diff --git a/package.json b/package.json
index 7d25f7fdf..d64617d85 100644
--- a/package.json
+++ b/package.json
@@ -15,6 +15,7 @@
     "libsodium": "^0.7.9",
     "libsodium-wrappers": "^0.7.9",
     "ts-proto": "^1.101.0",
+    "wasm-feature-detect": "^1.2.11",
     "zstddec": "^0.0.2"
   }
 }
diff --git a/src/codec.js b/src/codec.js
index 6dc7db92a..e35f72d85 100644
--- a/src/codec.js
+++ b/src/codec.js
@@ -19,14 +19,17 @@
   OGVDecoderVideoAV1MTW: 'ogv-decoder-video-av1-mt-wasm.js',
   OGVDecoderVideoAV1SIMDMTW: 'ogv-decoder-video-av1-simd-mt-wasm.js',
 */
+import { simd } from "wasm-feature-detect";
 
-export function loadVp9(callback) {
+export async function loadVp9(callback) {
   // Multithreading is used only if `options.threading` is true. 
   // This requires browser support for the new `SharedArrayBuffer` and `Atomics` APIs, 
   // currently available in Firefox and Chrome with experimental flags enabled.
   // 所有主流浏览器均默认于2018年1月5日禁用SharedArrayBuffer
+  const isSIMD = await simd();
+  console.log('isSIMD: ' + isSIMD);
   window.OGVLoader.loadClass(
-    "OGVDecoderVideoVP9SIMDW",
+    isSIMD ? "OGVDecoderVideoVP9SIMDW" : "OGVDecoderVideoVP9W",
     (videoCodecClass) => {
       window.videoCodecClass = videoCodecClass;
       videoCodecClass({ videoFormat: {} }).then((decoder) => {
diff --git a/src/common.ts b/src/common.ts
index 0355d3607..62438ddef 100644
--- a/src/common.ts
+++ b/src/common.ts
@@ -46,7 +46,7 @@ const zCode = "z".charCodeAt(0);
 const aCode = "a".charCodeAt(0);
 
 export function mapKey(name: string, isDesktop: Boolean) {
-  const tmp = KEY_MAP[name];
+  const tmp = KEY_MAP[name] || name;
   if (tmp.length == 1) {
     const chr = tmp.charCodeAt(0);
     if (!isDesktop && (chr > zCode || chr < aCode))
diff --git a/src/connection.ts b/src/connection.ts
index 6f75498c0..85249f45e 100644
--- a/src/connection.ts
+++ b/src/connection.ts
@@ -487,16 +487,16 @@ export default class Connection {
   ) {
     const key_event = mapKey(name, globals.isDesktop());
     if (!key_event) return;
-    if (alt && name == "VK_MENU") {
+    if (alt && (name == "Alt" || name == 'RAlt')) {
       alt = false;
     }
-    if (ctrl && name == "VK_CONTROL") {
+    if (ctrl && (name == "Control" || name == 'RControl')) {
       ctrl = false;
     }
-    if (shift && name == "VK_SHIFT") {
+    if (shift && (name == "Shift" || name == 'RShift')) {
       shift = false;
     }
-    if (command && name == "Meta") {
+    if (command && (name == "Meta" || name == 'RWin')) {
       command = false;
     }
     key_event.down = down;
diff --git a/src/globals.js b/src/globals.js
index 78f029043..a7ceaf2b3 100644
--- a/src/globals.js
+++ b/src/globals.js
@@ -20,7 +20,7 @@ window.isMobile = () => {
 }
 
 export function isDesktop() {
-  return !isMobile;
+  return !isMobile();
 }
 
 export function msgbox(type, title, text) {
@@ -184,7 +184,7 @@ window.setByName = (name, value) => {
       break;
     case 'input_key':
       value = JSON.parse(value);
-      curConn.inputKey(value.name, value.down || false, value.press || false, value.alt || false, value.ctrl || false, value.shift || false, value.command || false);
+      curConn.inputKey(value.name, value.down == 'true', value.press == 'true', value.alt == 'true', value.ctrl == 'true', value.shift == 'true', value.command == 'true');
       break;
     case 'input_string':
       curConn.inputString(value);
@@ -213,7 +213,7 @@ window.setByName = (name, value) => {
         case 'wheel':
           mask |= 4 << 3;
       }
-      curConn.inputMouse(mask, parseInt(value.x || '0'), parseInt(value.y || '0'), value.alt || false, value.ctrl || false, value.shift || false, value.command || false);
+      curConn.inputMouse(mask, parseInt(value.x || '0'), parseInt(value.y || '0'), value.alt == 'true', value.ctrl == 'true', value.shift == 'true', value.command == 'true');
       break;
     case 'option':
       value = JSON.parse(value);
diff --git a/yarn.lock b/yarn.lock
index 27da14f94..af7ff0b7d 100644
--- a/yarn.lock
+++ b/yarn.lock
@@ -358,6 +358,11 @@ vite@^2.7.2:
   optionalDependencies:
     fsevents "~2.3.2"
 
+wasm-feature-detect@^1.2.11:
+  version "1.2.11"
+  resolved "https://registry.yarnpkg.com/wasm-feature-detect/-/wasm-feature-detect-1.2.11.tgz#e21992fd1f1d41a47490e392a5893cb39d81e29e"
+  integrity sha512-HUqwaodrQGaZgz1lZaNioIkog9tkeEJjrM3eq4aUL04whXOVDRc/o2EGb/8kV0QX411iAYWEqq7fMBmJ6dKS6w==
+
 zstddec@^0.0.2:
   version "0.0.2"
   resolved "https://registry.yarnpkg.com/zstddec/-/zstddec-0.0.2.tgz#57e2f28dd1ff56b750e07d158a43f0611ad9eeb4"
diff --git a/yuv_rgb.c b/yuv_rgb.c
index fb4a703d1..fdf47269b 100644
--- a/yuv_rgb.c
+++ b/yuv_rgb.c
@@ -9,6 +9,12 @@
 
 typedef __i64x2 __m128i;
 #define _mm_load_si128 wasm_v128_load
+#define _mm_add_epi8 wasm_i8x16_add
+#define _mm_set1_epi8 wasm_i8x16_splat
+#define _mm_srai_epi16 wasm_i16x8_shr
+#define _mm_mullo_epi16 wasm_i16x8_mul
+#define _mm_sub_epi8 wasm_i8x16_sub
+#define _mm_setzero_si128 wasm_i64x2_const
 
 #include <stdio.h>
 
@@ -433,493 +439,6 @@ void nv21_rgb24_std(
 	}
 }
 
-
-#ifdef __SSE2__
-
-//see rgb.txt
-#define UNPACK_RGB24_32_STEP(RS1, RS2, RS3, RS4, RS5, RS6, RD1, RD2, RD3, RD4, RD5, RD6) \
-RD1 = _mm_unpacklo_epi8(RS1, RS4); \
-RD2 = _mm_unpackhi_epi8(RS1, RS4); \
-RD3 = _mm_unpacklo_epi8(RS2, RS5); \
-RD4 = _mm_unpackhi_epi8(RS2, RS5); \
-RD5 = _mm_unpacklo_epi8(RS3, RS6); \
-RD6 = _mm_unpackhi_epi8(RS3, RS6);
-
-#define RGB2YUV_16(R, G, B, Y, U, V) \
-Y = _mm_add_epi16(_mm_mullo_epi16(R, _mm_set1_epi16(param->r_factor)), \
-                  _mm_mullo_epi16(G, _mm_set1_epi16(param->g_factor))); \
-Y = _mm_add_epi16(Y, _mm_mullo_epi16(B, _mm_set1_epi16(param->b_factor))); \
-Y = _mm_srli_epi16(Y, 8); \
-U = _mm_mullo_epi16(_mm_sub_epi16(B, Y), _mm_set1_epi16(param->cb_factor)); \
-U = _mm_add_epi16(_mm_srai_epi16(U, 8), _mm_set1_epi16(128)); \
-V = _mm_mullo_epi16(_mm_sub_epi16(R, Y), _mm_set1_epi16(param->cr_factor)); \
-V = _mm_add_epi16(_mm_srai_epi16(V, 8), _mm_set1_epi16(128)); \
-Y = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(Y, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset));
-
-#define RGB2YUV_32 \
-	__m128i r_16, g_16, b_16; \
-	__m128i y1_16, y2_16, cb1_16, cb2_16, cr1_16, cr2_16, Y, cb, cr; \
-	__m128i tmp1, tmp2, tmp3, tmp4, tmp5, tmp6; \
-	__m128i rgb1 = LOAD_SI128((const __m128i*)(rgb_ptr1)), \
-		rgb2 = LOAD_SI128((const __m128i*)(rgb_ptr1+16)), \
-		rgb3 = LOAD_SI128((const __m128i*)(rgb_ptr1+32)), \
-		rgb4 = LOAD_SI128((const __m128i*)(rgb_ptr2)), \
-		rgb5 = LOAD_SI128((const __m128i*)(rgb_ptr2+16)), \
-		rgb6 = LOAD_SI128((const __m128i*)(rgb_ptr2+32)); \
-	/* unpack rgb24 data to r, g and b data in separate channels*/ \
-	/* see rgb.txt to get an idea of the algorithm, note that we only go to the next to last step*/ \
-	/* here, because averaging in horizontal direction is easier like this*/ \
-	/* The last step is applied further on the Y channel only*/ \
-	UNPACK_RGB24_32_STEP(rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6) \
-	UNPACK_RGB24_32_STEP(tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, rgb1, rgb2, rgb3, rgb4, rgb5, rgb6) \
-	UNPACK_RGB24_32_STEP(rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6) \
-	UNPACK_RGB24_32_STEP(tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, rgb1, rgb2, rgb3, rgb4, rgb5, rgb6) \
-	/* first compute Y', (B-Y') and (R-Y'), in 16bits values, for the first line */ \
-	/* Y is saved for each pixel, while only sums of (B-Y') and (R-Y') for pairs of adjacents pixels are saved*/ \
-	r_16 = _mm_unpacklo_epi8(rgb1, _mm_setzero_si128()); \
-	g_16 = _mm_unpacklo_epi8(rgb2, _mm_setzero_si128()); \
-	b_16 = _mm_unpacklo_epi8(rgb3, _mm_setzero_si128()); \
-	y1_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y1_16 = _mm_add_epi16(y1_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y1_16 = _mm_srli_epi16(y1_16, 8); \
-	cb1_16 = _mm_sub_epi16(b_16, y1_16); \
-	cr1_16 = _mm_sub_epi16(r_16, y1_16); \
-	r_16 = _mm_unpacklo_epi8(rgb4, _mm_setzero_si128()); \
-	g_16 = _mm_unpacklo_epi8(rgb5, _mm_setzero_si128()); \
-	b_16 = _mm_unpacklo_epi8(rgb6, _mm_setzero_si128()); \
-	y2_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y2_16 = _mm_add_epi16(y2_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y2_16 = _mm_srli_epi16(y2_16, 8); \
-	cb1_16 = _mm_add_epi16(cb1_16, _mm_sub_epi16(b_16, y2_16)); \
-	cr1_16 = _mm_add_epi16(cr1_16, _mm_sub_epi16(r_16, y2_16)); \
-	/* Rescale Y' to Y, pack it to 8bit values and save it */ \
-	y1_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y1_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	y2_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y2_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	Y = _mm_packus_epi16(y1_16, y2_16); \
-	Y = _mm_unpackhi_epi8(_mm_slli_si128(Y, 8), Y); \
-	SAVE_SI128((__m128i*)(y_ptr1), Y); \
-	/* same for the second line, compute Y', (B-Y') and (R-Y'), in 16bits values */ \
-	/* Y is saved for each pixel, while only sums of (B-Y') and (R-Y') for pairs of adjacents pixels are added to the previous values*/ \
-	r_16 = _mm_unpackhi_epi8(rgb1, _mm_setzero_si128()); \
-	g_16 = _mm_unpackhi_epi8(rgb2, _mm_setzero_si128()); \
-	b_16 = _mm_unpackhi_epi8(rgb3, _mm_setzero_si128()); \
-	y1_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y1_16 = _mm_add_epi16(y1_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y1_16 = _mm_srli_epi16(y1_16, 8); \
-	cb1_16 = _mm_add_epi16(cb1_16, _mm_sub_epi16(b_16, y1_16)); \
-	cr1_16 = _mm_add_epi16(cr1_16, _mm_sub_epi16(r_16, y1_16)); \
-	r_16 = _mm_unpackhi_epi8(rgb4, _mm_setzero_si128()); \
-	g_16 = _mm_unpackhi_epi8(rgb5, _mm_setzero_si128()); \
-	b_16 = _mm_unpackhi_epi8(rgb6, _mm_setzero_si128()); \
-	y2_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y2_16 = _mm_add_epi16(y2_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y2_16 = _mm_srli_epi16(y2_16, 8); \
-	cb1_16 = _mm_add_epi16(cb1_16, _mm_sub_epi16(b_16, y2_16)); \
-	cr1_16 = _mm_add_epi16(cr1_16, _mm_sub_epi16(r_16, y2_16)); \
-	/* Rescale Y' to Y, pack it to 8bit values and save it */ \
-	y1_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y1_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	y2_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y2_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	Y = _mm_packus_epi16(y1_16, y2_16); \
-	Y = _mm_unpackhi_epi8(_mm_slli_si128(Y, 8), Y); \
-	SAVE_SI128((__m128i*)(y_ptr2), Y); \
-	/* Rescale Cb and Cr to their final range */ \
-	cb1_16 = _mm_add_epi16(_mm_srai_epi16(_mm_mullo_epi16(_mm_srai_epi16(cb1_16, 2), _mm_set1_epi16(param->cb_factor)), 8), _mm_set1_epi16(128)); \
-	cr1_16 = _mm_add_epi16(_mm_srai_epi16(_mm_mullo_epi16(_mm_srai_epi16(cr1_16, 2), _mm_set1_epi16(param->cr_factor)), 8), _mm_set1_epi16(128)); \
-	\
-	/* do the same again with next data */ \
-	rgb1 = LOAD_SI128((const __m128i*)(rgb_ptr1+48)), \
-	rgb2 = LOAD_SI128((const __m128i*)(rgb_ptr1+64)), \
-	rgb3 = LOAD_SI128((const __m128i*)(rgb_ptr1+80)), \
-	rgb4 = LOAD_SI128((const __m128i*)(rgb_ptr2+48)), \
-	rgb5 = LOAD_SI128((const __m128i*)(rgb_ptr2+64)), \
-	rgb6 = LOAD_SI128((const __m128i*)(rgb_ptr2+80)); \
-	/* unpack rgb24 data to r, g and b data in separate channels*/ \
-	/* see rgb.txt to get an idea of the algorithm, note that we only go to the next to last step*/ \
-	/* here, because averaging in horizontal direction is easier like this*/ \
-	/* The last step is applied further on the Y channel only*/ \
-	UNPACK_RGB24_32_STEP(rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6) \
-	UNPACK_RGB24_32_STEP(tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, rgb1, rgb2, rgb3, rgb4, rgb5, rgb6) \
-	UNPACK_RGB24_32_STEP(rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6) \
-	UNPACK_RGB24_32_STEP(tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, rgb1, rgb2, rgb3, rgb4, rgb5, rgb6) \
-	/* first compute Y', (B-Y') and (R-Y'), in 16bits values, for the first line */ \
-	/* Y is saved for each pixel, while only sums of (B-Y') and (R-Y') for pairs of adjacents pixels are saved*/ \
-	r_16 = _mm_unpacklo_epi8(rgb1, _mm_setzero_si128()); \
-	g_16 = _mm_unpacklo_epi8(rgb2, _mm_setzero_si128()); \
-	b_16 = _mm_unpacklo_epi8(rgb3, _mm_setzero_si128()); \
-	y1_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y1_16 = _mm_add_epi16(y1_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y1_16 = _mm_srli_epi16(y1_16, 8); \
-	cb2_16 = _mm_sub_epi16(b_16, y1_16); \
-	cr2_16 = _mm_sub_epi16(r_16, y1_16); \
-	r_16 = _mm_unpacklo_epi8(rgb4, _mm_setzero_si128()); \
-	g_16 = _mm_unpacklo_epi8(rgb5, _mm_setzero_si128()); \
-	b_16 = _mm_unpacklo_epi8(rgb6, _mm_setzero_si128()); \
-	y2_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y2_16 = _mm_add_epi16(y2_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y2_16 = _mm_srli_epi16(y2_16, 8); \
-	cb2_16 = _mm_add_epi16(cb2_16, _mm_sub_epi16(b_16, y2_16)); \
-	cr2_16 = _mm_add_epi16(cr2_16, _mm_sub_epi16(r_16, y2_16)); \
-	/* Rescale Y' to Y, pack it to 8bit values and save it */ \
-	y1_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y1_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	y2_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y2_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	Y = _mm_packus_epi16(y1_16, y2_16); \
-	Y = _mm_unpackhi_epi8(_mm_slli_si128(Y, 8), Y); \
-	SAVE_SI128((__m128i*)(y_ptr1+16), Y); \
-	/* same for the second line, compute Y', (B-Y') and (R-Y'), in 16bits values */ \
-	/* Y is saved for each pixel, while only sums of (B-Y') and (R-Y') for pairs of adjacents pixels are added to the previous values*/ \
-	r_16 = _mm_unpackhi_epi8(rgb1, _mm_setzero_si128()); \
-	g_16 = _mm_unpackhi_epi8(rgb2, _mm_setzero_si128()); \
-	b_16 = _mm_unpackhi_epi8(rgb3, _mm_setzero_si128()); \
-	y1_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y1_16 = _mm_add_epi16(y1_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y1_16 = _mm_srli_epi16(y1_16, 8); \
-	cb2_16 = _mm_add_epi16(cb2_16, _mm_sub_epi16(b_16, y1_16)); \
-	cr2_16 = _mm_add_epi16(cr2_16, _mm_sub_epi16(r_16, y1_16)); \
-	r_16 = _mm_unpackhi_epi8(rgb4, _mm_setzero_si128()); \
-	g_16 = _mm_unpackhi_epi8(rgb5, _mm_setzero_si128()); \
-	b_16 = _mm_unpackhi_epi8(rgb6, _mm_setzero_si128()); \
-	y2_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y2_16 = _mm_add_epi16(y2_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y2_16 = _mm_srli_epi16(y2_16, 8); \
-	cb2_16 = _mm_add_epi16(cb2_16, _mm_sub_epi16(b_16, y2_16)); \
-	cr2_16 = _mm_add_epi16(cr2_16, _mm_sub_epi16(r_16, y2_16)); \
-	/* Rescale Y' to Y, pack it to 8bit values and save it */ \
-	y1_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y1_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	y2_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y2_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	Y = _mm_packus_epi16(y1_16, y2_16); \
-	Y = _mm_unpackhi_epi8(_mm_slli_si128(Y, 8), Y); \
-	SAVE_SI128((__m128i*)(y_ptr2+16), Y); \
-	/* Rescale Cb and Cr to their final range */ \
-	cb2_16 = _mm_add_epi16(_mm_srai_epi16(_mm_mullo_epi16(_mm_srai_epi16(cb2_16, 2), _mm_set1_epi16(param->cb_factor)), 8), _mm_set1_epi16(128)); \
-	cr2_16 = _mm_add_epi16(_mm_srai_epi16(_mm_mullo_epi16(_mm_srai_epi16(cr2_16, 2), _mm_set1_epi16(param->cr_factor)), 8), _mm_set1_epi16(128)); \
-	/* Pack and save Cb Cr */ \
-	cb = _mm_packus_epi16(cb1_16, cb2_16); \
-	cr = _mm_packus_epi16(cr1_16, cr2_16); \
-	SAVE_SI128((__m128i*)(u_ptr), cb); \
-	SAVE_SI128((__m128i*)(v_ptr), cr);
-
-
-void rgb24_yuv420_sse(uint32_t width, uint32_t height, 
-	const uint8_t *RGB, uint32_t RGB_stride, 
-	uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride, 
-	YCbCrType yuv_type)
-{
-	#define LOAD_SI128 _mm_load_si128
-	#define SAVE_SI128 _mm_stream_si128
-	const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]);
-	
-	uint32_t x, y;
-	for(y=0; y<(height-1); y+=2)
-	{
-		const uint8_t *rgb_ptr1=RGB+y*RGB_stride,
-			*rgb_ptr2=RGB+(y+1)*RGB_stride;
-		
-		uint8_t *y_ptr1=Y+y*Y_stride,
-			*y_ptr2=Y+(y+1)*Y_stride,
-			*u_ptr=U+(y/2)*UV_stride,
-			*v_ptr=V+(y/2)*UV_stride;
-		
-		for(x=0; x<(width-31); x+=32)
-		{
-			RGB2YUV_32
-			
-			rgb_ptr1+=96;
-			rgb_ptr2+=96;
-			y_ptr1+=32;
-			y_ptr2+=32;
-			u_ptr+=16; 
-			v_ptr+=16;
-		}
-	}
-	#undef LOAD_SI128
-	#undef SAVE_SI128
-}
-
-void rgb24_yuv420_sseu(uint32_t width, uint32_t height, 
-	const uint8_t *RGB, uint32_t RGB_stride, 
-	uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride, 
-	YCbCrType yuv_type)
-{
-	#define LOAD_SI128 _mm_loadu_si128
-	#define SAVE_SI128 _mm_storeu_si128
-	const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]);
-	
-	uint32_t x, y;
-	for(y=0; y<(height-1); y+=2)
-	{
-		const uint8_t *rgb_ptr1=RGB+y*RGB_stride,
-			*rgb_ptr2=RGB+(y+1)*RGB_stride;
-		
-		uint8_t *y_ptr1=Y+y*Y_stride,
-			*y_ptr2=Y+(y+1)*Y_stride,
-			*u_ptr=U+(y/2)*UV_stride,
-			*v_ptr=V+(y/2)*UV_stride;
-		
-		for(x=0; x<(width-31); x+=32)
-		{
-			RGB2YUV_32
-			
-			rgb_ptr1+=96;
-			rgb_ptr2+=96;
-			y_ptr1+=32;
-			y_ptr2+=32;
-			u_ptr+=16; 
-			v_ptr+=16;
-		}
-	}
-	#undef LOAD_SI128
-	#undef SAVE_SI128
-}
-
-
-// see rgba.txt
-#define UNPACK_RGB32_32_STEP(RS1, RS2, RS3, RS4, RS5, RS6, RS7, RS8, RD1, RD2, RD3, RD4, RD5, RD6, RD7, RD8) \
-RD1 = _mm_unpacklo_epi8(RS1, RS5); \
-RD2 = _mm_unpackhi_epi8(RS1, RS5); \
-RD3 = _mm_unpacklo_epi8(RS2, RS6); \
-RD4 = _mm_unpackhi_epi8(RS2, RS6); \
-RD5 = _mm_unpacklo_epi8(RS3, RS7); \
-RD6 = _mm_unpackhi_epi8(RS3, RS7); \
-RD7 = _mm_unpacklo_epi8(RS4, RS8); \
-RD8 = _mm_unpackhi_epi8(RS4, RS8);
-
-
-#define RGBA2YUV_32 \
-	__m128i r_16, g_16, b_16; \
-	__m128i y1_16, y2_16, cb1_16, cb2_16, cr1_16, cr2_16, Y, cb, cr; \
-	__m128i tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8; \
-	__m128i rgb1 = LOAD_SI128((const __m128i*)(rgb_ptr1)), \
-		rgb2 = LOAD_SI128((const __m128i*)(rgb_ptr1+16)), \
-		rgb3 = LOAD_SI128((const __m128i*)(rgb_ptr1+32)), \
-		rgb4 = LOAD_SI128((const __m128i*)(rgb_ptr1+48)), \
-		rgb5 = LOAD_SI128((const __m128i*)(rgb_ptr2)), \
-		rgb6 = LOAD_SI128((const __m128i*)(rgb_ptr2+16)), \
-		rgb7 = LOAD_SI128((const __m128i*)(rgb_ptr2+32)), \
-		rgb8 = LOAD_SI128((const __m128i*)(rgb_ptr2+48)); \
-	/* unpack rgb24 data to r, g and b data in separate channels*/ \
-	/* see rgb.txt to get an idea of the algorithm, note that we only go to the next to last step*/ \
-	/* here, because averaging in horizontal direction is easier like this*/ \
-	/* The last step is applied further on the Y channel only*/ \
-	UNPACK_RGB32_32_STEP(rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, rgb7, rgb8, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8) \
-	UNPACK_RGB32_32_STEP(tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8, rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, rgb7, rgb8) \
-	UNPACK_RGB32_32_STEP(rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, rgb7, rgb8, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8) \
-	UNPACK_RGB32_32_STEP(tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8, rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, rgb7, rgb8) \
-	/* first compute Y', (B-Y') and (R-Y'), in 16bits values, for the first line */ \
-	/* Y is saved for each pixel, while only sums of (B-Y') and (R-Y') for pairs of adjacents pixels are saved*/ \
-	r_16 = _mm_unpacklo_epi8(rgb1, _mm_setzero_si128()); \
-	g_16 = _mm_unpacklo_epi8(rgb2, _mm_setzero_si128()); \
-	b_16 = _mm_unpacklo_epi8(rgb3, _mm_setzero_si128()); \
-	y1_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y1_16 = _mm_add_epi16(y1_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y1_16 = _mm_srli_epi16(y1_16, 8); \
-	cb1_16 = _mm_sub_epi16(b_16, y1_16); \
-	cr1_16 = _mm_sub_epi16(r_16, y1_16); \
-	r_16 = _mm_unpacklo_epi8(rgb5, _mm_setzero_si128()); \
-	g_16 = _mm_unpacklo_epi8(rgb6, _mm_setzero_si128()); \
-	b_16 = _mm_unpacklo_epi8(rgb7, _mm_setzero_si128()); \
-	y2_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y2_16 = _mm_add_epi16(y2_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y2_16 = _mm_srli_epi16(y2_16, 8); \
-	cb1_16 = _mm_add_epi16(cb1_16, _mm_sub_epi16(b_16, y2_16)); \
-	cr1_16 = _mm_add_epi16(cr1_16, _mm_sub_epi16(r_16, y2_16)); \
-	/* Rescale Y' to Y, pack it to 8bit values and save it */ \
-	y1_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y1_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	y2_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y2_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	Y = _mm_packus_epi16(y1_16, y2_16); \
-	Y = _mm_unpackhi_epi8(_mm_slli_si128(Y, 8), Y); \
-	SAVE_SI128((__m128i*)(y_ptr1), Y); \
-	/* same for the second line, compute Y', (B-Y') and (R-Y'), in 16bits values */ \
-	/* Y is saved for each pixel, while only sums of (B-Y') and (R-Y') for pairs of adjacents pixels are added to the previous values*/ \
-	r_16 = _mm_unpackhi_epi8(rgb1, _mm_setzero_si128()); \
-	g_16 = _mm_unpackhi_epi8(rgb2, _mm_setzero_si128()); \
-	b_16 = _mm_unpackhi_epi8(rgb3, _mm_setzero_si128()); \
-	y1_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y1_16 = _mm_add_epi16(y1_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y1_16 = _mm_srli_epi16(y1_16, 8); \
-	cb1_16 = _mm_add_epi16(cb1_16, _mm_sub_epi16(b_16, y1_16)); \
-	cr1_16 = _mm_add_epi16(cr1_16, _mm_sub_epi16(r_16, y1_16)); \
-	r_16 = _mm_unpackhi_epi8(rgb5, _mm_setzero_si128()); \
-	g_16 = _mm_unpackhi_epi8(rgb6, _mm_setzero_si128()); \
-	b_16 = _mm_unpackhi_epi8(rgb7, _mm_setzero_si128()); \
-	y2_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y2_16 = _mm_add_epi16(y2_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y2_16 = _mm_srli_epi16(y2_16, 8); \
-	cb1_16 = _mm_add_epi16(cb1_16, _mm_sub_epi16(b_16, y2_16)); \
-	cr1_16 = _mm_add_epi16(cr1_16, _mm_sub_epi16(r_16, y2_16)); \
-	/* Rescale Y' to Y, pack it to 8bit values and save it */ \
-	y1_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y1_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	y2_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y2_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	Y = _mm_packus_epi16(y1_16, y2_16); \
-	Y = _mm_unpackhi_epi8(_mm_slli_si128(Y, 8), Y); \
-	SAVE_SI128((__m128i*)(y_ptr2), Y); \
-	/* Rescale Cb and Cr to their final range */ \
-	cb1_16 = _mm_add_epi16(_mm_srai_epi16(_mm_mullo_epi16(_mm_srai_epi16(cb1_16, 2), _mm_set1_epi16(param->cb_factor)), 8), _mm_set1_epi16(128)); \
-	cr1_16 = _mm_add_epi16(_mm_srai_epi16(_mm_mullo_epi16(_mm_srai_epi16(cr1_16, 2), _mm_set1_epi16(param->cr_factor)), 8), _mm_set1_epi16(128)); \
-	\
-	/* do the same again with next data */ \
-	rgb1 = LOAD_SI128((const __m128i*)(rgb_ptr1+64)), \
-	rgb2 = LOAD_SI128((const __m128i*)(rgb_ptr1+80)), \
-	rgb3 = LOAD_SI128((const __m128i*)(rgb_ptr1+96)), \
-	rgb4 = LOAD_SI128((const __m128i*)(rgb_ptr1+112)), \
-	rgb5 = LOAD_SI128((const __m128i*)(rgb_ptr2+64)), \
-	rgb6 = LOAD_SI128((const __m128i*)(rgb_ptr2+80)), \
-	rgb7 = LOAD_SI128((const __m128i*)(rgb_ptr2+96)), \
-	rgb8 = LOAD_SI128((const __m128i*)(rgb_ptr2+112)); \
-	/* unpack rgb24 data to r, g and b data in separate channels*/ \
-	/* see rgb.txt to get an idea of the algorithm, note that we only go to the next to last step*/ \
-	/* here, because averaging in horizontal direction is easier like this*/ \
-	/* The last step is applied further on the Y channel only*/ \
-	UNPACK_RGB32_32_STEP(rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, rgb7, rgb8, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8) \
-	UNPACK_RGB32_32_STEP(tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8, rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, rgb7, rgb8) \
-	UNPACK_RGB32_32_STEP(rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, rgb7, rgb8, tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8) \
-	UNPACK_RGB32_32_STEP(tmp1, tmp2, tmp3, tmp4, tmp5, tmp6, tmp7, tmp8, rgb1, rgb2, rgb3, rgb4, rgb5, rgb6, rgb7, rgb8) \
-	/* first compute Y', (B-Y') and (R-Y'), in 16bits values, for the first line */ \
-	/* Y is saved for each pixel, while only sums of (B-Y') and (R-Y') for pairs of adjacents pixels are saved*/ \
-	r_16 = _mm_unpacklo_epi8(rgb1, _mm_setzero_si128()); \
-	g_16 = _mm_unpacklo_epi8(rgb2, _mm_setzero_si128()); \
-	b_16 = _mm_unpacklo_epi8(rgb3, _mm_setzero_si128()); \
-	y1_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y1_16 = _mm_add_epi16(y1_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y1_16 = _mm_srli_epi16(y1_16, 8); \
-	cb2_16 = _mm_sub_epi16(b_16, y1_16); \
-	cr2_16 = _mm_sub_epi16(r_16, y1_16); \
-	r_16 = _mm_unpacklo_epi8(rgb5, _mm_setzero_si128()); \
-	g_16 = _mm_unpacklo_epi8(rgb6, _mm_setzero_si128()); \
-	b_16 = _mm_unpacklo_epi8(rgb7, _mm_setzero_si128()); \
-	y2_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y2_16 = _mm_add_epi16(y2_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y2_16 = _mm_srli_epi16(y2_16, 8); \
-	cb2_16 = _mm_add_epi16(cb2_16, _mm_sub_epi16(b_16, y2_16)); \
-	cr2_16 = _mm_add_epi16(cr2_16, _mm_sub_epi16(r_16, y2_16)); \
-	/* Rescale Y' to Y, pack it to 8bit values and save it */ \
-	y1_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y1_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	y2_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y2_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	Y = _mm_packus_epi16(y1_16, y2_16); \
-	Y = _mm_unpackhi_epi8(_mm_slli_si128(Y, 8), Y); \
-	SAVE_SI128((__m128i*)(y_ptr1+16), Y); \
-	/* same for the second line, compute Y', (B-Y') and (R-Y'), in 16bits values */ \
-	/* Y is saved for each pixel, while only sums of (B-Y') and (R-Y') for pairs of adjacents pixels are added to the previous values*/ \
-	r_16 = _mm_unpackhi_epi8(rgb1, _mm_setzero_si128()); \
-	g_16 = _mm_unpackhi_epi8(rgb2, _mm_setzero_si128()); \
-	b_16 = _mm_unpackhi_epi8(rgb3, _mm_setzero_si128()); \
-	y1_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y1_16 = _mm_add_epi16(y1_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y1_16 = _mm_srli_epi16(y1_16, 8); \
-	cb2_16 = _mm_add_epi16(cb2_16, _mm_sub_epi16(b_16, y1_16)); \
-	cr2_16 = _mm_add_epi16(cr2_16, _mm_sub_epi16(r_16, y1_16)); \
-	r_16 = _mm_unpackhi_epi8(rgb5, _mm_setzero_si128()); \
-	g_16 = _mm_unpackhi_epi8(rgb6, _mm_setzero_si128()); \
-	b_16 = _mm_unpackhi_epi8(rgb7, _mm_setzero_si128()); \
-	y2_16 = _mm_add_epi16(_mm_mullo_epi16(r_16, _mm_set1_epi16(param->r_factor)), \
-		_mm_mullo_epi16(g_16, _mm_set1_epi16(param->g_factor))); \
-	y2_16 = _mm_add_epi16(y2_16, _mm_mullo_epi16(b_16, _mm_set1_epi16(param->b_factor))); \
-	y2_16 = _mm_srli_epi16(y2_16, 8); \
-	cb2_16 = _mm_add_epi16(cb2_16, _mm_sub_epi16(b_16, y2_16)); \
-	cr2_16 = _mm_add_epi16(cr2_16, _mm_sub_epi16(r_16, y2_16)); \
-	/* Rescale Y' to Y, pack it to 8bit values and save it */ \
-	y1_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y1_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	y2_16 = _mm_add_epi16(_mm_srli_epi16(_mm_mullo_epi16(y2_16, _mm_set1_epi16(param->y_factor)), 7), _mm_set1_epi16(param->y_offset)); \
-	Y = _mm_packus_epi16(y1_16, y2_16); \
-	Y = _mm_unpackhi_epi8(_mm_slli_si128(Y, 8), Y); \
-	SAVE_SI128((__m128i*)(y_ptr2+16), Y); \
-	/* Rescale Cb and Cr to their final range */ \
-	cb2_16 = _mm_add_epi16(_mm_srai_epi16(_mm_mullo_epi16(_mm_srai_epi16(cb2_16, 2), _mm_set1_epi16(param->cb_factor)), 8), _mm_set1_epi16(128)); \
-	cr2_16 = _mm_add_epi16(_mm_srai_epi16(_mm_mullo_epi16(_mm_srai_epi16(cr2_16, 2), _mm_set1_epi16(param->cr_factor)), 8), _mm_set1_epi16(128)); \
-	/* Pack and save Cb Cr */ \
-	cb = _mm_packus_epi16(cb1_16, cb2_16); \
-	cr = _mm_packus_epi16(cr1_16, cr2_16); \
-	SAVE_SI128((__m128i*)(u_ptr), cb); \
-	SAVE_SI128((__m128i*)(v_ptr), cr);
-
-void rgb32_yuv420_sse(uint32_t width, uint32_t height, 
-	const uint8_t *RGBA, uint32_t RGBA_stride, 
-	uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride, 
-	YCbCrType yuv_type)
-{
-	#define LOAD_SI128 _mm_load_si128
-	#define SAVE_SI128 _mm_stream_si128
-	const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]);
-	
-	uint32_t x, y;
-	for(y=0; y<(height-1); y+=2)
-	{
-		const uint8_t *rgb_ptr1=RGBA+y*RGBA_stride,
-			*rgb_ptr2=RGBA+(y+1)*RGBA_stride;
-		
-		uint8_t *y_ptr1=Y+y*Y_stride,
-			*y_ptr2=Y+(y+1)*Y_stride,
-			*u_ptr=U+(y/2)*UV_stride,
-			*v_ptr=V+(y/2)*UV_stride;
-		
-		for(x=0; x<(width-31); x+=32)
-		{
-			RGBA2YUV_32
-			
-			rgb_ptr1+=128;
-			rgb_ptr2+=128;
-			y_ptr1+=32;
-			y_ptr2+=32;
-			u_ptr+=16; 
-			v_ptr+=16;
-		}
-	}
-	#undef LOAD_SI128
-	#undef SAVE_SI128
-}
-
-void rgb32_yuv420_sseu(uint32_t width, uint32_t height, 
-	const uint8_t *RGBA, uint32_t RGBA_stride, 
-	uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride, 
-	YCbCrType yuv_type)
-{
-	#define LOAD_SI128 _mm_loadu_si128
-	#define SAVE_SI128 _mm_storeu_si128
-	const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]);
-	
-	uint32_t x, y;
-	for(y=0; y<(height-1); y+=2)
-	{
-		const uint8_t *rgb_ptr1=RGBA+y*RGBA_stride,
-			*rgb_ptr2=RGBA+(y+1)*RGBA_stride;
-		
-		uint8_t *y_ptr1=Y+y*Y_stride,
-			*y_ptr2=Y+(y+1)*Y_stride,
-			*u_ptr=U+(y/2)*UV_stride,
-			*v_ptr=V+(y/2)*UV_stride;
-		
-		for(x=0; x<(width-31); x+=32)
-		{
-			RGBA2YUV_32
-			
-			rgb_ptr1+=128;
-			rgb_ptr2+=128;
-			y_ptr1+=32;
-			y_ptr2+=32;
-			u_ptr+=16; 
-			v_ptr+=16;
-		}
-	}
-	#undef LOAD_SI128
-	#undef SAVE_SI128
-}
-
-#endif
-
 #ifdef __SSE2__
 
 #define UV2RGB_16(U,V,R1,G1,B1,R2,G2,B2) \
@@ -965,22 +484,6 @@ PACK_RGB24_32_STEP(R1, R2, G1, G2, B1, B2, RGB1, RGB2, RGB3, RGB4, RGB5, RGB6) \
 	__m128i u = LOAD_SI128((const __m128i*)(u_ptr)); \
 	__m128i v = LOAD_SI128((const __m128i*)(v_ptr)); \
 
-#define LOAD_UV_NV12 \
-	__m128i uv1 = LOAD_SI128((const __m128i*)(uv_ptr)); \
-	__m128i uv2 = LOAD_SI128((const __m128i*)(uv_ptr+16)); \
-	__m128i u = _mm_packus_epi16(_mm_and_si128(uv1, _mm_set1_epi16(255)), _mm_and_si128(uv2, _mm_set1_epi16(255))); \
-	uv1 = _mm_srli_epi16(uv1, 8); \
-	uv2 = _mm_srli_epi16(uv2, 8); \
-	__m128i v = _mm_packus_epi16(_mm_and_si128(uv1, _mm_set1_epi16(255)), _mm_and_si128(uv2, _mm_set1_epi16(255))); \
-
-#define LOAD_UV_NV21 \
-	__m128i uv1 = LOAD_SI128((const __m128i*)(uv_ptr)); \
-	__m128i uv2 = LOAD_SI128((const __m128i*)(uv_ptr+16)); \
-	__m128i v = _mm_packus_epi16(_mm_and_si128(uv1, _mm_set1_epi16(255)), _mm_and_si128(uv2, _mm_set1_epi16(255))); \
-	uv1 = _mm_srli_epi16(uv1, 8); \
-	uv2 = _mm_srli_epi16(uv2, 8); \
-	__m128i u = _mm_packus_epi16(_mm_and_si128(uv1, _mm_set1_epi16(255)), _mm_and_si128(uv2, _mm_set1_epi16(255))); \
-
 #define YUV2RGB_32 \
 	__m128i r_tmp, g_tmp, b_tmp; \
 	__m128i r_16_1, g_16_1, b_16_1, r_16_2, g_16_2, b_16_2; \
@@ -1080,15 +583,6 @@ PACK_RGB24_32_STEP(R1, R2, G1, G2, B1, B2, RGB1, RGB2, RGB3, RGB4, RGB5, RGB6) \
 	LOAD_UV_PLANAR \
 	YUV2RGB_32
 
-#define YUV2RGB_32_NV12 \
-	LOAD_UV_NV12 \
-	YUV2RGB_32
-	
-#define YUV2RGB_32_NV21 \
-	LOAD_UV_NV21 \
-	YUV2RGB_32
-
-
 void yuv420_rgb24_sse(
 	uint32_t width, uint32_t height, 
 	const uint8_t *Y, const uint8_t *U, const uint8_t *V, uint32_t Y_stride, uint32_t UV_stride, 
@@ -1163,146 +657,4 @@ void yuv420_rgb24_sseu(
 	#undef SAVE_SI128
 }
 
-void nv12_rgb24_sse(
-	uint32_t width, uint32_t height, 
-	const uint8_t *Y, const uint8_t *UV, uint32_t Y_stride, uint32_t UV_stride, 
-	uint8_t *RGB, uint32_t RGB_stride, 
-	YCbCrType yuv_type)
-{
-	#define LOAD_SI128 _mm_load_si128
-	#define SAVE_SI128 _mm_stream_si128
-	const YUV2RGBParam *const param = &(YUV2RGB[yuv_type]);
-	
-	uint32_t x, y;
-	for(y=0; y<(height-1); y+=2)
-	{
-		const uint8_t *y_ptr1=Y+y*Y_stride,
-			*y_ptr2=Y+(y+1)*Y_stride,
-			*uv_ptr=UV+(y/2)*UV_stride;
-		
-		uint8_t *rgb_ptr1=RGB+y*RGB_stride,
-			*rgb_ptr2=RGB+(y+1)*RGB_stride;
-		
-		for(x=0; x<(width-31); x+=32)
-		{
-			YUV2RGB_32_NV12
-			
-			y_ptr1+=32;
-			y_ptr2+=32;
-			uv_ptr+=32; 
-			rgb_ptr1+=96;
-			rgb_ptr2+=96;
-		}
-	}
-	#undef LOAD_SI128
-	#undef SAVE_SI128
-}
-
-void nv12_rgb24_sseu(
-	uint32_t width, uint32_t height, 
-	const uint8_t *Y, const uint8_t *UV, uint32_t Y_stride, uint32_t UV_stride, 
-	uint8_t *RGB, uint32_t RGB_stride, 
-	YCbCrType yuv_type)
-{
-	#define LOAD_SI128 _mm_loadu_si128
-	#define SAVE_SI128 _mm_storeu_si128
-	const YUV2RGBParam *const param = &(YUV2RGB[yuv_type]);
-	
-	uint32_t x, y;
-	for(y=0; y<(height-1); y+=2)
-	{
-		const uint8_t *y_ptr1=Y+y*Y_stride,
-			*y_ptr2=Y+(y+1)*Y_stride,
-			*uv_ptr=UV+(y/2)*UV_stride;
-		
-		uint8_t *rgb_ptr1=RGB+y*RGB_stride,
-			*rgb_ptr2=RGB+(y+1)*RGB_stride;
-		
-		for(x=0; x<(width-31); x+=32)
-		{
-			YUV2RGB_32_NV12
-			
-			y_ptr1+=32;
-			y_ptr2+=32;
-			uv_ptr+=32; 
-			rgb_ptr1+=96;
-			rgb_ptr2+=96;
-		}
-	}
-	#undef LOAD_SI128
-	#undef SAVE_SI128
-}
-
-void nv21_rgb24_sse(
-	uint32_t width, uint32_t height, 
-	const uint8_t *Y, const uint8_t *UV, uint32_t Y_stride, uint32_t UV_stride, 
-	uint8_t *RGB, uint32_t RGB_stride, 
-	YCbCrType yuv_type)
-{
-	#define LOAD_SI128 _mm_load_si128
-	#define SAVE_SI128 _mm_stream_si128
-	const YUV2RGBParam *const param = &(YUV2RGB[yuv_type]);
-	
-	uint32_t x, y;
-	for(y=0; y<(height-1); y+=2)
-	{
-		const uint8_t *y_ptr1=Y+y*Y_stride,
-			*y_ptr2=Y+(y+1)*Y_stride,
-			*uv_ptr=UV+(y/2)*UV_stride;
-		
-		uint8_t *rgb_ptr1=RGB+y*RGB_stride,
-			*rgb_ptr2=RGB+(y+1)*RGB_stride;
-		
-		for(x=0; x<(width-31); x+=32)
-		{
-			YUV2RGB_32_NV21
-			
-			y_ptr1+=32;
-			y_ptr2+=32;
-			uv_ptr+=32; 
-			rgb_ptr1+=96;
-			rgb_ptr2+=96;
-		}
-	}
-	#undef LOAD_SI128
-	#undef SAVE_SI128
-}
-
-void nv21_rgb24_sseu(
-	uint32_t width, uint32_t height, 
-	const uint8_t *Y, const uint8_t *UV, uint32_t Y_stride, uint32_t UV_stride, 
-	uint8_t *RGB, uint32_t RGB_stride, 
-	YCbCrType yuv_type)
-{
-	#define LOAD_SI128 _mm_loadu_si128
-	#define SAVE_SI128 _mm_storeu_si128
-	const YUV2RGBParam *const param = &(YUV2RGB[yuv_type]);
-	
-	uint32_t x, y;
-	for(y=0; y<(height-1); y+=2)
-	{
-		const uint8_t *y_ptr1=Y+y*Y_stride,
-			*y_ptr2=Y+(y+1)*Y_stride,
-			*uv_ptr=UV+(y/2)*UV_stride;
-		
-		uint8_t *rgb_ptr1=RGB+y*RGB_stride,
-			*rgb_ptr2=RGB+(y+1)*RGB_stride;
-		
-		for(x=0; x<(width-31); x+=32)
-		{
-			YUV2RGB_32_NV21
-			
-			y_ptr1+=32;
-			y_ptr2+=32;
-			uv_ptr+=32; 
-			rgb_ptr1+=96;
-			rgb_ptr2+=96;
-		}
-	}
-	#undef LOAD_SI128
-	#undef SAVE_SI128
-}
-
-
-
 #endif //__SSE2__