Replace copy_partitioning use case with choose_partitioning

This commit replaces the use of copy_partitioning with
choose_partitioning based on the sse of subsamped pixels, which
provides significantly better coding performance and runs at
similar speed, as compared to copy_partitioning. It improves rtc
speed 5 coding performance by 3%.

Change-Id: I52d3682a12dce0147f5e52383a594fc242ca3228

diff --git a/vp9/encoder/vp9_encodeframe.c b/vp9/encoder/vp9_encodeframe.c
index b43ff974758e44bd97859ba0563f617a0817d6f3..193a3a5594df32db71afd075bb05191de226d8c8 100644 (file)
--- a/vp9/encoder/vp9_encodeframe.c
+++ b/vp9/encoder/vp9_encodeframe.c
@@ -3158,7 +3158,6 @@ static void encode_nonrd_sb_row(VP9_COMP *cpi, const TileInfo *const tile,
     int64_t dummy_dist = 0;
     const int idx_str = cm->mi_stride * mi_row + mi_col;
     MODE_INFO *mi = cm->mi + idx_str;
-    MODE_INFO *prev_mi = (cm->prev_mip + cm->mi_stride + 1 + idx_str)->src_mi;
     BLOCK_SIZE bsize;
     x->in_static_area = 0;
     x->source_variance = UINT_MAX;
@@ -3196,7 +3195,7 @@ static void encode_nonrd_sb_row(VP9_COMP *cpi, const TileInfo *const tile,
                                &dummy_rate, &dummy_dist, 1, INT64_MAX,
                                cpi->pc_root);
         } else {
-          copy_partitioning(cm, mi, prev_mi);
+          choose_partitioning(cpi, tile, mi_row, mi_col);
           nonrd_use_partition(cpi, tile, mi, tp, mi_row, mi_col,
                               BLOCK_64X64, 1, &dummy_rate, &dummy_dist,
                               cpi->pc_root);

diff --git a/vp9/encoder/vp9_speed_features.c b/vp9/encoder/vp9_speed_features.c
index 8788be645eebefc7e182e9f78ed6ea26958bc936..bec77d71fe8aea10e74bdea60e4421bfb7b93252 100644 (file)
--- a/vp9/encoder/vp9_speed_features.c
+++ b/vp9/encoder/vp9_speed_features.c
@@ -271,6 +271,10 @@ static void set_rt_speed_feature(VP9_COMP *cpi, SPEED_FEATURES *sf,
     sf->partition_search_type = REFERENCE_PARTITION;
     sf->use_nonrd_pick_mode = 1;
     sf->allow_skip_recode = 0;
+    sf->inter_mode_mask[BLOCK_32X32] = INTER_NEAREST_NEW_ZERO;
+    sf->inter_mode_mask[BLOCK_32X64] = INTER_NEAREST_NEW_ZERO;
+    sf->inter_mode_mask[BLOCK_64X32] = INTER_NEAREST_NEW_ZERO;
+    sf->inter_mode_mask[BLOCK_64X64] = INTER_NEAREST_NEW_ZERO;
   }
 
   if (speed >= 6) {
@@ -285,10 +289,6 @@ static void set_rt_speed_feature(VP9_COMP *cpi, SPEED_FEATURES *sf,
     sf->partition_search_type = VAR_BASED_PARTITION;
     sf->search_type_check_frequency = 50;
     sf->mv.search_method = NSTEP;
-    sf->inter_mode_mask[BLOCK_32X32] = INTER_NEAREST_NEW_ZERO;
-    sf->inter_mode_mask[BLOCK_32X64] = INTER_NEAREST_NEW_ZERO;
-    sf->inter_mode_mask[BLOCK_64X32] = INTER_NEAREST_NEW_ZERO;
-    sf->inter_mode_mask[BLOCK_64X64] = INTER_NEAREST_NEW_ZERO;
 
     sf->tx_size_search_method = is_keyframe ? USE_LARGESTALL : USE_TX_8X8;