Changed segmentation check order

In SPLITMV, the 8x8 segment will be checked first.  If the 8x8 rd
is better than the best, we check the other segments.  Otherwise
bail.  Adjustments to the thresh_mult were necessary to make
up for the initial quality loss.
The performance improved by 20% (average) for good quality,
speed 0 and speed 1, while the overall quality remained the same.

Change-Id: I717aef401323c8a254fba3e9777d2a316c774cc3

vp8/encoder/onyx_if.c

@@ -683,6 +683,32 @@ void vp8_set_speed_features(VP8_COMP *cpi)
         sf->thresh_mult[THR_NEARG    ] = 1000;
         sf->thresh_mult[THR_NEARA    ] = 1000;
 
+#if 1
+        sf->thresh_mult[THR_ZEROMV   ] = 0;
+        sf->thresh_mult[THR_ZEROG    ] = 0;
+        sf->thresh_mult[THR_ZEROA    ] = 0;
+        sf->thresh_mult[THR_NEARESTMV] = 0;
+        sf->thresh_mult[THR_NEARESTG ] = 0;
+        sf->thresh_mult[THR_NEARESTA ] = 0;
+        sf->thresh_mult[THR_NEARMV   ] = 0;
+        sf->thresh_mult[THR_NEARG    ] = 0;
+        sf->thresh_mult[THR_NEARA    ] = 0;
+
+//        sf->thresh_mult[THR_DC       ] = 0;
+
+//        sf->thresh_mult[THR_V_PRED   ] = 1000;
+//        sf->thresh_mult[THR_H_PRED   ] = 1000;
+//        sf->thresh_mult[THR_B_PRED   ] = 2000;
+//        sf->thresh_mult[THR_TM       ] = 1000;
+
+        sf->thresh_mult[THR_NEWMV    ] = 1000;
+        sf->thresh_mult[THR_NEWG     ] = 1000;
+        sf->thresh_mult[THR_NEWA     ] = 1000;
+
+        sf->thresh_mult[THR_SPLITMV  ] = 1700;
+        sf->thresh_mult[THR_SPLITG   ] = 4500;
+        sf->thresh_mult[THR_SPLITA   ] = 4500;
+#else
         sf->thresh_mult[THR_NEWMV    ] = 1500;
         sf->thresh_mult[THR_NEWG     ] = 1500;
         sf->thresh_mult[THR_NEWA     ] = 1500;
@@ -690,7 +716,7 @@ void vp8_set_speed_features(VP8_COMP *cpi)
         sf->thresh_mult[THR_SPLITMV  ] = 5000;
         sf->thresh_mult[THR_SPLITG   ] = 10000;
         sf->thresh_mult[THR_SPLITA   ] = 10000;
-
+#endif
         sf->full_freq[0] = 15;
         sf->full_freq[1] = 31;
 
@@ -731,7 +757,7 @@ void vp8_set_speed_features(VP8_COMP *cpi)
 
             cpi->mode_check_freq[THR_SPLITG] = 4;
             cpi->mode_check_freq[THR_SPLITA] = 4;
-            cpi->mode_check_freq[THR_SPLITMV] = 2;
+            cpi->mode_check_freq[THR_SPLITMV] = 0;
 
             sf->thresh_mult[THR_TM       ] = 1500;
             sf->thresh_mult[THR_V_PRED   ] = 1500;

vp8/encoder/rdopt.c

@@ -1038,7 +1038,7 @@ typedef struct
   int d;
   int segment_yrate;
   B_PREDICTION_MODE modes[16];
-  MV mvs[16];
+  int_mv mvs[16];
   unsigned char eobs[16];
 
   int mvthresh;
@@ -1276,7 +1276,7 @@ void vp8_rd_check_segment(VP8_COMP *cpi, MACROBLOCK *x, BEST_SEG_INFO *bsi,
         {
             BLOCKD *bd = &x->e_mbd.block[i];
 
-            bsi->mvs[i] = bd->bmi.mv.as_mv;
+            bsi->mvs[i].as_mv = bd->bmi.mv.as_mv;
             bsi->modes[i] = bd->bmi.mode;
             bsi->eobs[i] = bd->eob;
         }
@@ -1305,19 +1305,32 @@ static int vp8_rd_pick_best_mbsegmentation(VP8_COMP *cpi, MACROBLOCK *x,
     {
         bsi.modes[i] = ZERO4X4;
     }
-
+    if(cpi->compressor_speed == 0)
-    /* original */
+    {
-    vp8_rd_check_segment(cpi, x, &bsi, 0);
+        /* for now, we will keep the original segmentation order
-    vp8_rd_check_segment(cpi, x, &bsi, 1);
+           when in best quality mode */
-    vp8_rd_check_segment(cpi, x, &bsi, 2);
+        vp8_rd_check_segment(cpi, x, &bsi, BLOCK_16X8);
-    vp8_rd_check_segment(cpi, x, &bsi, 3);
+        vp8_rd_check_segment(cpi, x, &bsi, BLOCK_8X16);
+        vp8_rd_check_segment(cpi, x, &bsi, BLOCK_8X8);
+        vp8_rd_check_segment(cpi, x, &bsi, BLOCK_4X4);
+    }
+    else
+    {
+        vp8_rd_check_segment(cpi, x, &bsi, BLOCK_8X8);
+        if (bsi.segment_rd < best_rd)
+        {
+          vp8_rd_check_segment(cpi, x, &bsi, BLOCK_8X16);
+          vp8_rd_check_segment(cpi, x, &bsi, BLOCK_16X8);
+          vp8_rd_check_segment(cpi, x, &bsi, BLOCK_4X4);
+        }
+    }
 
     /* set it to the best */
     for (i = 0; i < 16; i++)
     {
         BLOCKD *bd = &x->e_mbd.block[i];
 
-        bd->bmi.mv.as_mv = bsi.mvs[i];
+        bd->bmi.mv.as_mv = bsi.mvs[i].as_mv;
         bd->bmi.mode = bsi.modes[i];
         bd->eob = bsi.eobs[i];
     }