Take out skip_recode speed feature
The assumption doesn't hold true in the current codebase. Remove this speed feature to simplify the codebase. Change-Id: I9b69f484c9b7cd612b825047cc5b2fce63ee0af7
This commit is contained in:
@@ -1707,7 +1707,6 @@ static void rd_pick_sb_modes(VP10_COMP *cpi,
|
||||
ctx->is_coded = 0;
|
||||
ctx->skippable = 0;
|
||||
ctx->pred_pixel_ready = 0;
|
||||
x->skip_recode = 0;
|
||||
|
||||
// Set to zero to make sure we do not use the previous encoded frame stats
|
||||
mbmi->skip = 0;
|
||||
@@ -2263,8 +2262,6 @@ static void encode_sb(VP10_COMP *cpi, ThreadData *td,
|
||||
|
||||
set_offsets(cpi, tile, x, mi_row, mi_col, bsize);
|
||||
if (!x->skip) {
|
||||
// TODO(geza.lore): Investigate if this can be relaxed
|
||||
x->skip_recode = 0;
|
||||
memset(x->skip_txfm, 0, sizeof(x->skip_txfm));
|
||||
|
||||
x->skip_optimize = 0;
|
||||
@@ -4970,10 +4967,7 @@ static void encode_superblock(VP10_COMP *cpi, ThreadData *td,
|
||||
const int mi_width = num_8x8_blocks_wide_lookup[bsize];
|
||||
const int mi_height = num_8x8_blocks_high_lookup[bsize];
|
||||
|
||||
x->skip_recode = 0;
|
||||
|
||||
if (!x->skip_recode)
|
||||
memset(x->skip_txfm, 0, sizeof(x->skip_txfm));
|
||||
memset(x->skip_txfm, 0, sizeof(x->skip_txfm));
|
||||
|
||||
x->skip_optimize = ctx->is_coded;
|
||||
ctx->is_coded = 1;
|
||||
|
||||
Reference in New Issue
Block a user