diff --git a/vp8/encoder/block.h b/vp8/encoder/block.h
index 82a93cd..0d14b54 100644
--- a/vp8/encoder/block.h
+++ b/vp8/encoder/block.h
@@ -65,7 +65,9 @@
 {
     DECLARE_ALIGNED(16, short, src_diff[400]);       // 16x16 Y 8x8 U 8x8 V 4x4 2nd Y
     DECLARE_ALIGNED(16, short, coeff[400]);     // 16x16 Y 8x8 U 8x8 V 4x4 2nd Y
+    DECLARE_ALIGNED(16, unsigned char, thismb[256]);
 
+    unsigned char *thismb_ptr;
     // 16 Y blocks, 4 U blocks, 4 V blocks, 1 DC 2nd order block each with 16 entries
     BLOCK block[25];
 
diff --git a/vp8/encoder/encodeframe.c b/vp8/encoder/encodeframe.c
index 132e508..0efb9f6 100644
--- a/vp8/encoder/encodeframe.c
+++ b/vp8/encoder/encodeframe.c
@@ -292,6 +292,9 @@
             xd->left_available = (mb_col != 0);
             recon_yoffset += 16;
 #endif
+            //Copy current mb to a buffer
+            RECON_INVOKE(&xd->rtcd->recon, copy16x16)(x->src.y_buffer, x->src.y_stride, x->thismb, 16);
+
             // measure activity
             mb_activity = mb_activity_measure( cpi, x, mb_row, mb_col );
 
@@ -432,6 +435,9 @@
         x->rddiv = cpi->RDDIV;
         x->rdmult = cpi->RDMULT;
 
+        //Copy current mb to a buffer
+        RECON_INVOKE(&xd->rtcd->recon, copy16x16)(x->src.y_buffer, x->src.y_stride, x->thismb, 16);
+
 #if CONFIG_MULTITHREAD
         if ((cpi->b_multi_threaded != 0) && (mb_row != 0))
         {
@@ -1015,14 +1021,18 @@
     vp8_build_block_doffsets(&x->e_mbd);
 
     // y blocks
+    x->thismb_ptr = &x->thismb[0];
     for (br = 0; br < 4; br++)
     {
         for (bc = 0; bc < 4; bc++)
         {
             BLOCK *this_block = &x->block[block];
-            this_block->base_src = &x->src.y_buffer;
-            this_block->src_stride = x->src.y_stride;
-            this_block->src = 4 * br * this_block->src_stride + 4 * bc;
+            //this_block->base_src = &x->src.y_buffer;
+            //this_block->src_stride = x->src.y_stride;
+            //this_block->src = 4 * br * this_block->src_stride + 4 * bc;
+            this_block->base_src = &x->thismb_ptr;
+            this_block->src_stride = 16;
+            this_block->src = 4 * br * 16 + 4 * bc;
             ++block;
         }
     }
diff --git a/vp8/encoder/encodeintra.c b/vp8/encoder/encodeintra.c
index f076bbb..59db025 100644
--- a/vp8/encoder/encodeintra.c
+++ b/vp8/encoder/encodeintra.c
@@ -91,11 +91,11 @@
 
 void vp8_encode_intra16x16mby(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
 {
-    int b;
+    BLOCK *b = &x->block[0];
 
     RECON_INVOKE(&rtcd->common->recon, build_intra_predictors_mby)(&x->e_mbd);
 
-    ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, x->src.y_buffer, x->e_mbd.predictor, x->src.y_stride);
+    ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, *(b->base_src), x->e_mbd.predictor, b->src_stride);
 
     vp8_transform_intra_mby(x);
 
diff --git a/vp8/encoder/encodemb.c b/vp8/encoder/encodemb.c
index fd42ee4..3ed16b6 100644
--- a/vp8/encoder/encodemb.c
+++ b/vp8/encoder/encodemb.c
@@ -101,7 +101,9 @@
 
 static void vp8_subtract_mb(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
 {
-    ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, x->src.y_buffer, x->e_mbd.predictor, x->src.y_stride);
+    BLOCK *b = &x->block[0];
+
+    ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, *(b->base_src), x->e_mbd.predictor, b->src_stride);
     ENCODEMB_INVOKE(&rtcd->encodemb, submbuv)(x->src_diff, x->src.u_buffer, x->src.v_buffer, x->e_mbd.predictor, x->src.uv_stride);
 }
 
@@ -598,9 +600,11 @@
 /* this funciton is used by first pass only */
 void vp8_encode_inter16x16y(const VP8_ENCODER_RTCD *rtcd, MACROBLOCK *x)
 {
+    BLOCK *b = &x->block[0];
+
     vp8_build_inter16x16_predictors_mby(&x->e_mbd);
 
-    ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, x->src.y_buffer, x->e_mbd.predictor, x->src.y_stride);
+    ENCODEMB_INVOKE(&rtcd->encodemb, submby)(x->src_diff, *(b->base_src), x->e_mbd.predictor, b->src_stride);
 
     transform_mby(x);
 
diff --git a/vp8/encoder/ethreading.c b/vp8/encoder/ethreading.c
index 420ed8e..1a37f03 100644
--- a/vp8/encoder/ethreading.c
+++ b/vp8/encoder/ethreading.c
@@ -148,6 +148,9 @@
                     x->rddiv = cpi->RDDIV;
                     x->rdmult = cpi->RDMULT;
 
+                    //Copy current mb to a buffer
+                    RECON_INVOKE(&xd->rtcd->recon, copy16x16)(x->src.y_buffer, x->src.y_stride, x->thismb, 16);
+
                     if (cpi->oxcf.tuning == VP8_TUNE_SSIM)
                         vp8_activity_masking(cpi, x);
 
diff --git a/vp8/encoder/firstpass.c b/vp8/encoder/firstpass.c
index 218fc06..4fc6a8a 100644
--- a/vp8/encoder/firstpass.c
+++ b/vp8/encoder/firstpass.c
@@ -579,6 +579,9 @@
             xd->dst.v_buffer = new_yv12->v_buffer + recon_uvoffset;
             xd->left_available = (mb_col != 0);
 
+            //Copy current mb to a buffer
+            RECON_INVOKE(&xd->rtcd->recon, copy16x16)(x->src.y_buffer, x->src.y_stride, x->thismb, 16);
+
             // do intra 16x16 prediction
             this_error = vp8_encode_intra(cpi, x, use_dc_pred);
 
diff --git a/vp8/encoder/pickinter.c b/vp8/encoder/pickinter.c
index 6c73511..fac068f 100644
--- a/vp8/encoder/pickinter.c
+++ b/vp8/encoder/pickinter.c
@@ -610,7 +610,7 @@
                 rate2 += rate;
                 distortion2 = VARIANCE_INVOKE
                                 (&cpi->rtcd.variance, var16x16)(
-                                    x->src.y_buffer, x->src.y_stride,
+                                    *(b->base_src), b->src_stride,
                                     x->e_mbd.predictor, 16, &sse);
                 this_rd = RDCOST(x->rdmult, x->rddiv, rate2, distortion2);
 
@@ -635,7 +635,7 @@
             RECON_INVOKE(&cpi->common.rtcd.recon, build_intra_predictors_mby)
                 (&x->e_mbd);
             distortion2 = VARIANCE_INVOKE(&cpi->rtcd.variance, var16x16)
-                                          (x->src.y_buffer, x->src.y_stride,
+                                          (*(b->base_src), b->src_stride,
                                           x->e_mbd.predictor, 16, &sse);
             rate2 += x->mbmode_cost[x->e_mbd.frame_type][x->e_mbd.mode_info_context->mbmi.mode];
             this_rd = RDCOST(x->rdmult, x->rddiv, rate2, distortion2);
@@ -908,6 +908,7 @@
     MB_PREDICTION_MODE mode, best_mode = DC_PRED;
     int this_rd;
     unsigned int sse;
+    BLOCK *b = &x->block[0];
 
     x->e_mbd.mode_info_context->mbmi.ref_frame = INTRA_FRAME;
 
@@ -919,7 +920,7 @@
         RECON_INVOKE(&cpi->common.rtcd.recon, build_intra_predictors_mby)
             (&x->e_mbd);
         distortion = VARIANCE_INVOKE(&cpi->rtcd.variance, var16x16)
-            (x->src.y_buffer, x->src.y_stride, x->e_mbd.predictor, 16, &sse);
+            (*(b->base_src), b->src_stride, x->e_mbd.predictor, 16, &sse);
         rate = x->mbmode_cost[x->e_mbd.frame_type][mode];
         this_rd = RDCOST(x->rdmult, x->rddiv, rate, distortion);
 
diff --git a/vp8/encoder/rdopt.c b/vp8/encoder/rdopt.c
index fa5bb07..355542e 100644
--- a/vp8/encoder/rdopt.c
+++ b/vp8/encoder/rdopt.c
@@ -543,8 +543,8 @@
     BLOCK *beptr;
     int d;
 
-    ENCODEMB_INVOKE(rtcd, submby)( mb->src_diff, mb->src.y_buffer,
-                                   mb->e_mbd.predictor, mb->src.y_stride );
+    ENCODEMB_INVOKE(rtcd, submby)( mb->src_diff, *(mb->block[0].base_src),
+                                   mb->e_mbd.predictor, mb->block[0].src_stride );
 
     // Fdct and building the 2nd order block
     for (beptr = mb->block; beptr < mb->block + 16; beptr += 2)
@@ -1637,6 +1637,8 @@
 {
 
     int near_sad[8] = {0}; // 0-cf above, 1-cf left, 2-cf aboveleft, 3-lf current, 4-lf above, 5-lf left, 6-lf right, 7-lf below
+    BLOCK *b = &x->block[0];
+    unsigned char *src_y_ptr = *(b->base_src);
 
     //calculate sad for current frame 3 nearby MBs.
     if( xd->mb_to_top_edge==0 && xd->mb_to_left_edge ==0)
@@ -1645,16 +1647,16 @@
     }else if(xd->mb_to_top_edge==0)
     {   //only has left MB for sad calculation.
         near_sad[0] = near_sad[2] = INT_MAX;
-        near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff);
+        near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff);
     }else if(xd->mb_to_left_edge ==0)
     {   //only has left MB for sad calculation.
         near_sad[1] = near_sad[2] = INT_MAX;
-        near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff);
+        near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff);
     }else
     {
-        near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff);
-        near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff);
-        near_sad[2] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, xd->dst.y_buffer - xd->dst.y_stride *16 -16,xd->dst.y_stride, 0x7fffffff);
+        near_sad[0] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16,xd->dst.y_stride, 0x7fffffff);
+        near_sad[1] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - 16,xd->dst.y_stride, 0x7fffffff);
+        near_sad[2] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, xd->dst.y_buffer - xd->dst.y_stride *16 -16,xd->dst.y_stride, 0x7fffffff);
     }
 
     if(cpi->common.last_frame_type != KEY_FRAME)
@@ -1669,14 +1671,14 @@
         if(xd->mb_to_bottom_edge==0) near_sad[7] = INT_MAX;
 
         if(near_sad[4] != INT_MAX)
-            near_sad[4] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer - pre_y_stride *16, pre_y_stride, 0x7fffffff);
+            near_sad[4] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer - pre_y_stride *16, pre_y_stride, 0x7fffffff);
         if(near_sad[5] != INT_MAX)
-            near_sad[5] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer - 16, pre_y_stride, 0x7fffffff);
-        near_sad[3] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer, pre_y_stride, 0x7fffffff);
+            near_sad[5] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer - 16, pre_y_stride, 0x7fffffff);
+        near_sad[3] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer, pre_y_stride, 0x7fffffff);
         if(near_sad[6] != INT_MAX)
-            near_sad[6] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer + 16, pre_y_stride, 0x7fffffff);
+            near_sad[6] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer + 16, pre_y_stride, 0x7fffffff);
         if(near_sad[7] != INT_MAX)
-            near_sad[7] = cpi->fn_ptr[BLOCK_16X16].sdf(x->src.y_buffer, x->src.y_stride, pre_y_buffer + pre_y_stride *16, pre_y_stride, 0x7fffffff);
+            near_sad[7] = cpi->fn_ptr[BLOCK_16X16].sdf(src_y_ptr, b->src_stride, pre_y_buffer + pre_y_stride *16, pre_y_stride, 0x7fffffff);
     }
 
     if(cpi->common.last_frame_type != KEY_FRAME)
@@ -2173,7 +2175,7 @@
                     threshold = x->encode_breakout;
 
                 var = VARIANCE_INVOKE(&cpi->rtcd.variance, var16x16)
-                        (x->src.y_buffer, x->src.y_stride,
+                        (*(b->base_src), b->src_stride,
                         x->e_mbd.predictor, 16, &sse);
 
                 if (sse < threshold)