diff --git a/av1/encoder/encodeframe.c b/av1/encoder/encodeframe.c
index 34c033d27ec19beadb71e3ab13bf29d69043f3b7..3764b275ce92541c6e1dea7a1f164a3245a6cc7d 100644
--- a/av1/encoder/encodeframe.c
+++ b/av1/encoder/encodeframe.c
@@ -3660,6 +3660,9 @@ static void rd_pick_partition(const AV1_COMP *const cpi, ThreadData *td,
         int use_activity_masking = 0;
         int64_t daala_dist;
         const int src_stride = x->plane[0].src.stride;
+#if CONFIG_PVQ
+        use_activity_masking = x->daala_enc.use_activity_masking;
+#endif
         daala_dist = av1_daala_dist(x->plane[0].src.buf - 4 * src_stride - 4,
                                     src_stride, x->decoded_8x8, 8, 8, 8, 8, 8,
                                     1, use_activity_masking, x->qindex)
@@ -3836,6 +3839,9 @@ static void rd_pick_partition(const AV1_COMP *const cpi, ThreadData *td,
         int use_activity_masking = 0;
         int64_t daala_dist;
         const int src_stride = x->plane[0].src.stride;
+#if CONFIG_PVQ
+        use_activity_masking = x->daala_enc.use_activity_masking;
+#endif
         daala_dist = av1_daala_dist(x->plane[0].src.buf - 4 * src_stride,
                                     src_stride, x->decoded_8x8, 8, 8, 8, 8, 8,
                                     1, use_activity_masking, x->qindex)
@@ -4008,6 +4014,9 @@ static void rd_pick_partition(const AV1_COMP *const cpi, ThreadData *td,
         int use_activity_masking = 0;
         int64_t daala_dist;
         const int src_stride = x->plane[0].src.stride;
+#if CONFIG_PVQ
+        use_activity_masking = x->daala_enc.use_activity_masking;
+#endif
         daala_dist =
             av1_daala_dist(x->plane[0].src.buf - 4, src_stride, x->decoded_8x8,
                            8, 8, 8, 8, 8, 1, use_activity_masking, x->qindex)
diff --git a/av1/encoder/rdopt.c b/av1/encoder/rdopt.c
index b1e1292fd75e0cd2f8b48dbf920730a2acb2c3fb..a3c084421b58979260c2fe6971eeaaf2259ca57b 100644
--- a/av1/encoder/rdopt.c
+++ b/av1/encoder/rdopt.c
@@ -1821,9 +1821,6 @@ static void daala_dist_sub8x8_txfm_rd(MACROBLOCK *x, BLOCK_SIZE bsize,
   const struct macroblockd_plane *const pd = &xd->plane[0];
   const struct macroblock_plane *const p = &x->plane[0];
   MB_MODE_INFO *const mbmi = &xd->mi[0]->mbmi;
-#if CONFIG_PVQ
-  use_activity_masking = x->daala_enc.use_activity_masking;
-#endif  // CONFIG_PVQ
   const int src_stride = p->src.stride;
   const int dst_stride = pd->dst.stride;
   const uint8_t *src = &p->src.buf[0];
@@ -1838,6 +1835,9 @@ static void daala_dist_sub8x8_txfm_rd(MACROBLOCK *x, BLOCK_SIZE bsize,
   int use_activity_masking = 0;
   unsigned int tmp1, tmp2;
   int qindex = x->qindex;
+#if CONFIG_PVQ
+  use_activity_masking = x->daala_enc.use_activity_masking;
+#endif
 
   assert((bw & 0x07) == 0);
   assert((bh & 0x07) == 0);
@@ -4081,7 +4081,6 @@ void av1_tx_block_rd_b(const AV1_COMP *cpi, MACROBLOCK *x, TX_SIZE tx_size,
   BLOCK_SIZE txm_bsize = txsize_to_bsize[tx_size];
   int bh = block_size_high[txm_bsize];
   int bw = block_size_wide[txm_bsize];
-
   int src_stride = p->src.stride;
   uint8_t *src =
       &p->src.buf[(blk_row * src_stride + blk_col) << tx_size_wide_log2[0]];
@@ -4137,6 +4136,23 @@ void av1_tx_block_rd_b(const AV1_COMP *cpi, MACROBLOCK *x, TX_SIZE tx_size,
   av1_inverse_transform_block(xd, dqcoeff, tx_type, tx_size, rec_buffer,
                               MAX_TX_SIZE, eob);
   if (eob > 0) {
+#if CONFIG_DAALA_DIST
+    if (plane == 0 && (bw < 8 && bh < 8)) {
+      // Save sub8x8 luma decoded pixels
+      // since 8x8 luma decoded pixels are not available for daala-dist
+      // after recursive split of BLOCK_8x8 is done.
+      const int pred_stride = block_size_wide[plane_bsize];
+      const int pred_idx = (blk_row * pred_stride + blk_col)
+                           << tx_size_wide_log2[0];
+      int16_t *decoded = &pd->pred[pred_idx];
+      int i, j;
+
+      // TODO(yushin): HBD support
+      for (j = 0; j < bh; j++)
+        for (i = 0; i < bw; i++)
+          decoded[j * pred_stride + i] = rec_buffer[j * MAX_TX_SIZE + i];
+    }
+#endif  // CONFIG_DAALA_DIST
     tmp = pixel_dist(cpi, x, plane, src, src_stride, rec_buffer, MAX_TX_SIZE,
                      blk_row, blk_col, plane_bsize, txm_bsize);
   }
@@ -4249,7 +4265,9 @@ static void select_tx_block(const AV1_COMP *cpi, MACROBLOCK *x, int blk_row,
     RD_STATS this_rd_stats;
     int this_cost_valid = 1;
     int64_t tmp_rd = 0;
-
+#if CONFIG_DAALA_DIST
+    int sub8x8_eob[4];
+#endif
     sum_rd_stats.rate =
         av1_cost_bit(cpi->common.fc->txfm_partition_prob[ctx], 1);
 
@@ -4265,13 +4283,76 @@ static void select_tx_block(const AV1_COMP *cpi, MACROBLOCK *x, int blk_row,
                       depth + 1, plane_bsize, ta, tl, tx_above, tx_left,
                       &this_rd_stats, ref_best_rd - tmp_rd, &this_cost_valid,
                       rd_stats_stack);
-
+#if CONFIG_DAALA_DIST
+      if (plane == 0 && tx_size == TX_8X8) {
+        sub8x8_eob[i] = p->eobs[block];
+      }
+#endif  // CONFIG_DAALA_DIST
       av1_merge_rd_stats(&sum_rd_stats, &this_rd_stats);
 
       tmp_rd = RDCOST(x->rdmult, sum_rd_stats.rate, sum_rd_stats.dist);
+#if !CONFIG_DAALA_DIST
       if (this_rd < tmp_rd) break;
+#endif
       block += sub_step;
     }
+#if CONFIG_DAALA_DIST
+    if (this_cost_valid && plane == 0 && tx_size == TX_8X8) {
+      const int src_stride = p->src.stride;
+      const int dst_stride = pd->dst.stride;
+
+      const uint8_t *src =
+          &p->src.buf[(blk_row * src_stride + blk_col) << tx_size_wide_log2[0]];
+      const uint8_t *dst =
+          &pd->dst
+               .buf[(blk_row * dst_stride + blk_col) << tx_size_wide_log2[0]];
+
+      int64_t daala_dist;
+      int qindex = x->qindex;
+      const int pred_stride = block_size_wide[plane_bsize];
+      const int pred_idx = (blk_row * pred_stride + blk_col)
+                           << tx_size_wide_log2[0];
+      int16_t *pred = &pd->pred[pred_idx];
+      int j;
+      int qm = OD_HVS_QM;
+      int use_activity_masking = 0;
+      int row, col;
+
+      DECLARE_ALIGNED(16, uint8_t, pred8[8 * 8]);
+
+#if CONFIG_PVQ
+      use_activity_masking = x->daala_enc.use_activity_masking;
+#endif
+      daala_dist = av1_daala_dist(src, src_stride, dst, dst_stride, 8, 8, 8, 8,
+                                  qm, use_activity_masking, qindex) *
+                   16;
+      sum_rd_stats.sse = daala_dist;
+
+      for (row = 0; row < 2; ++row) {
+        for (col = 0; col < 2; ++col) {
+          int idx = row * 2 + col;
+          int eob = sub8x8_eob[idx];
+
+          if (eob > 0) {
+            for (j = 0; j < 4; j++)
+              for (i = 0; i < 4; i++)
+                pred8[(row * 4 + j) * 8 + 4 * col + i] =
+                    pred[(row * 4 + j) * pred_stride + 4 * col + i];
+          } else {
+            for (j = 0; j < 4; j++)
+              for (i = 0; i < 4; i++)
+                pred8[(row * 4 + j) * 8 + 4 * col + i] =
+                    dst[(row * 4 + j) * dst_stride + 4 * col + i];
+          }
+        }
+      }
+      daala_dist = av1_daala_dist(src, src_stride, pred8, 8, 8, 8, 8, 8, qm,
+                                  use_activity_masking, qindex) *
+                   16;
+      sum_rd_stats.dist = daala_dist;
+      tmp_rd = RDCOST(x->rdmult, sum_rd_stats.rate, sum_rd_stats.dist);
+    }
+#endif  // CONFIG_DAALA_DIST
     if (this_cost_valid) sum_rd = tmp_rd;
   }