Reland: Add ETC1 powered SSE encoder for tile texture compression

cc/resources/texture_compressor_etc1_sse.cc

Index: cc/resources/texture_compressor_etc1_sse.cc
diff --git a/cc/resources/texture_compressor_etc1_sse.cc b/cc/resources/texture_compressor_etc1_sse.cc
new file mode 100644
index 0000000000000000000000000000000000000000..fc4dfb168557f83fc6031582eaf5f1819878aeee
--- /dev/null
+++ b/cc/resources/texture_compressor_etc1_sse.cc
@@ -0,0 +1,824 @@
+// Copyright 2015 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "cc/resources/texture_compressor_etc1_sse.h"
+
+#include <assert.h>

[reveman, 2015/05/07 16:26:29]

is this needed?

[radu.velea, 2015/05/08 10:52:11]

Done.

+#include <emmintrin.h>
+#include <cmath>
+#include <limits>
+
+#include "base/compiler_specific.h"
+#include "base/logging.h"
+// Using this header for common functions such as Color handling
+// and codeword table.
+#include "cc/resources/texture_compressor_etc1.h"
+
+namespace cc {
+
+namespace {
+
+inline uint32_t SetETC1MaxError(uint32_t avg_error) {
+  // ETC1 codeword table is sorted ascending.
+  // Our algorithm will try to identify the index that generates the minimum
+  // error.
+  // The min error calculated during ComputeLuminance main loop will converge
+  // towards that value.
+  // We use this threashold to determine when it doesn't make sense to iterate
+  // further through the array.
+  return avg_error + avg_error / 2 + 384;
+}
+
+struct __sse_data {
+  // This is used to store raw data.
+  uint8_t* block;
+  // This is used to store 8 bit packed values.
+  __m128i* packed;
+  // This is used to store 32 bit zero extended values into 4x4 arrays.
+  __m128i* blue;
+  __m128i* green;
+  __m128i* red;
+};
+
+// Commonly used registers throughout the code.
+static const __m128i __sse_zero = _mm_set1_epi32(0);
+static const __m128i __sse_max_int = _mm_set1_epi32(0x7FFFFFFF);
+
+inline __m128i AddAndClamp(const __m128i x, const __m128i y) {
+  static const __m128i color_max = _mm_set1_epi32(0xFF);
+  return _mm_max_epi16(__sse_zero,
+                       _mm_min_epi16(_mm_add_epi16(x, y), color_max));
+}
+
+inline __m128i GetColorErrorSSE(const __m128i x, const __m128i y) {
+  // Changed from _mm_mullo_epi32 (SSE4) to _mm_mullo_epi16 (SSE2).
+  __m128i ret = _mm_sub_epi16(x, y);
+  return _mm_mullo_epi16(ret, ret);
+}
+
+inline __m128i AddChannelError(const __m128i x,
+                               const __m128i y,
+                               const __m128i z) {
+  return _mm_add_epi32(x, _mm_add_epi32(y, z));
+}
+
+inline uint32_t SumSSE(const __m128i x) {
+  __m128i sum = _mm_add_epi32(x, _mm_shuffle_epi32(x, 0x4E));
+  sum = _mm_add_epi32(sum, _mm_shuffle_epi32(sum, 0xB1));
+
+  return _mm_cvtsi128_si32(sum);
+}
+
+inline uint32_t GetVerticalError(const __sse_data* data,
+                                 const __m128i* blue_avg,
+                                 const __m128i* green_avg,
+                                 const __m128i* red_avg,
+                                 uint32_t* verror) {
+  __m128i error = __sse_zero;
+
+  for (int i = 0; i < 4; i++) {
+    error = _mm_add_epi32(error, GetColorErrorSSE(data->blue[i], blue_avg[0]));
+    error =
+        _mm_add_epi32(error, GetColorErrorSSE(data->green[i], green_avg[0]));
+    error = _mm_add_epi32(error, GetColorErrorSSE(data->red[i], red_avg[0]));
+  }
+
+  error = _mm_add_epi32(error, _mm_shuffle_epi32(error, 0x4E));
+
+  verror[0] = _mm_cvtsi128_si32(error);
+  verror[1] = _mm_cvtsi128_si32(_mm_shuffle_epi32(error, 0xB1));
+
+  return verror[0] + verror[1];
+}
+
+inline uint32_t GetHorizontalError(const __sse_data* data,
+                                   const __m128i* blue_avg,
+                                   const __m128i* green_avg,
+                                   const __m128i* red_avg,
+                                   uint32_t* verror) {
+  __m128i error = __sse_zero;
+  int first_index, second_index;
+
+  for (int i = 0; i < 2; i++) {
+    first_index = 2 * i;
+    second_index = first_index + 1;
+
+    error = _mm_add_epi32(
+        error, GetColorErrorSSE(data->blue[first_index], blue_avg[i]));
+    error = _mm_add_epi32(
+        error, GetColorErrorSSE(data->blue[second_index], blue_avg[i]));
+    error = _mm_add_epi32(
+        error, GetColorErrorSSE(data->green[first_index], green_avg[i]));
+    error = _mm_add_epi32(
+        error, GetColorErrorSSE(data->green[second_index], green_avg[i]));
+    error = _mm_add_epi32(error,
+                          GetColorErrorSSE(data->red[first_index], red_avg[i]));
+    error = _mm_add_epi32(
+        error, GetColorErrorSSE(data->red[second_index], red_avg[i]));
+  }
+
+  error = _mm_add_epi32(error, _mm_shuffle_epi32(error, 0x4E));
+
+  verror[0] = _mm_cvtsi128_si32(error);
+  verror[1] = _mm_cvtsi128_si32(_mm_shuffle_epi32(error, 0xB1));
+
+  return verror[0] + verror[1];
+}
+
+inline void GetAvgColors(const __sse_data* data,
+                         float* output,
+                         bool* __sse_use_diff) {
+  __m128i sum[2], tmp;
+
+  // TODO(radu.velea): _mm_avg_epu8 on packed data maybe.
+
+  // Compute avg red value.
+  // [S0 S0 S1 S1]
+  sum[0] = _mm_add_epi32(data->red[0], data->red[1]);
+  sum[0] = _mm_add_epi32(sum[0], _mm_shuffle_epi32(sum[0], 0xB1));
+
+  // [S2 S2 S3 S3]
+  sum[1] = _mm_add_epi32(data->red[2], data->red[3]);
+  sum[1] = _mm_add_epi32(sum[1], _mm_shuffle_epi32(sum[1], 0xB1));
+
+  float hred[2], vred[2];
+  hred[0] = (_mm_cvtsi128_si32(
+                _mm_add_epi32(sum[0], _mm_shuffle_epi32(sum[0], 0x4E)))) /
+            8.0f;
+  hred[1] = (_mm_cvtsi128_si32(
+                _mm_add_epi32(sum[1], _mm_shuffle_epi32(sum[1], 0x4E)))) /
+            8.0f;
+
+  tmp = _mm_add_epi32(sum[0], sum[1]);
+  vred[0] = (_mm_cvtsi128_si32(tmp)) / 8.0f;
+  vred[1] = (_mm_cvtsi128_si32(_mm_shuffle_epi32(tmp, 0x2))) / 8.0f;
+
+  // Compute avg green value.
+  // [S0 S0 S1 S1]
+  sum[0] = _mm_add_epi32(data->green[0], data->green[1]);
+  sum[0] = _mm_add_epi32(sum[0], _mm_shuffle_epi32(sum[0], 0xB1));
+
+  // [S2 S2 S3 S3]
+  sum[1] = _mm_add_epi32(data->green[2], data->green[3]);
+  sum[1] = _mm_add_epi32(sum[1], _mm_shuffle_epi32(sum[1], 0xB1));
+
+  float hgreen[2], vgreen[2];
+  hgreen[0] = (_mm_cvtsi128_si32(
+                  _mm_add_epi32(sum[0], _mm_shuffle_epi32(sum[0], 0x4E)))) /
+              8.0f;
+  hgreen[1] = (_mm_cvtsi128_si32(
+                  _mm_add_epi32(sum[1], _mm_shuffle_epi32(sum[1], 0x4E)))) /
+              8.0f;
+
+  tmp = _mm_add_epi32(sum[0], sum[1]);
+  vgreen[0] = (_mm_cvtsi128_si32(tmp)) / 8.0f;
+  vgreen[1] = (_mm_cvtsi128_si32(_mm_shuffle_epi32(tmp, 0x2))) / 8.0f;
+
+  // Compute avg blue value.
+  // [S0 S0 S1 S1]
+  sum[0] = _mm_add_epi32(data->blue[0], data->blue[1]);
+  sum[0] = _mm_add_epi32(sum[0], _mm_shuffle_epi32(sum[0], 0xB1));
+
+  // [S2 S2 S3 S3]
+  sum[1] = _mm_add_epi32(data->blue[2], data->blue[3]);
+  sum[1] = _mm_add_epi32(sum[1], _mm_shuffle_epi32(sum[1], 0xB1));
+
+  float hblue[2], vblue[2];
+  hblue[0] = (_mm_cvtsi128_si32(
+                 _mm_add_epi32(sum[0], _mm_shuffle_epi32(sum[0], 0x4E)))) /
+             8.0f;
+  hblue[1] = (_mm_cvtsi128_si32(
+                 _mm_add_epi32(sum[1], _mm_shuffle_epi32(sum[1], 0x4E)))) /
+             8.0f;
+
+  tmp = _mm_add_epi32(sum[0], sum[1]);
+  vblue[0] = (_mm_cvtsi128_si32(tmp)) / 8.0f;
+  vblue[1] = (_mm_cvtsi128_si32(_mm_shuffle_epi32(tmp, 0x2))) / 8.0f;
+
+  // TODO(radu.velea): Return int's instead of floats, based on Quality.
+  output[0] = vblue[0];
+  output[1] = vgreen[0];
+  output[2] = vred[0];
+
+  output[3] = vblue[1];
+  output[4] = vgreen[1];
+  output[5] = vred[1];
+
+  output[6] = hblue[0];
+  output[7] = hgreen[0];
+  output[8] = hred[0];
+
+  output[9] = hblue[1];
+  output[10] = hgreen[1];
+  output[11] = hred[1];
+
+  __m128i threshold_upper = _mm_set1_epi32(3);
+  __m128i threshold_lower = _mm_set1_epi32(-4);
+
+  __m128 factor_v = _mm_set1_ps(31.0f / 255.0f);
+  __m128 rounding_v = _mm_set1_ps(0.5f);
+  __m128 h_avg_0 = _mm_set_ps(hblue[0], hgreen[0], hred[0], 0);
+  __m128 h_avg_1 = _mm_set_ps(hblue[1], hgreen[1], hred[1], 0);
+
+  __m128 v_avg_0 = _mm_set_ps(vblue[0], vgreen[0], vred[0], 0);
+  __m128 v_avg_1 = _mm_set_ps(vblue[1], vgreen[1], vred[1], 0);
+
+  h_avg_0 = _mm_mul_ps(h_avg_0, factor_v);
+  h_avg_1 = _mm_mul_ps(h_avg_1, factor_v);
+  v_avg_0 = _mm_mul_ps(v_avg_0, factor_v);
+  v_avg_1 = _mm_mul_ps(v_avg_1, factor_v);
+
+  h_avg_0 = _mm_add_ps(h_avg_0, rounding_v);
+  h_avg_1 = _mm_add_ps(h_avg_1, rounding_v);
+  v_avg_0 = _mm_add_ps(v_avg_0, rounding_v);
+  v_avg_1 = _mm_add_ps(v_avg_1, rounding_v);
+
+  __m128i h_avg_0i = _mm_cvttps_epi32(h_avg_0);
+  __m128i h_avg_1i = _mm_cvttps_epi32(h_avg_1);
+
+  __m128i v_avg_0i = _mm_cvttps_epi32(v_avg_0);
+  __m128i v_avg_1i = _mm_cvttps_epi32(v_avg_1);
+
+  h_avg_0i = _mm_sub_epi32(h_avg_1i, h_avg_0i);
+  v_avg_0i = _mm_sub_epi32(v_avg_1i, v_avg_0i);
+
+  __sse_use_diff[0] =
+      (0 == _mm_movemask_epi8(_mm_cmplt_epi32(v_avg_0i, threshold_lower)));
+  __sse_use_diff[0] &=
+      (0 == _mm_movemask_epi8(_mm_cmpgt_epi32(v_avg_0i, threshold_upper)));
+
+  __sse_use_diff[1] =
+      (0 == _mm_movemask_epi8(_mm_cmplt_epi32(h_avg_0i, threshold_lower)));
+  __sse_use_diff[1] &=
+      (0 == _mm_movemask_epi8(_mm_cmpgt_epi32(h_avg_0i, threshold_upper)));
+}
+
+void ComputeLuminance(uint8_t* block,
+                      const Color& base,
+                      const int sub_block_id,
+                      const uint8_t* idx_to_num_tab,
+                      const __sse_data* data,
+                      const uint32_t expected_error) {
+  uint8_t best_tbl_idx = 0;
+  uint32_t best_error = 0x7FFFFFFF;
+  uint8_t best_mod_idx[8][8];  // [table][texel]
+
+  const __m128i base_blue = _mm_set1_epi32(base.channels.b);
+  const __m128i base_green = _mm_set1_epi32(base.channels.g);
+  const __m128i base_red = _mm_set1_epi32(base.channels.r);
+
+  __m128i test_red, test_blue, test_green, tmp, tmp_blue, tmp_green, tmp_red;
+  __m128i block_error, mask;
+
+  // This will have the minimum errors for each 4 pixels.
+  __m128i first_half_min;
+  __m128i second_half_min;
+
+  // This will have the matching table index combo for each 4 pixels.
+  __m128i first_half_pattern;
+  __m128i second_half_pattern;
+
+  const __m128i first_blue_data_block = data->blue[2 * sub_block_id];
+  const __m128i first_green_data_block = data->green[2 * sub_block_id];
+  const __m128i first_red_data_block = data->red[2 * sub_block_id];
+
+  const __m128i second_blue_data_block = data->blue[2 * sub_block_id + 1];
+  const __m128i second_green_data_block = data->green[2 * sub_block_id + 1];
+  const __m128i second_red_data_block = data->red[2 * sub_block_id + 1];
+
+  uint32_t min;
+  // Fail early to increase speed.
+  long delta = INT32_MAX;
+  uint32_t last_min = INT32_MAX;
+
+  const uint8_t shuffle_mask[] = {
+      0x1B, 0x4E, 0xB1, 0xE4};  // Important they are sorted ascending.
+
+  for (unsigned int tbl_idx = 0; tbl_idx < 8; ++tbl_idx) {
+    tmp = _mm_set_epi32(
+        g_codeword_tables[tbl_idx][3], g_codeword_tables[tbl_idx][2],
+        g_codeword_tables[tbl_idx][1], g_codeword_tables[tbl_idx][0]);
+
+    test_blue = AddAndClamp(tmp, base_blue);
+    test_green = AddAndClamp(tmp, base_green);
+    test_red = AddAndClamp(tmp, base_red);
+
+    first_half_min = __sse_max_int;
+    second_half_min = __sse_max_int;
+
+    first_half_pattern = __sse_zero;
+    second_half_pattern = __sse_zero;
+
+    for (uint8_t imm8 : shuffle_mask) {
+      switch (imm8) {
+        case 0x1B:
+          tmp_blue = _mm_shuffle_epi32(test_blue, 0x1B);
+          tmp_green = _mm_shuffle_epi32(test_green, 0x1B);
+          tmp_red = _mm_shuffle_epi32(test_red, 0x1B);
+          break;
+        case 0x4E:
+          tmp_blue = _mm_shuffle_epi32(test_blue, 0x4E);
+          tmp_green = _mm_shuffle_epi32(test_green, 0x4E);
+          tmp_red = _mm_shuffle_epi32(test_red, 0x4E);
+          break;
+        case 0xB1:
+          tmp_blue = _mm_shuffle_epi32(test_blue, 0xB1);
+          tmp_green = _mm_shuffle_epi32(test_green, 0xB1);
+          tmp_red = _mm_shuffle_epi32(test_red, 0xB1);
+          break;
+        case 0xE4:
+          tmp_blue = _mm_shuffle_epi32(test_blue, 0xE4);
+          tmp_green = _mm_shuffle_epi32(test_green, 0xE4);
+          tmp_red = _mm_shuffle_epi32(test_red, 0xE4);
+          break;
+        default:
+          tmp_blue = test_blue;
+          tmp_green = test_green;
+          tmp_red = test_red;
+      }
+
+      tmp = _mm_set1_epi32(imm8);
+
+      block_error =
+          AddChannelError(GetColorErrorSSE(tmp_blue, first_blue_data_block),
+                          GetColorErrorSSE(tmp_green, first_green_data_block),
+                          GetColorErrorSSE(tmp_red, first_red_data_block));
+
+      // Save winning pattern.
+      first_half_pattern = _mm_max_epi16(
+          first_half_pattern,
+          _mm_and_si128(tmp, _mm_cmpgt_epi32(first_half_min, block_error)));
+      // Should use _mm_min_epi32(first_half_min, block_error); from SSE4
+      // otherwise we have a small performance penalty.
+      mask = _mm_cmplt_epi32(block_error, first_half_min);
+      first_half_min = _mm_add_epi32(_mm_and_si128(mask, block_error),
+                                     _mm_andnot_si128(mask, first_half_min));
+
+      // Compute second part of the block.
+      block_error =
+          AddChannelError(GetColorErrorSSE(tmp_blue, second_blue_data_block),
+                          GetColorErrorSSE(tmp_green, second_green_data_block),
+                          GetColorErrorSSE(tmp_red, second_red_data_block));
+
+      // Save winning pattern.
+      second_half_pattern = _mm_max_epi16(
+          second_half_pattern,
+          _mm_and_si128(tmp, _mm_cmpgt_epi32(second_half_min, block_error)));
+      // Should use _mm_min_epi32(second_half_min, block_error); from SSE4
+      // otherwise we have a small performance penalty.
+      mask = _mm_cmplt_epi32(block_error, second_half_min);
+      second_half_min = _mm_add_epi32(_mm_and_si128(mask, block_error),
+                                      _mm_andnot_si128(mask, second_half_min));
+    }
+
+    first_half_min = _mm_add_epi32(first_half_min, second_half_min);
+    first_half_min =
+        _mm_add_epi32(first_half_min, _mm_shuffle_epi32(first_half_min, 0x4E));
+    first_half_min =
+        _mm_add_epi32(first_half_min, _mm_shuffle_epi32(first_half_min, 0xB1));
+
+    min = _mm_cvtsi128_si32(first_half_min);
+
+    delta = min - last_min;
+    last_min = min;
+
+    if (min < best_error) {
+      best_tbl_idx = tbl_idx;
+      best_error = min;
+
+      best_mod_idx[tbl_idx][0] =
+          (_mm_cvtsi128_si32(first_half_pattern) >> (0)) & 3;
+      best_mod_idx[tbl_idx][4] =
+          (_mm_cvtsi128_si32(second_half_pattern) >> (0)) & 3;
+
+      best_mod_idx[tbl_idx][1] =
+          (_mm_cvtsi128_si32(_mm_shuffle_epi32(first_half_pattern, 0x1)) >>
+           (2)) &
+          3;
+      best_mod_idx[tbl_idx][5] =
+          (_mm_cvtsi128_si32(_mm_shuffle_epi32(second_half_pattern, 0x1)) >>
+           (2)) &
+          3;
+
+      best_mod_idx[tbl_idx][2] =
+          (_mm_cvtsi128_si32(_mm_shuffle_epi32(first_half_pattern, 0x2)) >>
+           (4)) &
+          3;
+      best_mod_idx[tbl_idx][6] =
+          (_mm_cvtsi128_si32(_mm_shuffle_epi32(second_half_pattern, 0x2)) >>
+           (4)) &
+          3;
+
+      best_mod_idx[tbl_idx][3] =
+          (_mm_cvtsi128_si32(_mm_shuffle_epi32(first_half_pattern, 0x3)) >>
+           (6)) &
+          3;
+      best_mod_idx[tbl_idx][7] =
+          (_mm_cvtsi128_si32(_mm_shuffle_epi32(second_half_pattern, 0x3)) >>
+           (6)) &
+          3;
+
+      if (best_error == 0) {
+        break;
+      }
+    } else if (delta > 0 && expected_error < min) {
+      // The error is growing and is well beyond expected threshold.
+      break;
+    }
+  }
+
+  WriteCodewordTable(block, sub_block_id, best_tbl_idx);
+
+  uint32_t pix_data = 0;
+  uint8_t mod_idx;
+  uint8_t pix_idx;
+  uint32_t lsb;
+  uint32_t msb;
+  int texel_num;
+
+  for (unsigned int i = 0; i < 8; ++i) {
+    mod_idx = best_mod_idx[best_tbl_idx][i];
+    pix_idx = g_mod_to_pix[mod_idx];
+
+    lsb = pix_idx & 0x1;
+    msb = pix_idx >> 1;
+
+    // Obtain the texel number as specified in the standard.
+    texel_num = idx_to_num_tab[i];
+    pix_data |= msb << (texel_num + 16);
+    pix_data |= lsb << (texel_num);
+  }
+
+  WritePixelData(block, pix_data);
+}
+
+void CompressBlock(uint8_t* dst, __sse_data* data) {
+  // First 3 values are for vertical 1, second 3 vertical 2, third 3 horizontal
+  // 1, last 3
+  // horizontal 2.
+  float __sse_avg_colors[12] = {
+      0,
+  };
+  bool use_differential[2] = {true, true};
+  GetAvgColors(data, __sse_avg_colors, use_differential);
+  Color sub_block_avg[4];
+
+  // TODO(radu.velea): Remove floating point operations and use only int's +
+  // normal rounding and shifts for reduced Quality.
+  for (int i = 0, j = 1; i < 4; i += 2, j += 2) {
+    if (use_differential[i / 2] == false) {
+      sub_block_avg[i] = MakeColor444(&__sse_avg_colors[i * 3]);
+      sub_block_avg[j] = MakeColor444(&__sse_avg_colors[j * 3]);
+    } else {
+      sub_block_avg[i] = MakeColor555(&__sse_avg_colors[i * 3]);
+      sub_block_avg[j] = MakeColor555(&__sse_avg_colors[j * 3]);
+    }
+  }
+
+  __m128i red_avg[2], green_avg[2], blue_avg[2];
+
+  // TODO(radu.velea): Perfect accuracy, maybe skip floating variables.
+  blue_avg[0] = _mm_set_epi32(static_cast<int>(__sse_avg_colors[3]),
+                              static_cast<int>(__sse_avg_colors[3]),
+                              static_cast<int>(__sse_avg_colors[0]),
+                              static_cast<int>(__sse_avg_colors[0]));
+
+  green_avg[0] = _mm_set_epi32(static_cast<int>(__sse_avg_colors[4]),
+                               static_cast<int>(__sse_avg_colors[4]),
+                               static_cast<int>(__sse_avg_colors[1]),
+                               static_cast<int>(__sse_avg_colors[1]));
+
+  red_avg[0] = _mm_set_epi32(static_cast<int>(__sse_avg_colors[5]),
+                             static_cast<int>(__sse_avg_colors[5]),
+                             static_cast<int>(__sse_avg_colors[2]),
+                             static_cast<int>(__sse_avg_colors[2]));
+
+  uint32_t vertical_error[2];
+  GetVerticalError(data, blue_avg, green_avg, red_avg, vertical_error);
+
+  // TODO(radu.velea): Perfect accuracy, maybe skip floating variables.
+  blue_avg[0] = _mm_set1_epi32(static_cast<int>(__sse_avg_colors[6]));
+  blue_avg[1] = _mm_set1_epi32(static_cast<int>(__sse_avg_colors[9]));
+
+  green_avg[0] = _mm_set1_epi32(static_cast<int>(__sse_avg_colors[7]));
+  green_avg[1] = _mm_set1_epi32(static_cast<int>(__sse_avg_colors[10]));
+
+  red_avg[0] = _mm_set1_epi32(static_cast<int>(__sse_avg_colors[8]));
+  red_avg[1] = _mm_set1_epi32(static_cast<int>(__sse_avg_colors[11]));
+
+  uint32_t horizontal_error[2];
+  GetHorizontalError(data, blue_avg, green_avg, red_avg, horizontal_error);
+
+  bool flip = (horizontal_error[0] + horizontal_error[1]) <
+              (vertical_error[0] + vertical_error[1]);
+  uint32_t* expected_errors = flip ? horizontal_error : vertical_error;
+
+  // Clear destination buffer so that we can "or" in the results.
+  memset(dst, 0, 8);
+
+  WriteDiff(dst, use_differential[!!flip]);
+  WriteFlip(dst, flip);
+
+  uint8_t sub_block_off_0 = flip ? 2 : 0;
+  uint8_t sub_block_off_1 = sub_block_off_0 + 1;
+
+  if (use_differential[!!flip]) {
+    WriteColors555(dst, sub_block_avg[sub_block_off_0],
+                   sub_block_avg[sub_block_off_1]);
+  } else {
+    WriteColors444(dst, sub_block_avg[sub_block_off_0],
+                   sub_block_avg[sub_block_off_1]);
+  }
+
+  if (!flip) {
+    // Transpose vertical data into horizontal lines.
+    __m128i tmp;
+    for (int i = 0; i < 4; i += 2) {
+      tmp = data->blue[i];
+      data->blue[i] = _mm_add_epi32(
+          _mm_move_epi64(data->blue[i]),
+          _mm_shuffle_epi32(_mm_move_epi64(data->blue[i + 1]), 0x4E));
+      data->blue[i + 1] = _mm_add_epi32(
+          _mm_move_epi64(_mm_shuffle_epi32(tmp, 0x4E)),
+          _mm_shuffle_epi32(
+              _mm_move_epi64(_mm_shuffle_epi32(data->blue[i + 1], 0x4E)),
+              0x4E));
+
+      tmp = data->green[i];
+      data->green[i] = _mm_add_epi32(
+          _mm_move_epi64(data->green[i]),
+          _mm_shuffle_epi32(_mm_move_epi64(data->green[i + 1]), 0x4E));
+      data->green[i + 1] = _mm_add_epi32(
+          _mm_move_epi64(_mm_shuffle_epi32(tmp, 0x4E)),
+          _mm_shuffle_epi32(
+              _mm_move_epi64(_mm_shuffle_epi32(data->green[i + 1], 0x4E)),
+              0x4E));
+
+      tmp = data->red[i];
+      data->red[i] = _mm_add_epi32(
+          _mm_move_epi64(data->red[i]),
+          _mm_shuffle_epi32(_mm_move_epi64(data->red[i + 1]), 0x4E));
+      data->red[i + 1] = _mm_add_epi32(
+          _mm_move_epi64(_mm_shuffle_epi32(tmp, 0x4E)),
+          _mm_shuffle_epi32(
+              _mm_move_epi64(_mm_shuffle_epi32(data->red[i + 1], 0x4E)), 0x4E));
+    }
+
+    tmp = data->blue[1];
+    data->blue[1] = data->blue[2];
+    data->blue[2] = tmp;
+
+    tmp = data->green[1];
+    data->green[1] = data->green[2];
+    data->green[2] = tmp;
+
+    tmp = data->red[1];
+    data->red[1] = data->red[2];
+    data->red[2] = tmp;
+  }
+
+  // Compute luminance for the first sub block.
+  ComputeLuminance(dst, sub_block_avg[sub_block_off_0], 0,
+                   g_idx_to_num[sub_block_off_0], data,
+                   SetETC1MaxError(expected_errors[0]));
+  // Compute luminance for the second sub block.
+  ComputeLuminance(dst, sub_block_avg[sub_block_off_1], 1,
+                   g_idx_to_num[sub_block_off_1], data,
+                   SetETC1MaxError(expected_errors[1]));
+}
+
+static void ExtractBlock(uint8_t* dst, const uint8_t* src, int width) {
+  for (int j = 0; j < 4; ++j) {
+    memcpy(&dst[j * 4 * 4], src, 4 * 4);
+    src += width * 4;
+  }
+}
+
+inline bool TransposeBlock(uint8_t* block, __m128i* transposed) {
+  // This function transforms an incommig block of RGBA or GBRA pixels into 4
+  // registers, each containing the data corresponding for a single channel.
+  // Ex: transposed[0] will have all the R values for a RGBA block,
+  // transposed[1] will have G, etc.
+  // The values are packed as 8 bit unsigned values in the SSE registers.
+
+  // Before doing any work we check if the block is solid.
+  __m128i tmp3, tmp2, tmp1, tmp0;
+  __m128i test_solid = _mm_set1_epi32(*((uint32_t*)block));
+  uint16_t mask = 0xFFFF;
+
+  // a0,a1,a2,...a7, ...a15
+  transposed[0] = _mm_loadu_si128((__m128i*)(block));
+  // b0, b1,b2,...b7.... b15
+  transposed[1] = _mm_loadu_si128((__m128i*)(block + 16));
+  // c0, c1,c2,...c7....c15
+  transposed[2] = _mm_loadu_si128((__m128i*)(block + 32));
+  // d0,d1,d2,...d7....d15
+  transposed[3] = _mm_loadu_si128((__m128i*)(block + 48));
+
+  for (int i = 0; i < 4; i++) {
+    mask &= _mm_movemask_epi8(_mm_cmpeq_epi8(transposed[i], test_solid));
+  }
+
+  if (mask == 0xFFFF) {
+    // Block is solid, no need to do any more work.
+    return false;
+  }
+
+  // a0,b0, a1,b1, a2,b2, a3,b3,....a7,b7
+  tmp0 = _mm_unpacklo_epi8(transposed[0], transposed[1]);
+  // c0,d0, c1,d1, c2,d2, c3,d3,... c7,d7
+  tmp1 = _mm_unpacklo_epi8(transposed[2], transposed[3]);
+  // a8,b8, a9,b9, a10,b10, a11,b11,...a15,b15
+  tmp2 = _mm_unpackhi_epi8(transposed[0], transposed[1]);
+  // c8,d8, c9,d9, c10,d10, c11,d11,...c15,d15
+  tmp3 = _mm_unpackhi_epi8(transposed[2], transposed[3]);
+
+  // a0,a8, b0,b8,  a1,a9, b1,b9, ....a3,a11, b3,b11
+  transposed[0] = _mm_unpacklo_epi8(tmp0, tmp2);
+  // a4,a12, b4,b12, a5,a13, b5,b13,....a7,a15,b7,b15
+  transposed[1] = _mm_unpackhi_epi8(tmp0, tmp2);
+  // c0,c8, d0,d8, c1,c9, d1,d9.....d3,d11
+  transposed[2] = _mm_unpacklo_epi8(tmp1, tmp3);
+  // c4,c12,d4,d12, c5,c13, d5,d13,....d7,d15
+  transposed[3] = _mm_unpackhi_epi8(tmp1, tmp3);
+
+  // a0,a8, b0,b8, c0,c8, d0,d8, a1,a9, b1,b9, c1,c9, d1,d9
+  tmp0 = _mm_unpacklo_epi32(transposed[0], transposed[2]);
+  // a2,a10, b2,b10, c2,c10, d2,d10, a3,a11, b3,b11, c3,c11, d3,d11
+  tmp1 = _mm_unpackhi_epi32(transposed[0], transposed[2]);
+  // a4,a12, b4,b12, c4,c12, d4,d12, a5,a13, b5,b13, c5,c13, d5,d13
+  tmp2 = _mm_unpacklo_epi32(transposed[1], transposed[3]);
+  // a6,a14, b6,b14, c6,c14, d6,d14, a7,a15, b7,b15, c7,c15, d7,d15
+  tmp3 = _mm_unpackhi_epi32(transposed[1], transposed[3]);
+
+  // a0,a4, a8,a12, b0,b4, b8,b12,  c0,c4, c8,c12, d0,d4, d8,d12
+  transposed[0] = _mm_unpacklo_epi8(tmp0, tmp2);
+  // a1,a5, a9,a13, b1,b5, b9,b13,  c1,c5, c9,c13, d1,d5, d9,d13
+  transposed[1] = _mm_unpackhi_epi8(tmp0, tmp2);
+  // a2,a6, a10,a14, b2,b6, b10,b14, c2,c6, c10,c14, d2,d6, d10,d14
+  transposed[2] = _mm_unpacklo_epi8(tmp1, tmp3);
+  // a3,a7, a11,a15, b3,b7, b11,b15, c3,c7, c11,c15, d3,d7, d11,d15
+  transposed[3] = _mm_unpackhi_epi8(tmp1, tmp3);
+
+  return true;
+}
+
+inline void UnpackBlock(__m128i* packed,
+                        __m128i* red,
+                        __m128i* green,
+                        __m128i* blue,
+                        __m128i* alpha) {
+  const __m128i zero = _mm_set1_epi8(0);
+  __m128i tmp_low, tmp_high;
+
+  // Unpack red.
+  tmp_low = _mm_unpacklo_epi8(packed[0], zero);
+  tmp_high = _mm_unpackhi_epi8(packed[0], zero);
+
+  red[0] = _mm_unpacklo_epi16(tmp_low, zero);
+  red[1] = _mm_unpackhi_epi16(tmp_low, zero);
+
+  red[2] = _mm_unpacklo_epi16(tmp_high, zero);
+  red[3] = _mm_unpackhi_epi16(tmp_high, zero);
+
+  // Unpack green.
+  tmp_low = _mm_unpacklo_epi8(packed[1], zero);
+  tmp_high = _mm_unpackhi_epi8(packed[1], zero);
+
+  green[0] = _mm_unpacklo_epi16(tmp_low, zero);
+  green[1] = _mm_unpackhi_epi16(tmp_low, zero);
+
+  green[2] = _mm_unpacklo_epi16(tmp_high, zero);
+  green[3] = _mm_unpackhi_epi16(tmp_high, zero);
+
+  // Unpack blue.
+  tmp_low = _mm_unpacklo_epi8(packed[2], zero);
+  tmp_high = _mm_unpackhi_epi8(packed[2], zero);
+
+  blue[0] = _mm_unpacklo_epi16(tmp_low, zero);
+  blue[1] = _mm_unpackhi_epi16(tmp_low, zero);
+
+  blue[2] = _mm_unpacklo_epi16(tmp_high, zero);
+  blue[3] = _mm_unpackhi_epi16(tmp_high, zero);
+
+  // Unpack alpha - unused for ETC1.
+  tmp_low = _mm_unpacklo_epi8(packed[3], zero);
+  tmp_high = _mm_unpackhi_epi8(packed[3], zero);
+
+  alpha[0] = _mm_unpacklo_epi16(tmp_low, zero);
+  alpha[1] = _mm_unpackhi_epi16(tmp_low, zero);
+
+  alpha[2] = _mm_unpacklo_epi16(tmp_high, zero);
+  alpha[3] = _mm_unpackhi_epi16(tmp_high, zero);
+}
+
+inline void CompressSolid(uint8_t* dst, uint8_t* block) {
+  // Clear destination buffer so that we can "or" in the results.
+  memset(dst, 0, 8);
+
+  const float src_color_float[3] = {static_cast<float>(block[0]),
+                                    static_cast<float>(block[1]),
+                                    static_cast<float>(block[2])};
+  const Color base = MakeColor555(src_color_float);
+  const __m128i base_v =
+      _mm_set_epi32(0, base.channels.r, base.channels.g, base.channels.b);
+
+  const __m128i constant = _mm_set_epi32(0, block[2], block[1], block[0]);
+  __m128i lum;
+  __m128i colors[4];
+  static const __m128i rgb =
+      _mm_set_epi32(0, 0xFFFFFFFF, 0xFFFFFFFF, 0xFFFFFFFF);
+
+  WriteDiff(dst, true);
+  WriteFlip(dst, false);
+
+  WriteColors555(dst, base, base);
+
+  uint8_t best_tbl_idx = 0;
+  uint8_t best_mod_idx = 0;
+  uint32_t best_mod_err = INT32_MAX;
+
+  for (unsigned int tbl_idx = 0; tbl_idx < 8; ++tbl_idx) {
+    lum = _mm_set_epi32(
+        g_codeword_tables[tbl_idx][3], g_codeword_tables[tbl_idx][2],
+        g_codeword_tables[tbl_idx][1], g_codeword_tables[tbl_idx][0]);
+    colors[0] = AddAndClamp(base_v, _mm_shuffle_epi32(lum, 0x0));
+    colors[1] = AddAndClamp(base_v, _mm_shuffle_epi32(lum, 0x55));
+    colors[2] = AddAndClamp(base_v, _mm_shuffle_epi32(lum, 0xAA));
+    colors[3] = AddAndClamp(base_v, _mm_shuffle_epi32(lum, 0xFF));
+
+    for (int i = 0; i < 4; i++) {
+      uint32_t mod_err =
+          SumSSE(GetColorErrorSSE(constant, _mm_and_si128(colors[i], rgb)));
+      colors[i] = _mm_and_si128(colors[i], rgb);
+      if (mod_err < best_mod_err) {
+        best_tbl_idx = tbl_idx;
+        best_mod_idx = i;
+        best_mod_err = mod_err;
+
+        if (mod_err == 0) {
+          break;  // We cannot do any better than this.
+        }
+      }
+    }
+  }
+
+  WriteCodewordTable(dst, 0, best_tbl_idx);
+  WriteCodewordTable(dst, 1, best_tbl_idx);
+
+  uint8_t pix_idx = g_mod_to_pix[best_mod_idx];
+  uint32_t lsb = pix_idx & 0x1;
+  uint32_t msb = pix_idx >> 1;
+
+  uint32_t pix_data = 0;
+  for (unsigned int i = 0; i < 2; ++i) {
+    for (unsigned int j = 0; j < 8; ++j) {
+      // Obtain the texel number as specified in the standard.
+      int texel_num = g_idx_to_num[i][j];
+      pix_data |= msb << (texel_num + 16);
+      pix_data |= lsb << (texel_num);
+    }
+  }
+
+  WritePixelData(dst, pix_data);
+}
+
+}  // namespace
+
+void TextureCompressorETC1SSE::Compress(const uint8_t* src,
+                                        uint8_t* dst,
+                                        int width,
+                                        int height,
+                                        Quality quality) {
+  DCHECK_GE(width, 4);
+  DCHECK_EQ((width & 3), 0);
+  DCHECK_GE(height, 4);
+  DCHECK_EQ((height & 3), 0);
+
+  ALIGNAS(16) uint8_t block[64];
+  __m128i packed[4];
+  __m128i red[4], green[4], blue[4], alpha[4];
+  __sse_data data;
+
+  for (int y = 0; y < height; y += 4, src += width * 4 * 4) {
+    for (int x = 0; x < width; x += 4, dst += 8) {
+      ExtractBlock(block, src + x * 4, width);
+      if (TransposeBlock(block, packed) == false) {
+        CompressSolid(dst, block);
+      } else {
+        UnpackBlock(packed, blue, green, red, alpha);
+
+        data.block = block;
+        data.packed = packed;
+        data.red = red;
+        data.blue = blue;
+        data.green = green;
+
+        CompressBlock(dst, &data);
+      }
+    }
+  }
+}
+
+}  // namespace cc