Incorporate WRAPLOW macro into non-highbitdepth tx

Incorporates the WRAPLOW macro into the non-highbitdepth transforms
to aid hardware verification between a software C model and an
intended hardware implementation though the use of the configure
options: --enable-experimental --enable-emulate-hardware.
Note that to avoid further discrepancies between the sse/sse2
implementations of the transforms and the C implementation, when the
emulate hardware option is invoked, we also disable sse/sse2/etc.

Also incudes some minor cleanups/renaming etc.

Change-Id: Ib864d8493313927d429cce402982f1c8e45b3287

6 files changed, 1141 insertions, 1048 deletions

diff --git a/vp9/common/vp9_common.h b/vp9/common/vp9_common.h
index 530d86f34..6801dd3a2 100644
--- a/vp9/common/vp9_common.h
+++ b/vp9/common/vp9_common.h
@@ -65,7 +65,7 @@ static INLINE int get_unsigned_bits(unsigned int num_values) {
 }
 
 #if CONFIG_VP9_HIGHBITDEPTH
-static INLINE uint16_t clip_pixel_high(int val, int bd) {
+static INLINE uint16_t clip_pixel_highbd(int val, int bd) {
   switch (bd) {
     case 8:
     default:
diff --git a/vp9/common/vp9_convolve.c b/vp9/common/vp9_convolve.c
index ad70e59bf..e616047ab 100644
--- a/vp9/common/vp9_convolve.c
+++ b/vp9/common/vp9_convolve.c
@@ -301,7 +301,7 @@ static void high_convolve_horiz(const uint8_t *src8, ptrdiff_t src_stride,
       int k, sum = 0;
       for (k = 0; k < SUBPEL_TAPS; ++k)
         sum += src_x[k] * x_filter[k];
-      dst[x] = clip_pixel_high(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
+      dst[x] = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
       x_q4 += x_step_q4;
     }
     src += src_stride;
@@ -327,7 +327,7 @@ static void high_convolve_avg_horiz(const uint8_t *src8, ptrdiff_t src_stride,
       for (k = 0; k < SUBPEL_TAPS; ++k)
         sum += src_x[k] * x_filter[k];
       dst[x] = ROUND_POWER_OF_TWO(dst[x] +
-          clip_pixel_high(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd), 1);
+          clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd), 1);
       x_q4 += x_step_q4;
     }
     src += src_stride;
@@ -352,7 +352,7 @@ static void high_convolve_vert(const uint8_t *src8, ptrdiff_t src_stride,
       int k, sum = 0;
       for (k = 0; k < SUBPEL_TAPS; ++k)
         sum += src_y[k * src_stride] * y_filter[k];
-      dst[y * dst_stride] = clip_pixel_high(
+      dst[y * dst_stride] = clip_pixel_highbd(
           ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
       y_q4 += y_step_q4;
     }
@@ -379,7 +379,7 @@ static void high_convolve_avg_vert(const uint8_t *src8, ptrdiff_t src_stride,
       for (k = 0; k < SUBPEL_TAPS; ++k)
         sum += src_y[k * src_stride] * y_filter[k];
       dst[y * dst_stride] = ROUND_POWER_OF_TWO(dst[y * dst_stride] +
-          clip_pixel_high(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd), 1);
+          clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd), 1);
       y_q4 += y_step_q4;
     }
     ++src;
diff --git a/vp9/common/vp9_idct.c b/vp9/common/vp9_idct.c
index 5b031752b..093160fc3 100644
--- a/vp9/common/vp9_idct.c
+++ b/vp9/common/vp9_idct.c
@@ -15,39 +15,49 @@
 #include "vp9/common/vp9_blockd.h"
 #include "vp9/common/vp9_idct.h"
 
-#if CONFIG_EMULATE_HARDWARE_HIGHBITDEPTH
-// When CONFIG_EMULATE_HW_HIGHBITDEPTH is 1 the transform performs strict
-// overflow wrapping to match expected hardware implementations.
+#if CONFIG_EMULATE_HARDWARE
+// When CONFIG_EMULATE_HARDWARE is 1 the transform performs a
+// non-normative method to handle overflows. A stream that causes
+// overflows  in the inverse transform is considered invalid in VP9,
+// and a hardware implementer is free to choose any reasonable
+// method to handle overflows. However to aid in hardware
+// verification they can use a specific implementation of the
+// WRAPLOW() macro below that is identical to their intended
+// hardware implementation (and also use configure options to trigger
+// the C-implementation of the transform).
+//
+// The particular WRAPLOW implementation below performs strict
+// overflow wrapping to match common hardware implementations.
 // bd of 8 uses trans_low with 16bits, need to remove 16bits
 // bd of 10 uses trans_low with 18bits, need to remove 14bits
 // bd of 12 uses trans_low with 20bits, need to remove 12bits
 // bd of x uses trans_low with 8+x bits, need to remove 24-x bits
-#define WRAPLOW(x) ((((int32_t)x) << (24 - bd)) >> (24 - bd))
+#define WRAPLOW(x, bd) ((((int32_t)(x)) << (24 - bd)) >> (24 - bd))
 #else
-#define WRAPLOW(x) (x)
-#endif  // CONFIG_EMULATE_HARDWARE_HIGHBITDEPTH
+#define WRAPLOW(x, bd) (x)
+#endif  // CONFIG_EMULATE_HARDWARE
 
 #if CONFIG_VP9_HIGHBITDEPTH
-static INLINE tran_low_t clamp_high(tran_high_t value, tran_low_t low,
-                                    tran_low_t high) {
-  return value < low ? low : (value > high ? high : value);
-}
-
-static INLINE tran_low_t clip_pixel_bd_high(tran_high_t dest,
-                                            tran_high_t trans, int bd) {
-  trans = WRAPLOW(trans);
+static INLINE tran_low_t highbd_clip_pixel_add(tran_high_t dest,
+                                               tran_high_t trans, int bd) {
+  trans = WRAPLOW(trans, bd);
   switch (bd) {
     case 8:
     default:
-      return clamp_high(WRAPLOW(dest + trans), 0, 255);
+      return clamp(WRAPLOW(dest + trans, bd), 0, 255);
     case 10:
-      return clamp_high(WRAPLOW(dest + trans), 0, 1023);
+      return clamp(WRAPLOW(dest + trans, bd), 0, 1023);
     case 12:
-      return clamp_high(WRAPLOW(dest + trans), 0, 4095);
+      return clamp(WRAPLOW(dest + trans, bd), 0, 4095);
   }
 }
 #endif  // CONFIG_VP9_HIGHBITDEPTH
 
+static INLINE tran_low_t clip_pixel_add(tran_high_t dest, tran_high_t trans) {
+  trans = WRAPLOW(trans, 8);
+  return clamp(WRAPLOW(dest + trans, 8), 0, 255);
+}
+
 void vp9_iwht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int stride) {
 /* 4-point reversible, orthonormal inverse Walsh-Hadamard in 3.5 adds,
    0.5 shifts per pixel. */
@@ -69,10 +79,10 @@ void vp9_iwht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int stride) {
     c1 = e1 - c1;
     a1 -= b1;
     d1 += c1;
-    op[0] = a1;
-    op[1] = b1;
-    op[2] = c1;
-    op[3] = d1;
+    op[0] = WRAPLOW(a1, 8);
+    op[1] = WRAPLOW(b1, 8);
+    op[2] = WRAPLOW(c1, 8);
+    op[3] = WRAPLOW(d1, 8);
     ip += 4;
     op += 4;
   }
@@ -90,10 +100,10 @@ void vp9_iwht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int stride) {
     c1 = e1 - c1;
     a1 -= b1;
     d1 += c1;
-    dest[stride * 0] = clip_pixel(dest[stride * 0] + a1);
-    dest[stride * 1] = clip_pixel(dest[stride * 1] + b1);
-    dest[stride * 2] = clip_pixel(dest[stride * 2] + c1);
-    dest[stride * 3] = clip_pixel(dest[stride * 3] + d1);
+    dest[stride * 0] = clip_pixel_add(dest[stride * 0], a1);
+    dest[stride * 1] = clip_pixel_add(dest[stride * 1], b1);
+    dest[stride * 2] = clip_pixel_add(dest[stride * 2], c1);
+    dest[stride * 3] = clip_pixel_add(dest[stride * 3], d1);
 
     ip++;
     dest++;
@@ -110,17 +120,17 @@ void vp9_iwht4x4_1_add_c(const tran_low_t *in, uint8_t *dest, int dest_stride) {
   a1 = ip[0] >> UNIT_QUANT_SHIFT;
   e1 = a1 >> 1;
   a1 -= e1;
-  op[0] = a1;
-  op[1] = op[2] = op[3] = e1;
+  op[0] = WRAPLOW(a1, 8);
+  op[1] = op[2] = op[3] = WRAPLOW(e1, 8);
 
   ip = tmp;
   for (i = 0; i < 4; i++) {
     e1 = ip[0] >> 1;
     a1 = ip[0] - e1;
-    dest[dest_stride * 0] = clip_pixel(dest[dest_stride * 0] + a1);
-    dest[dest_stride * 1] = clip_pixel(dest[dest_stride * 1] + e1);
-    dest[dest_stride * 2] = clip_pixel(dest[dest_stride * 2] + e1);
-    dest[dest_stride * 3] = clip_pixel(dest[dest_stride * 3] + e1);
+    dest[dest_stride * 0] = clip_pixel_add(dest[dest_stride * 0], a1);
+    dest[dest_stride * 1] = clip_pixel_add(dest[dest_stride * 1], e1);
+    dest[dest_stride * 2] = clip_pixel_add(dest[dest_stride * 2], e1);
+    dest[dest_stride * 3] = clip_pixel_add(dest[dest_stride * 3], e1);
     ip++;
     dest++;
   }
@@ -132,18 +142,18 @@ static void idct4(const tran_low_t *input, tran_low_t *output) {
   // stage 1
   temp1 = (input[0] + input[2]) * cospi_16_64;
   temp2 = (input[0] - input[2]) * cospi_16_64;
-  step[0] = dct_const_round_shift(temp1);
-  step[1] = dct_const_round_shift(temp2);
+  step[0] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step[1] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = input[1] * cospi_24_64 - input[3] * cospi_8_64;
   temp2 = input[1] * cospi_8_64 + input[3] * cospi_24_64;
-  step[2] = dct_const_round_shift(temp1);
-  step[3] = dct_const_round_shift(temp2);
+  step[2] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step[3] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   // stage 2
-  output[0] = step[0] + step[3];
-  output[1] = step[1] + step[2];
-  output[2] = step[1] - step[2];
-  output[3] = step[0] - step[3];
+  output[0] = WRAPLOW(step[0] + step[3], 8);
+  output[1] = WRAPLOW(step[1] + step[2], 8);
+  output[2] = WRAPLOW(step[1] - step[2], 8);
+  output[3] = WRAPLOW(step[0] - step[3], 8);
 }
 
 void vp9_idct4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int stride) {
@@ -164,9 +174,10 @@ void vp9_idct4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int stride) {
     for (j = 0; j < 4; ++j)
       temp_in[j] = out[j * 4 + i];
     idct4(temp_in, temp_out);
-    for (j = 0; j < 4; ++j)
-      dest[j * stride + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 4)
-                                  + dest[j * stride + i]);
+    for (j = 0; j < 4; ++j) {
+      dest[j * stride + i] = clip_pixel_add(ROUND_POWER_OF_TWO(temp_out[j], 4),
+                                            dest[j * stride + i]);
+    }
   }
 }
 
@@ -174,15 +185,15 @@ void vp9_idct4x4_1_add_c(const tran_low_t *input, uint8_t *dest,
                          int dest_stride) {
   int i;
   tran_high_t a1;
-  tran_low_t out = dct_const_round_shift(input[0] * cospi_16_64);
-  out = dct_const_round_shift(out * cospi_16_64);
+  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64), 8);
+  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64), 8);
   a1 = ROUND_POWER_OF_TWO(out, 4);
 
   for (i = 0; i < 4; i++) {
-    dest[0] = clip_pixel(dest[0] + a1);
-    dest[1] = clip_pixel(dest[1] + a1);
-    dest[2] = clip_pixel(dest[2] + a1);
-    dest[3] = clip_pixel(dest[3] + a1);
+    dest[0] = clip_pixel_add(dest[0], a1);
+    dest[1] = clip_pixel_add(dest[1], a1);
+    dest[2] = clip_pixel_add(dest[2], a1);
+    dest[3] = clip_pixel_add(dest[3], a1);
     dest += dest_stride;
   }
 }
@@ -197,39 +208,39 @@ static void idct8(const tran_low_t *input, tran_low_t *output) {
   step1[3] = input[6];
   temp1 = input[1] * cospi_28_64 - input[7] * cospi_4_64;
   temp2 = input[1] * cospi_4_64 + input[7] * cospi_28_64;
-  step1[4] = dct_const_round_shift(temp1);
-  step1[7] = dct_const_round_shift(temp2);
+  step1[4] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[7] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = input[5] * cospi_12_64 - input[3] * cospi_20_64;
   temp2 = input[5] * cospi_20_64 + input[3] * cospi_12_64;
-  step1[5] = dct_const_round_shift(temp1);
-  step1[6] = dct_const_round_shift(temp2);
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   // stage 2 & stage 3 - even half
   idct4(step1, step1);
 
   // stage 2 - odd half
-  step2[4] = step1[4] + step1[5];
-  step2[5] = step1[4] - step1[5];
-  step2[6] = -step1[6] + step1[7];
-  step2[7] = step1[6] + step1[7];
+  step2[4] = WRAPLOW(step1[4] + step1[5], 8);
+  step2[5] = WRAPLOW(step1[4] - step1[5], 8);
+  step2[6] = WRAPLOW(-step1[6] + step1[7], 8);
+  step2[7] = WRAPLOW(step1[6] + step1[7], 8);
 
   // stage 3 -odd half
   step1[4] = step2[4];
   temp1 = (step2[6] - step2[5]) * cospi_16_64;
   temp2 = (step2[5] + step2[6]) * cospi_16_64;
-  step1[5] = dct_const_round_shift(temp1);
-  step1[6] = dct_const_round_shift(temp2);
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), 8);
   step1[7] = step2[7];
 
   // stage 4
-  output[0] = step1[0] + step1[7];
-  output[1] = step1[1] + step1[6];
-  output[2] = step1[2] + step1[5];
-  output[3] = step1[3] + step1[4];
-  output[4] = step1[3] - step1[4];
-  output[5] = step1[2] - step1[5];
-  output[6] = step1[1] - step1[6];
-  output[7] = step1[0] - step1[7];
+  output[0] = WRAPLOW(step1[0] + step1[7], 8);
+  output[1] = WRAPLOW(step1[1] + step1[6], 8);
+  output[2] = WRAPLOW(step1[2] + step1[5], 8);
+  output[3] = WRAPLOW(step1[3] + step1[4], 8);
+  output[4] = WRAPLOW(step1[3] - step1[4], 8);
+  output[5] = WRAPLOW(step1[2] - step1[5], 8);
+  output[6] = WRAPLOW(step1[1] - step1[6], 8);
+  output[7] = WRAPLOW(step1[0] - step1[7], 8);
 }
 
 void vp9_idct8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int stride) {
@@ -250,21 +261,22 @@ void vp9_idct8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int stride) {
     for (j = 0; j < 8; ++j)
       temp_in[j] = out[j * 8 + i];
     idct8(temp_in, temp_out);
-    for (j = 0; j < 8; ++j)
-      dest[j * stride + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 5)
-                                  + dest[j * stride + i]);
+    for (j = 0; j < 8; ++j) {
+      dest[j * stride + i] = clip_pixel_add(ROUND_POWER_OF_TWO(temp_out[j], 5),
+                                            dest[j * stride + i]);
+    }
   }
 }
 
 void vp9_idct8x8_1_add_c(const tran_low_t *input, uint8_t *dest, int stride) {
   int i, j;
   tran_high_t a1;
-  tran_low_t out = dct_const_round_shift(input[0] * cospi_16_64);
-  out = dct_const_round_shift(out * cospi_16_64);
+  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64), 8);
+  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64), 8);
   a1 = ROUND_POWER_OF_TWO(out, 5);
   for (j = 0; j < 8; ++j) {
     for (i = 0; i < 8; ++i)
-      dest[i] = clip_pixel(dest[i] + a1);
+      dest[i] = clip_pixel_add(dest[i], a1);
     dest += stride;
   }
 }
@@ -305,10 +317,10 @@ static void iadst4(const tran_low_t *input, tran_low_t *output) {
   // The overall dynamic range is 14b (input) + 14b (multiplication scaling)
   // + 1b (addition) = 29b.
   // Hence the output bit depth is 15b.
-  output[0] = dct_const_round_shift(s0);
-  output[1] = dct_const_round_shift(s1);
-  output[2] = dct_const_round_shift(s2);
-  output[3] = dct_const_round_shift(s3);
+  output[0] = WRAPLOW(dct_const_round_shift(s0), 8);
+  output[1] = WRAPLOW(dct_const_round_shift(s1), 8);
+  output[2] = WRAPLOW(dct_const_round_shift(s2), 8);
+  output[3] = WRAPLOW(dct_const_round_shift(s3), 8);
 }
 
 void vp9_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int stride,
@@ -337,11 +349,13 @@ void vp9_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int stride,
     for (j = 0; j < 4; ++j)
       temp_in[j] = out[j * 4 + i];
     IHT_4[tx_type].cols(temp_in, temp_out);
-    for (j = 0; j < 4; ++j)
-      dest[j * stride + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 4)
-                                  + dest[j * stride + i]);
+    for (j = 0; j < 4; ++j) {
+      dest[j * stride + i] = clip_pixel_add(ROUND_POWER_OF_TWO(temp_out[j], 4),
+                                            dest[j * stride + i]);
+    }
   }
 }
+
 static void iadst8(const tran_low_t *input, tran_low_t *output) {
   int s0, s1, s2, s3, s4, s5, s6, s7;
 
@@ -370,14 +384,14 @@ static void iadst8(const tran_low_t *input, tran_low_t *output) {
   s6 = cospi_26_64 * x6 + cospi_6_64  * x7;
   s7 = cospi_6_64  * x6 - cospi_26_64 * x7;
 
-  x0 = dct_const_round_shift(s0 + s4);
-  x1 = dct_const_round_shift(s1 + s5);
-  x2 = dct_const_round_shift(s2 + s6);
-  x3 = dct_const_round_shift(s3 + s7);
-  x4 = dct_const_round_shift(s0 - s4);
-  x5 = dct_const_round_shift(s1 - s5);
-  x6 = dct_const_round_shift(s2 - s6);
-  x7 = dct_const_round_shift(s3 - s7);
+  x0 = WRAPLOW(dct_const_round_shift(s0 + s4), 8);
+  x1 = WRAPLOW(dct_const_round_shift(s1 + s5), 8);
+  x2 = WRAPLOW(dct_const_round_shift(s2 + s6), 8);
+  x3 = WRAPLOW(dct_const_round_shift(s3 + s7), 8);
+  x4 = WRAPLOW(dct_const_round_shift(s0 - s4), 8);
+  x5 = WRAPLOW(dct_const_round_shift(s1 - s5), 8);
+  x6 = WRAPLOW(dct_const_round_shift(s2 - s6), 8);
+  x7 = WRAPLOW(dct_const_round_shift(s3 - s7), 8);
 
   // stage 2
   s0 = x0;
@@ -389,14 +403,14 @@ static void iadst8(const tran_low_t *input, tran_low_t *output) {
   s6 = -cospi_24_64 * x6 + cospi_8_64  * x7;
   s7 =  cospi_8_64  * x6 + cospi_24_64 * x7;
 
-  x0 = s0 + s2;
-  x1 = s1 + s3;
-  x2 = s0 - s2;
-  x3 = s1 - s3;
-  x4 = dct_const_round_shift(s4 + s6);
-  x5 = dct_const_round_shift(s5 + s7);
-  x6 = dct_const_round_shift(s4 - s6);
-  x7 = dct_const_round_shift(s5 - s7);
+  x0 = WRAPLOW(s0 + s2, 8);
+  x1 = WRAPLOW(s1 + s3, 8);
+  x2 = WRAPLOW(s0 - s2, 8);
+  x3 = WRAPLOW(s1 - s3, 8);
+  x4 = WRAPLOW(dct_const_round_shift(s4 + s6), 8);
+  x5 = WRAPLOW(dct_const_round_shift(s5 + s7), 8);
+  x6 = WRAPLOW(dct_const_round_shift(s4 - s6), 8);
+  x7 = WRAPLOW(dct_const_round_shift(s5 - s7), 8);
 
   // stage 3
   s2 = cospi_16_64 * (x2 + x3);
@@ -404,19 +418,19 @@ static void iadst8(const tran_low_t *input, tran_low_t *output) {
   s6 = cospi_16_64 * (x6 + x7);
   s7 = cospi_16_64 * (x6 - x7);
 
-  x2 = dct_const_round_shift(s2);
-  x3 = dct_const_round_shift(s3);
-  x6 = dct_const_round_shift(s6);
-  x7 = dct_const_round_shift(s7);
+  x2 = WRAPLOW(dct_const_round_shift(s2), 8);
+  x3 = WRAPLOW(dct_const_round_shift(s3), 8);
+  x6 = WRAPLOW(dct_const_round_shift(s6), 8);
+  x7 = WRAPLOW(dct_const_round_shift(s7), 8);
 
-  output[0] =  x0;
-  output[1] = -x4;
-  output[2] =  x6;
-  output[3] = -x2;
-  output[4] =  x3;
-  output[5] = -x7;
-  output[6] =  x5;
-  output[7] = -x1;
+  output[0] = WRAPLOW(x0, 8);
+  output[1] = WRAPLOW(-x4, 8);
+  output[2] = WRAPLOW(x6, 8);
+  output[3] = WRAPLOW(-x2, 8);
+  output[4] = WRAPLOW(x3, 8);
+  output[5] = WRAPLOW(-x7, 8);
+  output[6] = WRAPLOW(x5, 8);
+  output[7] = WRAPLOW(-x1, 8);
 }
 
 static const transform_2d IHT_8[] = {
@@ -446,9 +460,10 @@ void vp9_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int stride,
     for (j = 0; j < 8; ++j)
       temp_in[j] = out[j * 8 + i];
     ht.cols(temp_in, temp_out);
-    for (j = 0; j < 8; ++j)
-      dest[j * stride + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 5)
-                                  + dest[j * stride + i]);
+    for (j = 0; j < 8; ++j) {
+      dest[j * stride + i] = clip_pixel_add(ROUND_POWER_OF_TWO(temp_out[j], 5),
+                                            dest[j * stride + i]);
+    }
   }
 }
 
@@ -471,9 +486,10 @@ void vp9_idct8x8_12_add_c(const tran_low_t *input, uint8_t *dest, int stride) {
     for (j = 0; j < 8; ++j)
       temp_in[j] = out[j * 8 + i];
     idct8(temp_in, temp_out);
-    for (j = 0; j < 8; ++j)
-      dest[j * stride + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 5)
-                                  + dest[j * stride + i]);
+    for (j = 0; j < 8; ++j) {
+      dest[j * stride + i] = clip_pixel_add(ROUND_POWER_OF_TWO(temp_out[j], 5),
+                                            dest[j * stride + i]);
+    }
   }
 }
 
@@ -511,23 +527,23 @@ static void idct16(const tran_low_t *input, tran_low_t *output) {
 
   temp1 = step1[8] * cospi_30_64 - step1[15] * cospi_2_64;
   temp2 = step1[8] * cospi_2_64 + step1[15] * cospi_30_64;
-  step2[8] = dct_const_round_shift(temp1);
-  step2[15] = dct_const_round_shift(temp2);
+  step2[8] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[15] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = step1[9] * cospi_14_64 - step1[14] * cospi_18_64;
   temp2 = step1[9] * cospi_18_64 + step1[14] * cospi_14_64;
-  step2[9] = dct_const_round_shift(temp1);
-  step2[14] = dct_const_round_shift(temp2);
+  step2[9] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[14] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = step1[10] * cospi_22_64 - step1[13] * cospi_10_64;
   temp2 = step1[10] * cospi_10_64 + step1[13] * cospi_22_64;
-  step2[10] = dct_const_round_shift(temp1);
-  step2[13] = dct_const_round_shift(temp2);
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = step1[11] * cospi_6_64 - step1[12] * cospi_26_64;
   temp2 = step1[11] * cospi_26_64 + step1[12] * cospi_6_64;
-  step2[11] = dct_const_round_shift(temp1);
-  step2[12] = dct_const_round_shift(temp2);
+  step2[11] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[12] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   // stage 3
   step1[0] = step2[0];
@@ -537,109 +553,109 @@ static void idct16(const tran_low_t *input, tran_low_t *output) {
 
   temp1 = step2[4] * cospi_28_64 - step2[7] * cospi_4_64;
   temp2 = step2[4] * cospi_4_64 + step2[7] * cospi_28_64;
-  step1[4] = dct_const_round_shift(temp1);
-  step1[7] = dct_const_round_shift(temp2);
+  step1[4] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[7] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = step2[5] * cospi_12_64 - step2[6] * cospi_20_64;
   temp2 = step2[5] * cospi_20_64 + step2[6] * cospi_12_64;
-  step1[5] = dct_const_round_shift(temp1);
-  step1[6] = dct_const_round_shift(temp2);
-
-  step1[8] = step2[8] + step2[9];
-  step1[9] = step2[8] - step2[9];
-  step1[10] = -step2[10] + step2[11];
-  step1[11] = step2[10] + step2[11];
-  step1[12] = step2[12] + step2[13];
-  step1[13] = step2[12] - step2[13];
-  step1[14] = -step2[14] + step2[15];
-  step1[15] = step2[14] + step2[15];
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), 8);
+
+  step1[8] = WRAPLOW(step2[8] + step2[9], 8);
+  step1[9] = WRAPLOW(step2[8] - step2[9], 8);
+  step1[10] = WRAPLOW(-step2[10] + step2[11], 8);
+  step1[11] = WRAPLOW(step2[10] + step2[11], 8);
+  step1[12] = WRAPLOW(step2[12] + step2[13], 8);
+  step1[13] = WRAPLOW(step2[12] - step2[13], 8);
+  step1[14] = WRAPLOW(-step2[14] + step2[15], 8);
+  step1[15] = WRAPLOW(step2[14] + step2[15], 8);
 
   // stage 4
   temp1 = (step1[0] + step1[1]) * cospi_16_64;
   temp2 = (step1[0] - step1[1]) * cospi_16_64;
-  step2[0] = dct_const_round_shift(temp1);
-  step2[1] = dct_const_round_shift(temp2);
+  step2[0] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[1] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = step1[2] * cospi_24_64 - step1[3] * cospi_8_64;
   temp2 = step1[2] * cospi_8_64 + step1[3] * cospi_24_64;
-  step2[2] = dct_const_round_shift(temp1);
-  step2[3] = dct_const_round_shift(temp2);
-  step2[4] = step1[4] + step1[5];
-  step2[5] = step1[4] - step1[5];
-  step2[6] = -step1[6] + step1[7];
-  step2[7] = step1[6] + step1[7];
+  step2[2] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[3] = WRAPLOW(dct_const_round_shift(temp2), 8);
+  step2[4] = WRAPLOW(step1[4] + step1[5], 8);
+  step2[5] = WRAPLOW(step1[4] - step1[5], 8);
+  step2[6] = WRAPLOW(-step1[6] + step1[7], 8);
+  step2[7] = WRAPLOW(step1[6] + step1[7], 8);
 
   step2[8] = step1[8];
   step2[15] = step1[15];
   temp1 = -step1[9] * cospi_8_64 + step1[14] * cospi_24_64;
   temp2 = step1[9] * cospi_24_64 + step1[14] * cospi_8_64;
-  step2[9] = dct_const_round_shift(temp1);
-  step2[14] = dct_const_round_shift(temp2);
+  step2[9] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[14] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = -step1[10] * cospi_24_64 - step1[13] * cospi_8_64;
   temp2 = -step1[10] * cospi_8_64 + step1[13] * cospi_24_64;
-  step2[10] = dct_const_round_shift(temp1);
-  step2[13] = dct_const_round_shift(temp2);
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), 8);
   step2[11] = step1[11];
   step2[12] = step1[12];
 
   // stage 5
-  step1[0] = step2[0] + step2[3];
-  step1[1] = step2[1] + step2[2];
-  step1[2] = step2[1] - step2[2];
-  step1[3] = step2[0] - step2[3];
+  step1[0] = WRAPLOW(step2[0] + step2[3], 8);
+  step1[1] = WRAPLOW(step2[1] + step2[2], 8);
+  step1[2] = WRAPLOW(step2[1] - step2[2], 8);
+  step1[3] = WRAPLOW(step2[0] - step2[3], 8);
   step1[4] = step2[4];
   temp1 = (step2[6] - step2[5]) * cospi_16_64;
   temp2 = (step2[5] + step2[6]) * cospi_16_64;
-  step1[5] = dct_const_round_shift(temp1);
-  step1[6] = dct_const_round_shift(temp2);
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), 8);
   step1[7] = step2[7];
 
-  step1[8] = step2[8] + step2[11];
-  step1[9] = step2[9] + step2[10];
-  step1[10] = step2[9] - step2[10];
-  step1[11] = step2[8] - step2[11];
-  step1[12] = -step2[12] + step2[15];
-  step1[13] = -step2[13] + step2[14];
-  step1[14] = step2[13] + step2[14];
-  step1[15] = step2[12] + step2[15];
+  step1[8] = WRAPLOW(step2[8] + step2[11], 8);
+  step1[9] = WRAPLOW(step2[9] + step2[10], 8);
+  step1[10] = WRAPLOW(step2[9] - step2[10], 8);
+  step1[11] = WRAPLOW(step2[8] - step2[11], 8);
+  step1[12] = WRAPLOW(-step2[12] + step2[15], 8);
+  step1[13] = WRAPLOW(-step2[13] + step2[14], 8);
+  step1[14] = WRAPLOW(step2[13] + step2[14], 8);
+  step1[15] = WRAPLOW(step2[12] + step2[15], 8);
 
   // stage 6
-  step2[0] = step1[0] + step1[7];
-  step2[1] = step1[1] + step1[6];
-  step2[2] = step1[2] + step1[5];
-  step2[3] = step1[3] + step1[4];
-  step2[4] = step1[3] - step1[4];
-  step2[5] = step1[2] - step1[5];
-  step2[6] = step1[1] - step1[6];
-  step2[7] = step1[0] - step1[7];
+  step2[0] = WRAPLOW(step1[0] + step1[7], 8);
+  step2[1] = WRAPLOW(step1[1] + step1[6], 8);
+  step2[2] = WRAPLOW(step1[2] + step1[5], 8);
+  step2[3] = WRAPLOW(step1[3] + step1[4], 8);
+  step2[4] = WRAPLOW(step1[3] - step1[4], 8);
+  step2[5] = WRAPLOW(step1[2] - step1[5], 8);
+  step2[6] = WRAPLOW(step1[1] - step1[6], 8);
+  step2[7] = WRAPLOW(step1[0] - step1[7], 8);
   step2[8] = step1[8];
   step2[9] = step1[9];
   temp1 = (-step1[10] + step1[13]) * cospi_16_64;
   temp2 = (step1[10] + step1[13]) * cospi_16_64;
-  step2[10] = dct_const_round_shift(temp1);
-  step2[13] = dct_const_round_shift(temp2);
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = (-step1[11] + step1[12]) * cospi_16_64;
   temp2 = (step1[11] + step1[12]) * cospi_16_64;
-  step2[11] = dct_const_round_shift(temp1);
-  step2[12] = dct_const_round_shift(temp2);
+  step2[11] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[12] = WRAPLOW(dct_const_round_shift(temp2), 8);
   step2[14] = step1[14];
   step2[15] = step1[15];
 
   // stage 7
-  output[0] = step2[0] + step2[15];
-  output[1] = step2[1] + step2[14];
-  output[2] = step2[2] + step2[13];
-  output[3] = step2[3] + step2[12];
-  output[4] = step2[4] + step2[11];
-  output[5] = step2[5] + step2[10];
-  output[6] = step2[6] + step2[9];
-  output[7] = step2[7] + step2[8];
-  output[8] = step2[7] - step2[8];
-  output[9] = step2[6] - step2[9];
-  output[10] = step2[5] - step2[10];
-  output[11] = step2[4] - step2[11];
-  output[12] = step2[3] - step2[12];
-  output[13] = step2[2] - step2[13];
-  output[14] = step2[1] - step2[14];
-  output[15] = step2[0] - step2[15];
+  output[0] = WRAPLOW(step2[0] + step2[15], 8);
+  output[1] = WRAPLOW(step2[1] + step2[14], 8);
+  output[2] = WRAPLOW(step2[2] + step2[13], 8);
+  output[3] = WRAPLOW(step2[3] + step2[12], 8);
+  output[4] = WRAPLOW(step2[4] + step2[11], 8);
+  output[5] = WRAPLOW(step2[5] + step2[10], 8);
+  output[6] = WRAPLOW(step2[6] + step2[9], 8);
+  output[7] = WRAPLOW(step2[7] + step2[8], 8);
+  output[8] = WRAPLOW(step2[7] - step2[8], 8);
+  output[9] = WRAPLOW(step2[6] - step2[9], 8);
+  output[10] = WRAPLOW(step2[5] - step2[10], 8);
+  output[11] = WRAPLOW(step2[4] - step2[11], 8);
+  output[12] = WRAPLOW(step2[3] - step2[12], 8);
+  output[13] = WRAPLOW(step2[2] - step2[13], 8);
+  output[14] = WRAPLOW(step2[1] - step2[14], 8);
+  output[15] = WRAPLOW(step2[0] - step2[15], 8);
 }
 
 void vp9_idct16x16_256_add_c(const tran_low_t *input, uint8_t *dest,
@@ -661,9 +677,10 @@ void vp9_idct16x16_256_add_c(const tran_low_t *input, uint8_t *dest,
     for (j = 0; j < 16; ++j)
       temp_in[j] = out[j * 16 + i];
     idct16(temp_in, temp_out);
-    for (j = 0; j < 16; ++j)
-      dest[j * stride + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 6)
-                                  + dest[j * stride + i]);
+    for (j = 0; j < 16; ++j) {
+      dest[j * stride + i] = clip_pixel_add(ROUND_POWER_OF_TWO(temp_out[j], 6),
+                                            dest[j * stride + i]);
+    }
   }
 }
 
@@ -715,22 +732,22 @@ static void iadst16(const tran_low_t *input, tran_low_t *output) {
   s14 = x14 * cospi_29_64 + x15 * cospi_3_64;
   s15 = x14 * cospi_3_64  - x15 * cospi_29_64;
 
-  x0 = dct_const_round_shift(s0 + s8);
-  x1 = dct_const_round_shift(s1 + s9);
-  x2 = dct_const_round_shift(s2 + s10);
-  x3 = dct_const_round_shift(s3 + s11);
-  x4 = dct_const_round_shift(s4 + s12);
-  x5 = dct_const_round_shift(s5 + s13);
-  x6 = dct_const_round_shift(s6 + s14);
-  x7 = dct_const_round_shift(s7 + s15);
-  x8  = dct_const_round_shift(s0 - s8);
-  x9  = dct_const_round_shift(s1 - s9);
-  x10 = dct_const_round_shift(s2 - s10);
-  x11 = dct_const_round_shift(s3 - s11);
-  x12 = dct_const_round_shift(s4 - s12);
-  x13 = dct_const_round_shift(s5 - s13);
-  x14 = dct_const_round_shift(s6 - s14);
-  x15 = dct_const_round_shift(s7 - s15);
+  x0 = WRAPLOW(dct_const_round_shift(s0 + s8), 8);
+  x1 = WRAPLOW(dct_const_round_shift(s1 + s9), 8);
+  x2 = WRAPLOW(dct_const_round_shift(s2 + s10), 8);
+  x3 = WRAPLOW(dct_const_round_shift(s3 + s11), 8);
+  x4 = WRAPLOW(dct_const_round_shift(s4 + s12), 8);
+  x5 = WRAPLOW(dct_const_round_shift(s5 + s13), 8);
+  x6 = WRAPLOW(dct_const_round_shift(s6 + s14), 8);
+  x7 = WRAPLOW(dct_const_round_shift(s7 + s15), 8);
+  x8 = WRAPLOW(dct_const_round_shift(s0 - s8), 8);
+  x9 = WRAPLOW(dct_const_round_shift(s1 - s9), 8);
+  x10 = WRAPLOW(dct_const_round_shift(s2 - s10), 8);
+  x11 = WRAPLOW(dct_const_round_shift(s3 - s11), 8);
+  x12 = WRAPLOW(dct_const_round_shift(s4 - s12), 8);
+  x13 = WRAPLOW(dct_const_round_shift(s5 - s13), 8);
+  x14 = WRAPLOW(dct_const_round_shift(s6 - s14), 8);
+  x15 = WRAPLOW(dct_const_round_shift(s7 - s15), 8);
 
   // stage 2
   s0 = x0;
@@ -750,22 +767,22 @@ static void iadst16(const tran_low_t *input, tran_low_t *output) {
   s14 = - x14 * cospi_12_64 + x15 * cospi_20_64;
   s15 =   x14 * cospi_20_64 + x15 * cospi_12_64;
 
-  x0 = s0 + s4;
-  x1 = s1 + s5;
-  x2 = s2 + s6;
-  x3 = s3 + s7;
-  x4 = s0 - s4;
-  x5 = s1 - s5;
-  x6 = s2 - s6;
-  x7 = s3 - s7;
-  x8 = dct_const_round_shift(s8 + s12);
-  x9 = dct_const_round_shift(s9 + s13);
-  x10 = dct_const_round_shift(s10 + s14);
-  x11 = dct_const_round_shift(s11 + s15);
-  x12 = dct_const_round_shift(s8 - s12);
-  x13 = dct_const_round_shift(s9 - s13);
-  x14 = dct_const_round_shift(s10 - s14);
-  x15 = dct_const_round_shift(s11 - s15);
+  x0 = WRAPLOW(s0 + s4, 8);
+  x1 = WRAPLOW(s1 + s5, 8);
+  x2 = WRAPLOW(s2 + s6, 8);
+  x3 = WRAPLOW(s3 + s7, 8);
+  x4 = WRAPLOW(s0 - s4, 8);
+  x5 = WRAPLOW(s1 - s5, 8);
+  x6 = WRAPLOW(s2 - s6, 8);
+  x7 = WRAPLOW(s3 - s7, 8);
+  x8 = WRAPLOW(dct_const_round_shift(s8 + s12), 8);
+  x9 = WRAPLOW(dct_const_round_shift(s9 + s13), 8);
+  x10 = WRAPLOW(dct_const_round_shift(s10 + s14), 8);
+  x11 = WRAPLOW(dct_const_round_shift(s11 + s15), 8);
+  x12 = WRAPLOW(dct_const_round_shift(s8 - s12), 8);
+  x13 = WRAPLOW(dct_const_round_shift(s9 - s13), 8);
+  x14 = WRAPLOW(dct_const_round_shift(s10 - s14), 8);
+  x15 = WRAPLOW(dct_const_round_shift(s11 - s15), 8);
 
   // stage 3
   s0 = x0;
@@ -785,22 +802,22 @@ static void iadst16(const tran_low_t *input, tran_low_t *output) {
   s14 = - x14 * cospi_24_64 + x15 * cospi_8_64;
   s15 =   x14 * cospi_8_64  + x15 * cospi_24_64;
 
-  x0 = s0 + s2;
-  x1 = s1 + s3;
-  x2 = s0 - s2;
-  x3 = s1 - s3;
-  x4 = dct_const_round_shift(s4 + s6);
-  x5 = dct_const_round_shift(s5 + s7);
-  x6 = dct_const_round_shift(s4 - s6);
-  x7 = dct_const_round_shift(s5 - s7);
-  x8 = s8 + s10;
-  x9 = s9 + s11;
-  x10 = s8 - s10;
-  x11 = s9 - s11;
-  x12 = dct_const_round_shift(s12 + s14);
-  x13 = dct_const_round_shift(s13 + s15);
-  x14 = dct_const_round_shift(s12 - s14);
-  x15 = dct_const_round_shift(s13 - s15);
+  x0 = WRAPLOW(s0 + s2, 8);
+  x1 = WRAPLOW(s1 + s3, 8);
+  x2 = WRAPLOW(s0 - s2, 8);
+  x3 = WRAPLOW(s1 - s3, 8);
+  x4 = WRAPLOW(dct_const_round_shift(s4 + s6), 8);
+  x5 = WRAPLOW(dct_const_round_shift(s5 + s7), 8);
+  x6 = WRAPLOW(dct_const_round_shift(s4 - s6), 8);
+  x7 = WRAPLOW(dct_const_round_shift(s5 - s7), 8);
+  x8 = WRAPLOW(s8 + s10, 8);
+  x9 = WRAPLOW(s9 + s11, 8);
+  x10 = WRAPLOW(s8 - s10, 8);
+  x11 = WRAPLOW(s9 - s11, 8);
+  x12 = WRAPLOW(dct_const_round_shift(s12 + s14), 8);
+  x13 = WRAPLOW(dct_const_round_shift(s13 + s15), 8);
+  x14 = WRAPLOW(dct_const_round_shift(s12 - s14), 8);
+  x15 = WRAPLOW(dct_const_round_shift(s13 - s15), 8);
 
   // stage 4
   s2 = (- cospi_16_64) * (x2 + x3);
@@ -812,31 +829,31 @@ static void iadst16(const tran_low_t *input, tran_low_t *output) {
   s14 = (- cospi_16_64) * (x14 + x15);
   s15 = cospi_16_64 * (x14 - x15);
 
-  x2 = dct_const_round_shift(s2);
-  x3 = dct_const_round_shift(s3);
-  x6 = dct_const_round_shift(s6);
-  x7 = dct_const_round_shift(s7);
-  x10 = dct_const_round_shift(s10);
-  x11 = dct_const_round_shift(s11);
-  x14 = dct_const_round_shift(s14);
-  x15 = dct_const_round_shift(s15);
-
-  output[0] =  x0;
-  output[1] = -x8;
-  output[2] =  x12;
-  output[3] = -x4;
-  output[4] =  x6;
-  output[5] =  x14;
-  output[6] =  x10;
-  output[7] =  x2;
-  output[8] =  x3;
-  output[9] =  x11;
-  output[10] =  x15;
-  output[11] =  x7;
-  output[12] =  x5;
-  output[13] = -x13;
-  output[14] =  x9;
-  output[15] = -x1;
+  x2 = WRAPLOW(dct_const_round_shift(s2), 8);
+  x3 = WRAPLOW(dct_const_round_shift(s3), 8);
+  x6 = WRAPLOW(dct_const_round_shift(s6), 8);
+  x7 = WRAPLOW(dct_const_round_shift(s7), 8);
+  x10 = WRAPLOW(dct_const_round_shift(s10), 8);
+  x11 = WRAPLOW(dct_const_round_shift(s11), 8);
+  x14 = WRAPLOW(dct_const_round_shift(s14), 8);
+  x15 = WRAPLOW(dct_const_round_shift(s15), 8);
+
+  output[0] = WRAPLOW(x0, 8);
+  output[1] = WRAPLOW(-x8, 8);
+  output[2] = WRAPLOW(x12, 8);
+  output[3] = WRAPLOW(-x4, 8);
+  output[4] = WRAPLOW(x6, 8);
+  output[5] = WRAPLOW(x14, 8);
+  output[6] = WRAPLOW(x10, 8);
+  output[7] = WRAPLOW(x2, 8);
+  output[8] = WRAPLOW(x3, 8);
+  output[9] = WRAPLOW(x11, 8);
+  output[10] = WRAPLOW(x15, 8);
+  output[11] = WRAPLOW(x7, 8);
+  output[12] = WRAPLOW(x5, 8);
+  output[13] = WRAPLOW(-x13, 8);
+  output[14] = WRAPLOW(x9, 8);
+  output[15] = WRAPLOW(-x1, 8);
 }
 
 static const transform_2d IHT_16[] = {
@@ -866,9 +883,10 @@ void vp9_iht16x16_256_add_c(const tran_low_t *input, uint8_t *dest, int stride,
     for (j = 0; j < 16; ++j)
       temp_in[j] = out[j * 16 + i];
     ht.cols(temp_in, temp_out);
-    for (j = 0; j < 16; ++j)
-      dest[j * stride + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 6)
-                                        + dest[j * stride + i]);
+    for (j = 0; j < 16; ++j) {
+      dest[j * stride + i] = clip_pixel_add(ROUND_POWER_OF_TWO(temp_out[j], 6),
+                                            dest[j * stride + i]);
+    }
   }
 }
 
@@ -892,21 +910,22 @@ void vp9_idct16x16_10_add_c(const tran_low_t *input, uint8_t *dest,
     for (j = 0; j < 16; ++j)
       temp_in[j] = out[j*16 + i];
     idct16(temp_in, temp_out);
-    for (j = 0; j < 16; ++j)
-      dest[j * stride + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 6)
-                                  + dest[j * stride + i]);
+    for (j = 0; j < 16; ++j) {
+      dest[j * stride + i] = clip_pixel_add(ROUND_POWER_OF_TWO(temp_out[j], 6),
+                                            dest[j * stride + i]);
+    }
   }
 }
 
 void vp9_idct16x16_1_add_c(const tran_low_t *input, uint8_t *dest, int stride) {
   int i, j;
   tran_high_t a1;
-  tran_low_t out = dct_const_round_shift(input[0] * cospi_16_64);
-  out = dct_const_round_shift(out * cospi_16_64);
+  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64), 8);
+  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64), 8);
   a1 = ROUND_POWER_OF_TWO(out, 6);
   for (j = 0; j < 16; ++j) {
     for (i = 0; i < 16; ++i)
-      dest[i] = clip_pixel(dest[i] + a1);
+      dest[i] = clip_pixel_add(dest[i], a1);
     dest += stride;
   }
 }
@@ -935,43 +954,43 @@ static void idct32(const tran_low_t *input, tran_low_t *output) {
 
   temp1 = input[1] * cospi_31_64 - input[31] * cospi_1_64;
   temp2 = input[1] * cospi_1_64 + input[31] * cospi_31_64;
-  step1[16] = dct_const_round_shift(temp1);
-  step1[31] = dct_const_round_shift(temp2);
+  step1[16] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[31] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = input[17] * cospi_15_64 - input[15] * cospi_17_64;
   temp2 = input[17] * cospi_17_64 + input[15] * cospi_15_64;
-  step1[17] = dct_const_round_shift(temp1);
-  step1[30] = dct_const_round_shift(temp2);
+  step1[17] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[30] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = input[9] * cospi_23_64 - input[23] * cospi_9_64;
   temp2 = input[9] * cospi_9_64 + input[23] * cospi_23_64;
-  step1[18] = dct_const_round_shift(temp1);
-  step1[29] = dct_const_round_shift(temp2);
+  step1[18] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[29] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = input[25] * cospi_7_64 - input[7] * cospi_25_64;
   temp2 = input[25] * cospi_25_64 + input[7] * cospi_7_64;
-  step1[19] = dct_const_round_shift(temp1);
-  step1[28] = dct_const_round_shift(temp2);
+  step1[19] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[28] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = input[5] * cospi_27_64 - input[27] * cospi_5_64;
   temp2 = input[5] * cospi_5_64 + input[27] * cospi_27_64;
-  step1[20] = dct_const_round_shift(temp1);
-  step1[27] = dct_const_round_shift(temp2);
+  step1[20] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[27] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = input[21] * cospi_11_64 - input[11] * cospi_21_64;
   temp2 = input[21] * cospi_21_64 + input[11] * cospi_11_64;
-  step1[21] = dct_const_round_shift(temp1);
-  step1[26] = dct_const_round_shift(temp2);
+  step1[21] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[26] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = input[13] * cospi_19_64 - input[19] * cospi_13_64;
   temp2 = input[13] * cospi_13_64 + input[19] * cospi_19_64;
-  step1[22] = dct_const_round_shift(temp1);
-  step1[25] = dct_const_round_shift(temp2);
+  step1[22] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[25] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = input[29] * cospi_3_64 - input[3] * cospi_29_64;
   temp2 = input[29] * cospi_29_64 + input[3] * cospi_3_64;
-  step1[23] = dct_const_round_shift(temp1);
-  step1[24] = dct_const_round_shift(temp2);
+  step1[23] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[24] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   // stage 2
   step2[0] = step1[0];
@@ -985,40 +1004,40 @@ static void idct32(const tran_low_t *input, tran_low_t *output) {
 
   temp1 = step1[8] * cospi_30_64 - step1[15] * cospi_2_64;
   temp2 = step1[8] * cospi_2_64 + step1[15] * cospi_30_64;
-  step2[8] = dct_const_round_shift(temp1);
-  step2[15] = dct_const_round_shift(temp2);
+  step2[8] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[15] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = step1[9] * cospi_14_64 - step1[14] * cospi_18_64;
   temp2 = step1[9] * cospi_18_64 + step1[14] * cospi_14_64;
-  step2[9] = dct_const_round_shift(temp1);
-  step2[14] = dct_const_round_shift(temp2);
+  step2[9] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[14] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = step1[10] * cospi_22_64 - step1[13] * cospi_10_64;
   temp2 = step1[10] * cospi_10_64 + step1[13] * cospi_22_64;
-  step2[10] = dct_const_round_shift(temp1);
-  step2[13] = dct_const_round_shift(temp2);
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), 8);
 
   temp1 = step1[11] * cospi_6_64 - step1[12] * cospi_26_64;
   temp2 = step1[11] * cospi_26_64 + step1[12] * cospi_6_64;
-  step2[11] = dct_const_round_shift(temp1);
-  step2[12] = dct_const_round_shift(temp2);
-
-  step2[16] = step1[16] + step1[17];
-  step2[17] = step1[16] - step1[17];
-  step2[18] = -step1[18] + step1[19];
-  step2[19] = step1[18] + step1[19];
-  step2[20] = step1[20] + step1[21];
-  step2[21] = step1[20] - step1[21];
-  step2[22] = -step1[22] + step1[23];
-  step2[23] = step1[22] + step1[23];
-  step2[24] = step1[24] + step1[25];
-  step2[25] = step1[24] - step1[25];
-  step2[26] = -step1[26] + step1[27];
-  step2[27] = step1[26] + step1[27];
-  step2[28] = step1[28] + step1[29];
-  step2[29] = step1[28] - step1[29];
-  step2[30] = -step1[30] + step1[31];
-  step2[31] = step1[30] + step1[31];
+  step2[11] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[12] = WRAPLOW(dct_const_round_shift(temp2), 8);
+
+  step2[16] = WRAPLOW(step1[16] + step1[17], 8);
+  step2[17] = WRAPLOW(step1[16] - step1[17], 8);
+  step2[18] = WRAPLOW(-step1[18] + step1[19], 8);
+  step2[19] = WRAPLOW(step1[18] + step1[19], 8);
+  step2[20] = WRAPLOW(step1[20] + step1[21], 8);
+  step2[21] = WRAPLOW(step1[20] - step1[21], 8);
+  step2[22] = WRAPLOW(-step1[22] + step1[23], 8);
+  step2[23] = WRAPLOW(step1[22] + step1[23], 8);
+  step2[24] = WRAPLOW(step1[24] + step1[25], 8);
+  step2[25] = WRAPLOW(step1[24] - step1[25], 8);
+  step2[26] = WRAPLOW(-step1[26] + step1[27], 8);
+  step2[27] = WRAPLOW(step1[26] + step1[27], 8);
+  step2[28] = WRAPLOW(step1[28] + step1[29], 8);
+  step2[29] = WRAPLOW(step1[28] - step1[29], 8);
+  step2[30] = WRAPLOW(-step1[30] + step1[31], 8);
+  step2[31] = WRAPLOW(step1[30] + step1[31], 8);
 
   // stage 3
   step1[0] = step2[0];
@@ -1028,42 +1047,42 @@ static void idct32(const tran_low_t *input, tran_low_t *output) {
 
   temp1 = step2[4] * cospi_28_64 - step2[7] * cospi_4_64;
   temp2 = step2[4] * cospi_4_64 + step2[7] * cospi_28_64;
-  step1[4] = dct_const_round_shift(temp1);
-  step1[7] = dct_const_round_shift(temp2);
+  step1[4] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[7] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = step2[5] * cospi_12_64 - step2[6] * cospi_20_64;
   temp2 = step2[5] * cospi_20_64 + step2[6] * cospi_12_64;
-  step1[5] = dct_const_round_shift(temp1);
-  step1[6] = dct_const_round_shift(temp2);
-
-  step1[8] = step2[8] + step2[9];
-  step1[9] = step2[8] - step2[9];
-  step1[10] = -step2[10] + step2[11];
-  step1[11] = step2[10] + step2[11];
-  step1[12] = step2[12] + step2[13];
-  step1[13] = step2[12] - step2[13];
-  step1[14] = -step2[14] + step2[15];
-  step1[15] = step2[14] + step2[15];
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), 8);
+
+  step1[8] = WRAPLOW(step2[8] + step2[9], 8);
+  step1[9] = WRAPLOW(step2[8] - step2[9], 8);
+  step1[10] = WRAPLOW(-step2[10] + step2[11], 8);
+  step1[11] = WRAPLOW(step2[10] + step2[11], 8);
+  step1[12] = WRAPLOW(step2[12] + step2[13], 8);
+  step1[13] = WRAPLOW(step2[12] - step2[13], 8);
+  step1[14] = WRAPLOW(-step2[14] + step2[15], 8);
+  step1[15] = WRAPLOW(step2[14] + step2[15], 8);
 
   step1[16] = step2[16];
   step1[31] = step2[31];
   temp1 = -step2[17] * cospi_4_64 + step2[30] * cospi_28_64;
   temp2 = step2[17] * cospi_28_64 + step2[30] * cospi_4_64;
-  step1[17] = dct_const_round_shift(temp1);
-  step1[30] = dct_const_round_shift(temp2);
+  step1[17] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[30] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = -step2[18] * cospi_28_64 - step2[29] * cospi_4_64;
   temp2 = -step2[18] * cospi_4_64 + step2[29] * cospi_28_64;
-  step1[18] = dct_const_round_shift(temp1);
-  step1[29] = dct_const_round_shift(temp2);
+  step1[18] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[29] = WRAPLOW(dct_const_round_shift(temp2), 8);
   step1[19] = step2[19];
   step1[20] = step2[20];
   temp1 = -step2[21] * cospi_20_64 + step2[26] * cospi_12_64;
   temp2 = step2[21] * cospi_12_64 + step2[26] * cospi_20_64;
-  step1[21] = dct_const_round_shift(temp1);
-  step1[26] = dct_const_round_shift(temp2);
+  step1[21] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[26] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = -step2[22] * cospi_12_64 - step2[25] * cospi_20_64;
   temp2 = -step2[22] * cospi_20_64 + step2[25] * cospi_12_64;
-  step1[22] = dct_const_round_shift(temp1);
-  step1[25] = dct_const_round_shift(temp2);
+  step1[22] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[25] = WRAPLOW(dct_const_round_shift(temp2), 8);
   step1[23] = step2[23];
   step1[24] = step2[24];
   step1[27] = step2[27];
@@ -1072,87 +1091,87 @@ static void idct32(const tran_low_t *input, tran_low_t *output) {
   // stage 4
   temp1 = (step1[0] + step1[1]) * cospi_16_64;
   temp2 = (step1[0] - step1[1]) * cospi_16_64;
-  step2[0] = dct_const_round_shift(temp1);
-  step2[1] = dct_const_round_shift(temp2);
+  step2[0] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[1] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = step1[2] * cospi_24_64 - step1[3] * cospi_8_64;
   temp2 = step1[2] * cospi_8_64 + step1[3] * cospi_24_64;
-  step2[2] = dct_const_round_shift(temp1);
-  step2[3] = dct_const_round_shift(temp2);
-  step2[4] = step1[4] + step1[5];
-  step2[5] = step1[4] - step1[5];
-  step2[6] = -step1[6] + step1[7];
-  step2[7] = step1[6] + step1[7];
+  step2[2] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[3] = WRAPLOW(dct_const_round_shift(temp2), 8);
+  step2[4] = WRAPLOW(step1[4] + step1[5], 8);
+  step2[5] = WRAPLOW(step1[4] - step1[5], 8);
+  step2[6] = WRAPLOW(-step1[6] + step1[7], 8);
+  step2[7] = WRAPLOW(step1[6] + step1[7], 8);
 
   step2[8] = step1[8];
   step2[15] = step1[15];
   temp1 = -step1[9] * cospi_8_64 + step1[14] * cospi_24_64;
   temp2 = step1[9] * cospi_24_64 + step1[14] * cospi_8_64;
-  step2[9] = dct_const_round_shift(temp1);
-  step2[14] = dct_const_round_shift(temp2);
+  step2[9] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[14] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = -step1[10] * cospi_24_64 - step1[13] * cospi_8_64;
   temp2 = -step1[10] * cospi_8_64 + step1[13] * cospi_24_64;
-  step2[10] = dct_const_round_shift(temp1);
-  step2[13] = dct_const_round_shift(temp2);
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), 8);
   step2[11] = step1[11];
   step2[12] = step1[12];
 
-  step2[16] = step1[16] + step1[19];
-  step2[17] = step1[17] + step1[18];
-  step2[18] = step1[17] - step1[18];
-  step2[19] = step1[16] - step1[19];
-  step2[20] = -step1[20] + step1[23];
-  step2[21] = -step1[21] + step1[22];
-  step2[22] = step1[21] + step1[22];
-  step2[23] = step1[20] + step1[23];
-
-  step2[24] = step1[24] + step1[27];
-  step2[25] = step1[25] + step1[26];
-  step2[26] = step1[25] - step1[26];
-  step2[27] = step1[24] - step1[27];
-  step2[28] = -step1[28] + step1[31];
-  step2[29] = -step1[29] + step1[30];
-  step2[30] = step1[29] + step1[30];
-  step2[31] = step1[28] + step1[31];
+  step2[16] = WRAPLOW(step1[16] + step1[19], 8);
+  step2[17] = WRAPLOW(step1[17] + step1[18], 8);
+  step2[18] = WRAPLOW(step1[17] - step1[18], 8);
+  step2[19] = WRAPLOW(step1[16] - step1[19], 8);
+  step2[20] = WRAPLOW(-step1[20] + step1[23], 8);
+  step2[21] = WRAPLOW(-step1[21] + step1[22], 8);
+  step2[22] = WRAPLOW(step1[21] + step1[22], 8);
+  step2[23] = WRAPLOW(step1[20] + step1[23], 8);
+
+  step2[24] = WRAPLOW(step1[24] + step1[27], 8);
+  step2[25] = WRAPLOW(step1[25] + step1[26], 8);
+  step2[26] = WRAPLOW(step1[25] - step1[26], 8);
+  step2[27] = WRAPLOW(step1[24] - step1[27], 8);
+  step2[28] = WRAPLOW(-step1[28] + step1[31], 8);
+  step2[29] = WRAPLOW(-step1[29] + step1[30], 8);
+  step2[30] = WRAPLOW(step1[29] + step1[30], 8);
+  step2[31] = WRAPLOW(step1[28] + step1[31], 8);
 
   // stage 5
-  step1[0] = step2[0] + step2[3];
-  step1[1] = step2[1] + step2[2];
-  step1[2] = step2[1] - step2[2];
-  step1[3] = step2[0] - step2[3];
+  step1[0] = WRAPLOW(step2[0] + step2[3], 8);
+  step1[1] = WRAPLOW(step2[1] + step2[2], 8);
+  step1[2] = WRAPLOW(step2[1] - step2[2], 8);
+  step1[3] = WRAPLOW(step2[0] - step2[3], 8);
   step1[4] = step2[4];
   temp1 = (step2[6] - step2[5]) * cospi_16_64;
   temp2 = (step2[5] + step2[6]) * cospi_16_64;
-  step1[5] = dct_const_round_shift(temp1);
-  step1[6] = dct_const_round_shift(temp2);
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), 8);
   step1[7] = step2[7];
 
-  step1[8] = step2[8] + step2[11];
-  step1[9] = step2[9] + step2[10];
-  step1[10] = step2[9] - step2[10];
-  step1[11] = step2[8] - step2[11];
-  step1[12] = -step2[12] + step2[15];
-  step1[13] = -step2[13] + step2[14];
-  step1[14] = step2[13] + step2[14];
-  step1[15] = step2[12] + step2[15];
+  step1[8] = WRAPLOW(step2[8] + step2[11], 8);
+  step1[9] = WRAPLOW(step2[9] + step2[10], 8);
+  step1[10] = WRAPLOW(step2[9] - step2[10], 8);
+  step1[11] = WRAPLOW(step2[8] - step2[11], 8);
+  step1[12] = WRAPLOW(-step2[12] + step2[15], 8);
+  step1[13] = WRAPLOW(-step2[13] + step2[14], 8);
+  step1[14] = WRAPLOW(step2[13] + step2[14], 8);
+  step1[15] = WRAPLOW(step2[12] + step2[15], 8);
 
   step1[16] = step2[16];
   step1[17] = step2[17];
   temp1 = -step2[18] * cospi_8_64 + step2[29] * cospi_24_64;
   temp2 = step2[18] * cospi_24_64 + step2[29] * cospi_8_64;
-  step1[18] = dct_const_round_shift(temp1);
-  step1[29] = dct_const_round_shift(temp2);
+  step1[18] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[29] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = -step2[19] * cospi_8_64 + step2[28] * cospi_24_64;
   temp2 = step2[19] * cospi_24_64 + step2[28] * cospi_8_64;
-  step1[19] = dct_const_round_shift(temp1);
-  step1[28] = dct_const_round_shift(temp2);
+  step1[19] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[28] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = -step2[20] * cospi_24_64 - step2[27] * cospi_8_64;
   temp2 = -step2[20] * cospi_8_64 + step2[27] * cospi_24_64;
-  step1[20] = dct_const_round_shift(temp1);
-  step1[27] = dct_const_round_shift(temp2);
+  step1[20] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[27] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = -step2[21] * cospi_24_64 - step2[26] * cospi_8_64;
   temp2 = -step2[21] * cospi_8_64 + step2[26] * cospi_24_64;
-  step1[21] = dct_const_round_shift(temp1);
-  step1[26] = dct_const_round_shift(temp2);
+  step1[21] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[26] = WRAPLOW(dct_const_round_shift(temp2), 8);
   step1[22] = step2[22];
   step1[23] = step2[23];
   step1[24] = step2[24];
@@ -1161,62 +1180,62 @@ static void idct32(const tran_low_t *input, tran_low_t *output) {
   step1[31] = step2[31];
 
   // stage 6
-  step2[0] = step1[0] + step1[7];
-  step2[1] = step1[1] + step1[6];
-  step2[2] = step1[2] + step1[5];
-  step2[3] = step1[3] + step1[4];
-  step2[4] = step1[3] - step1[4];
-  step2[5] = step1[2] - step1[5];
-  step2[6] = step1[1] - step1[6];
-  step2[7] = step1[0] - step1[7];
+  step2[0] = WRAPLOW(step1[0] + step1[7], 8);
+  step2[1] = WRAPLOW(step1[1] + step1[6], 8);
+  step2[2] = WRAPLOW(step1[2] + step1[5], 8);
+  step2[3] = WRAPLOW(step1[3] + step1[4], 8);
+  step2[4] = WRAPLOW(step1[3] - step1[4], 8);
+  step2[5] = WRAPLOW(step1[2] - step1[5], 8);
+  step2[6] = WRAPLOW(step1[1] - step1[6], 8);
+  step2[7] = WRAPLOW(step1[0] - step1[7], 8);
   step2[8] = step1[8];
   step2[9] = step1[9];
   temp1 = (-step1[10] + step1[13]) * cospi_16_64;
   temp2 = (step1[10] + step1[13]) * cospi_16_64;
-  step2[10] = dct_const_round_shift(temp1);
-  step2[13] = dct_const_round_shift(temp2);
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = (-step1[11] + step1[12]) * cospi_16_64;
   temp2 = (step1[11] + step1[12]) * cospi_16_64;
-  step2[11] = dct_const_round_shift(temp1);
-  step2[12] = dct_const_round_shift(temp2);
+  step2[11] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step2[12] = WRAPLOW(dct_const_round_shift(temp2), 8);
   step2[14] = step1[14];
   step2[15] = step1[15];
 
-  step2[16] = step1[16] + step1[23];
-  step2[17] = step1[17] + step1[22];
-  step2[18] = step1[18] + step1[21];
-  step2[19] = step1[19] + step1[20];
-  step2[20] = step1[19] - step1[20];
-  step2[21] = step1[18] - step1[21];
-  step2[22] = step1[17] - step1[22];
-  step2[23] = step1[16] - step1[23];
-
-  step2[24] = -step1[24] + step1[31];
-  step2[25] = -step1[25] + step1[30];
-  step2[26] = -step1[26] + step1[29];
-  step2[27] = -step1[27] + step1[28];
-  step2[28] = step1[27] + step1[28];
-  step2[29] = step1[26] + step1[29];
-  step2[30] = step1[25] + step1[30];
-  step2[31] = step1[24] + step1[31];
+  step2[16] = WRAPLOW(step1[16] + step1[23], 8);
+  step2[17] = WRAPLOW(step1[17] + step1[22], 8);
+  step2[18] = WRAPLOW(step1[18] + step1[21], 8);
+  step2[19] = WRAPLOW(step1[19] + step1[20], 8);
+  step2[20] = WRAPLOW(step1[19] - step1[20], 8);
+  step2[21] = WRAPLOW(step1[18] - step1[21], 8);
+  step2[22] = WRAPLOW(step1[17] - step1[22], 8);
+  step2[23] = WRAPLOW(step1[16] - step1[23], 8);
+
+  step2[24] = WRAPLOW(-step1[24] + step1[31], 8);
+  step2[25] = WRAPLOW(-step1[25] + step1[30], 8);
+  step2[26] = WRAPLOW(-step1[26] + step1[29], 8);
+  step2[27] = WRAPLOW(-step1[27] + step1[28], 8);
+  step2[28] = WRAPLOW(step1[27] + step1[28], 8);
+  step2[29] = WRAPLOW(step1[26] + step1[29], 8);
+  step2[30] = WRAPLOW(step1[25] + step1[30], 8);
+  step2[31] = WRAPLOW(step1[24] + step1[31], 8);
 
   // stage 7
-  step1[0] = step2[0] + step2[15];
-  step1[1] = step2[1] + step2[14];
-  step1[2] = step2[2] + step2[13];
-  step1[3] = step2[3] + step2[12];
-  step1[4] = step2[4] + step2[11];
-  step1[5] = step2[5] + step2[10];
-  step1[6] = step2[6] + step2[9];
-  step1[7] = step2[7] + step2[8];
-  step1[8] = step2[7] - step2[8];
-  step1[9] = step2[6] - step2[9];
-  step1[10] = step2[5] - step2[10];
-  step1[11] = step2[4] - step2[11];
-  step1[12] = step2[3] - step2[12];
-  step1[13] = step2[2] - step2[13];
-  step1[14] = step2[1] - step2[14];
-  step1[15] = step2[0] - step2[15];
+  step1[0] = WRAPLOW(step2[0] + step2[15], 8);
+  step1[1] = WRAPLOW(step2[1] + step2[14], 8);
+  step1[2] = WRAPLOW(step2[2] + step2[13], 8);
+  step1[3] = WRAPLOW(step2[3] + step2[12], 8);
+  step1[4] = WRAPLOW(step2[4] + step2[11], 8);
+  step1[5] = WRAPLOW(step2[5] + step2[10], 8);
+  step1[6] = WRAPLOW(step2[6] + step2[9], 8);
+  step1[7] = WRAPLOW(step2[7] + step2[8], 8);
+  step1[8] = WRAPLOW(step2[7] - step2[8], 8);
+  step1[9] = WRAPLOW(step2[6] - step2[9], 8);
+  step1[10] = WRAPLOW(step2[5] - step2[10], 8);
+  step1[11] = WRAPLOW(step2[4] - step2[11], 8);
+  step1[12] = WRAPLOW(step2[3] - step2[12], 8);
+  step1[13] = WRAPLOW(step2[2] - step2[13], 8);
+  step1[14] = WRAPLOW(step2[1] - step2[14], 8);
+  step1[15] = WRAPLOW(step2[0] - step2[15], 8);
 
   step1[16] = step2[16];
   step1[17] = step2[17];
@@ -1224,58 +1243,58 @@ static void idct32(const tran_low_t *input, tran_low_t *output) {
   step1[19] = step2[19];
   temp1 = (-step2[20] + step2[27]) * cospi_16_64;
   temp2 = (step2[20] + step2[27]) * cospi_16_64;
-  step1[20] = dct_const_round_shift(temp1);
-  step1[27] = dct_const_round_shift(temp2);
+  step1[20] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[27] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = (-step2[21] + step2[26]) * cospi_16_64;
   temp2 = (step2[21] + step2[26]) * cospi_16_64;
-  step1[21] = dct_const_round_shift(temp1);
-  step1[26] = dct_const_round_shift(temp2);
+  step1[21] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[26] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = (-step2[22] + step2[25]) * cospi_16_64;
   temp2 = (step2[22] + step2[25]) * cospi_16_64;
-  step1[22] = dct_const_round_shift(temp1);
-  step1[25] = dct_const_round_shift(temp2);
+  step1[22] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[25] = WRAPLOW(dct_const_round_shift(temp2), 8);
   temp1 = (-step2[23] + step2[24]) * cospi_16_64;
   temp2 = (step2[23] + step2[24]) * cospi_16_64;
-  step1[23] = dct_const_round_shift(temp1);
-  step1[24] = dct_const_round_shift(temp2);
+  step1[23] = WRAPLOW(dct_const_round_shift(temp1), 8);
+  step1[24] = WRAPLOW(dct_const_round_shift(temp2), 8);
   step1[28] = step2[28];
   step1[29] = step2[29];
   step1[30] = step2[30];
   step1[31] = step2[31];
 
   // final stage
-  output[0] = step1[0] + step1[31];
-  output[1] = step1[1] + step1[30];
-  output[2] = step1[2] + step1[29];
-  output[3] = step1[3] + step1[28];
-  output[4] = step1[4] + step1[27];
-  output[5] = step1[5] + step1[26];
-  output[6] = step1[6] + step1[25];
-  output[7] = step1[7] + step1[24];
-  output[8] = step1[8] + step1[23];
-  output[9] = step1[9] + step1[22];
-  output[10] = step1[10] + step1[21];
-  output[11] = step1[11] + step1[20];
-  output[12] = step1[12] + step1[19];
-  output[13] = step1[13] + step1[18];
-  output[14] = step1[14] + step1[17];
-  output[15] = step1[15] + step1[16];
-  output[16] = step1[15] - step1[16];
-  output[17] = step1[14] - step1[17];
-  output[18] = step1[13] - step1[18];
-  output[19] = step1[12] - step1[19];
-  output[20] = step1[11] - step1[20];
-  output[21] = step1[10] - step1[21];
-  output[22] = step1[9] - step1[22];
-  output[23] = step1[8] - step1[23];
-  output[24] = step1[7] - step1[24];
-  output[25] = step1[6] - step1[25];
-  output[26] = step1[5] - step1[26];
-  output[27] = step1[4] - step1[27];
-  output[28] = step1[3] - step1[28];
-  output[29] = step1[2] - step1[29];
-  output[30] = step1[1] - step1[30];
-  output[31] = step1[0] - step1[31];
+  output[0] = WRAPLOW(step1[0] + step1[31], 8);
+  output[1] = WRAPLOW(step1[1] + step1[30], 8);
+  output[2] = WRAPLOW(step1[2] + step1[29], 8);
+  output[3] = WRAPLOW(step1[3] + step1[28], 8);
+  output[4] = WRAPLOW(step1[4] + step1[27], 8);
+  output[5] = WRAPLOW(step1[5] + step1[26], 8);
+  output[6] = WRAPLOW(step1[6] + step1[25], 8);
+  output[7] = WRAPLOW(step1[7] + step1[24], 8);
+  output[8] = WRAPLOW(step1[8] + step1[23], 8);
+  output[9] = WRAPLOW(step1[9] + step1[22], 8);
+  output[10] = WRAPLOW(step1[10] + step1[21], 8);
+  output[11] = WRAPLOW(step1[11] + step1[20], 8);
+  output[12] = WRAPLOW(step1[12] + step1[19], 8);
+  output[13] = WRAPLOW(step1[13] + step1[18], 8);
+  output[14] = WRAPLOW(step1[14] + step1[17], 8);
+  output[15] = WRAPLOW(step1[15] + step1[16], 8);
+  output[16] = WRAPLOW(step1[15] - step1[16], 8);
+  output[17] = WRAPLOW(step1[14] - step1[17], 8);
+  output[18] = WRAPLOW(step1[13] - step1[18], 8);
+  output[19] = WRAPLOW(step1[12] - step1[19], 8);
+  output[20] = WRAPLOW(step1[11] - step1[20], 8);
+  output[21] = WRAPLOW(step1[10] - step1[21], 8);
+  output[22] = WRAPLOW(step1[9] - step1[22], 8);
+  output[23] = WRAPLOW(step1[8] - step1[23], 8);
+  output[24] = WRAPLOW(step1[7] - step1[24], 8);
+  output[25] = WRAPLOW(step1[6] - step1[25], 8);
+  output[26] = WRAPLOW(step1[5] - step1[26], 8);
+  output[27] = WRAPLOW(step1[4] - step1[27], 8);
+  output[28] = WRAPLOW(step1[3] - step1[28], 8);
+  output[29] = WRAPLOW(step1[2] - step1[29], 8);
+  output[30] = WRAPLOW(step1[1] - step1[30], 8);
+  output[31] = WRAPLOW(step1[0] - step1[31], 8);
 }
 
 void vp9_idct32x32_1024_add_c(const tran_low_t *input, uint8_t *dest,
@@ -1310,9 +1329,10 @@ void vp9_idct32x32_1024_add_c(const tran_low_t *input, uint8_t *dest,
     for (j = 0; j < 32; ++j)
       temp_in[j] = out[j * 32 + i];
     idct32(temp_in, temp_out);
-    for (j = 0; j < 32; ++j)
-      dest[j * stride + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 6)
-                                        + dest[j * stride + i]);
+    for (j = 0; j < 32; ++j) {
+      dest[j * stride + i] = clip_pixel_add(ROUND_POWER_OF_TWO(temp_out[j], 6),
+                                            dest[j * stride + i]);
+    }
   }
 }
 
@@ -1336,9 +1356,10 @@ void vp9_idct32x32_34_add_c(const tran_low_t *input, uint8_t *dest,
     for (j = 0; j < 32; ++j)
       temp_in[j] = out[j * 32 + i];
     idct32(temp_in, temp_out);
-    for (j = 0; j < 32; ++j)
-      dest[j * stride + i] = clip_pixel(ROUND_POWER_OF_TWO(temp_out[j], 6)
-                                  + dest[j * stride + i]);
+    for (j = 0; j < 32; ++j) {
+      dest[j * stride + i] = clip_pixel_add(ROUND_POWER_OF_TWO(temp_out[j], 6),
+                                            dest[j * stride + i]);
+    }
   }
 }
 
@@ -1346,13 +1367,13 @@ void vp9_idct32x32_1_add_c(const tran_low_t *input, uint8_t *dest, int stride) {
   int i, j;
   tran_high_t a1;
 
-  tran_low_t out = dct_const_round_shift(input[0] * cospi_16_64);
-  out = dct_const_round_shift(out * cospi_16_64);
+  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64), 8);
+  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64), 8);
   a1 = ROUND_POWER_OF_TWO(out, 6);
 
   for (j = 0; j < 32; ++j) {
     for (i = 0; i < 32; ++i)
-      dest[i] = clip_pixel(dest[i] + a1);
+      dest[i] = clip_pixel_add(dest[i], a1);
     dest += stride;
   }
 }
@@ -1468,10 +1489,10 @@ void vp9_high_iwht4x4_16_add_c(const tran_low_t *input, uint8_t *dest8,
     c1 = e1 - c1;
     a1 -= b1;
     d1 += c1;
-    op[0] = WRAPLOW(a1);
-    op[1] = WRAPLOW(b1);
-    op[2] = WRAPLOW(c1);
-    op[3] = WRAPLOW(d1);
+    op[0] = WRAPLOW(a1, bd);
+    op[1] = WRAPLOW(b1, bd);
+    op[2] = WRAPLOW(c1, bd);
+    op[3] = WRAPLOW(d1, bd);
     ip += 4;
     op += 4;
   }
@@ -1489,37 +1510,16 @@ void vp9_high_iwht4x4_16_add_c(const tran_low_t *input, uint8_t *dest8,
     c1 = e1 - c1;
     a1 -= b1;
     d1 += c1;
-    dest[stride * 0] = clip_pixel_bd_high(dest[stride * 0], a1, bd);
-    dest[stride * 1] = clip_pixel_bd_high(dest[stride * 1], b1, bd);
-    dest[stride * 2] = clip_pixel_bd_high(dest[stride * 2], c1, bd);
-    dest[stride * 3] = clip_pixel_bd_high(dest[stride * 3], d1, bd);
+    dest[stride * 0] = highbd_clip_pixel_add(dest[stride * 0], a1, bd);
+    dest[stride * 1] = highbd_clip_pixel_add(dest[stride * 1], b1, bd);
+    dest[stride * 2] = highbd_clip_pixel_add(dest[stride * 2], c1, bd);
+    dest[stride * 3] = highbd_clip_pixel_add(dest[stride * 3], d1, bd);
 
     ip++;
     dest++;
   }
 }
 
-static void high_idct4(const tran_low_t *input, tran_low_t *output, int bd) {
-  tran_low_t step[4];
-  tran_high_t temp1, temp2;
-  (void) bd;
-  // stage 1
-  temp1 = (input[0] + input[2]) * cospi_16_64;
-  temp2 = (input[0] - input[2]) * cospi_16_64;
-  step[0] = WRAPLOW(dct_const_round_shift(temp1));
-  step[1] = WRAPLOW(dct_const_round_shift(temp2));
-  temp1 = input[1] * cospi_24_64 - input[3] * cospi_8_64;
-  temp2 = input[1] * cospi_8_64 + input[3] * cospi_24_64;
-  step[2] = WRAPLOW(dct_const_round_shift(temp1));
-  step[3] = WRAPLOW(dct_const_round_shift(temp2));
-
-  // stage 2
-  output[0] = WRAPLOW(step[0] + step[3]);
-  output[1] = WRAPLOW(step[1] + step[2]);
-  output[2] = WRAPLOW(step[1] - step[2]);
-  output[3] = WRAPLOW(step[0] - step[3]);
-}
-
 void vp9_high_iwht4x4_1_add_c(const tran_low_t *in, uint8_t *dest8,
                               int dest_stride, int bd) {
   int i;
@@ -1533,22 +1533,47 @@ void vp9_high_iwht4x4_1_add_c(const tran_low_t *in, uint8_t *dest8,
   a1 = ip[0] >> UNIT_QUANT_SHIFT;
   e1 = a1 >> 1;
   a1 -= e1;
-  op[0] = WRAPLOW(a1);
-  op[1] = op[2] = op[3] = WRAPLOW(e1);
+  op[0] = WRAPLOW(a1, bd);
+  op[1] = op[2] = op[3] = WRAPLOW(e1, bd);
 
   ip = tmp;
   for (i = 0; i < 4; i++) {
     e1 = ip[0] >> 1;
     a1 = ip[0] - e1;
-    dest[dest_stride * 0] = clip_pixel_bd_high(dest[dest_stride * 0], a1, bd);
-    dest[dest_stride * 1] = clip_pixel_bd_high(dest[dest_stride * 1], e1, bd);
-    dest[dest_stride * 2] = clip_pixel_bd_high(dest[dest_stride * 2], e1, bd);
-    dest[dest_stride * 3] = clip_pixel_bd_high(dest[dest_stride * 3], e1, bd);
+    dest[dest_stride * 0] = highbd_clip_pixel_add(
+        dest[dest_stride * 0], a1, bd);
+    dest[dest_stride * 1] = highbd_clip_pixel_add(
+        dest[dest_stride * 1], e1, bd);
+    dest[dest_stride * 2] = highbd_clip_pixel_add(
+        dest[dest_stride * 2], e1, bd);
+    dest[dest_stride * 3] = highbd_clip_pixel_add(
+        dest[dest_stride * 3], e1, bd);
     ip++;
     dest++;
   }
 }
 
+static void highbd_idct4(const tran_low_t *input, tran_low_t *output, int bd) {
+  tran_low_t step[4];
+  tran_high_t temp1, temp2;
+  (void) bd;
+  // stage 1
+  temp1 = (input[0] + input[2]) * cospi_16_64;
+  temp2 = (input[0] - input[2]) * cospi_16_64;
+  step[0] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step[1] = WRAPLOW(dct_const_round_shift(temp2), bd);
+  temp1 = input[1] * cospi_24_64 - input[3] * cospi_8_64;
+  temp2 = input[1] * cospi_8_64 + input[3] * cospi_24_64;
+  step[2] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step[3] = WRAPLOW(dct_const_round_shift(temp2), bd);
+
+  // stage 2
+  output[0] = WRAPLOW(step[0] + step[3], bd);
+  output[1] = WRAPLOW(step[1] + step[2], bd);
+  output[2] = WRAPLOW(step[1] - step[2], bd);
+  output[3] = WRAPLOW(step[0] - step[3], bd);
+}
+
 void vp9_high_idct4x4_16_add_c(const tran_low_t *input, uint8_t *dest8,
                                int stride, int bd) {
   tran_low_t out[4 * 4];
@@ -1559,7 +1584,7 @@ void vp9_high_idct4x4_16_add_c(const tran_low_t *input, uint8_t *dest8,
 
   // Rows
   for (i = 0; i < 4; ++i) {
-    high_idct4(input, outptr, bd);
+    highbd_idct4(input, outptr, bd);
     input += 4;
     outptr += 4;
   }
@@ -1568,10 +1593,11 @@ void vp9_high_idct4x4_16_add_c(const tran_low_t *input, uint8_t *dest8,
   for (i = 0; i < 4; ++i) {
     for (j = 0; j < 4; ++j)
       temp_in[j] = out[j * 4 + i];
-    high_idct4(temp_in, temp_out, bd);
-    for (j = 0; j < 4; ++j)
-      dest[j * stride + i] = clip_pixel_bd_high(
+    highbd_idct4(temp_in, temp_out, bd);
+    for (j = 0; j < 4; ++j) {
+      dest[j * stride + i] = highbd_clip_pixel_add(
           dest[j * stride + i], ROUND_POWER_OF_TWO(temp_out[j], 4), bd);
+    }
   }
 }
 
@@ -1579,22 +1605,22 @@ void vp9_high_idct4x4_1_add_c(const tran_low_t *input, uint8_t *dest8,
                               int dest_stride, int bd) {
   int i;
   tran_high_t a1;
-  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64));
+  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64), bd);
   uint16_t *dest = CONVERT_TO_SHORTPTR(dest8);
 
-  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64));
+  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64), bd);
   a1 = ROUND_POWER_OF_TWO(out, 4);
 
   for (i = 0; i < 4; i++) {
-    dest[0] = clip_pixel_bd_high(dest[0], a1, bd);
-    dest[1] = clip_pixel_bd_high(dest[1], a1, bd);
-    dest[2] = clip_pixel_bd_high(dest[2], a1, bd);
-    dest[3] = clip_pixel_bd_high(dest[3], a1, bd);
+    dest[0] = highbd_clip_pixel_add(dest[0], a1, bd);
+    dest[1] = highbd_clip_pixel_add(dest[1], a1, bd);
+    dest[2] = highbd_clip_pixel_add(dest[2], a1, bd);
+    dest[3] = highbd_clip_pixel_add(dest[3], a1, bd);
     dest += dest_stride;
   }
 }
 
-static void high_idct8(const tran_low_t *input, tran_low_t *output, int bd) {
+static void highbd_idct8(const tran_low_t *input, tran_low_t *output, int bd) {
   tran_low_t step1[8], step2[8];
   tran_high_t temp1, temp2;
   // stage 1
@@ -1604,39 +1630,39 @@ static void high_idct8(const tran_low_t *input, tran_low_t *output, int bd) {
   step1[3] = input[6];
   temp1 = input[1] * cospi_28_64 - input[7] * cospi_4_64;
   temp2 = input[1] * cospi_4_64 + input[7] * cospi_28_64;
-  step1[4] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[7] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[4] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[7] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = input[5] * cospi_12_64 - input[3] * cospi_20_64;
   temp2 = input[5] * cospi_20_64 + input[3] * cospi_12_64;
-  step1[5] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[6] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   // stage 2 & stage 3 - even half
-  high_idct4(step1, step1, bd);
+  highbd_idct4(step1, step1, bd);
 
   // stage 2 - odd half
-  step2[4] = WRAPLOW(step1[4] + step1[5]);
-  step2[5] = WRAPLOW(step1[4] - step1[5]);
-  step2[6] = WRAPLOW(-step1[6] + step1[7]);
-  step2[7] = WRAPLOW(step1[6] + step1[7]);
+  step2[4] = WRAPLOW(step1[4] + step1[5], bd);
+  step2[5] = WRAPLOW(step1[4] - step1[5], bd);
+  step2[6] = WRAPLOW(-step1[6] + step1[7], bd);
+  step2[7] = WRAPLOW(step1[6] + step1[7], bd);
 
   // stage 3 - odd half
   step1[4] = step2[4];
   temp1 = (step2[6] - step2[5]) * cospi_16_64;
   temp2 = (step2[5] + step2[6]) * cospi_16_64;
-  step1[5] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[6] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), bd);
   step1[7] = step2[7];
 
   // stage 4
-  output[0] = WRAPLOW(step1[0] + step1[7]);
-  output[1] = WRAPLOW(step1[1] + step1[6]);
-  output[2] = WRAPLOW(step1[2] + step1[5]);
-  output[3] = WRAPLOW(step1[3] + step1[4]);
-  output[4] = WRAPLOW(step1[3] - step1[4]);
-  output[5] = WRAPLOW(step1[2] - step1[5]);
-  output[6] = WRAPLOW(step1[1] - step1[6]);
-  output[7] = WRAPLOW(step1[0] - step1[7]);
+  output[0] = WRAPLOW(step1[0] + step1[7], bd);
+  output[1] = WRAPLOW(step1[1] + step1[6], bd);
+  output[2] = WRAPLOW(step1[2] + step1[5], bd);
+  output[3] = WRAPLOW(step1[3] + step1[4], bd);
+  output[4] = WRAPLOW(step1[3] - step1[4], bd);
+  output[5] = WRAPLOW(step1[2] - step1[5], bd);
+  output[6] = WRAPLOW(step1[1] - step1[6], bd);
+  output[7] = WRAPLOW(step1[0] - step1[7], bd);
 }
 
 void vp9_high_idct8x8_64_add_c(const tran_low_t *input, uint8_t *dest8,
@@ -1649,7 +1675,7 @@ void vp9_high_idct8x8_64_add_c(const tran_low_t *input, uint8_t *dest8,
 
   // First transform rows.
   for (i = 0; i < 8; ++i) {
-    high_idct8(input, outptr, bd);
+    highbd_idct8(input, outptr, bd);
     input += 8;
     outptr += 8;
   }
@@ -1658,11 +1684,11 @@ void vp9_high_idct8x8_64_add_c(const tran_low_t *input, uint8_t *dest8,
   for (i = 0; i < 8; ++i) {
     for (j = 0; j < 8; ++j)
       temp_in[j] = out[j * 8 + i];
-    high_idct8(temp_in, temp_out, bd);
-    for (j = 0; j < 8; ++j)
-      dest[j * stride + i] = clip_pixel_bd_high(dest[j * stride + i],
-                                        ROUND_POWER_OF_TWO(temp_out[j], 5),
-                                        bd);
+    highbd_idct8(temp_in, temp_out, bd);
+    for (j = 0; j < 8; ++j) {
+      dest[j * stride + i] = highbd_clip_pixel_add(
+          dest[j * stride + i], ROUND_POWER_OF_TWO(temp_out[j], 5), bd);
+    }
   }
 }
 
@@ -1670,18 +1696,18 @@ void vp9_high_idct8x8_1_add_c(const tran_low_t *input, uint8_t *dest8,
                               int stride, int bd) {
   int i, j;
   tran_high_t a1;
-  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64));
+  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64), bd);
   uint16_t *dest = CONVERT_TO_SHORTPTR(dest8);
-  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64));
+  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64), bd);
   a1 = ROUND_POWER_OF_TWO(out, 5);
   for (j = 0; j < 8; ++j) {
     for (i = 0; i < 8; ++i)
-      dest[i] = clip_pixel_bd_high(dest[i], a1, bd);
+      dest[i] = highbd_clip_pixel_add(dest[i], a1, bd);
     dest += stride;
   }
 }
 
-static void high_iadst4(const tran_low_t *input, tran_low_t *output, int bd) {
+static void highbd_iadst4(const tran_low_t *input, tran_low_t *output, int bd) {
   tran_high_t s0, s1, s2, s3, s4, s5, s6, s7;
 
   tran_high_t x0 = input[0];
@@ -1718,19 +1744,19 @@ static void high_iadst4(const tran_low_t *input, tran_low_t *output, int bd) {
   // The overall dynamic range is 14b (input) + 14b (multiplication scaling)
   // + 1b (addition) = 29b.
   // Hence the output bit depth is 15b.
-  output[0] = WRAPLOW(dct_const_round_shift(s0));
-  output[1] = WRAPLOW(dct_const_round_shift(s1));
-  output[2] = WRAPLOW(dct_const_round_shift(s2));
-  output[3] = WRAPLOW(dct_const_round_shift(s3));
+  output[0] = WRAPLOW(dct_const_round_shift(s0), bd);
+  output[1] = WRAPLOW(dct_const_round_shift(s1), bd);
+  output[2] = WRAPLOW(dct_const_round_shift(s2), bd);
+  output[3] = WRAPLOW(dct_const_round_shift(s3), bd);
 }
 
 void vp9_high_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest8,
                               int stride, int tx_type, int bd) {
   const high_transform_2d IHT_4[] = {
-    { high_idct4, high_idct4  },    // DCT_DCT  = 0
-    { high_iadst4, high_idct4 },    // ADST_DCT = 1
-    { high_idct4, high_iadst4 },    // DCT_ADST = 2
-    { high_iadst4, high_iadst4 }    // ADST_ADST = 3
+    { highbd_idct4, highbd_idct4  },    // DCT_DCT  = 0
+    { highbd_iadst4, highbd_idct4 },    // ADST_DCT = 1
+    { highbd_idct4, highbd_iadst4 },    // DCT_ADST = 2
+    { highbd_iadst4, highbd_iadst4 }    // ADST_ADST = 3
   };
   uint16_t *dest = CONVERT_TO_SHORTPTR(dest8);
 
@@ -1751,13 +1777,14 @@ void vp9_high_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest8,
     for (j = 0; j < 4; ++j)
       temp_in[j] = out[j * 4 + i];
     IHT_4[tx_type].cols(temp_in, temp_out, bd);
-    for (j = 0; j < 4; ++j)
-      dest[j * stride + i] = clip_pixel_bd_high(
+    for (j = 0; j < 4; ++j) {
+      dest[j * stride + i] = highbd_clip_pixel_add(
           dest[j * stride + i], ROUND_POWER_OF_TWO(temp_out[j], 4), bd);
+    }
   }
 }
 
-static void high_iadst8(const tran_low_t *input, tran_low_t *output, int bd) {
+static void highbd_iadst8(const tran_low_t *input, tran_low_t *output, int bd) {
   tran_high_t s0, s1, s2, s3, s4, s5, s6, s7;
 
   tran_high_t x0 = input[7];
@@ -1785,14 +1812,14 @@ static void high_iadst8(const tran_low_t *input, tran_low_t *output, int bd) {
   s6 = cospi_26_64 * x6 + cospi_6_64  * x7;
   s7 = cospi_6_64  * x6 - cospi_26_64 * x7;
 
-  x0 = WRAPLOW(dct_const_round_shift(s0 + s4));
-  x1 = WRAPLOW(dct_const_round_shift(s1 + s5));
-  x2 = WRAPLOW(dct_const_round_shift(s2 + s6));
-  x3 = WRAPLOW(dct_const_round_shift(s3 + s7));
-  x4 = WRAPLOW(dct_const_round_shift(s0 - s4));
-  x5 = WRAPLOW(dct_const_round_shift(s1 - s5));
-  x6 = WRAPLOW(dct_const_round_shift(s2 - s6));
-  x7 = WRAPLOW(dct_const_round_shift(s3 - s7));
+  x0 = WRAPLOW(dct_const_round_shift(s0 + s4), bd);
+  x1 = WRAPLOW(dct_const_round_shift(s1 + s5), bd);
+  x2 = WRAPLOW(dct_const_round_shift(s2 + s6), bd);
+  x3 = WRAPLOW(dct_const_round_shift(s3 + s7), bd);
+  x4 = WRAPLOW(dct_const_round_shift(s0 - s4), bd);
+  x5 = WRAPLOW(dct_const_round_shift(s1 - s5), bd);
+  x6 = WRAPLOW(dct_const_round_shift(s2 - s6), bd);
+  x7 = WRAPLOW(dct_const_round_shift(s3 - s7), bd);
 
   // stage 2
   s0 = x0;
@@ -1804,14 +1831,14 @@ static void high_iadst8(const tran_low_t *input, tran_low_t *output, int bd) {
   s6 = -cospi_24_64 * x6 + cospi_8_64  * x7;
   s7 =  cospi_8_64  * x6 + cospi_24_64 * x7;
 
-  x0 = s0 + s2;
-  x1 = s1 + s3;
-  x2 = s0 - s2;
-  x3 = s1 - s3;
-  x4 = WRAPLOW(dct_const_round_shift(s4 + s6));
-  x5 = WRAPLOW(dct_const_round_shift(s5 + s7));
-  x6 = WRAPLOW(dct_const_round_shift(s4 - s6));
-  x7 = WRAPLOW(dct_const_round_shift(s5 - s7));
+  x0 = WRAPLOW(s0 + s2, bd);
+  x1 = WRAPLOW(s1 + s3, bd);
+  x2 = WRAPLOW(s0 - s2, bd);
+  x3 = WRAPLOW(s1 - s3, bd);
+  x4 = WRAPLOW(dct_const_round_shift(s4 + s6), bd);
+  x5 = WRAPLOW(dct_const_round_shift(s5 + s7), bd);
+  x6 = WRAPLOW(dct_const_round_shift(s4 - s6), bd);
+  x7 = WRAPLOW(dct_const_round_shift(s5 - s7), bd);
 
   // stage 3
   s2 = cospi_16_64 * (x2 + x3);
@@ -1819,26 +1846,26 @@ static void high_iadst8(const tran_low_t *input, tran_low_t *output, int bd) {
   s6 = cospi_16_64 * (x6 + x7);
   s7 = cospi_16_64 * (x6 - x7);
 
-  x2 = WRAPLOW(dct_const_round_shift(s2));
-  x3 = WRAPLOW(dct_const_round_shift(s3));
-  x6 = WRAPLOW(dct_const_round_shift(s6));
-  x7 = WRAPLOW(dct_const_round_shift(s7));
+  x2 = WRAPLOW(dct_const_round_shift(s2), bd);
+  x3 = WRAPLOW(dct_const_round_shift(s3), bd);
+  x6 = WRAPLOW(dct_const_round_shift(s6), bd);
+  x7 = WRAPLOW(dct_const_round_shift(s7), bd);
 
-  output[0] = WRAPLOW(x0);
-  output[1] = WRAPLOW(-x4);
-  output[2] = WRAPLOW(x6);
-  output[3] = WRAPLOW(-x2);
-  output[4] = WRAPLOW(x3);
-  output[5] = WRAPLOW(-x7);
-  output[6] = WRAPLOW(x5);
-  output[7] = WRAPLOW(-x1);
+  output[0] = WRAPLOW(x0, bd);
+  output[1] = WRAPLOW(-x4, bd);
+  output[2] = WRAPLOW(x6, bd);
+  output[3] = WRAPLOW(-x2, bd);
+  output[4] = WRAPLOW(x3, bd);
+  output[5] = WRAPLOW(-x7, bd);
+  output[6] = WRAPLOW(x5, bd);
+  output[7] = WRAPLOW(-x1, bd);
 }
 
 static const high_transform_2d HIGH_IHT_8[] = {
-  { high_idct8,  high_idct8  },  // DCT_DCT  = 0
-  { high_iadst8, high_idct8  },  // ADST_DCT = 1
-  { high_idct8,  high_iadst8 },  // DCT_ADST = 2
-  { high_iadst8, high_iadst8 }   // ADST_ADST = 3
+  { highbd_idct8,  highbd_idct8  },  // DCT_DCT  = 0
+  { highbd_iadst8, highbd_idct8  },  // ADST_DCT = 1
+  { highbd_idct8,  highbd_iadst8 },  // DCT_ADST = 2
+  { highbd_iadst8, highbd_iadst8 }   // ADST_ADST = 3
 };
 
 void vp9_high_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest8,
@@ -1862,9 +1889,10 @@ void vp9_high_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest8,
     for (j = 0; j < 8; ++j)
       temp_in[j] = out[j * 8 + i];
     ht.cols(temp_in, temp_out, bd);
-    for (j = 0; j < 8; ++j)
-      dest[j * stride + i] = clip_pixel_bd_high(
+    for (j = 0; j < 8; ++j) {
+      dest[j * stride + i] = highbd_clip_pixel_add(
           dest[j * stride + i], ROUND_POWER_OF_TWO(temp_out[j], 5), bd);
+    }
   }
 }
 
@@ -1879,7 +1907,7 @@ void vp9_high_idct8x8_10_add_c(const tran_low_t *input, uint8_t *dest8,
   // First transform rows.
   // Only first 4 row has non-zero coefs.
   for (i = 0; i < 4; ++i) {
-    high_idct8(input, outptr, bd);
+    highbd_idct8(input, outptr, bd);
     input += 8;
     outptr += 8;
   }
@@ -1887,14 +1915,15 @@ void vp9_high_idct8x8_10_add_c(const tran_low_t *input, uint8_t *dest8,
   for (i = 0; i < 8; ++i) {
     for (j = 0; j < 8; ++j)
       temp_in[j] = out[j * 8 + i];
-    high_idct8(temp_in, temp_out, bd);
-    for (j = 0; j < 8; ++j)
-      dest[j * stride + i] = clip_pixel_bd_high(
+    highbd_idct8(temp_in, temp_out, bd);
+    for (j = 0; j < 8; ++j) {
+      dest[j * stride + i] = highbd_clip_pixel_add(
           dest[j * stride + i], ROUND_POWER_OF_TWO(temp_out[j], 5), bd);
+    }
   }
 }
 
-static void high_idct16(const tran_low_t *input, tran_low_t *output, int bd) {
+static void highbd_idct16(const tran_low_t *input, tran_low_t *output, int bd) {
   tran_low_t step1[16], step2[16];
   tran_high_t temp1, temp2;
   (void) bd;
@@ -1929,23 +1958,23 @@ static void high_idct16(const tran_low_t *input, tran_low_t *output, int bd) {
 
   temp1 = step1[8] * cospi_30_64 - step1[15] * cospi_2_64;
   temp2 = step1[8] * cospi_2_64 + step1[15] * cospi_30_64;
-  step2[8] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[15] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[8] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[15] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = step1[9] * cospi_14_64 - step1[14] * cospi_18_64;
   temp2 = step1[9] * cospi_18_64 + step1[14] * cospi_14_64;
-  step2[9] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[14] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[9] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[14] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = step1[10] * cospi_22_64 - step1[13] * cospi_10_64;
   temp2 = step1[10] * cospi_10_64 + step1[13] * cospi_22_64;
-  step2[10] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[13] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = step1[11] * cospi_6_64 - step1[12] * cospi_26_64;
   temp2 = step1[11] * cospi_26_64 + step1[12] * cospi_6_64;
-  step2[11] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[12] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[11] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[12] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   // stage 3
   step1[0] = step2[0];
@@ -1955,109 +1984,109 @@ static void high_idct16(const tran_low_t *input, tran_low_t *output, int bd) {
 
   temp1 = step2[4] * cospi_28_64 - step2[7] * cospi_4_64;
   temp2 = step2[4] * cospi_4_64 + step2[7] * cospi_28_64;
-  step1[4] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[7] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[4] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[7] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = step2[5] * cospi_12_64 - step2[6] * cospi_20_64;
   temp2 = step2[5] * cospi_20_64 + step2[6] * cospi_12_64;
-  step1[5] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[6] = WRAPLOW(dct_const_round_shift(temp2));
-
-  step1[8] = WRAPLOW(step2[8] + step2[9]);
-  step1[9] = WRAPLOW(step2[8] - step2[9]);
-  step1[10] = WRAPLOW(-step2[10] + step2[11]);
-  step1[11] = WRAPLOW(step2[10] + step2[11]);
-  step1[12] = WRAPLOW(step2[12] + step2[13]);
-  step1[13] = WRAPLOW(step2[12] - step2[13]);
-  step1[14] = WRAPLOW(-step2[14] + step2[15]);
-  step1[15] = WRAPLOW(step2[14] + step2[15]);
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), bd);
+
+  step1[8] = WRAPLOW(step2[8] + step2[9], bd);
+  step1[9] = WRAPLOW(step2[8] - step2[9], bd);
+  step1[10] = WRAPLOW(-step2[10] + step2[11], bd);
+  step1[11] = WRAPLOW(step2[10] + step2[11], bd);
+  step1[12] = WRAPLOW(step2[12] + step2[13], bd);
+  step1[13] = WRAPLOW(step2[12] - step2[13], bd);
+  step1[14] = WRAPLOW(-step2[14] + step2[15], bd);
+  step1[15] = WRAPLOW(step2[14] + step2[15], bd);
 
   // stage 4
   temp1 = (step1[0] + step1[1]) * cospi_16_64;
   temp2 = (step1[0] - step1[1]) * cospi_16_64;
-  step2[0] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[1] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[0] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[1] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = step1[2] * cospi_24_64 - step1[3] * cospi_8_64;
   temp2 = step1[2] * cospi_8_64 + step1[3] * cospi_24_64;
-  step2[2] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[3] = WRAPLOW(dct_const_round_shift(temp2));
-  step2[4] = WRAPLOW(step1[4] + step1[5]);
-  step2[5] = WRAPLOW(step1[4] - step1[5]);
-  step2[6] = WRAPLOW(-step1[6] + step1[7]);
-  step2[7] = WRAPLOW(step1[6] + step1[7]);
+  step2[2] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[3] = WRAPLOW(dct_const_round_shift(temp2), bd);
+  step2[4] = WRAPLOW(step1[4] + step1[5], bd);
+  step2[5] = WRAPLOW(step1[4] - step1[5], bd);
+  step2[6] = WRAPLOW(-step1[6] + step1[7], bd);
+  step2[7] = WRAPLOW(step1[6] + step1[7], bd);
 
   step2[8] = step1[8];
   step2[15] = step1[15];
   temp1 = -step1[9] * cospi_8_64 + step1[14] * cospi_24_64;
   temp2 = step1[9] * cospi_24_64 + step1[14] * cospi_8_64;
-  step2[9] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[14] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[9] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[14] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = -step1[10] * cospi_24_64 - step1[13] * cospi_8_64;
   temp2 = -step1[10] * cospi_8_64 + step1[13] * cospi_24_64;
-  step2[10] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[13] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), bd);
   step2[11] = step1[11];
   step2[12] = step1[12];
 
   // stage 5
-  step1[0] = WRAPLOW(step2[0] + step2[3]);
-  step1[1] = WRAPLOW(step2[1] + step2[2]);
-  step1[2] = WRAPLOW(step2[1] - step2[2]);
-  step1[3] = WRAPLOW(step2[0] - step2[3]);
+  step1[0] = WRAPLOW(step2[0] + step2[3], bd);
+  step1[1] = WRAPLOW(step2[1] + step2[2], bd);
+  step1[2] = WRAPLOW(step2[1] - step2[2], bd);
+  step1[3] = WRAPLOW(step2[0] - step2[3], bd);
   step1[4] = step2[4];
   temp1 = (step2[6] - step2[5]) * cospi_16_64;
   temp2 = (step2[5] + step2[6]) * cospi_16_64;
-  step1[5] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[6] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), bd);
   step1[7] = step2[7];
 
-  step1[8] = WRAPLOW(step2[8] + step2[11]);
-  step1[9] = WRAPLOW(step2[9] + step2[10]);
-  step1[10] = WRAPLOW(step2[9] - step2[10]);
-  step1[11] = WRAPLOW(step2[8] - step2[11]);
-  step1[12] = WRAPLOW(-step2[12] + step2[15]);
-  step1[13] = WRAPLOW(-step2[13] + step2[14]);
-  step1[14] = WRAPLOW(step2[13] + step2[14]);
-  step1[15] = WRAPLOW(step2[12] + step2[15]);
+  step1[8] = WRAPLOW(step2[8] + step2[11], bd);
+  step1[9] = WRAPLOW(step2[9] + step2[10], bd);
+  step1[10] = WRAPLOW(step2[9] - step2[10], bd);
+  step1[11] = WRAPLOW(step2[8] - step2[11], bd);
+  step1[12] = WRAPLOW(-step2[12] + step2[15], bd);
+  step1[13] = WRAPLOW(-step2[13] + step2[14], bd);
+  step1[14] = WRAPLOW(step2[13] + step2[14], bd);
+  step1[15] = WRAPLOW(step2[12] + step2[15], bd);
 
   // stage 6
-  step2[0] = WRAPLOW(step1[0] + step1[7]);
-  step2[1] = WRAPLOW(step1[1] + step1[6]);
-  step2[2] = WRAPLOW(step1[2] + step1[5]);
-  step2[3] = WRAPLOW(step1[3] + step1[4]);
-  step2[4] = WRAPLOW(step1[3] - step1[4]);
-  step2[5] = WRAPLOW(step1[2] - step1[5]);
-  step2[6] = WRAPLOW(step1[1] - step1[6]);
-  step2[7] = WRAPLOW(step1[0] - step1[7]);
+  step2[0] = WRAPLOW(step1[0] + step1[7], bd);
+  step2[1] = WRAPLOW(step1[1] + step1[6], bd);
+  step2[2] = WRAPLOW(step1[2] + step1[5], bd);
+  step2[3] = WRAPLOW(step1[3] + step1[4], bd);
+  step2[4] = WRAPLOW(step1[3] - step1[4], bd);
+  step2[5] = WRAPLOW(step1[2] - step1[5], bd);
+  step2[6] = WRAPLOW(step1[1] - step1[6], bd);
+  step2[7] = WRAPLOW(step1[0] - step1[7], bd);
   step2[8] = step1[8];
   step2[9] = step1[9];
   temp1 = (-step1[10] + step1[13]) * cospi_16_64;
   temp2 = (step1[10] + step1[13]) * cospi_16_64;
-  step2[10] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[13] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = (-step1[11] + step1[12]) * cospi_16_64;
   temp2 = (step1[11] + step1[12]) * cospi_16_64;
-  step2[11] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[12] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[11] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[12] = WRAPLOW(dct_const_round_shift(temp2), bd);
   step2[14] = step1[14];
   step2[15] = step1[15];
 
   // stage 7
-  output[0] = WRAPLOW(step2[0] + step2[15]);
-  output[1] = WRAPLOW(step2[1] + step2[14]);
-  output[2] = WRAPLOW(step2[2] + step2[13]);
-  output[3] = WRAPLOW(step2[3] + step2[12]);
-  output[4] = WRAPLOW(step2[4] + step2[11]);
-  output[5] = WRAPLOW(step2[5] + step2[10]);
-  output[6] = WRAPLOW(step2[6] + step2[9]);
-  output[7] = WRAPLOW(step2[7] + step2[8]);
-  output[8] = WRAPLOW(step2[7] - step2[8]);
-  output[9] = WRAPLOW(step2[6] - step2[9]);
-  output[10] = WRAPLOW(step2[5] - step2[10]);
-  output[11] = WRAPLOW(step2[4] - step2[11]);
-  output[12] = WRAPLOW(step2[3] - step2[12]);
-  output[13] = WRAPLOW(step2[2] - step2[13]);
-  output[14] = WRAPLOW(step2[1] - step2[14]);
-  output[15] = WRAPLOW(step2[0] - step2[15]);
+  output[0] = WRAPLOW(step2[0] + step2[15], bd);
+  output[1] = WRAPLOW(step2[1] + step2[14], bd);
+  output[2] = WRAPLOW(step2[2] + step2[13], bd);
+  output[3] = WRAPLOW(step2[3] + step2[12], bd);
+  output[4] = WRAPLOW(step2[4] + step2[11], bd);
+  output[5] = WRAPLOW(step2[5] + step2[10], bd);
+  output[6] = WRAPLOW(step2[6] + step2[9], bd);
+  output[7] = WRAPLOW(step2[7] + step2[8], bd);
+  output[8] = WRAPLOW(step2[7] - step2[8], bd);
+  output[9] = WRAPLOW(step2[6] - step2[9], bd);
+  output[10] = WRAPLOW(step2[5] - step2[10], bd);
+  output[11] = WRAPLOW(step2[4] - step2[11], bd);
+  output[12] = WRAPLOW(step2[3] - step2[12], bd);
+  output[13] = WRAPLOW(step2[2] - step2[13], bd);
+  output[14] = WRAPLOW(step2[1] - step2[14], bd);
+  output[15] = WRAPLOW(step2[0] - step2[15], bd);
 }
 
 void vp9_high_idct16x16_256_add_c(const tran_low_t *input, uint8_t *dest8,
@@ -2070,7 +2099,7 @@ void vp9_high_idct16x16_256_add_c(const tran_low_t *input, uint8_t *dest8,
 
   // First transform rows.
   for (i = 0; i < 16; ++i) {
-    high_idct16(input, outptr, bd);
+    highbd_idct16(input, outptr, bd);
     input += 16;
     outptr += 16;
   }
@@ -2079,14 +2108,16 @@ void vp9_high_idct16x16_256_add_c(const tran_low_t *input, uint8_t *dest8,
   for (i = 0; i < 16; ++i) {
     for (j = 0; j < 16; ++j)
       temp_in[j] = out[j * 16 + i];
-    high_idct16(temp_in, temp_out, bd);
-    for (j = 0; j < 16; ++j)
-      dest[j * stride + i] = clip_pixel_bd_high(
+    highbd_idct16(temp_in, temp_out, bd);
+    for (j = 0; j < 16; ++j) {
+      dest[j * stride + i] = highbd_clip_pixel_add(
           dest[j * stride + i], ROUND_POWER_OF_TWO(temp_out[j], 6), bd);
+    }
   }
 }
 
-static void high_iadst16(const tran_low_t *input, tran_low_t *output, int bd) {
+static void highbd_iadst16(const tran_low_t *input, tran_low_t *output,
+                           int bd) {
   tran_high_t s0, s1, s2, s3, s4, s5, s6, s7, s8;
   tran_high_t s9, s10, s11, s12, s13, s14, s15;
 
@@ -2132,22 +2163,22 @@ static void high_iadst16(const tran_low_t *input, tran_low_t *output, int bd) {
   s14 = x14 * cospi_29_64 + x15 * cospi_3_64;
   s15 = x14 * cospi_3_64  - x15 * cospi_29_64;
 
-  x0 = WRAPLOW(dct_const_round_shift(s0 + s8));
-  x1 = WRAPLOW(dct_const_round_shift(s1 + s9));
-  x2 = WRAPLOW(dct_const_round_shift(s2 + s10));
-  x3 = WRAPLOW(dct_const_round_shift(s3 + s11));
-  x4 = WRAPLOW(dct_const_round_shift(s4 + s12));
-  x5 = WRAPLOW(dct_const_round_shift(s5 + s13));
-  x6 = WRAPLOW(dct_const_round_shift(s6 + s14));
-  x7 = WRAPLOW(dct_const_round_shift(s7 + s15));
-  x8  = WRAPLOW(dct_const_round_shift(s0 - s8));
-  x9  = WRAPLOW(dct_const_round_shift(s1 - s9));
-  x10 = WRAPLOW(dct_const_round_shift(s2 - s10));
-  x11 = WRAPLOW(dct_const_round_shift(s3 - s11));
-  x12 = WRAPLOW(dct_const_round_shift(s4 - s12));
-  x13 = WRAPLOW(dct_const_round_shift(s5 - s13));
-  x14 = WRAPLOW(dct_const_round_shift(s6 - s14));
-  x15 = WRAPLOW(dct_const_round_shift(s7 - s15));
+  x0 = WRAPLOW(dct_const_round_shift(s0 + s8), bd);
+  x1 = WRAPLOW(dct_const_round_shift(s1 + s9), bd);
+  x2 = WRAPLOW(dct_const_round_shift(s2 + s10), bd);
+  x3 = WRAPLOW(dct_const_round_shift(s3 + s11), bd);
+  x4 = WRAPLOW(dct_const_round_shift(s4 + s12), bd);
+  x5 = WRAPLOW(dct_const_round_shift(s5 + s13), bd);
+  x6 = WRAPLOW(dct_const_round_shift(s6 + s14), bd);
+  x7 = WRAPLOW(dct_const_round_shift(s7 + s15), bd);
+  x8  = WRAPLOW(dct_const_round_shift(s0 - s8), bd);
+  x9  = WRAPLOW(dct_const_round_shift(s1 - s9), bd);
+  x10 = WRAPLOW(dct_const_round_shift(s2 - s10), bd);
+  x11 = WRAPLOW(dct_const_round_shift(s3 - s11), bd);
+  x12 = WRAPLOW(dct_const_round_shift(s4 - s12), bd);
+  x13 = WRAPLOW(dct_const_round_shift(s5 - s13), bd);
+  x14 = WRAPLOW(dct_const_round_shift(s6 - s14), bd);
+  x15 = WRAPLOW(dct_const_round_shift(s7 - s15), bd);
 
   // stage 2
   s0 = x0;
@@ -2167,22 +2198,22 @@ static void high_iadst16(const tran_low_t *input, tran_low_t *output, int bd) {
   s14 = -x14 * cospi_12_64 + x15 * cospi_20_64;
   s15 = x14 * cospi_20_64 + x15 * cospi_12_64;
 
-  x0 = WRAPLOW(s0 + s4);
-  x1 = WRAPLOW(s1 + s5);
-  x2 = WRAPLOW(s2 + s6);
-  x3 = WRAPLOW(s3 + s7);
-  x4 = WRAPLOW(s0 - s4);
-  x5 = WRAPLOW(s1 - s5);
-  x6 = WRAPLOW(s2 - s6);
-  x7 = WRAPLOW(s3 - s7);
-  x8 = WRAPLOW(dct_const_round_shift(s8 + s12));
-  x9 = WRAPLOW(dct_const_round_shift(s9 + s13));
-  x10 = WRAPLOW(dct_const_round_shift(s10 + s14));
-  x11 = WRAPLOW(dct_const_round_shift(s11 + s15));
-  x12 = WRAPLOW(dct_const_round_shift(s8 - s12));
-  x13 = WRAPLOW(dct_const_round_shift(s9 - s13));
-  x14 = WRAPLOW(dct_const_round_shift(s10 - s14));
-  x15 = WRAPLOW(dct_const_round_shift(s11 - s15));
+  x0 = WRAPLOW(s0 + s4, bd);
+  x1 = WRAPLOW(s1 + s5, bd);
+  x2 = WRAPLOW(s2 + s6, bd);
+  x3 = WRAPLOW(s3 + s7, bd);
+  x4 = WRAPLOW(s0 - s4, bd);
+  x5 = WRAPLOW(s1 - s5, bd);
+  x6 = WRAPLOW(s2 - s6, bd);
+  x7 = WRAPLOW(s3 - s7, bd);
+  x8 = WRAPLOW(dct_const_round_shift(s8 + s12), bd);
+  x9 = WRAPLOW(dct_const_round_shift(s9 + s13), bd);
+  x10 = WRAPLOW(dct_const_round_shift(s10 + s14), bd);
+  x11 = WRAPLOW(dct_const_round_shift(s11 + s15), bd);
+  x12 = WRAPLOW(dct_const_round_shift(s8 - s12), bd);
+  x13 = WRAPLOW(dct_const_round_shift(s9 - s13), bd);
+  x14 = WRAPLOW(dct_const_round_shift(s10 - s14), bd);
+  x15 = WRAPLOW(dct_const_round_shift(s11 - s15), bd);
 
   // stage 3
   s0 = x0;
@@ -2202,22 +2233,22 @@ static void high_iadst16(const tran_low_t *input, tran_low_t *output, int bd) {
   s14 = -x14 * cospi_24_64 + x15 * cospi_8_64;
   s15 = x14 * cospi_8_64 + x15 * cospi_24_64;
 
-  x0 = WRAPLOW(s0 + s2);
-  x1 = WRAPLOW(s1 + s3);
-  x2 = WRAPLOW(s0 - s2);
-  x3 = WRAPLOW(s1 - s3);
-  x4 = WRAPLOW(dct_const_round_shift(s4 + s6));
-  x5 = WRAPLOW(dct_const_round_shift(s5 + s7));
-  x6 = WRAPLOW(dct_const_round_shift(s4 - s6));
-  x7 = WRAPLOW(dct_const_round_shift(s5 - s7));
-  x8 = WRAPLOW(s8 + s10);
-  x9 = WRAPLOW(s9 + s11);
-  x10 = WRAPLOW(s8 - s10);
-  x11 = WRAPLOW(s9 - s11);
-  x12 = WRAPLOW(dct_const_round_shift(s12 + s14));
-  x13 = WRAPLOW(dct_const_round_shift(s13 + s15));
-  x14 = WRAPLOW(dct_const_round_shift(s12 - s14));
-  x15 = WRAPLOW(dct_const_round_shift(s13 - s15));
+  x0 = WRAPLOW(s0 + s2, bd);
+  x1 = WRAPLOW(s1 + s3, bd);
+  x2 = WRAPLOW(s0 - s2, bd);
+  x3 = WRAPLOW(s1 - s3, bd);
+  x4 = WRAPLOW(dct_const_round_shift(s4 + s6), bd);
+  x5 = WRAPLOW(dct_const_round_shift(s5 + s7), bd);
+  x6 = WRAPLOW(dct_const_round_shift(s4 - s6), bd);
+  x7 = WRAPLOW(dct_const_round_shift(s5 - s7), bd);
+  x8 = WRAPLOW(s8 + s10, bd);
+  x9 = WRAPLOW(s9 + s11, bd);
+  x10 = WRAPLOW(s8 - s10, bd);
+  x11 = WRAPLOW(s9 - s11, bd);
+  x12 = WRAPLOW(dct_const_round_shift(s12 + s14), bd);
+  x13 = WRAPLOW(dct_const_round_shift(s13 + s15), bd);
+  x14 = WRAPLOW(dct_const_round_shift(s12 - s14), bd);
+  x15 = WRAPLOW(dct_const_round_shift(s13 - s15), bd);
 
   // stage 4
   s2 = (- cospi_16_64) * (x2 + x3);
@@ -2229,38 +2260,38 @@ static void high_iadst16(const tran_low_t *input, tran_low_t *output, int bd) {
   s14 = (- cospi_16_64) * (x14 + x15);
   s15 = cospi_16_64 * (x14 - x15);
 
-  x2 = WRAPLOW(dct_const_round_shift(s2));
-  x3 = WRAPLOW(dct_const_round_shift(s3));
-  x6 = WRAPLOW(dct_const_round_shift(s6));
-  x7 = WRAPLOW(dct_const_round_shift(s7));
-  x10 = WRAPLOW(dct_const_round_shift(s10));
-  x11 = WRAPLOW(dct_const_round_shift(s11));
-  x14 = WRAPLOW(dct_const_round_shift(s14));
-  x15 = WRAPLOW(dct_const_round_shift(s15));
-
-  output[0] = WRAPLOW(x0);
-  output[1] = WRAPLOW(-x8);
-  output[2] = WRAPLOW(x12);
-  output[3] = WRAPLOW(-x4);
-  output[4] = WRAPLOW(x6);
-  output[5] = WRAPLOW(x14);
-  output[6] = WRAPLOW(x10);
-  output[7] = WRAPLOW(x2);
-  output[8] = WRAPLOW(x3);
-  output[9] = WRAPLOW(x11);
-  output[10] = WRAPLOW(x15);
-  output[11] = WRAPLOW(x7);
-  output[12] = WRAPLOW(x5);
-  output[13] = WRAPLOW(-x13);
-  output[14] = WRAPLOW(x9);
-  output[15] = WRAPLOW(-x1);
+  x2 = WRAPLOW(dct_const_round_shift(s2), bd);
+  x3 = WRAPLOW(dct_const_round_shift(s3), bd);
+  x6 = WRAPLOW(dct_const_round_shift(s6), bd);
+  x7 = WRAPLOW(dct_const_round_shift(s7), bd);
+  x10 = WRAPLOW(dct_const_round_shift(s10), bd);
+  x11 = WRAPLOW(dct_const_round_shift(s11), bd);
+  x14 = WRAPLOW(dct_const_round_shift(s14), bd);
+  x15 = WRAPLOW(dct_const_round_shift(s15), bd);
+
+  output[0] = WRAPLOW(x0, bd);
+  output[1] = WRAPLOW(-x8, bd);
+  output[2] = WRAPLOW(x12, bd);
+  output[3] = WRAPLOW(-x4, bd);
+  output[4] = WRAPLOW(x6, bd);
+  output[5] = WRAPLOW(x14, bd);
+  output[6] = WRAPLOW(x10, bd);
+  output[7] = WRAPLOW(x2, bd);
+  output[8] = WRAPLOW(x3, bd);
+  output[9] = WRAPLOW(x11, bd);
+  output[10] = WRAPLOW(x15, bd);
+  output[11] = WRAPLOW(x7, bd);
+  output[12] = WRAPLOW(x5, bd);
+  output[13] = WRAPLOW(-x13, bd);
+  output[14] = WRAPLOW(x9, bd);
+  output[15] = WRAPLOW(-x1, bd);
 }
 
 static const high_transform_2d HIGH_IHT_16[] = {
-  { high_idct16,  high_idct16  },  // DCT_DCT  = 0
-  { high_iadst16, high_idct16  },  // ADST_DCT = 1
-  { high_idct16,  high_iadst16 },  // DCT_ADST = 2
-  { high_iadst16, high_iadst16 }   // ADST_ADST = 3
+  { highbd_idct16,  highbd_idct16  },  // DCT_DCT  = 0
+  { highbd_iadst16, highbd_idct16  },  // ADST_DCT = 1
+  { highbd_idct16,  highbd_iadst16 },  // DCT_ADST = 2
+  { highbd_iadst16, highbd_iadst16 }   // ADST_ADST = 3
 };
 
 void vp9_high_iht16x16_256_add_c(const tran_low_t *input, uint8_t *dest8,
@@ -2284,9 +2315,10 @@ void vp9_high_iht16x16_256_add_c(const tran_low_t *input, uint8_t *dest8,
     for (j = 0; j < 16; ++j)
       temp_in[j] = out[j * 16 + i];
     ht.cols(temp_in, temp_out, bd);
-    for (j = 0; j < 16; ++j)
-      dest[j * stride + i] = clip_pixel_bd_high(
+    for (j = 0; j < 16; ++j) {
+      dest[j * stride + i] = highbd_clip_pixel_add(
           dest[j * stride + i], ROUND_POWER_OF_TWO(temp_out[j], 6), bd);
+    }
   }
 }
 
@@ -2301,7 +2333,7 @@ void vp9_high_idct16x16_10_add_c(const tran_low_t *input, uint8_t *dest8,
   // First transform rows. Since all non-zero dct coefficients are in
   // upper-left 4x4 area, we only need to calculate first 4 rows here.
   for (i = 0; i < 4; ++i) {
-    high_idct16(input, outptr, bd);
+    highbd_idct16(input, outptr, bd);
     input += 16;
     outptr += 16;
   }
@@ -2310,10 +2342,11 @@ void vp9_high_idct16x16_10_add_c(const tran_low_t *input, uint8_t *dest8,
   for (i = 0; i < 16; ++i) {
     for (j = 0; j < 16; ++j)
       temp_in[j] = out[j*16 + i];
-    high_idct16(temp_in, temp_out, bd);
-    for (j = 0; j < 16; ++j)
-      dest[j * stride + i] = clip_pixel_bd_high(
+    highbd_idct16(temp_in, temp_out, bd);
+    for (j = 0; j < 16; ++j) {
+      dest[j * stride + i] = highbd_clip_pixel_add(
           dest[j * stride + i], ROUND_POWER_OF_TWO(temp_out[j], 6), bd);
+    }
   }
 }
 
@@ -2321,19 +2354,19 @@ void vp9_high_idct16x16_1_add_c(const tran_low_t *input, uint8_t *dest8,
                                 int stride, int bd) {
   int i, j;
   tran_high_t a1;
-  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64));
+  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64), bd);
   uint16_t *dest = CONVERT_TO_SHORTPTR(dest8);
 
-  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64));
+  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64), bd);
   a1 = ROUND_POWER_OF_TWO(out, 6);
   for (j = 0; j < 16; ++j) {
     for (i = 0; i < 16; ++i)
-      dest[i] = clip_pixel_bd_high(dest[i], a1, bd);
+      dest[i] = highbd_clip_pixel_add(dest[i], a1, bd);
     dest += stride;
   }
 }
 
-static void high_idct32(const tran_low_t *input, tran_low_t *output, int bd) {
+static void highbd_idct32(const tran_low_t *input, tran_low_t *output, int bd) {
   tran_low_t step1[32], step2[32];
   tran_high_t temp1, temp2;
   (void) bd;
@@ -2358,43 +2391,43 @@ static void high_idct32(const tran_low_t *input, tran_low_t *output, int bd) {
 
   temp1 = input[1] * cospi_31_64 - input[31] * cospi_1_64;
   temp2 = input[1] * cospi_1_64 + input[31] * cospi_31_64;
-  step1[16] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[31] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[16] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[31] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = input[17] * cospi_15_64 - input[15] * cospi_17_64;
   temp2 = input[17] * cospi_17_64 + input[15] * cospi_15_64;
-  step1[17] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[30] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[17] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[30] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = input[9] * cospi_23_64 - input[23] * cospi_9_64;
   temp2 = input[9] * cospi_9_64 + input[23] * cospi_23_64;
-  step1[18] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[29] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[18] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[29] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = input[25] * cospi_7_64 - input[7] * cospi_25_64;
   temp2 = input[25] * cospi_25_64 + input[7] * cospi_7_64;
-  step1[19] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[28] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[19] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[28] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = input[5] * cospi_27_64 - input[27] * cospi_5_64;
   temp2 = input[5] * cospi_5_64 + input[27] * cospi_27_64;
-  step1[20] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[27] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[20] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[27] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = input[21] * cospi_11_64 - input[11] * cospi_21_64;
   temp2 = input[21] * cospi_21_64 + input[11] * cospi_11_64;
-  step1[21] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[26] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[21] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[26] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = input[13] * cospi_19_64 - input[19] * cospi_13_64;
   temp2 = input[13] * cospi_13_64 + input[19] * cospi_19_64;
-  step1[22] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[25] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[22] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[25] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = input[29] * cospi_3_64 - input[3] * cospi_29_64;
   temp2 = input[29] * cospi_29_64 + input[3] * cospi_3_64;
-  step1[23] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[24] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[23] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[24] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   // stage 2
   step2[0] = step1[0];
@@ -2408,40 +2441,40 @@ static void high_idct32(const tran_low_t *input, tran_low_t *output, int bd) {
 
   temp1 = step1[8] * cospi_30_64 - step1[15] * cospi_2_64;
   temp2 = step1[8] * cospi_2_64 + step1[15] * cospi_30_64;
-  step2[8] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[15] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[8] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[15] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = step1[9] * cospi_14_64 - step1[14] * cospi_18_64;
   temp2 = step1[9] * cospi_18_64 + step1[14] * cospi_14_64;
-  step2[9] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[14] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[9] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[14] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = step1[10] * cospi_22_64 - step1[13] * cospi_10_64;
   temp2 = step1[10] * cospi_10_64 + step1[13] * cospi_22_64;
-  step2[10] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[13] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), bd);
 
   temp1 = step1[11] * cospi_6_64 - step1[12] * cospi_26_64;
   temp2 = step1[11] * cospi_26_64 + step1[12] * cospi_6_64;
-  step2[11] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[12] = WRAPLOW(dct_const_round_shift(temp2));
-
-  step2[16] = WRAPLOW(step1[16] + step1[17]);
-  step2[17] = WRAPLOW(step1[16] - step1[17]);
-  step2[18] = WRAPLOW(-step1[18] + step1[19]);
-  step2[19] = WRAPLOW(step1[18] + step1[19]);
-  step2[20] = WRAPLOW(step1[20] + step1[21]);
-  step2[21] = WRAPLOW(step1[20] - step1[21]);
-  step2[22] = WRAPLOW(-step1[22] + step1[23]);
-  step2[23] = WRAPLOW(step1[22] + step1[23]);
-  step2[24] = WRAPLOW(step1[24] + step1[25]);
-  step2[25] = WRAPLOW(step1[24] - step1[25]);
-  step2[26] = WRAPLOW(-step1[26] + step1[27]);
-  step2[27] = WRAPLOW(step1[26] + step1[27]);
-  step2[28] = WRAPLOW(step1[28] + step1[29]);
-  step2[29] = WRAPLOW(step1[28] - step1[29]);
-  step2[30] = WRAPLOW(-step1[30] + step1[31]);
-  step2[31] = WRAPLOW(step1[30] + step1[31]);
+  step2[11] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[12] = WRAPLOW(dct_const_round_shift(temp2), bd);
+
+  step2[16] = WRAPLOW(step1[16] + step1[17], bd);
+  step2[17] = WRAPLOW(step1[16] - step1[17], bd);
+  step2[18] = WRAPLOW(-step1[18] + step1[19], bd);
+  step2[19] = WRAPLOW(step1[18] + step1[19], bd);
+  step2[20] = WRAPLOW(step1[20] + step1[21], bd);
+  step2[21] = WRAPLOW(step1[20] - step1[21], bd);
+  step2[22] = WRAPLOW(-step1[22] + step1[23], bd);
+  step2[23] = WRAPLOW(step1[22] + step1[23], bd);
+  step2[24] = WRAPLOW(step1[24] + step1[25], bd);
+  step2[25] = WRAPLOW(step1[24] - step1[25], bd);
+  step2[26] = WRAPLOW(-step1[26] + step1[27], bd);
+  step2[27] = WRAPLOW(step1[26] + step1[27], bd);
+  step2[28] = WRAPLOW(step1[28] + step1[29], bd);
+  step2[29] = WRAPLOW(step1[28] - step1[29], bd);
+  step2[30] = WRAPLOW(-step1[30] + step1[31], bd);
+  step2[31] = WRAPLOW(step1[30] + step1[31], bd);
 
   // stage 3
   step1[0] = step2[0];
@@ -2451,42 +2484,42 @@ static void high_idct32(const tran_low_t *input, tran_low_t *output, int bd) {
 
   temp1 = step2[4] * cospi_28_64 - step2[7] * cospi_4_64;
   temp2 = step2[4] * cospi_4_64 + step2[7] * cospi_28_64;
-  step1[4] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[7] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[4] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[7] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = step2[5] * cospi_12_64 - step2[6] * cospi_20_64;
   temp2 = step2[5] * cospi_20_64 + step2[6] * cospi_12_64;
-  step1[5] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[6] = WRAPLOW(dct_const_round_shift(temp2));
-
-  step1[8] = WRAPLOW(step2[8] + step2[9]);
-  step1[9] = WRAPLOW(step2[8] - step2[9]);
-  step1[10] = WRAPLOW(-step2[10] + step2[11]);
-  step1[11] = WRAPLOW(step2[10] + step2[11]);
-  step1[12] = WRAPLOW(step2[12] + step2[13]);
-  step1[13] = WRAPLOW(step2[12] - step2[13]);
-  step1[14] = WRAPLOW(-step2[14] + step2[15]);
-  step1[15] = WRAPLOW(step2[14] + step2[15]);
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), bd);
+
+  step1[8] = WRAPLOW(step2[8] + step2[9], bd);
+  step1[9] = WRAPLOW(step2[8] - step2[9], bd);
+  step1[10] = WRAPLOW(-step2[10] + step2[11], bd);
+  step1[11] = WRAPLOW(step2[10] + step2[11], bd);
+  step1[12] = WRAPLOW(step2[12] + step2[13], bd);
+  step1[13] = WRAPLOW(step2[12] - step2[13], bd);
+  step1[14] = WRAPLOW(-step2[14] + step2[15], bd);
+  step1[15] = WRAPLOW(step2[14] + step2[15], bd);
 
   step1[16] = step2[16];
   step1[31] = step2[31];
   temp1 = -step2[17] * cospi_4_64 + step2[30] * cospi_28_64;
   temp2 = step2[17] * cospi_28_64 + step2[30] * cospi_4_64;
-  step1[17] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[30] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[17] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[30] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = -step2[18] * cospi_28_64 - step2[29] * cospi_4_64;
   temp2 = -step2[18] * cospi_4_64 + step2[29] * cospi_28_64;
-  step1[18] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[29] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[18] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[29] = WRAPLOW(dct_const_round_shift(temp2), bd);
   step1[19] = step2[19];
   step1[20] = step2[20];
   temp1 = -step2[21] * cospi_20_64 + step2[26] * cospi_12_64;
   temp2 = step2[21] * cospi_12_64 + step2[26] * cospi_20_64;
-  step1[21] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[26] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[21] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[26] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = -step2[22] * cospi_12_64 - step2[25] * cospi_20_64;
   temp2 = -step2[22] * cospi_20_64 + step2[25] * cospi_12_64;
-  step1[22] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[25] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[22] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[25] = WRAPLOW(dct_const_round_shift(temp2), bd);
   step1[23] = step2[23];
   step1[24] = step2[24];
   step1[27] = step2[27];
@@ -2495,87 +2528,87 @@ static void high_idct32(const tran_low_t *input, tran_low_t *output, int bd) {
   // stage 4
   temp1 = (step1[0] + step1[1]) * cospi_16_64;
   temp2 = (step1[0] - step1[1]) * cospi_16_64;
-  step2[0] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[1] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[0] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[1] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = step1[2] * cospi_24_64 - step1[3] * cospi_8_64;
   temp2 = step1[2] * cospi_8_64 + step1[3] * cospi_24_64;
-  step2[2] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[3] = WRAPLOW(dct_const_round_shift(temp2));
-  step2[4] = WRAPLOW(step1[4] + step1[5]);
-  step2[5] = WRAPLOW(step1[4] - step1[5]);
-  step2[6] = WRAPLOW(-step1[6] + step1[7]);
-  step2[7] = WRAPLOW(step1[6] + step1[7]);
+  step2[2] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[3] = WRAPLOW(dct_const_round_shift(temp2), bd);
+  step2[4] = WRAPLOW(step1[4] + step1[5], bd);
+  step2[5] = WRAPLOW(step1[4] - step1[5], bd);
+  step2[6] = WRAPLOW(-step1[6] + step1[7], bd);
+  step2[7] = WRAPLOW(step1[6] + step1[7], bd);
 
   step2[8] = step1[8];
   step2[15] = step1[15];
   temp1 = -step1[9] * cospi_8_64 + step1[14] * cospi_24_64;
   temp2 = step1[9] * cospi_24_64 + step1[14] * cospi_8_64;
-  step2[9] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[14] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[9] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[14] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = -step1[10] * cospi_24_64 - step1[13] * cospi_8_64;
   temp2 = -step1[10] * cospi_8_64 + step1[13] * cospi_24_64;
-  step2[10] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[13] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), bd);
   step2[11] = step1[11];
   step2[12] = step1[12];
 
-  step2[16] = WRAPLOW(step1[16] + step1[19]);
-  step2[17] = WRAPLOW(step1[17] + step1[18]);
-  step2[18] = WRAPLOW(step1[17] - step1[18]);
-  step2[19] = WRAPLOW(step1[16] - step1[19]);
-  step2[20] = WRAPLOW(-step1[20] + step1[23]);
-  step2[21] = WRAPLOW(-step1[21] + step1[22]);
-  step2[22] = WRAPLOW(step1[21] + step1[22]);
-  step2[23] = WRAPLOW(step1[20] + step1[23]);
-
-  step2[24] = WRAPLOW(step1[24] + step1[27]);
-  step2[25] = WRAPLOW(step1[25] + step1[26]);
-  step2[26] = WRAPLOW(step1[25] - step1[26]);
-  step2[27] = WRAPLOW(step1[24] - step1[27]);
-  step2[28] = WRAPLOW(-step1[28] + step1[31]);
-  step2[29] = WRAPLOW(-step1[29] + step1[30]);
-  step2[30] = WRAPLOW(step1[29] + step1[30]);
-  step2[31] = WRAPLOW(step1[28] + step1[31]);
+  step2[16] = WRAPLOW(step1[16] + step1[19], bd);
+  step2[17] = WRAPLOW(step1[17] + step1[18], bd);
+  step2[18] = WRAPLOW(step1[17] - step1[18], bd);
+  step2[19] = WRAPLOW(step1[16] - step1[19], bd);
+  step2[20] = WRAPLOW(-step1[20] + step1[23], bd);
+  step2[21] = WRAPLOW(-step1[21] + step1[22], bd);
+  step2[22] = WRAPLOW(step1[21] + step1[22], bd);
+  step2[23] = WRAPLOW(step1[20] + step1[23], bd);
+
+  step2[24] = WRAPLOW(step1[24] + step1[27], bd);
+  step2[25] = WRAPLOW(step1[25] + step1[26], bd);
+  step2[26] = WRAPLOW(step1[25] - step1[26], bd);
+  step2[27] = WRAPLOW(step1[24] - step1[27], bd);
+  step2[28] = WRAPLOW(-step1[28] + step1[31], bd);
+  step2[29] = WRAPLOW(-step1[29] + step1[30], bd);
+  step2[30] = WRAPLOW(step1[29] + step1[30], bd);
+  step2[31] = WRAPLOW(step1[28] + step1[31], bd);
 
   // stage 5
-  step1[0] = WRAPLOW(step2[0] + step2[3]);
-  step1[1] = WRAPLOW(step2[1] + step2[2]);
-  step1[2] = WRAPLOW(step2[1] - step2[2]);
-  step1[3] = WRAPLOW(step2[0] - step2[3]);
+  step1[0] = WRAPLOW(step2[0] + step2[3], bd);
+  step1[1] = WRAPLOW(step2[1] + step2[2], bd);
+  step1[2] = WRAPLOW(step2[1] - step2[2], bd);
+  step1[3] = WRAPLOW(step2[0] - step2[3], bd);
   step1[4] = step2[4];
   temp1 = (step2[6] - step2[5]) * cospi_16_64;
   temp2 = (step2[5] + step2[6]) * cospi_16_64;
-  step1[5] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[6] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[5] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[6] = WRAPLOW(dct_const_round_shift(temp2), bd);
   step1[7] = step2[7];
 
-  step1[8] = WRAPLOW(step2[8] + step2[11]);
-  step1[9] = WRAPLOW(step2[9] + step2[10]);
-  step1[10] = WRAPLOW(step2[9] - step2[10]);
-  step1[11] = WRAPLOW(step2[8] - step2[11]);
-  step1[12] = WRAPLOW(-step2[12] + step2[15]);
-  step1[13] = WRAPLOW(-step2[13] + step2[14]);
-  step1[14] = WRAPLOW(step2[13] + step2[14]);
-  step1[15] = WRAPLOW(step2[12] + step2[15]);
+  step1[8] = WRAPLOW(step2[8] + step2[11], bd);
+  step1[9] = WRAPLOW(step2[9] + step2[10], bd);
+  step1[10] = WRAPLOW(step2[9] - step2[10], bd);
+  step1[11] = WRAPLOW(step2[8] - step2[11], bd);
+  step1[12] = WRAPLOW(-step2[12] + step2[15], bd);
+  step1[13] = WRAPLOW(-step2[13] + step2[14], bd);
+  step1[14] = WRAPLOW(step2[13] + step2[14], bd);
+  step1[15] = WRAPLOW(step2[12] + step2[15], bd);
 
   step1[16] = step2[16];
   step1[17] = step2[17];
   temp1 = -step2[18] * cospi_8_64 + step2[29] * cospi_24_64;
   temp2 = step2[18] * cospi_24_64 + step2[29] * cospi_8_64;
-  step1[18] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[29] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[18] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[29] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = -step2[19] * cospi_8_64 + step2[28] * cospi_24_64;
   temp2 = step2[19] * cospi_24_64 + step2[28] * cospi_8_64;
-  step1[19] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[28] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[19] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[28] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = -step2[20] * cospi_24_64 - step2[27] * cospi_8_64;
   temp2 = -step2[20] * cospi_8_64 + step2[27] * cospi_24_64;
-  step1[20] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[27] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[20] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[27] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = -step2[21] * cospi_24_64 - step2[26] * cospi_8_64;
   temp2 = -step2[21] * cospi_8_64 + step2[26] * cospi_24_64;
-  step1[21] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[26] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[21] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[26] = WRAPLOW(dct_const_round_shift(temp2), bd);
   step1[22] = step2[22];
   step1[23] = step2[23];
   step1[24] = step2[24];
@@ -2584,62 +2617,62 @@ static void high_idct32(const tran_low_t *input, tran_low_t *output, int bd) {
   step1[31] = step2[31];
 
   // stage 6
-  step2[0] = WRAPLOW(step1[0] + step1[7]);
-  step2[1] = WRAPLOW(step1[1] + step1[6]);
-  step2[2] = WRAPLOW(step1[2] + step1[5]);
-  step2[3] = WRAPLOW(step1[3] + step1[4]);
-  step2[4] = WRAPLOW(step1[3] - step1[4]);
-  step2[5] = WRAPLOW(step1[2] - step1[5]);
-  step2[6] = WRAPLOW(step1[1] - step1[6]);
-  step2[7] = WRAPLOW(step1[0] - step1[7]);
+  step2[0] = WRAPLOW(step1[0] + step1[7], bd);
+  step2[1] = WRAPLOW(step1[1] + step1[6], bd);
+  step2[2] = WRAPLOW(step1[2] + step1[5], bd);
+  step2[3] = WRAPLOW(step1[3] + step1[4], bd);
+  step2[4] = WRAPLOW(step1[3] - step1[4], bd);
+  step2[5] = WRAPLOW(step1[2] - step1[5], bd);
+  step2[6] = WRAPLOW(step1[1] - step1[6], bd);
+  step2[7] = WRAPLOW(step1[0] - step1[7], bd);
   step2[8] = step1[8];
   step2[9] = step1[9];
   temp1 = (-step1[10] + step1[13]) * cospi_16_64;
   temp2 = (step1[10] + step1[13]) * cospi_16_64;
-  step2[10] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[13] = WRAPLOW(dct_const_round_shift(temp2));
+  step2[10] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[13] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = (-step1[11] + step1[12]) * cospi_16_64;
   temp2 = (step1[11] + step1[12]) * cospi_16_64;
-  step2[11] = WRAPLOW(dct_const_round_shift(temp1));
-  step2[12] = WRAPLOW(dct_const_round_shift(temp2));
-  step2[14] = WRAPLOW(step1[14]);
-  step2[15] = WRAPLOW(step1[15]);
-
-  step2[16] = WRAPLOW(step1[16] + step1[23]);
-  step2[17] = WRAPLOW(step1[17] + step1[22]);
-  step2[18] = WRAPLOW(step1[18] + step1[21]);
-  step2[19] = WRAPLOW(step1[19] + step1[20]);
-  step2[20] = WRAPLOW(step1[19] - step1[20]);
-  step2[21] = WRAPLOW(step1[18] - step1[21]);
-  step2[22] = WRAPLOW(step1[17] - step1[22]);
-  step2[23] = WRAPLOW(step1[16] - step1[23]);
-
-  step2[24] = WRAPLOW(-step1[24] + step1[31]);
-  step2[25] = WRAPLOW(-step1[25] + step1[30]);
-  step2[26] = WRAPLOW(-step1[26] + step1[29]);
-  step2[27] = WRAPLOW(-step1[27] + step1[28]);
-  step2[28] = WRAPLOW(step1[27] + step1[28]);
-  step2[29] = WRAPLOW(step1[26] + step1[29]);
-  step2[30] = WRAPLOW(step1[25] + step1[30]);
-  step2[31] = WRAPLOW(step1[24] + step1[31]);
+  step2[11] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step2[12] = WRAPLOW(dct_const_round_shift(temp2), bd);
+  step2[14] = step1[14];
+  step2[15] = step1[15];
+
+  step2[16] = WRAPLOW(step1[16] + step1[23], bd);
+  step2[17] = WRAPLOW(step1[17] + step1[22], bd);
+  step2[18] = WRAPLOW(step1[18] + step1[21], bd);
+  step2[19] = WRAPLOW(step1[19] + step1[20], bd);
+  step2[20] = WRAPLOW(step1[19] - step1[20], bd);
+  step2[21] = WRAPLOW(step1[18] - step1[21], bd);
+  step2[22] = WRAPLOW(step1[17] - step1[22], bd);
+  step2[23] = WRAPLOW(step1[16] - step1[23], bd);
+
+  step2[24] = WRAPLOW(-step1[24] + step1[31], bd);
+  step2[25] = WRAPLOW(-step1[25] + step1[30], bd);
+  step2[26] = WRAPLOW(-step1[26] + step1[29], bd);
+  step2[27] = WRAPLOW(-step1[27] + step1[28], bd);
+  step2[28] = WRAPLOW(step1[27] + step1[28], bd);
+  step2[29] = WRAPLOW(step1[26] + step1[29], bd);
+  step2[30] = WRAPLOW(step1[25] + step1[30], bd);
+  step2[31] = WRAPLOW(step1[24] + step1[31], bd);
 
   // stage 7
-  step1[0] = WRAPLOW(step2[0] + step2[15]);
-  step1[1] = WRAPLOW(step2[1] + step2[14]);
-  step1[2] = WRAPLOW(step2[2] + step2[13]);
-  step1[3] = WRAPLOW(step2[3] + step2[12]);
-  step1[4] = WRAPLOW(step2[4] + step2[11]);
-  step1[5] = WRAPLOW(step2[5] + step2[10]);
-  step1[6] = WRAPLOW(step2[6] + step2[9]);
-  step1[7] = WRAPLOW(step2[7] + step2[8]);
-  step1[8] = WRAPLOW(step2[7] - step2[8]);
-  step1[9] = WRAPLOW(step2[6] - step2[9]);
-  step1[10] = WRAPLOW(step2[5] - step2[10]);
-  step1[11] = WRAPLOW(step2[4] - step2[11]);
-  step1[12] = WRAPLOW(step2[3] - step2[12]);
-  step1[13] = WRAPLOW(step2[2] - step2[13]);
-  step1[14] = WRAPLOW(step2[1] - step2[14]);
-  step1[15] = WRAPLOW(step2[0] - step2[15]);
+  step1[0] = WRAPLOW(step2[0] + step2[15], bd);
+  step1[1] = WRAPLOW(step2[1] + step2[14], bd);
+  step1[2] = WRAPLOW(step2[2] + step2[13], bd);
+  step1[3] = WRAPLOW(step2[3] + step2[12], bd);
+  step1[4] = WRAPLOW(step2[4] + step2[11], bd);
+  step1[5] = WRAPLOW(step2[5] + step2[10], bd);
+  step1[6] = WRAPLOW(step2[6] + step2[9], bd);
+  step1[7] = WRAPLOW(step2[7] + step2[8], bd);
+  step1[8] = WRAPLOW(step2[7] - step2[8], bd);
+  step1[9] = WRAPLOW(step2[6] - step2[9], bd);
+  step1[10] = WRAPLOW(step2[5] - step2[10], bd);
+  step1[11] = WRAPLOW(step2[4] - step2[11], bd);
+  step1[12] = WRAPLOW(step2[3] - step2[12], bd);
+  step1[13] = WRAPLOW(step2[2] - step2[13], bd);
+  step1[14] = WRAPLOW(step2[1] - step2[14], bd);
+  step1[15] = WRAPLOW(step2[0] - step2[15], bd);
 
   step1[16] = step2[16];
   step1[17] = step2[17];
@@ -2647,58 +2680,58 @@ static void high_idct32(const tran_low_t *input, tran_low_t *output, int bd) {
   step1[19] = step2[19];
   temp1 = (-step2[20] + step2[27]) * cospi_16_64;
   temp2 = (step2[20] + step2[27]) * cospi_16_64;
-  step1[20] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[27] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[20] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[27] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = (-step2[21] + step2[26]) * cospi_16_64;
   temp2 = (step2[21] + step2[26]) * cospi_16_64;
-  step1[21] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[26] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[21] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[26] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = (-step2[22] + step2[25]) * cospi_16_64;
   temp2 = (step2[22] + step2[25]) * cospi_16_64;
-  step1[22] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[25] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[22] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[25] = WRAPLOW(dct_const_round_shift(temp2), bd);
   temp1 = (-step2[23] + step2[24]) * cospi_16_64;
   temp2 = (step2[23] + step2[24]) * cospi_16_64;
-  step1[23] = WRAPLOW(dct_const_round_shift(temp1));
-  step1[24] = WRAPLOW(dct_const_round_shift(temp2));
+  step1[23] = WRAPLOW(dct_const_round_shift(temp1), bd);
+  step1[24] = WRAPLOW(dct_const_round_shift(temp2), bd);
   step1[28] = step2[28];
   step1[29] = step2[29];
   step1[30] = step2[30];
   step1[31] = step2[31];
 
   // final stage
-  output[0] = WRAPLOW(step1[0] + step1[31]);
-  output[1] = WRAPLOW(step1[1] + step1[30]);
-  output[2] = WRAPLOW(step1[2] + step1[29]);
-  output[3] = WRAPLOW(step1[3] + step1[28]);
-  output[4] = WRAPLOW(step1[4] + step1[27]);
-  output[5] = WRAPLOW(step1[5] + step1[26]);
-  output[6] = WRAPLOW(step1[6] + step1[25]);
-  output[7] = WRAPLOW(step1[7] + step1[24]);
-  output[8] = WRAPLOW(step1[8] + step1[23]);
-  output[9] = WRAPLOW(step1[9] + step1[22]);
-  output[10] = WRAPLOW(step1[10] + step1[21]);
-  output[11] = WRAPLOW(step1[11] + step1[20]);
-  output[12] = WRAPLOW(step1[12] + step1[19]);
-  output[13] = WRAPLOW(step1[13] + step1[18]);
-  output[14] = WRAPLOW(step1[14] + step1[17]);
-  output[15] = WRAPLOW(step1[15] + step1[16]);
-  output[16] = WRAPLOW(step1[15] - step1[16]);
-  output[17] = WRAPLOW(step1[14] - step1[17]);
-  output[18] = WRAPLOW(step1[13] - step1[18]);
-  output[19] = WRAPLOW(step1[12] - step1[19]);
-  output[20] = WRAPLOW(step1[11] - step1[20]);
-  output[21] = WRAPLOW(step1[10] - step1[21]);
-  output[22] = WRAPLOW(step1[9] - step1[22]);
-  output[23] = WRAPLOW(step1[8] - step1[23]);
-  output[24] = WRAPLOW(step1[7] - step1[24]);
-  output[25] = WRAPLOW(step1[6] - step1[25]);
-  output[26] = WRAPLOW(step1[5] - step1[26]);
-  output[27] = WRAPLOW(step1[4] - step1[27]);
-  output[28] = WRAPLOW(step1[3] - step1[28]);
-  output[29] = WRAPLOW(step1[2] - step1[29]);
-  output[30] = WRAPLOW(step1[1] - step1[30]);
-  output[31] = WRAPLOW(step1[0] - step1[31]);
+  output[0] = WRAPLOW(step1[0] + step1[31], bd);
+  output[1] = WRAPLOW(step1[1] + step1[30], bd);
+  output[2] = WRAPLOW(step1[2] + step1[29], bd);
+  output[3] = WRAPLOW(step1[3] + step1[28], bd);
+  output[4] = WRAPLOW(step1[4] + step1[27], bd);
+  output[5] = WRAPLOW(step1[5] + step1[26], bd);
+  output[6] = WRAPLOW(step1[6] + step1[25], bd);
+  output[7] = WRAPLOW(step1[7] + step1[24], bd);
+  output[8] = WRAPLOW(step1[8] + step1[23], bd);
+  output[9] = WRAPLOW(step1[9] + step1[22], bd);
+  output[10] = WRAPLOW(step1[10] + step1[21], bd);
+  output[11] = WRAPLOW(step1[11] + step1[20], bd);
+  output[12] = WRAPLOW(step1[12] + step1[19], bd);
+  output[13] = WRAPLOW(step1[13] + step1[18], bd);
+  output[14] = WRAPLOW(step1[14] + step1[17], bd);
+  output[15] = WRAPLOW(step1[15] + step1[16], bd);
+  output[16] = WRAPLOW(step1[15] - step1[16], bd);
+  output[17] = WRAPLOW(step1[14] - step1[17], bd);
+  output[18] = WRAPLOW(step1[13] - step1[18], bd);
+  output[19] = WRAPLOW(step1[12] - step1[19], bd);
+  output[20] = WRAPLOW(step1[11] - step1[20], bd);
+  output[21] = WRAPLOW(step1[10] - step1[21], bd);
+  output[22] = WRAPLOW(step1[9] - step1[22], bd);
+  output[23] = WRAPLOW(step1[8] - step1[23], bd);
+  output[24] = WRAPLOW(step1[7] - step1[24], bd);
+  output[25] = WRAPLOW(step1[6] - step1[25], bd);
+  output[26] = WRAPLOW(step1[5] - step1[26], bd);
+  output[27] = WRAPLOW(step1[4] - step1[27], bd);
+  output[28] = WRAPLOW(step1[3] - step1[28], bd);
+  output[29] = WRAPLOW(step1[2] - step1[29], bd);
+  output[30] = WRAPLOW(step1[1] - step1[30], bd);
+  output[31] = WRAPLOW(step1[0] - step1[31], bd);
 }
 
 void vp9_high_idct32x32_1024_add_c(const tran_low_t *input, uint8_t *dest8,
@@ -2722,7 +2755,7 @@ void vp9_high_idct32x32_1024_add_c(const tran_low_t *input, uint8_t *dest8,
       zero_coeff[j] = zero_coeff[2 * j] | zero_coeff[2 * j + 1];
 
     if (zero_coeff[0] | zero_coeff[1])
-      high_idct32(input, outptr, bd);
+      highbd_idct32(input, outptr, bd);
     else
       vpx_memset(outptr, 0, sizeof(tran_low_t) * 32);
     input += 32;
@@ -2733,10 +2766,11 @@ void vp9_high_idct32x32_1024_add_c(const tran_low_t *input, uint8_t *dest8,
   for (i = 0; i < 32; ++i) {
     for (j = 0; j < 32; ++j)
       temp_in[j] = out[j * 32 + i];
-    high_idct32(temp_in, temp_out, bd);
-    for (j = 0; j < 32; ++j)
-      dest[j * stride + i] = clip_pixel_bd_high(
+    highbd_idct32(temp_in, temp_out, bd);
+    for (j = 0; j < 32; ++j) {
+      dest[j * stride + i] = highbd_clip_pixel_add(
           dest[j * stride + i], ROUND_POWER_OF_TWO(temp_out[j], 6), bd);
+    }
   }
 }
 
@@ -2751,7 +2785,7 @@ void vp9_high_idct32x32_34_add_c(const tran_low_t *input, uint8_t *dest8,
   // Rows
   // Only upper-left 8x8 has non-zero coeff.
   for (i = 0; i < 8; ++i) {
-    high_idct32(input, outptr, bd);
+    highbd_idct32(input, outptr, bd);
     input += 32;
     outptr += 32;
   }
@@ -2759,10 +2793,11 @@ void vp9_high_idct32x32_34_add_c(const tran_low_t *input, uint8_t *dest8,
   for (i = 0; i < 32; ++i) {
     for (j = 0; j < 32; ++j)
       temp_in[j] = out[j * 32 + i];
-    high_idct32(temp_in, temp_out, bd);
-    for (j = 0; j < 32; ++j)
-      dest[j * stride + i] = clip_pixel_bd_high(
+    highbd_idct32(temp_in, temp_out, bd);
+    for (j = 0; j < 32; ++j) {
+      dest[j * stride + i] = highbd_clip_pixel_add(
           dest[j * stride + i], ROUND_POWER_OF_TWO(temp_out[j], 6), bd);
+    }
   }
 }
 
@@ -2772,13 +2807,13 @@ void vp9_high_idct32x32_1_add_c(const tran_low_t *input, uint8_t *dest8,
   int a1;
   uint16_t *dest = CONVERT_TO_SHORTPTR(dest8);
 
-  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64));
-  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64));
+  tran_low_t out = WRAPLOW(dct_const_round_shift(input[0] * cospi_16_64), bd);
+  out = WRAPLOW(dct_const_round_shift(out * cospi_16_64), bd);
   a1 = ROUND_POWER_OF_TWO(out, 6);
 
   for (j = 0; j < 32; ++j) {
     for (i = 0; i < 32; ++i)
-      dest[i] = clip_pixel_bd_high(dest[i], a1, bd);
+      dest[i] = highbd_clip_pixel_add(dest[i], a1, bd);
     dest += stride;
   }
 }
diff --git a/vp9/common/vp9_reconintra.c b/vp9/common/vp9_reconintra.c
index 7ebd2ea87..84d9d34b7 100644
--- a/vp9/common/vp9_reconintra.c
+++ b/vp9/common/vp9_reconintra.c
@@ -251,7 +251,7 @@ static INLINE void high_tm_predictor(uint16_t *dst, ptrdiff_t stride, int bs,
 
   for (r = 0; r < bs; r++) {
     for (c = 0; c < bs; c++)
-      dst[c] = clip_pixel_high(left[r] + above[c] - ytop_left, bd);
+      dst[c] = clip_pixel_highbd(left[r] + above[c] - ytop_left, bd);
     dst += stride;
   }
 }
diff --git a/vp9/common/vp9_rtcd_defs.pl b/vp9/common/vp9_rtcd_defs.pl
index 9451a1c8f..1d9a20670 100644
--- a/vp9/common/vp9_rtcd_defs.pl
+++ b/vp9/common/vp9_rtcd_defs.pl
@@ -331,6 +331,8 @@ $vp9_convolve8_avg_vert_neon_asm=vp9_convolve8_avg_vert_neon;
 # dct
 #
 if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
+  # Note as optimized versions of these functions are added we need to add a check to ensure
+  # that when CONFIG_EMULATE_HARDWARE is on, it defaults to the C versions only.
   add_proto qw/void vp9_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
   specialize qw/vp9_idct4x4_1_add/;
 
@@ -380,69 +382,123 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
 
   add_proto qw/void vp9_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
   specialize qw/vp9_iwht4x4_16_add/;
+
 } else {
-  add_proto qw/void vp9_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_idct4x4_1_add sse2 neon_asm dspr2/;
-  $vp9_idct4x4_1_add_neon_asm=vp9_idct4x4_1_add_neon;
+  # Force C versions if CONFIG_EMULATE_HARDWARE is 1
+  if (vpx_config("CONFIG_EMULATE_HARDWARE") eq "yes") {
+    add_proto qw/void vp9_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct4x4_1_add/;
 
-  add_proto qw/void vp9_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_idct4x4_16_add sse2 neon_asm dspr2/;
-  $vp9_idct4x4_16_add_neon_asm=vp9_idct4x4_16_add_neon;
+    add_proto qw/void vp9_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct4x4_16_add/;
 
-  add_proto qw/void vp9_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_idct8x8_1_add sse2 neon_asm dspr2/;
-  $vp9_idct8x8_1_add_neon_asm=vp9_idct8x8_1_add_neon;
+    add_proto qw/void vp9_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct8x8_1_add/;
 
-  add_proto qw/void vp9_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_idct8x8_64_add sse2 neon_asm dspr2/, "$ssse3_x86_64";
-  $vp9_idct8x8_64_add_neon_asm=vp9_idct8x8_64_add_neon;
+    add_proto qw/void vp9_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct8x8_64_add/;
 
-  add_proto qw/void vp9_idct8x8_12_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_idct8x8_12_add sse2 neon_asm dspr2/, "$ssse3_x86_64";
-  $vp9_idct8x8_12_add_neon_asm=vp9_idct8x8_12_add_neon;
+    add_proto qw/void vp9_idct8x8_12_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct8x8_12_add/;
 
-  add_proto qw/void vp9_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_idct16x16_1_add sse2 neon_asm dspr2/;
-  $vp9_idct16x16_1_add_neon_asm=vp9_idct16x16_1_add_neon;
+    add_proto qw/void vp9_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct16x16_1_add/;
 
-  add_proto qw/void vp9_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_idct16x16_256_add sse2 ssse3 neon_asm dspr2/;
-  $vp9_idct16x16_256_add_neon_asm=vp9_idct16x16_256_add_neon;
+    add_proto qw/void vp9_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct16x16_256_add/;
 
-  add_proto qw/void vp9_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_idct16x16_10_add sse2 ssse3 neon_asm dspr2/;
-  $vp9_idct16x16_10_add_neon_asm=vp9_idct16x16_10_add_neon;
+    add_proto qw/void vp9_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct16x16_10_add/;
 
-  add_proto qw/void vp9_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_idct32x32_1024_add sse2 neon_asm dspr2/;
-  $vp9_idct32x32_1024_add_neon_asm=vp9_idct32x32_1024_add_neon;
+    add_proto qw/void vp9_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct32x32_1024_add/;
 
-  add_proto qw/void vp9_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_idct32x32_34_add sse2 neon_asm dspr2/;
-  $vp9_idct32x32_34_add_neon_asm=vp9_idct32x32_1024_add_neon;
+    add_proto qw/void vp9_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct32x32_34_add/;
 
-  add_proto qw/void vp9_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_idct32x32_1_add sse2 neon_asm dspr2/;
-  $vp9_idct32x32_1_add_neon_asm=vp9_idct32x32_1_add_neon;
+    add_proto qw/void vp9_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct32x32_1_add/;
 
-  add_proto qw/void vp9_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
-  specialize qw/vp9_iht4x4_16_add sse2 neon_asm dspr2/;
-  $vp9_iht4x4_16_add_neon_asm=vp9_iht4x4_16_add_neon;
+    add_proto qw/void vp9_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
+    specialize qw/vp9_iht4x4_16_add/;
 
-  add_proto qw/void vp9_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
-  specialize qw/vp9_iht8x8_64_add sse2 neon_asm dspr2/;
-  $vp9_iht8x8_64_add_neon_asm=vp9_iht8x8_64_add_neon;
+    add_proto qw/void vp9_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
+    specialize qw/vp9_iht8x8_64_add/;
 
-  add_proto qw/void vp9_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, int tx_type";
-  specialize qw/vp9_iht16x16_256_add sse2 dspr2/;
+    add_proto qw/void vp9_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, int tx_type";
+    specialize qw/vp9_iht16x16_256_add/;
 
-  # dct and add
+    # dct and add
 
-  add_proto qw/void vp9_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_iwht4x4_1_add/;
+    add_proto qw/void vp9_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_iwht4x4_1_add/;
 
-  add_proto qw/void vp9_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
-  specialize qw/vp9_iwht4x4_16_add/;
+    add_proto qw/void vp9_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_iwht4x4_16_add/;
+  } else {
+    add_proto qw/void vp9_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct4x4_1_add sse2 neon_asm dspr2/;
+    $vp9_idct4x4_1_add_neon_asm=vp9_idct4x4_1_add_neon;
+
+    add_proto qw/void vp9_idct4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct4x4_16_add sse2 neon_asm dspr2/;
+    $vp9_idct4x4_16_add_neon_asm=vp9_idct4x4_16_add_neon;
+
+    add_proto qw/void vp9_idct8x8_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct8x8_1_add sse2 neon_asm dspr2/;
+    $vp9_idct8x8_1_add_neon_asm=vp9_idct8x8_1_add_neon;
+
+    add_proto qw/void vp9_idct8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct8x8_64_add sse2 neon_asm dspr2/, "$ssse3_x86_64";
+    $vp9_idct8x8_64_add_neon_asm=vp9_idct8x8_64_add_neon;
+
+    add_proto qw/void vp9_idct8x8_12_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct8x8_12_add sse2 neon_asm dspr2/, "$ssse3_x86_64";
+    $vp9_idct8x8_12_add_neon_asm=vp9_idct8x8_12_add_neon;
+
+    add_proto qw/void vp9_idct16x16_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct16x16_1_add sse2 neon_asm dspr2/;
+    $vp9_idct16x16_1_add_neon_asm=vp9_idct16x16_1_add_neon;
+
+    add_proto qw/void vp9_idct16x16_256_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct16x16_256_add sse2 ssse3 neon_asm dspr2/;
+    $vp9_idct16x16_256_add_neon_asm=vp9_idct16x16_256_add_neon;
+
+    add_proto qw/void vp9_idct16x16_10_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct16x16_10_add sse2 ssse3 neon_asm dspr2/;
+    $vp9_idct16x16_10_add_neon_asm=vp9_idct16x16_10_add_neon;
+
+    add_proto qw/void vp9_idct32x32_1024_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct32x32_1024_add sse2 neon_asm dspr2/;
+    $vp9_idct32x32_1024_add_neon_asm=vp9_idct32x32_1024_add_neon;
+
+    add_proto qw/void vp9_idct32x32_34_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct32x32_34_add sse2 neon_asm dspr2/;
+    $vp9_idct32x32_34_add_neon_asm=vp9_idct32x32_1024_add_neon;
+
+    add_proto qw/void vp9_idct32x32_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_idct32x32_1_add sse2 neon_asm dspr2/;
+    $vp9_idct32x32_1_add_neon_asm=vp9_idct32x32_1_add_neon;
+
+    add_proto qw/void vp9_iht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
+    specialize qw/vp9_iht4x4_16_add sse2 neon_asm dspr2/;
+    $vp9_iht4x4_16_add_neon_asm=vp9_iht4x4_16_add_neon;
+
+    add_proto qw/void vp9_iht8x8_64_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type";
+    specialize qw/vp9_iht8x8_64_add sse2 neon_asm dspr2/;
+    $vp9_iht8x8_64_add_neon_asm=vp9_iht8x8_64_add_neon;
+
+    add_proto qw/void vp9_iht16x16_256_add/, "const tran_low_t *input, uint8_t *output, int pitch, int tx_type";
+    specialize qw/vp9_iht16x16_256_add sse2 dspr2/;
+
+    # dct and add
+
+    add_proto qw/void vp9_iwht4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_iwht4x4_1_add/;
+
+    add_proto qw/void vp9_iwht4x4_16_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride";
+    specialize qw/vp9_iwht4x4_16_add/;
+  }
 }
 
 # High bitdepth functions
@@ -689,6 +745,8 @@ if (vpx_config("CONFIG_VP9_HIGHBITDEPTH") eq "yes") {
   #
   # dct
   #
+  # Note as optimized versions of these functions are added we need to add a check to ensure
+  # that when CONFIG_EMULATE_HARDWARE is on, it defaults to the C versions only.
   add_proto qw/void vp9_high_idct4x4_1_add/, "const tran_low_t *input, uint8_t *dest, int dest_stride, int bd";
   specialize qw/vp9_high_idct4x4_1_add/;
 
diff --git a/vp9/encoder/vp9_resize.c b/vp9/encoder/vp9_resize.c
index 4a8a52156..3d361d4f2 100644
--- a/vp9/encoder/vp9_resize.c
+++ b/vp9/encoder/vp9_resize.c
@@ -571,7 +571,7 @@ static void highbd_interpolate(const uint16_t *const input, int inlength,
         sum += filter[k] *
             input[(pk < 0 ? 0 : (pk >= inlength ? inlength - 1 : pk))];
       }
-      *optr++ = clip_pixel_high(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
+      *optr++ = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
     }
   } else {
     // Initial part.
@@ -585,7 +585,7 @@ static void highbd_interpolate(const uint16_t *const input, int inlength,
         sum += filter[k] *
             input[(int_pel - INTERP_TAPS / 2 + 1 + k < 0 ?
                    0 : int_pel - INTERP_TAPS / 2 + 1 + k)];
-      *optr++ = clip_pixel_high(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
+      *optr++ = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
     }
     // Middle part.
     for (; x <= x2; ++x, y += delta) {
@@ -596,7 +596,7 @@ static void highbd_interpolate(const uint16_t *const input, int inlength,
       sum = 0;
       for (k = 0; k < INTERP_TAPS; ++k)
         sum += filter[k] * input[int_pel - INTERP_TAPS / 2 + 1 + k];
-      *optr++ = clip_pixel_high(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
+      *optr++ = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
     }
     // End part.
     for (; x < outlength; ++x, y += delta) {
@@ -609,7 +609,7 @@ static void highbd_interpolate(const uint16_t *const input, int inlength,
         sum += filter[k] * input[(int_pel - INTERP_TAPS / 2 + 1 + k >=
                                   inlength ?  inlength - 1 :
                                   int_pel - INTERP_TAPS / 2 + 1 + k)];
-      *optr++ = clip_pixel_high(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
+      *optr++ = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
     }
   }
 }
@@ -635,7 +635,7 @@ static void highbd_down2_symeven(const uint16_t *const input, int length,
             filter[j];
       }
       sum >>= FILTER_BITS;
-      *optr++ = clip_pixel_high(sum, bd);
+      *optr++ = clip_pixel_highbd(sum, bd);
     }
   } else {
     // Initial part.
@@ -645,7 +645,7 @@ static void highbd_down2_symeven(const uint16_t *const input, int length,
         sum += (input[(i - j < 0 ? 0 : i - j)] + input[i + 1 + j]) * filter[j];
       }
       sum >>= FILTER_BITS;
-      *optr++ = clip_pixel_high(sum, bd);
+      *optr++ = clip_pixel_highbd(sum, bd);
     }
     // Middle part.
     for (; i < l2; i += 2) {
@@ -654,7 +654,7 @@ static void highbd_down2_symeven(const uint16_t *const input, int length,
         sum += (input[i - j] + input[i + 1 + j]) * filter[j];
       }
       sum >>= FILTER_BITS;
-      *optr++ = clip_pixel_high(sum, bd);
+      *optr++ = clip_pixel_highbd(sum, bd);
     }
     // End part.
     for (; i < length; i += 2) {
@@ -665,7 +665,7 @@ static void highbd_down2_symeven(const uint16_t *const input, int length,
             filter[j];
       }
       sum >>= FILTER_BITS;
-      *optr++ = clip_pixel_high(sum, bd);
+      *optr++ = clip_pixel_highbd(sum, bd);
     }
   }
 }
@@ -691,7 +691,7 @@ static void highbd_down2_symodd(const uint16_t *const input, int length,
             filter[j];
       }
       sum >>= FILTER_BITS;
-      *optr++ = clip_pixel_high(sum, bd);
+      *optr++ = clip_pixel_highbd(sum, bd);
     }
   } else {
     // Initial part.
@@ -701,7 +701,7 @@ static void highbd_down2_symodd(const uint16_t *const input, int length,
         sum += (input[(i - j < 0 ? 0 : i - j)] + input[i + j]) * filter[j];
       }
       sum >>= FILTER_BITS;
-      *optr++ = clip_pixel_high(sum, bd);
+      *optr++ = clip_pixel_highbd(sum, bd);
     }
     // Middle part.
     for (; i < l2; i += 2) {
@@ -710,7 +710,7 @@ static void highbd_down2_symodd(const uint16_t *const input, int length,
         sum += (input[i - j] + input[i + j]) * filter[j];
       }
       sum >>= FILTER_BITS;
-      *optr++ = clip_pixel_high(sum, bd);
+      *optr++ = clip_pixel_highbd(sum, bd);
     }
     // End part.
     for (; i < length; i += 2) {
@@ -720,7 +720,7 @@ static void highbd_down2_symodd(const uint16_t *const input, int length,
             filter[j];
       }
       sum >>= FILTER_BITS;
-      *optr++ = clip_pixel_high(sum, bd);
+      *optr++ = clip_pixel_highbd(sum, bd);
     }
   }
 }