < 8; i++ )
+            dst[i] = cm[ dst[i] + dc ];
+        dst += stride;
+    }
+}