Reduce dqcoeff array size in decoder

The decoding process handles detokenization and reconstruction per
transform block sequentially. There is no need to offset the dqcoeff
buffer according to the transform block index. This allows to
reduce the memory spill and improve cache performance.

Change-Id: Ibb8bfe532a7a08fcabaf6d42cbec1e986901d32d

2 files changed, 2 insertions, 2 deletions

diff --git a/vp9/decoder/vp9_decodeframe.c b/vp9/decoder/vp9_decodeframe.c
index e81fa0478..6ddfcbb4b 100644
--- a/vp9/decoder/vp9_decodeframe.c
+++ b/vp9/decoder/vp9_decodeframe.c
@@ -188,7 +188,7 @@ static void inverse_transform_block(MACROBLOCKD* xd, int plane, int block,
   struct macroblockd_plane *const pd = &xd->plane[plane];
   if (eob > 0) {
     TX_TYPE tx_type = DCT_DCT;
-    tran_low_t *const dqcoeff = BLOCK_OFFSET(pd->dqcoeff, block);
+    tran_low_t *const dqcoeff = pd->dqcoeff;
 #if CONFIG_VP9_HIGHBITDEPTH
     if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
       if (xd->lossless) {
diff --git a/vp9/decoder/vp9_detokenize.c b/vp9/decoder/vp9_detokenize.c
index 3304e64b2..2e2605903 100644
--- a/vp9/decoder/vp9_detokenize.c
+++ b/vp9/decoder/vp9_detokenize.c
@@ -217,7 +217,7 @@ int vp9_decode_block_tokens(MACROBLOCKD *xd,
                                                pd->left_context + y);
   const scan_order *so = get_scan(xd, tx_size, pd->plane_type, block);
   const int eob = decode_coefs(xd, pd->plane_type,
-                               BLOCK_OFFSET(pd->dqcoeff, block), tx_size,
+                               pd->dqcoeff, tx_size,
                                dequant, ctx, so->scan, so->neighbors, r);
   vp9_set_contexts(xd, pd, plane_bsize, tx_size, eob > 0, x, y);
   return eob;