SSSE3/SSE4 Intrinsics Optimizations

Change-Id: I55c74d9b2a0b323886d8a80eaad899b187c22cd7

8 files changed, 223 insertions, 425 deletions

diff --git a/common/x86/ih264_chroma_intra_pred_filters_ssse3.c b/common/x86/ih264_chroma_intra_pred_filters_ssse3.c
index 45101a4..d43ce20 100644
--- a/common/x86/ih264_chroma_intra_pred_filters_ssse3.c
+++ b/common/x86/ih264_chroma_intra_pred_filters_ssse3.c
@@ -103,47 +103,35 @@ void ih264_intra_pred_chroma_8x8_mode_horz_ssse3(UWORD8 *pu1_src,
     UWORD8 *pu1_left; /* Pointer to start of top predictors */
     WORD32 dst_strd2;
 
-    __m128i left_16x8b, left_sh_16x8b;
     __m128i row1_16x8b, row2_16x8b;
-    __m128i const_14_15_16x8b;
 
     UNUSED(src_strd);
     UNUSED(ngbr_avail);
 
     pu1_left = pu1_src + 2 * BLK8x8SIZE - 2;
 
-    left_16x8b = _mm_loadu_si128((__m128i *)(pu1_left - 14));
-
-    const_14_15_16x8b = _mm_set1_epi16(0x0f0e);
 
     dst_strd2 = dst_strd << 1;
-    left_sh_16x8b = _mm_slli_si128(left_16x8b, 2);
-    row1_16x8b = _mm_shuffle_epi8(left_16x8b, const_14_15_16x8b);
-    row2_16x8b = _mm_shuffle_epi8(left_sh_16x8b, const_14_15_16x8b);
+    row1_16x8b = _mm_set1_epi16(*((WORD16 *)(pu1_left)));
+    row2_16x8b = _mm_set1_epi16(*((WORD16 *)(pu1_left - 2)));
     _mm_storeu_si128((__m128i *)pu1_dst, row1_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), row2_16x8b);
 
-    left_16x8b = _mm_slli_si128(left_16x8b, 4);
-    left_sh_16x8b = _mm_slli_si128(left_sh_16x8b, 4);
     pu1_dst += dst_strd2;
-    row1_16x8b = _mm_shuffle_epi8(left_16x8b, const_14_15_16x8b);
-    row2_16x8b = _mm_shuffle_epi8(left_sh_16x8b, const_14_15_16x8b);
+    row1_16x8b = _mm_set1_epi16(*((WORD16 *)(pu1_left - 4)));
+    row2_16x8b = _mm_set1_epi16(*((WORD16 *)(pu1_left - 6)));
     _mm_storeu_si128((__m128i *)pu1_dst, row1_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), row2_16x8b);
 
-    left_16x8b = _mm_slli_si128(left_16x8b, 4);
-    left_sh_16x8b = _mm_slli_si128(left_sh_16x8b, 4);
     pu1_dst += dst_strd2;
-    row1_16x8b = _mm_shuffle_epi8(left_16x8b, const_14_15_16x8b);
-    row2_16x8b = _mm_shuffle_epi8(left_sh_16x8b, const_14_15_16x8b);
+    row1_16x8b = _mm_set1_epi16(*((WORD16 *)(pu1_left - 8)));
+    row2_16x8b = _mm_set1_epi16(*((WORD16 *)(pu1_left - 10)));
     _mm_storeu_si128((__m128i *)pu1_dst, row1_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), row2_16x8b);
 
-    left_16x8b = _mm_slli_si128(left_16x8b, 4);
-    left_sh_16x8b = _mm_slli_si128(left_sh_16x8b, 4);
     pu1_dst += dst_strd2;
-    row1_16x8b = _mm_shuffle_epi8(left_16x8b, const_14_15_16x8b);
-    row2_16x8b = _mm_shuffle_epi8(left_sh_16x8b, const_14_15_16x8b);
+    row1_16x8b = _mm_set1_epi16(*((WORD16 *)(pu1_left - 12)));
+    row2_16x8b = _mm_set1_epi16(*((WORD16 *)(pu1_left - 14)));
     _mm_storeu_si128((__m128i *)pu1_dst, row1_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), row2_16x8b);
 }
@@ -273,7 +261,6 @@ void ih264_intra_pred_chroma_8x8_mode_plane_ssse3(UWORD8 *pu1_src,
     //calculating a, b and c
     {
         WORD32 h_u, h_v, v_u, v_v;
-        WORD32 temp1, temp2;
 
         __m128i h_val1_16x8b, h_val2_16x8b;
         __m128i h_val1_8x16b, h_val2_8x16b, h_val_4x32b;
@@ -302,13 +289,10 @@ void ih264_intra_pred_chroma_8x8_mode_plane_ssse3(UWORD8 *pu1_src,
         h_val_4x32b = _mm_madd_epi16(mul_8x16b, h_val1_8x16b);
         v_val_4x32b = _mm_madd_epi16(mul_8x16b, v_val1_8x16b);
 
-        temp1 = _mm_extract_epi16(h_val1_16x8b, 3);
-        temp2 = _mm_extract_epi16(v_val1_16x8b, 3);
-
         hv_val_4x32b = _mm_hadd_epi32(h_val_4x32b, v_val_4x32b);
 
-        a_u = ((temp1 & 0xff) + (temp2 & 0xff)) << 4;
-        a_v = ((temp1 >> 8) + (temp2 >> 8)) << 4;
+        a_u = (pu1_left[7 * (-2)] + pu1_top[14]) << 4;
+        a_v = (pu1_left[7 * (-2) + 1] + pu1_top[15]) << 4;
 
         h_u = _mm_extract_epi16(hv_val_4x32b, 0);
         h_v = _mm_extract_epi16(hv_val_4x32b, 2);
diff --git a/common/x86/ih264_inter_pred_filters_ssse3.c b/common/x86/ih264_inter_pred_filters_ssse3.c
index 6d318c9..c8537da 100644
--- a/common/x86/ih264_inter_pred_filters_ssse3.c
+++ b/common/x86/ih264_inter_pred_filters_ssse3.c
@@ -111,23 +111,12 @@ void ih264_inter_pred_luma_copy_ssse3(UWORD8 *pu1_src,
 
     if(wd == 4)
     {
-        __m128i mask_full_128b, mask_low_32b;
-
-        mask_full_128b = _mm_set1_epi8(0xff);
-        mask_low_32b = _mm_srli_si128(mask_full_128b, 12);
-        // mask for first four bytes
-
         do
         {
-            y_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src);
-            y_1_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + src_strd));
-            y_2_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + src_strd2));
-            y_3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + src_strd3));
-
-            _mm_maskmoveu_si128(y_0_16x8b, mask_low_32b, (char*)pu1_dst);
-            _mm_maskmoveu_si128(y_1_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd));
-            _mm_maskmoveu_si128(y_2_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd2));
-            _mm_maskmoveu_si128(y_3_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd3));
+            *((WORD32 *)(pu1_dst)) =  *((WORD32 *)(pu1_src));
+            *((WORD32 *)(pu1_dst + dst_strd)) = *((WORD32 *)(pu1_src + src_strd));
+            *((WORD32 *)(pu1_dst + dst_strd2)) = *((WORD32 *)(pu1_src + src_strd2));
+            *((WORD32 *)(pu1_dst + dst_strd3)) = *((WORD32 *)(pu1_src + src_strd3));
 
             ht -= 4;
             pu1_src += src_strd4;
@@ -255,11 +244,6 @@ void ih264_inter_pred_luma_horz_ssse3(UWORD8 *pu1_src,
         __m128i res_r0r1_t1_8x16b, res_r0r1_t2_8x16b, res_r0r1_t3_8x16b;
         __m128i res_r0r1_16x8b;
 
-        __m128i mask_full_16x8b, mask_low32b;
-
-        mask_full_16x8b = _mm_set1_epi8(0xff);
-        mask_low32b = _mm_srli_si128(mask_full_16x8b, 12); // mask for first four bytes
-
         //Row0 : a0 a1 a2 a3 a4 a5 a6 a7 a8 a9.....
         //Row1 : b0 b1 b2 b3 b4 b5 b6 b7 b8 b9.....
 
@@ -307,9 +291,9 @@ void ih264_inter_pred_luma_horz_ssse3(UWORD8 *pu1_src,
 
             res_r0r1_16x8b = _mm_packus_epi16(res_r0r1_t1_8x16b, res_r0r1_t1_8x16b);
 
-            _mm_maskmoveu_si128(res_r0r1_16x8b, mask_low32b, (char*)pu1_dst);
+            *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(res_r0r1_16x8b);
             res_r0r1_16x8b = _mm_srli_si128(res_r0r1_16x8b, 4);
-            _mm_maskmoveu_si128(res_r0r1_16x8b, mask_low32b, (char*)(pu1_dst + dst_strd));
+            *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(res_r0r1_16x8b);
 
             ht -= 2;
             pu1_src += src_strd << 1;
@@ -525,10 +509,6 @@ void ih264_inter_pred_luma_vert_ssse3(UWORD8 *pu1_src,
 
     if(wd == 4)
     {
-        __m128i mask_low32b;
-
-        mask_low32b = _mm_set1_epi8(0xff);
-
         //Epilogue: Load all the pred rows except sixth and seventh row
         //          for the first and second row processing.
         src_r0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src);
@@ -542,8 +522,6 @@ void ih264_inter_pred_luma_vert_ssse3(UWORD8 *pu1_src,
         src_r4_16x8b = _mm_loadl_epi64((__m128i *)pu1_src);
         pu1_src += src_strd;
 
-        mask_low32b = _mm_srli_si128(mask_low32b, 12); // mask for first four bytes
-
         src_r0_16x8b = _mm_unpacklo_epi32(src_r0_16x8b, src_r1_16x8b);
         src_r1_16x8b = _mm_unpacklo_epi32(src_r1_16x8b, src_r2_16x8b);
         src_r2_16x8b = _mm_unpacklo_epi32(src_r2_16x8b, src_r3_16x8b);
@@ -572,9 +550,9 @@ void ih264_inter_pred_luma_vert_ssse3(UWORD8 *pu1_src,
             res_t1_8x16b = _mm_srai_epi16(res_t1_8x16b, 5); //shifting right by 5 bits.
             res_16x8b = _mm_packus_epi16(res_t1_8x16b, res_t1_8x16b);
 
-            _mm_maskmoveu_si128(res_16x8b, mask_low32b, (char*)pu1_dst);
+            *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(res_16x8b);
             res_16x8b = _mm_srli_si128(res_16x8b, 4);
-            _mm_maskmoveu_si128(res_16x8b, mask_low32b, (char*)(pu1_dst + dst_strd));
+            *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(res_16x8b);
 
             src_r0_16x8b = src_r2_16x8b;
             src_r1_16x8b = src_r3_16x8b;
@@ -893,15 +871,12 @@ void ih264_inter_pred_luma_horz_hpel_vert_hpel_ssse3(UWORD8 *pu1_src,
             __m128i res_8x16b, res_16x8b;
 
             __m128i coeff0_1_8x16b, coeff2_3_8x16b, coeff4_5_8x16b;
-            __m128i const_val512_4x32b, mask_low32b;
-
-            mask_low32b = _mm_set1_epi8(0xff);
+            __m128i const_val512_4x32b;
 
             coeff0_1_8x16b = _mm_set1_epi32(0xFFFB0001);
             coeff2_3_8x16b = _mm_set1_epi32(0x00140014);
             coeff4_5_8x16b = _mm_set1_epi32(0x0001FFFB);
 
-            mask_low32b = _mm_srli_si128(mask_low32b, 12);
             const_val512_4x32b = _mm_set1_epi32(512);
 
             src_r0_8x16b = _mm_loadl_epi64((__m128i *)(pi2_temp));
@@ -947,9 +922,9 @@ void ih264_inter_pred_luma_horz_hpel_vert_hpel_ssse3(UWORD8 *pu1_src,
                 res_8x16b = _mm_packs_epi32(res_t0_4x32b, res_t1_4x32b);
                 res_16x8b = _mm_packus_epi16(res_8x16b, res_8x16b);
 
-                _mm_maskmoveu_si128(res_16x8b, mask_low32b, (char*)pu1_dst);
+                *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(res_16x8b);
                 res_16x8b = _mm_srli_si128(res_16x8b, 4);
-                _mm_maskmoveu_si128(res_16x8b, mask_low32b, (char*)(pu1_dst + dst_strd));
+                *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(res_16x8b);
 
                 src_r0_8x16b = src_r2_8x16b;
                 src_r1_8x16b = src_r3_8x16b;
@@ -1551,11 +1526,6 @@ void ih264_inter_pred_luma_horz_qpel_ssse3(UWORD8 *pu1_src,
         __m128i res_r0r1_t1_8x16b, res_r0r1_t2_8x16b, res_r0r1_t3_8x16b;
         __m128i res_r0r1_16x8b;
 
-        __m128i mask_full_16x8b, mask_low32b;
-
-        mask_full_16x8b = _mm_set1_epi8(0xff);
-        mask_low32b = _mm_srli_si128(mask_full_16x8b, 12); // mask for first four bytes
-
         //Row0 : a0 a1 a2 a3 a4 a5 a6 a7 a8 a9.....
         //Row1 : b0 b1 b2 b3 b4 b5 b6 b7 b8 b9.....
 
@@ -1607,9 +1577,9 @@ void ih264_inter_pred_luma_horz_qpel_ssse3(UWORD8 *pu1_src,
             res_r0r1_16x8b = _mm_packus_epi16(res_r0r1_t1_8x16b, res_r0r1_t1_8x16b);
             res_r0r1_16x8b = _mm_avg_epu8(src_r0r1_16x8b, res_r0r1_16x8b);              //computing q-pel
 
-            _mm_maskmoveu_si128(res_r0r1_16x8b, mask_low32b, (char*)pu1_dst);
+            *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(res_r0r1_16x8b);
             res_r0r1_16x8b = _mm_srli_si128(res_r0r1_16x8b, 4);
-            _mm_maskmoveu_si128(res_r0r1_16x8b, mask_low32b, (char*)(pu1_dst + dst_strd));
+            *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(res_r0r1_16x8b);
 
             ht -= 2;
             pu1_src += src_strd << 1;
@@ -1849,10 +1819,6 @@ void ih264_inter_pred_luma_vert_qpel_ssse3(UWORD8 *pu1_src,
 
     if(wd == 4)
     {
-        __m128i mask_low32b;
-
-        mask_low32b = _mm_set1_epi8(0xff);
-
         //Epilogue: Load all the pred rows except sixth and seventh row
         //          for the first and second row processing.
         src_r0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src);
@@ -1866,8 +1832,6 @@ void ih264_inter_pred_luma_vert_qpel_ssse3(UWORD8 *pu1_src,
         src_r4_16x8b = _mm_loadl_epi64((__m128i *)pu1_src);
         pu1_src += src_strd;
 
-        mask_low32b = _mm_srli_si128(mask_low32b, 12); // mask for first four bytes
-
         src_r0_16x8b = _mm_unpacklo_epi32(src_r0_16x8b, src_r1_16x8b);
         src_r1_16x8b = _mm_unpacklo_epi32(src_r1_16x8b, src_r2_16x8b);
         src_r2_16x8b = _mm_unpacklo_epi32(src_r2_16x8b, src_r3_16x8b);
@@ -1904,9 +1868,9 @@ void ih264_inter_pred_luma_vert_qpel_ssse3(UWORD8 *pu1_src,
 
             res_16x8b = _mm_avg_epu8(src_r0r1_16x8b, res_16x8b); //computing q-pel
 
-            _mm_maskmoveu_si128(res_16x8b, mask_low32b, (char*)pu1_dst);
+            *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(res_16x8b);
             res_16x8b = _mm_srli_si128(res_16x8b, 4);
-            _mm_maskmoveu_si128(res_16x8b, mask_low32b, (char*)(pu1_dst + dst_strd));
+            *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(res_16x8b);
 
             src_r0_16x8b = src_r2_16x8b;
             src_r1_16x8b = src_r3_16x8b;
@@ -2257,11 +2221,6 @@ void ih264_inter_pred_luma_horz_qpel_vert_qpel_ssse3(UWORD8 *pu1_src,
             __m128i res_r0r1_t1_8x16b, res_r0r1_t2_8x16b, res_r0r1_t3_8x16b;
             __m128i res_r0r1_16x8b;
 
-            __m128i mask_low32b;
-
-            mask_low32b = _mm_set1_epi8(0xff);
-            mask_low32b = _mm_srli_si128(mask_low32b, 12);
-
             //Row0 : a0 a1 a2 a3 a4 a5 a6 a7 a8 a9.....
             //Row1 : b0 b1 b2 b3 b4 b5 b6 b7 b8 b9.....
 
@@ -2313,9 +2272,9 @@ void ih264_inter_pred_luma_horz_qpel_vert_qpel_ssse3(UWORD8 *pu1_src,
 
                 res_r0r1_16x8b = _mm_avg_epu8(res_r0r1_16x8b,src_r0r1_vpel_16x8b);
 
-                _mm_maskmoveu_si128(res_r0r1_16x8b, mask_low32b, (char*)pu1_dst);
+                *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(res_r0r1_16x8b);
                 res_r0r1_16x8b = _mm_srli_si128(res_r0r1_16x8b, 4);
-                _mm_maskmoveu_si128(res_r0r1_16x8b, mask_low32b, (char*)(pu1_dst + dst_strd));
+                *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(res_r0r1_16x8b);
 
                 ht -= 2;
                 pu1_pred_horiz += src_strd << 1;
@@ -2852,16 +2811,11 @@ void ih264_inter_pred_luma_horz_qpel_vert_hpel_ssse3(UWORD8 *pu1_src,
 
             __m128i coeff0_1_8x16b, coeff2_3_8x16b, coeff4_5_8x16b;
             __m128i const_val512_4x32b, const_val16_8x16b;
-            __m128i mask_low32b;
-
-            mask_low32b = _mm_set1_epi8(0xff);
 
             coeff0_1_8x16b = _mm_set1_epi32(0xFFFB0001);
             coeff2_3_8x16b = _mm_set1_epi32(0x00140014);
             coeff4_5_8x16b = _mm_set1_epi32(0x0001FFFB);
 
-            mask_low32b = _mm_srli_si128(mask_low32b, 12);
-
             const_val512_4x32b = _mm_set1_epi32(512);
             const_val16_8x16b = _mm_set1_epi16(16);
 
@@ -2897,7 +2851,7 @@ void ih264_inter_pred_luma_horz_qpel_vert_hpel_ssse3(UWORD8 *pu1_src,
 
                 res_16x8b = _mm_avg_epu8(res_16x8b, src_hpel_16x8b);
 
-                _mm_maskmoveu_si128(res_16x8b, mask_low32b, (char*)pu1_dst);
+                *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(res_16x8b);
 
                 ht--;
                 pi2_temp2 = pi2_temp2 + 4 + 5;
@@ -3424,12 +3378,9 @@ void ih264_inter_pred_luma_horz_hpel_vert_qpel_ssse3(UWORD8 *pu1_src,
 
             __m128i coeff0_1_8x16b, coeff2_3_8x16b, coeff4_5_8x16b;
             __m128i const_val512_4x32b, const_val16_8x16b;
-            __m128i mask_low32b;
 
-            mask_low32b = _mm_set1_epi8(0xff);
             const_val512_4x32b = _mm_set1_epi32(512);
             const_val16_8x16b = _mm_set1_epi16(16);
-            mask_low32b = _mm_srli_si128(mask_low32b, 12);
 
             coeff0_1_8x16b = _mm_set1_epi32(0xFFFB0001);
             coeff2_3_8x16b = _mm_set1_epi32(0x00140014);
@@ -3483,9 +3434,9 @@ void ih264_inter_pred_luma_horz_hpel_vert_qpel_ssse3(UWORD8 *pu1_src,
 
                 res_16x8b = _mm_avg_epu8(res_16x8b, src_hpel_16x8b);
 
-                _mm_maskmoveu_si128(res_16x8b, mask_low32b, (char *)(pu1_dst));
+                *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(res_16x8b);
                 res_16x8b = _mm_srli_si128(res_16x8b, 4);
-                _mm_maskmoveu_si128(res_16x8b, mask_low32b, (char *)(pu1_dst + dst_strd));
+                *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(res_16x8b);
 
                 src_r0_8x16b = src_r2_8x16b;
                 src_r1_8x16b = src_r3_8x16b;
@@ -4106,65 +4057,6 @@ void ih264_inter_pred_chroma_ssse3(UWORD8 *pu1_src,
         }
         while(ht > 0);
 
-        /*
-        WORD32 AB, CD;
-
-        __m128i src_r1_16x8b, src_r2_16x8b, src_r3_16x8b;
-        __m128i src_r1r2_16x8b, src_r2r3_16x8b;
-        __m128i res_AB_8x16b, res_CD_8x16b, res_8x16b, res_16x8b;
-        __m128i mask_low32b;
-
-        __m128i coeffAB_16x8b, coeffCD_16x8b, round_add32_8x16b;
-        __m128i const_shuff_16x8b;
-
-        AB = (B << 8) + A;
-        CD = (D << 8) + C;
-
-        coeffAB_16x8b = _mm_set1_epi16(AB);
-        coeffCD_16x8b = _mm_set1_epi16(CD);
-
-        round_add32_8x16b = _mm_set1_epi16(32);
-
-        mask_low32b = _mm_set1_epi8(0xff);
-        src_r1_16x8b = _mm_loadl_epi64((__m128i *)pu1_src);                       //u1[0] v1[0] u1[1] v1[1] u1[2] v1[2] u1[3] v1[3]
-        pu1_src += src_strd;
-
-        const_shuff_16x8b = _mm_setr_epi32(0x03010200, 0x05030402, 0x0b090a08, 0x0d0b0c0a);
-        mask_low32b = _mm_srli_si128(mask_low32b, 12);
-
-        do
-        {
-            src_r2_16x8b = _mm_loadl_epi64((__m128i *)pu1_src);                   //u2[0] v2[0] u2[1] v2[1] u1[2] v2[2] u2[3] v2[3]
-            src_r3_16x8b = _mm_loadl_epi64((__m128i *)(pu1_src + src_strd));      //u3[0] v3[0] u3[1] v3[1] u3[2] v3[2] u3[3] v3[3]
-
-            src_r1r2_16x8b = _mm_unpacklo_epi64(src_r1_16x8b, src_r2_16x8b);
-            src_r2r3_16x8b = _mm_unpacklo_epi64(src_r2_16x8b, src_r3_16x8b);
-
-            src_r1r2_16x8b = _mm_shuffle_epi8(src_r1r2_16x8b, const_shuff_16x8b); //u1[0] u1[1] v1[0] v1[1] u1[1] u1[2] v1[1] v1[2]
-                                                                                  //u2[0] u2[1] v2[0] v2[1] u2[1] u2[2] v2[1] v2[2]
-            src_r2r3_16x8b = _mm_shuffle_epi8(src_r2r3_16x8b, const_shuff_16x8b); //u2[0] u2[1] v2[0] v2[1] u2[1] u2[2] v2[1] v2[2]
-                                                                                  //u3[0] u3[1] v3[0] v3[1] u3[1] u3[2] v3[1] v3[2]
-            res_AB_8x16b = _mm_maddubs_epi16(src_r1r2_16x8b, coeffAB_16x8b);
-            res_CD_8x16b = _mm_maddubs_epi16(src_r2r3_16x8b, coeffCD_16x8b);
-
-            res_8x16b = _mm_add_epi16(res_AB_8x16b, round_add32_8x16b);
-            res_8x16b = _mm_add_epi16(res_8x16b, res_CD_8x16b);
-            res_8x16b = _mm_srai_epi16(res_8x16b, 6);
-            res_16x8b = _mm_packus_epi16(res_8x16b, res_8x16b);
-
-            _mm_maskmoveu_si128(res_16x8b, mask_low32b, (char *)pu1_dst);
-
-            ht -= 2;
-            pu1_src += src_strd << 1;
-            res_16x8b = _mm_srli_si128(res_16x8b, 4);
-            src_r1_16x8b = src_r3_16x8b;
-
-            _mm_maskmoveu_si128(res_16x8b, mask_low32b, (char *)(pu1_dst + dst_strd));
-
-            pu1_dst += dst_strd << 1;
-        }
-        while(ht > 0);
-        */
     }
     else if(wd == 4)
     {
diff --git a/common/x86/ih264_iquant_itrans_recon_dc_ssse3.c b/common/x86/ih264_iquant_itrans_recon_dc_ssse3.c
index 565cc75..83a23ac 100644
--- a/common/x86/ih264_iquant_itrans_recon_dc_ssse3.c
+++ b/common/x86/ih264_iquant_itrans_recon_dc_ssse3.c
@@ -30,8 +30,8 @@
  *  Mohit [100664]
  *
  * @par List of Functions:
- *  - ihevc_iquant_itrans_recon_4x4_dc_ssse3()
- *  - ihevc_iquant_itrans_recon_8x8_dc_ssse3()
+ *  - ih264_iquant_itrans_recon_4x4_dc_ssse3()
+ *  - ih264_iquant_itrans_recon_8x8_dc_ssse3()
  *
  * @remarks
  *  None
@@ -397,6 +397,7 @@ void ih264_iquant_itrans_recon_chroma_4x4_dc_ssse3(WORD16 *pi2_src,
     __m128i zero_8x16b = _mm_setzero_si128();          // all bits reset to zero
     __m128i chroma_mask = _mm_set1_epi16 (0xFF);
     __m128i value_add = _mm_set1_epi16(i_macro);
+    __m128i out_r0, out_r1, out_r2, out_r3;
 
     UNUSED (pi2_src);
     UNUSED (pu2_iscal_mat);
@@ -438,12 +439,26 @@ void ih264_iquant_itrans_recon_chroma_4x4_dc_ssse3(WORD16 *pi2_src,
     pred_r2 = _mm_unpacklo_epi8(pred_r2, zero_8x16b); //p20 p21 p22 p23 -- all 16 bits
     pred_r3 = _mm_unpacklo_epi8(pred_r3, zero_8x16b); //p30 p31 p32 p33 -- all 16 bits
 
-    chroma_mask = _mm_unpacklo_epi64(chroma_mask, zero_8x16b);  //1 0 1 0 1 0 1 0 0 0 0 0 0 0 0 0  -- 8 bits
-
-    _mm_maskmoveu_si128(pred_r0, chroma_mask, (char *)(&pu1_out[0]));
-    _mm_maskmoveu_si128(pred_r1, chroma_mask, (char *)(&pu1_out[out_strd]));
-    _mm_maskmoveu_si128(pred_r2, chroma_mask, (char *)(&pu1_out[2*out_strd]));
-    _mm_maskmoveu_si128(pred_r3, chroma_mask, (char *)(&pu1_out[3*out_strd]));
+    chroma_mask = _mm_set1_epi16 (0xFF00);
+    out_r0 = _mm_loadl_epi64((__m128i *) (&pu1_out[0]));
+    out_r1 = _mm_loadl_epi64((__m128i *) (&pu1_out[out_strd]));
+    out_r2 = _mm_loadl_epi64((__m128i *) (&pu1_out[2 * out_strd]));
+    out_r3 = _mm_loadl_epi64((__m128i *) (&pu1_out[3 * out_strd]));
+
+    out_r0 = _mm_and_si128(out_r0, chroma_mask);
+    out_r1 = _mm_and_si128(out_r1, chroma_mask);
+    out_r2 = _mm_and_si128(out_r2, chroma_mask);
+    out_r3 = _mm_and_si128(out_r3, chroma_mask);
+
+    out_r0 = _mm_add_epi8(out_r0, pred_r0);
+    out_r1 = _mm_add_epi8(out_r1, pred_r1);
+    out_r2 = _mm_add_epi8(out_r2, pred_r2);
+    out_r3 = _mm_add_epi8(out_r3, pred_r3);
+
+    _mm_storel_epi64((__m128i *)(&pu1_out[0]), out_r0);
+    _mm_storel_epi64((__m128i *)(&pu1_out[out_strd]), out_r1);
+    _mm_storel_epi64((__m128i *)(&pu1_out[2 * out_strd]), out_r2);
+    _mm_storel_epi64((__m128i *)(&pu1_out[3 * out_strd]), out_r3);
 }
 
 
diff --git a/common/x86/ih264_iquant_itrans_recon_sse42.c b/common/x86/ih264_iquant_itrans_recon_sse42.c
index 6399b65..f27111f 100644
--- a/common/x86/ih264_iquant_itrans_recon_sse42.c
+++ b/common/x86/ih264_iquant_itrans_recon_sse42.c
@@ -30,8 +30,8 @@
  *  Mohit [100664]
  *
  * @par List of Functions:
- *  - ihevc_iquant_itrans_recon_4x4_sse42()
- *  - ihevc_iquant_itrans_recon_chroma_4x4_sse42()
+ *  - ih264_iquant_itrans_recon_4x4_sse42()
+ *  - ih264_iquant_itrans_recon_chroma_4x4_sse42()
  *
  * @remarks
  *  None
@@ -370,6 +370,7 @@ void ih264_iquant_itrans_recon_chroma_4x4_sse42(WORD16 *pi2_src,
     __m128i add_rshift = _mm_set1_epi32((1 << (3 - u4_qp_div_6)));
     __m128i value_32 = _mm_set1_epi32(32);
     __m128i chroma_mask = _mm_set1_epi16 (0xFF);
+    __m128i out_r0, out_r1, out_r2, out_r3;
     UNUSED (pi2_tmp);
 
     /*************************************************************/
@@ -548,10 +549,24 @@ void ih264_iquant_itrans_recon_chroma_4x4_sse42(WORD16 *pi2_src,
     resq_r2 = _mm_cvtepu8_epi16(resq_r2); //p20 p21 p22 p23 -- all 16 bits
     resq_r3 = _mm_cvtepu8_epi16(resq_r3); //p30 p31 p32 p33 -- all 16 bits
 
-    chroma_mask = _mm_unpacklo_epi64(chroma_mask, zero_8x16b);
-
-    _mm_maskmoveu_si128(resq_r0, chroma_mask, (char *)(&pu1_out[0]));
-    _mm_maskmoveu_si128(resq_r1, chroma_mask, (char *)(&pu1_out[out_strd]));
-    _mm_maskmoveu_si128(resq_r2, chroma_mask, (char *)(&pu1_out[2*out_strd]));
-    _mm_maskmoveu_si128(resq_r3, chroma_mask, (char *)(&pu1_out[3*out_strd]));
+    chroma_mask = _mm_set1_epi16 (0xFF00);
+    out_r0 = _mm_loadl_epi64((__m128i *) (&pu1_out[0]));
+    out_r1 = _mm_loadl_epi64((__m128i *) (&pu1_out[out_strd]));
+    out_r2 = _mm_loadl_epi64((__m128i *) (&pu1_out[2 * out_strd]));
+    out_r3 = _mm_loadl_epi64((__m128i *) (&pu1_out[3 * out_strd]));
+
+    out_r0 = _mm_and_si128(out_r0, chroma_mask);
+    out_r1 = _mm_and_si128(out_r1, chroma_mask);
+    out_r2 = _mm_and_si128(out_r2, chroma_mask);
+    out_r3 = _mm_and_si128(out_r3, chroma_mask);
+
+    out_r0 = _mm_add_epi8(out_r0, resq_r0);
+    out_r1 = _mm_add_epi8(out_r1, resq_r1);
+    out_r2 = _mm_add_epi8(out_r2, resq_r2);
+    out_r3 = _mm_add_epi8(out_r3, resq_r3);
+
+    _mm_storel_epi64((__m128i *)(&pu1_out[0]), out_r0);
+    _mm_storel_epi64((__m128i *)(&pu1_out[out_strd]), out_r1);
+    _mm_storel_epi64((__m128i *)(&pu1_out[2 * out_strd]), out_r2);
+    _mm_storel_epi64((__m128i *)(&pu1_out[3 * out_strd]), out_r3);
 }
diff --git a/common/x86/ih264_iquant_itrans_recon_ssse3.c b/common/x86/ih264_iquant_itrans_recon_ssse3.c
index 388cafe..30f7e59 100644
--- a/common/x86/ih264_iquant_itrans_recon_ssse3.c
+++ b/common/x86/ih264_iquant_itrans_recon_ssse3.c
@@ -30,8 +30,8 @@
  *  Mohit [100664]
  *
  * @par List of Functions:
- *  - ihevc_iquant_itrans_recon_4x4_ssse3()
- *  - ihevc_iquant_itrans_recon_8x8_ssse3()
+ *  - ih264_iquant_itrans_recon_4x4_ssse3()
+ *  - ih264_iquant_itrans_recon_8x8_ssse3()
  *
  * @remarks
  *  None
diff --git a/common/x86/ih264_luma_intra_pred_filters_ssse3.c b/common/x86/ih264_luma_intra_pred_filters_ssse3.c
index 5a35372..a1721d5 100644
--- a/common/x86/ih264_luma_intra_pred_filters_ssse3.c
+++ b/common/x86/ih264_luma_intra_pred_filters_ssse3.c
@@ -122,28 +122,22 @@ void ih264_intra_pred_luma_4x4_mode_vert_ssse3(UWORD8 *pu1_src,
 {
     UWORD8 *pu1_top;
     WORD32 dst_strd2, dst_strd3;
-
-    __m128i top_16x8b;
-    __m128i mask_full_128b, mask_low_32b;
+    WORD32 i4_top;
 
     UNUSED(src_strd);
     UNUSED(ngbr_avail);
 
-    mask_full_128b = _mm_set1_epi8(0xff);
-
     pu1_top = pu1_src + BLK_SIZE + 1;
 
-    mask_low_32b = _mm_srli_si128(mask_full_128b, 12);
-
-    top_16x8b = _mm_loadl_epi64((__m128i *)pu1_top);
+    i4_top = *((WORD32 *)pu1_top);
 
     dst_strd2 = dst_strd << 1;
     dst_strd3 = dst_strd + dst_strd2;
 
-    _mm_maskmoveu_si128(top_16x8b, mask_low_32b, (char*)pu1_dst);
-    _mm_maskmoveu_si128(top_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd));
-    _mm_maskmoveu_si128(top_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd2));
-    _mm_maskmoveu_si128(top_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd3));
+    *((WORD32 *)(pu1_dst)) = i4_top;
+    *((WORD32 *)(pu1_dst + dst_strd)) = i4_top;
+    *((WORD32 *)(pu1_dst + dst_strd2)) = i4_top;
+    *((WORD32 *)(pu1_dst + dst_strd3)) = i4_top;
 }
 
 /**
@@ -185,39 +179,31 @@ void ih264_intra_pred_luma_4x4_mode_horz_ssse3(UWORD8 *pu1_src,
                                                WORD32 dst_strd,
                                                WORD32 ngbr_avail)
 {
-    UWORD8 *pu1_left;
+    UWORD8 *pu1_left = NULL; /* Pointer to start of left predictors */
+    WORD32 row1,row2,row3,row4;
+    UWORD8 val;
     WORD32 dst_strd2, dst_strd3;
-    WORD32 val1, val2;
-
-    __m128i left_16x8b;
-    __m128i row1_16x8b, row2_16x8b, row3_16x8b, row4_16x8b;
-    __m128i mask_full_128b, mask_low_32b;
 
     UNUSED(src_strd);
     UNUSED(ngbr_avail);
-
-    mask_full_128b = _mm_set1_epi8(0xff);
-
     pu1_left = pu1_src + BLK_SIZE - 1;
 
-    mask_low_32b = _mm_srli_si128(mask_full_128b, 12);
-    left_16x8b = _mm_loadl_epi64((__m128i *)(pu1_left - 3));
-
-    val1 = _mm_extract_epi16(left_16x8b, 1);
-    val2 = _mm_extract_epi16(left_16x8b, 0);
-
-    row1_16x8b = _mm_set1_epi8(val1 >> 8);
-    row2_16x8b = _mm_set1_epi8(val1 & 0xff);
-    row3_16x8b = _mm_set1_epi8(val2 >> 8);
-    row4_16x8b = _mm_set1_epi8(val2 & 0xff);
+    val  = *pu1_left;
+    row1 = val + (val << 8) + (val << 16) + (val << 24);
+    val  = *(pu1_left - 1);
+    row2 = val + (val << 8) + (val << 16) + (val << 24);
+    val  = *(pu1_left - 2);
+    row3 = val + (val << 8) + (val << 16) + (val << 24);
+    val  = *(pu1_left - 3);
+    row4 = val + (val << 8) + (val << 16) + (val << 24);
 
     dst_strd2 = dst_strd << 1;
     dst_strd3 = dst_strd + dst_strd2;
 
-    _mm_maskmoveu_si128(row1_16x8b, mask_low_32b, (char*)pu1_dst);
-    _mm_maskmoveu_si128(row2_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd));
-    _mm_maskmoveu_si128(row3_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd2));
-    _mm_maskmoveu_si128(row4_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd3));
+    *((WORD32 *)(pu1_dst)) = row1;
+    *((WORD32 *)(pu1_dst + dst_strd)) = row2;
+    *((WORD32 *)(pu1_dst + dst_strd2)) = row3;
+    *((WORD32 *)(pu1_dst + dst_strd3)) = row4;
 }
 
 /**
@@ -259,72 +245,43 @@ void ih264_intra_pred_luma_4x4_mode_dc_ssse3(UWORD8 *pu1_src,
                                              WORD32 ngbr_avail)
 {
     UWORD8 u1_useleft; /* availability of left predictors (only for DC) */
-    UWORD8 u1_usetop;  /* availability of top predictors (only for DC) */
-    UWORD8 *pu1_left, *pu1_top;
-    WORD32 dc_val, flag;
+    UWORD8 u1_usetop; /* availability of top predictors (only for DC) */
+    UWORD8 *pu1_left = NULL; /* Pointer to start of left predictors */
+    UWORD8 *pu1_top = NULL; /* Pointer to start of top predictors */
     WORD32 dst_strd2, dst_strd3;
-
-    __m128i mask_full_128b, mask_low_32b;
-    __m128i dcval_16x8b;
-
+    WORD32 val = 0;
     UNUSED(src_strd);
     UNUSED(ngbr_avail);
-
-    mask_full_128b = _mm_set1_epi8(0xff);
-
     u1_useleft = BOOLEAN(ngbr_avail & LEFT_MB_AVAILABLE_MASK);
     u1_usetop = BOOLEAN(ngbr_avail & TOP_MB_AVAILABLE_MASK);
-
-    pu1_left = pu1_src + BLK_SIZE - 1;
     pu1_top = pu1_src + BLK_SIZE + 1;
+    pu1_left = pu1_src + BLK_SIZE - 1;
 
-    mask_low_32b = _mm_srli_si128(mask_full_128b, 12);
-
-    flag = u1_useleft + u1_usetop;
-
-    if(flag)
+    if(u1_useleft)
     {
-        WORD32 shft, ofst = 0;
-
-        __m128i left_16x8b, top_16x8b, val_16x8b, tmp_8x16b, zero_vector;
-
-        if(u1_useleft)
-        {
-            left_16x8b = _mm_loadl_epi64((__m128i *)(pu1_left - 3));
-            ofst += 2;
-        }
-        else
-            left_16x8b = _mm_setzero_si128();
-
-        zero_vector = _mm_setzero_si128();
-
-        if(u1_usetop)
-        {
-            top_16x8b = _mm_loadl_epi64((__m128i *)pu1_top);
-            ofst += 2;
-        }
-        else
-            top_16x8b = _mm_setzero_si128();
-
-        shft = flag + 1;
-        val_16x8b = _mm_unpacklo_epi32(left_16x8b, top_16x8b);
-        tmp_8x16b = _mm_sad_epu8(val_16x8b, zero_vector);
-
-        dc_val = _mm_extract_epi16(tmp_8x16b, 0);
-        dc_val = (dc_val + ofst) >> shft;
+        val += *pu1_left--;
+        val += *pu1_left--;
+        val += *pu1_left--;
+        val += *pu1_left + 2;
     }
-    else
-        dc_val = 128;
+    if(u1_usetop)
+    {
+        val += *pu1_top + *(pu1_top + 1) + *(pu1_top + 2) + *(pu1_top + 3)
+                        + 2;
+    }
+    /* Since 2 is added if either left/top pred is there,
+     val still being zero implies both preds are not there */
+    val = (val) ? (val >> (1 + u1_useleft + u1_usetop)) : 128;
+
+    val = val + (val << 8) + (val << 16) + (val << 24);
 
     dst_strd2 = dst_strd << 1;
     dst_strd3 = dst_strd + dst_strd2;
 
-    dcval_16x8b = _mm_set1_epi8(dc_val);
-
-    _mm_maskmoveu_si128(dcval_16x8b, mask_low_32b, (char*)pu1_dst);
-    _mm_maskmoveu_si128(dcval_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd));
-    _mm_maskmoveu_si128(dcval_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd2));
-    _mm_maskmoveu_si128(dcval_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd3));
+    *((WORD32 *)(pu1_dst)) = val;
+    *((WORD32 *)(pu1_dst + dst_strd)) = val;
+    *((WORD32 *)(pu1_dst + dst_strd2)) = val;
+    *((WORD32 *)(pu1_dst + dst_strd3)) = val;
 }
 
 /**
@@ -371,7 +328,7 @@ void ih264_intra_pred_luma_4x4_mode_diag_dl_ssse3(UWORD8 *pu1_src,
     __m128i top_16x8b, top_8x16b, top_sh_8x16b;
     __m128i res1_8x16b, res2_8x16b, res_16x8b;
     __m128i zero_vector, const_2_8x16b;
-    __m128i mask_full_128b, mask_low_32b;
+    WORD32 row1,row2,row3,row4;
 
     UNUSED(src_strd);
     UNUSED(ngbr_avail);
@@ -382,13 +339,11 @@ void ih264_intra_pred_luma_4x4_mode_diag_dl_ssse3(UWORD8 *pu1_src,
     zero_vector = _mm_setzero_si128();
     top_8x16b = _mm_unpacklo_epi8(top_16x8b, zero_vector);    //t0 t1 t2 t3 t4 t5 t6 t7
 
-    mask_full_128b = _mm_set1_epi8(0xff);
     top_sh_8x16b = _mm_srli_si128(top_8x16b, 2);              //t1 t2 t3 t4 t5 t6 t7 0
     const_2_8x16b = _mm_set1_epi16(2);
 
     top_sh_8x16b = _mm_shufflehi_epi16(top_sh_8x16b, 0xa4);   //t1 t2 t3 t4 t5 t6 t7 t7
     res1_8x16b = _mm_add_epi16(top_8x16b, top_sh_8x16b);
-    mask_low_32b = _mm_srli_si128(mask_full_128b, 12);
     res2_8x16b = _mm_srli_si128(res1_8x16b, 2);
 
     res1_8x16b = _mm_add_epi16(res1_8x16b, const_2_8x16b);
@@ -399,13 +354,18 @@ void ih264_intra_pred_luma_4x4_mode_diag_dl_ssse3(UWORD8 *pu1_src,
     dst_strd3 = dst_strd + dst_strd2;
 
     res_16x8b = _mm_packus_epi16(res1_8x16b, res1_8x16b);
-    _mm_maskmoveu_si128(res_16x8b, mask_low_32b, (char*)pu1_dst);
+    row1 = _mm_cvtsi128_si32(res_16x8b);
     res_16x8b = _mm_srli_si128(res_16x8b, 1);
-    _mm_maskmoveu_si128(res_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd));
+    row2 = _mm_cvtsi128_si32(res_16x8b);
     res_16x8b = _mm_srli_si128(res_16x8b, 1);
-    _mm_maskmoveu_si128(res_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd2));
+    row3 = _mm_cvtsi128_si32(res_16x8b);
     res_16x8b = _mm_srli_si128(res_16x8b, 1);
-    _mm_maskmoveu_si128(res_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd3));
+    row4 = _mm_cvtsi128_si32(res_16x8b);
+
+    *((WORD32 *)(pu1_dst)) = row1;
+    *((WORD32 *)(pu1_dst + dst_strd)) = row2;
+    *((WORD32 *)(pu1_dst + dst_strd2)) = row3;
+    *((WORD32 *)(pu1_dst + dst_strd3)) = row4;
 }
 
 /**
@@ -454,7 +414,7 @@ void ih264_intra_pred_luma_4x4_mode_diag_dr_ssse3(UWORD8 *pu1_src,
     __m128i res1_8x16b, res2_8x16b;
     __m128i res1_16x8b, res2_16x8b;
     __m128i zero_vector, const_2_8x16b;
-    __m128i mask_full_128b, mask_low_32b;
+    WORD32 row1,row2,row3,row4;
 
     UNUSED(src_strd);
     UNUSED(ngbr_avail);
@@ -468,13 +428,11 @@ void ih264_intra_pred_luma_4x4_mode_diag_dr_ssse3(UWORD8 *pu1_src,
     top_left_8x16b = _mm_unpacklo_epi8(top_left_16x8b, zero_vector);
     top_left_sh_8x16b = _mm_unpacklo_epi8(top_left_sh_16x8b, zero_vector);
 
-    mask_full_128b = _mm_set1_epi8(0xff);
     res1_8x16b = _mm_add_epi16(top_left_8x16b, top_left_sh_8x16b);           //l3+l2 l2+l1 l1+l0 l0+tl tl+t0 t0+t1 t1+t2 t2+t3...
     const_2_8x16b = _mm_set1_epi16(2);
     res2_8x16b = _mm_srli_si128(res1_8x16b, 2);                              //l2+l1 l1+l0 l0+tl tl+t0 t0+t1 t1+t2 t2+t3...
 
     res1_8x16b = _mm_add_epi16(res1_8x16b, const_2_8x16b);
-    mask_low_32b = _mm_srli_si128(mask_full_128b, 12);
     res1_8x16b = _mm_add_epi16(res2_8x16b, res1_8x16b);                      //l3+2*l2+l1+2 l2+2*l1+l0+2...
     res1_8x16b = _mm_srai_epi16(res1_8x16b, 2);
     res1_16x8b = _mm_packus_epi16(res1_8x16b, res1_8x16b);
@@ -483,12 +441,18 @@ void ih264_intra_pred_luma_4x4_mode_diag_dr_ssse3(UWORD8 *pu1_src,
     dst_strd3 = dst_strd + dst_strd2;
 
     res2_16x8b = _mm_srli_si128(res1_16x8b, 3);
-    _mm_maskmoveu_si128(res2_16x8b, mask_low_32b, (char*)pu1_dst);
+
+    row1 = _mm_cvtsi128_si32(res2_16x8b);
     res2_16x8b = _mm_srli_si128(res1_16x8b, 2);
-    _mm_maskmoveu_si128(res2_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd));
+    row2 = _mm_cvtsi128_si32(res2_16x8b);
     res2_16x8b = _mm_srli_si128(res1_16x8b, 1);
-    _mm_maskmoveu_si128(res2_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd2));
-    _mm_maskmoveu_si128(res1_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd3));
+    row3 = _mm_cvtsi128_si32(res2_16x8b);
+    row4 = _mm_cvtsi128_si32(res1_16x8b);
+
+    *((WORD32 *)(pu1_dst)) = row1;
+    *((WORD32 *)(pu1_dst + dst_strd)) = row2;
+    *((WORD32 *)(pu1_dst + dst_strd2)) = row3;
+    *((WORD32 *)(pu1_dst + dst_strd3)) = row4;
 }
 
 /**
@@ -537,14 +501,11 @@ void ih264_intra_pred_luma_4x4_mode_vert_r_ssse3(UWORD8 *pu1_src,
     __m128i w121_a1_8x16b, w121_a2_8x16b, w121_sh_8x16b;
     __m128i row1_16x8b, row2_16x8b, row3_16x8b, row4_16x8b;
     __m128i zero_vector, const_2_8x16b;
-    __m128i mask_full_128b, mask_low_32b;
+    WORD32 row1,row2,row3,row4;
 
     UNUSED(src_strd);
     UNUSED(ngbr_avail);
 
-    mask_full_128b = _mm_set1_epi8(0xff);
-    mask_low_32b = _mm_srli_si128(mask_full_128b, 12);
-
     pu1_left = pu1_src + BLK_SIZE - 1;
 
     val_16x8b = _mm_loadl_epi64((__m128i *)(pu1_left - 2));
@@ -575,10 +536,15 @@ void ih264_intra_pred_luma_4x4_mode_vert_r_ssse3(UWORD8 *pu1_src,
     dst_strd2 = dst_strd << 1;
     dst_strd3 = dst_strd + dst_strd2;
 
-    _mm_maskmoveu_si128(row1_16x8b, mask_low_32b, (char*)pu1_dst);
-    _mm_maskmoveu_si128(row2_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd));
-    _mm_maskmoveu_si128(row3_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd2));
-    _mm_maskmoveu_si128(row4_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd3));
+    row1 = _mm_cvtsi128_si32(row1_16x8b);
+    row2 = _mm_cvtsi128_si32(row2_16x8b);
+    row3 = _mm_cvtsi128_si32(row3_16x8b);
+    row4 = _mm_cvtsi128_si32(row4_16x8b);
+
+    *((WORD32 *)(pu1_dst)) = row1;
+    *((WORD32 *)(pu1_dst + dst_strd)) = row2;
+    *((WORD32 *)(pu1_dst + dst_strd2)) = row3;
+    *((WORD32 *)(pu1_dst + dst_strd3)) = row4;
 }
 
 /*
@@ -629,14 +595,11 @@ void ih264_intra_pred_luma_4x4_mode_horz_d_ssse3(UWORD8 *pu1_src,
     __m128i row1_16x8b, row2_16x8b, row3_16x8b, row4_16x8b;
 
     __m128i zero_vector, const_2_8x16b;
-    __m128i mask_full_128b, mask_low_32b;
+    WORD32 row1,row2,row3,row4;
 
     UNUSED(src_strd);
     UNUSED(ngbr_avail);
 
-    mask_full_128b = _mm_set1_epi8(0xff);
-    mask_low_32b = _mm_srli_si128(mask_full_128b, 12);
-
     pu1_left = pu1_src + BLK_SIZE - 1;
 
     val_16x8b = _mm_loadl_epi64((__m128i *)(pu1_left - 3));
@@ -669,10 +632,15 @@ void ih264_intra_pred_luma_4x4_mode_horz_d_ssse3(UWORD8 *pu1_src,
     row2_16x8b = _mm_srli_si128(row4_16x8b, 4);
     row3_16x8b = _mm_srli_si128(row4_16x8b, 2);
 
-    _mm_maskmoveu_si128(row1_16x8b, mask_low_32b, (char*)pu1_dst);
-    _mm_maskmoveu_si128(row2_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd));
-    _mm_maskmoveu_si128(row3_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd2));
-    _mm_maskmoveu_si128(row4_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd3));
+    row1 = _mm_cvtsi128_si32(row1_16x8b);
+    row2 = _mm_cvtsi128_si32(row2_16x8b);
+    row3 = _mm_cvtsi128_si32(row3_16x8b);
+    row4 = _mm_cvtsi128_si32(row4_16x8b);
+
+    *((WORD32 *)(pu1_dst)) = row1;
+    *((WORD32 *)(pu1_dst + dst_strd)) = row2;
+    *((WORD32 *)(pu1_dst + dst_strd2)) = row3;
+    *((WORD32 *)(pu1_dst + dst_strd3)) = row4;
 }
 
 /**
@@ -721,14 +689,11 @@ void ih264_intra_pred_luma_4x4_mode_vert_l_ssse3(UWORD8 *pu1_src,
     __m128i row1_16x8b, row2_16x8b, row3_16x8b, row4_16x8b;
 
     __m128i zero_vector, const_2_8x16b;
-    __m128i mask_full_128b, mask_low_32b;
+    WORD32 row1,row2,row3,row4;
 
     UNUSED(src_strd);
     UNUSED(ngbr_avail);
 
-    mask_full_128b = _mm_set1_epi8(0xff);
-    mask_low_32b = _mm_srli_si128(mask_full_128b, 12);
-
     pu1_top = pu1_src +BLK_SIZE + 1;
 
     val_16x8b = _mm_loadl_epi64((__m128i *)pu1_top);
@@ -756,10 +721,15 @@ void ih264_intra_pred_luma_4x4_mode_vert_l_ssse3(UWORD8 *pu1_src,
     row3_16x8b = _mm_srli_si128(row1_16x8b, 1);
     row4_16x8b = _mm_srli_si128(row2_16x8b, 1);
 
-    _mm_maskmoveu_si128(row1_16x8b, mask_low_32b, (char*)pu1_dst);
-    _mm_maskmoveu_si128(row2_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd));
-    _mm_maskmoveu_si128(row3_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd2));
-    _mm_maskmoveu_si128(row4_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd3));
+    row1 = _mm_cvtsi128_si32(row1_16x8b);
+    row2 = _mm_cvtsi128_si32(row2_16x8b);
+    row3 = _mm_cvtsi128_si32(row3_16x8b);
+    row4 = _mm_cvtsi128_si32(row4_16x8b);
+
+    *((WORD32 *)(pu1_dst)) = row1;
+    *((WORD32 *)(pu1_dst + dst_strd)) = row2;
+    *((WORD32 *)(pu1_dst + dst_strd2)) = row3;
+    *((WORD32 *)(pu1_dst + dst_strd3)) = row4;
 }
 
 /**
@@ -809,14 +779,11 @@ void ih264_intra_pred_luma_4x4_mode_horz_u_ssse3(UWORD8 *pu1_src,
     __m128i row1_16x8b, row2_16x8b, row3_16x8b, row4_16x8b;
 
     __m128i zero_vector, const_2_8x16b, rev_16x8b;
-    __m128i mask_full_128b, mask_low_32b;
+    WORD32 row1,row2,row3,row4;
 
     UNUSED(src_strd);
     UNUSED(ngbr_avail);
 
-    mask_full_128b = _mm_set1_epi8(0xff);
-    mask_low_32b = _mm_srli_si128(mask_full_128b, 12);
-
     pu1_left = pu1_src + BLK_SIZE - 1;
 
     zero_vector = _mm_setzero_si128();
@@ -851,10 +818,15 @@ void ih264_intra_pred_luma_4x4_mode_horz_u_ssse3(UWORD8 *pu1_src,
     row3_16x8b = _mm_srli_si128(row1_16x8b, 4);
     row4_16x8b = _mm_srli_si128(row1_16x8b, 6);
 
-    _mm_maskmoveu_si128(row1_16x8b, mask_low_32b, (char*)pu1_dst);
-    _mm_maskmoveu_si128(row2_16x8b, mask_low_32b, (char*)(pu1_dst + dst_strd));
-    _mm_maskmoveu_si128(row3_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd2));
-    _mm_maskmoveu_si128(row4_16x8b, mask_low_32b, (char*)(pu1_dst  + dst_strd3));
+    row1 = _mm_cvtsi128_si32(row1_16x8b);
+    row2 = _mm_cvtsi128_si32(row2_16x8b);
+    row3 = _mm_cvtsi128_si32(row3_16x8b);
+    row4 = _mm_cvtsi128_si32(row4_16x8b);
+
+    *((WORD32 *)(pu1_dst)) = row1;
+    *((WORD32 *)(pu1_dst + dst_strd)) = row2;
+    *((WORD32 *)(pu1_dst + dst_strd2)) = row3;
+    *((WORD32 *)(pu1_dst + dst_strd3)) = row4;
 }
 
 /*******************    8x8 Modes    *******************/
@@ -1814,9 +1786,7 @@ void ih264_intra_pred_luma_16x16_mode_horz_ssse3(UWORD8 *pu1_src,
 {
     UWORD8 *pu1_left;
     WORD32 dst_strd2, dst_strd3, dst_strd4;
-    WORD32 val1, val2;
 
-    __m128i val_16x8b;
     __m128i row1_16x8b, row2_16x8b, row3_16x8b, row4_16x8b;
 
     UNUSED(src_strd);
@@ -1826,60 +1796,46 @@ void ih264_intra_pred_luma_16x16_mode_horz_ssse3(UWORD8 *pu1_src,
 
     dst_strd4 = dst_strd << 2;
 
-    val_16x8b = _mm_loadu_si128((__m128i *)(pu1_left - 15));
-
     dst_strd2 = dst_strd << 1;
     dst_strd3 = dst_strd4 - dst_strd;
 
-    val1 =  _mm_extract_epi16(val_16x8b, 7);
-    val2 =  _mm_extract_epi16(val_16x8b, 6);
-
-    row1_16x8b = _mm_set1_epi8(val1 >> 8);
-    row2_16x8b = _mm_set1_epi8(val1 & 0xff);
-    row3_16x8b = _mm_set1_epi8(val2 >> 8);
-    row4_16x8b = _mm_set1_epi8(val2 & 0xff);
+    row1_16x8b = _mm_set1_epi8(*(pu1_left));
+    row2_16x8b = _mm_set1_epi8(*(pu1_left - 1));
+    row3_16x8b = _mm_set1_epi8(*(pu1_left - 2));
+    row4_16x8b = _mm_set1_epi8(*(pu1_left - 3));
 
     _mm_storeu_si128((__m128i *)pu1_dst, row1_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), row2_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd2), row3_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd3), row4_16x8b);
 
-    val1 =  _mm_extract_epi16(val_16x8b, 5);
-    val2 =  _mm_extract_epi16(val_16x8b, 4);
-
     pu1_dst += dst_strd4;
-    row1_16x8b = _mm_set1_epi8(val1 >> 8);
-    row2_16x8b = _mm_set1_epi8(val1 & 0xff);
-    row3_16x8b = _mm_set1_epi8(val2 >> 8);
-    row4_16x8b = _mm_set1_epi8(val2 & 0xff);
+    row1_16x8b = _mm_set1_epi8(*(pu1_left - 4));
+    row2_16x8b = _mm_set1_epi8(*(pu1_left - 5));
+    row3_16x8b = _mm_set1_epi8(*(pu1_left - 6));
+    row4_16x8b = _mm_set1_epi8(*(pu1_left - 7));
 
     _mm_storeu_si128((__m128i *)pu1_dst, row1_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), row2_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd2), row3_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd3), row4_16x8b);
 
-    val1 =  _mm_extract_epi16(val_16x8b, 3);
-    val2 =  _mm_extract_epi16(val_16x8b, 2);
-
     pu1_dst += dst_strd4;
-    row1_16x8b = _mm_set1_epi8(val1 >> 8);
-    row2_16x8b = _mm_set1_epi8(val1 & 0xff);
-    row3_16x8b = _mm_set1_epi8(val2 >> 8);
-    row4_16x8b = _mm_set1_epi8(val2 & 0xff);
+    row1_16x8b = _mm_set1_epi8(*(pu1_left - 8));
+    row2_16x8b = _mm_set1_epi8(*(pu1_left - 9));
+    row3_16x8b = _mm_set1_epi8(*(pu1_left - 10));
+    row4_16x8b = _mm_set1_epi8(*(pu1_left - 11));
 
     _mm_storeu_si128((__m128i *)pu1_dst, row1_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), row2_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd2), row3_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd3), row4_16x8b);
 
-    val1 =  _mm_extract_epi16(val_16x8b, 1);
-    val2 =  _mm_extract_epi16(val_16x8b, 0);
-
     pu1_dst += dst_strd4;
-    row1_16x8b = _mm_set1_epi8(val1 >> 8);
-    row2_16x8b = _mm_set1_epi8(val1 & 0xff);
-    row3_16x8b = _mm_set1_epi8(val2 >> 8);
-    row4_16x8b = _mm_set1_epi8(val2 & 0xff);
+    row1_16x8b = _mm_set1_epi8(*(pu1_left - 12));
+    row2_16x8b = _mm_set1_epi8(*(pu1_left - 13));
+    row3_16x8b = _mm_set1_epi8(*(pu1_left - 14));
+    row4_16x8b = _mm_set1_epi8(*(pu1_left - 15));
 
     _mm_storeu_si128((__m128i *)pu1_dst, row1_16x8b);
     _mm_storeu_si128((__m128i *)(pu1_dst + dst_strd), row2_16x8b);
diff --git a/common/x86/ih264_padding_ssse3.c b/common/x86/ih264_padding_ssse3.c
index 6dadd39..43ded8e 100644
--- a/common/x86/ih264_padding_ssse3.c
+++ b/common/x86/ih264_padding_ssse3.c
@@ -97,9 +97,6 @@ void ih264_pad_left_luma_ssse3(UWORD8 *pu1_src,
     WORD32 row;
     WORD32 i;
     UWORD8 *pu1_dst;
-    __m128i const0_16x8b;
-
-    const0_16x8b = _mm_setzero_si128();
 
     ASSERT(pad_size % 8 == 0);
 
@@ -107,9 +104,8 @@ void ih264_pad_left_luma_ssse3(UWORD8 *pu1_src,
     {
         __m128i src_temp0_16x8b;
 
-        src_temp0_16x8b =  _mm_loadu_si128((__m128i *)pu1_src);
         pu1_dst = pu1_src - pad_size;
-        src_temp0_16x8b = _mm_shuffle_epi8(src_temp0_16x8b, const0_16x8b);
+        src_temp0_16x8b = _mm_set1_epi8(*pu1_src);
         for(i = 0; i < pad_size; i += 8)
         {
             _mm_storel_epi64((__m128i *)(pu1_dst + i), src_temp0_16x8b);
@@ -168,20 +164,14 @@ void ih264_pad_left_chroma_ssse3(UWORD8 *pu1_src,
     WORD32 row;
     WORD32 col;
     UWORD8 *pu1_dst;
-    __m128i const0_16x8b, const1_16x8b;
-    const0_16x8b = _mm_setzero_si128();
-    const1_16x8b = _mm_set1_epi8(1);
-    const0_16x8b = _mm_unpacklo_epi8(const0_16x8b, const1_16x8b);
 
     ASSERT(pad_size % 8 == 0);
     for(row = 0; row < ht; row++)
     {
         __m128i src_temp0_16x8b;
 
-        src_temp0_16x8b =  _mm_loadu_si128((__m128i *)pu1_src);
         pu1_dst = pu1_src - pad_size;
-        src_temp0_16x8b = _mm_shuffle_epi8(src_temp0_16x8b, const0_16x8b);
-
+        src_temp0_16x8b = _mm_set1_epi16(*((UWORD16 *)pu1_src));
         for(col = 0; col < pad_size; col += 8)
         {
             _mm_storel_epi64((__m128i *)(pu1_dst + col), src_temp0_16x8b);
@@ -240,7 +230,6 @@ void ih264_pad_right_luma_ssse3(UWORD8 *pu1_src,
     WORD32 row;
     WORD32 col;
     UWORD8 *pu1_dst;
-    __m128i const0_16x8b;
 
     ASSERT(pad_size % 8 == 0);
 
@@ -248,10 +237,8 @@ void ih264_pad_right_luma_ssse3(UWORD8 *pu1_src,
     {
         __m128i src_temp0_16x8b;
 
-        src_temp0_16x8b =  _mm_loadu_si128((__m128i *)(pu1_src - 1));
-        const0_16x8b = _mm_setzero_si128();
         pu1_dst = pu1_src;
-        src_temp0_16x8b = _mm_shuffle_epi8(src_temp0_16x8b, const0_16x8b);
+        src_temp0_16x8b = _mm_set1_epi8(*(pu1_src - 1));
         for(col = 0; col < pad_size; col += 8)
         {
             _mm_storel_epi64((__m128i *)(pu1_dst + col), src_temp0_16x8b);
@@ -310,10 +297,6 @@ void ih264_pad_right_chroma_ssse3(UWORD8 *pu1_src,
     WORD32 row;
     WORD32 col;
     UWORD8 *pu1_dst;
-    __m128i const0_16x8b, const1_16x8b;
-    const0_16x8b = _mm_setzero_si128();
-    const1_16x8b = _mm_set1_epi8(1);
-    const0_16x8b = _mm_unpacklo_epi8(const0_16x8b, const1_16x8b);
 
     ASSERT(pad_size % 8 == 0);
 
@@ -321,9 +304,8 @@ void ih264_pad_right_chroma_ssse3(UWORD8 *pu1_src,
     {
         __m128i src_temp0_16x8b;
 
-        src_temp0_16x8b =  _mm_loadu_si128((__m128i *)(pu1_src - 2));
         pu1_dst = pu1_src;
-        src_temp0_16x8b = _mm_shuffle_epi8(src_temp0_16x8b, const0_16x8b);
+        src_temp0_16x8b = _mm_set1_epi16(*((UWORD16 *)(pu1_src - 2)));
         for(col = 0; col < pad_size; col += 8)
         {
             _mm_storel_epi64((__m128i *)(pu1_dst + col), src_temp0_16x8b);
diff --git a/common/x86/ih264_weighted_pred_sse42.c b/common/x86/ih264_weighted_pred_sse42.c
index b1684b7..48f1f54 100644
--- a/common/x86/ih264_weighted_pred_sse42.c
+++ b/common/x86/ih264_weighted_pred_sse42.c
@@ -96,12 +96,6 @@ void ih264_default_weighted_pred_luma_sse42(UWORD8 *pu1_src1,
 
     if(wd == 4)
     {
-        __m128i mask_full_16x8b, mask_ll4B_16x8b;
-
-        mask_full_16x8b = _mm_set1_epi8(0xff);
-        mask_ll4B_16x8b = _mm_srli_si128(mask_full_16x8b, 12);
-        // mask for first four bytes
-
         do
         {
             y0_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1);
@@ -121,13 +115,10 @@ void ih264_default_weighted_pred_luma_sse42(UWORD8 *pu1_src1,
             y0_2_16x8b = _mm_avg_epu8(y0_2_16x8b, y1_2_16x8b);
             y0_3_16x8b = _mm_avg_epu8(y0_3_16x8b, y1_3_16x8b);
 
-            _mm_maskmoveu_si128(y0_0_16x8b, mask_ll4B_16x8b, (char*)pu1_dst);
-            _mm_maskmoveu_si128(y0_1_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + dst_strd));
-            _mm_maskmoveu_si128(y0_2_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + (dst_strd << 1)));
-            _mm_maskmoveu_si128(y0_3_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + dst_strd * 3));
+            *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(y0_0_16x8b);
+            *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(y0_1_16x8b);
+            *((WORD32 *)(pu1_dst + (dst_strd << 1))) = _mm_cvtsi128_si32(y0_2_16x8b);
+            *((WORD32 *)(pu1_dst + dst_strd * 3)) = _mm_cvtsi128_si32(y0_3_16x8b);
 
             ht -= 4;
             pu1_src1 += src_strd1 << 2;
@@ -268,12 +259,6 @@ void ih264_default_weighted_pred_chroma_sse42(UWORD8 *pu1_src1,
 
     if(wd == 2)
     {
-        __m128i mask_full_16x8b, mask_ll4B_16x8b;
-
-        mask_full_16x8b = _mm_set1_epi8(0xff);
-        mask_ll4B_16x8b = _mm_srli_si128(mask_full_16x8b, 12);
-        // mask for first four bytes
-
         do
         {
             uv0_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1);
@@ -285,9 +270,8 @@ void ih264_default_weighted_pred_chroma_sse42(UWORD8 *pu1_src1,
             uv0_0_16x8b = _mm_avg_epu8(uv0_0_16x8b, uv1_0_16x8b);
             uv0_1_16x8b = _mm_avg_epu8(uv0_1_16x8b, uv1_1_16x8b);
 
-            _mm_maskmoveu_si128(uv0_0_16x8b, mask_ll4B_16x8b, (char*)pu1_dst);
-            _mm_maskmoveu_si128(uv0_1_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + dst_strd));
+            *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(uv0_0_16x8b);
+            *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(uv0_1_16x8b);
 
             ht -= 2;
             pu1_src1 += src_strd1 << 1;
@@ -419,12 +403,6 @@ void ih264_weighted_pred_luma_sse42(UWORD8 *pu1_src,
     {
         __m128i y_0_8x16b, y_2_8x16b;
 
-        __m128i mask_full_16x8b, mask_ll4B_16x8b;
-
-        mask_full_16x8b = _mm_set1_epi8(0xff);
-        mask_ll4B_16x8b = _mm_srli_si128(mask_full_16x8b, 12);
-        // mask for first four bytes
-
         do
         {
             y_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src);
@@ -455,13 +433,10 @@ void ih264_weighted_pred_luma_sse42(UWORD8 *pu1_src,
             y_2_16x8b = _mm_srli_si128(y_0_16x8b, 8);
             y_3_16x8b = _mm_srli_si128(y_0_16x8b, 12);
 
-            _mm_maskmoveu_si128(y_0_16x8b, mask_ll4B_16x8b, (char*)pu1_dst);
-            _mm_maskmoveu_si128(y_1_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + dst_strd));
-            _mm_maskmoveu_si128(y_2_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + (dst_strd << 1)));
-            _mm_maskmoveu_si128(y_3_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + dst_strd * 3));
+            *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(y_0_16x8b);
+            *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(y_1_16x8b);
+            *((WORD32 *)(pu1_dst + (dst_strd << 1))) = _mm_cvtsi128_si32(y_2_16x8b);
+            *((WORD32 *)(pu1_dst + dst_strd * 3)) = _mm_cvtsi128_si32(y_3_16x8b);
 
             ht -= 4;
             pu1_src += src_strd << 2;
@@ -660,12 +635,6 @@ void ih264_weighted_pred_chroma_sse42(UWORD8 *pu1_src,
     {
         __m128i y_0_8x16b;
 
-        __m128i mask_full_16x8b, mask_ll4B_16x8b;
-
-        mask_full_16x8b = _mm_set1_epi8(0xff);
-        mask_ll4B_16x8b = _mm_srli_si128(mask_full_16x8b, 12);
-        // mask for first four bytes
-
         do
         {
             y_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src);
@@ -686,9 +655,8 @@ void ih264_weighted_pred_chroma_sse42(UWORD8 *pu1_src,
             y_0_16x8b = _mm_packus_epi16(y_0_8x16b, y_0_8x16b);
             y_1_16x8b = _mm_srli_si128(y_0_16x8b, 4);
 
-            _mm_maskmoveu_si128(y_0_16x8b, mask_ll4B_16x8b, (char*)pu1_dst);
-            _mm_maskmoveu_si128(y_1_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + dst_strd));
+            *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(y_0_16x8b);
+            *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(y_1_16x8b);
 
             ht -= 2;
             pu1_src += src_strd << 1;
@@ -890,12 +858,6 @@ void ih264_weighted_bi_pred_luma_sse42(UWORD8 *pu1_src1,
         __m128i y1_0_8x16b, y1_2_8x16b;
         __m128i y2_0_8x16b, y2_2_8x16b;
 
-        __m128i mask_ll4B_16x8b;
-
-        mask_ll4B_16x8b = _mm_set1_epi8(0xff);
-        mask_ll4B_16x8b = _mm_srli_si128(mask_ll4B_16x8b, 12);
-        // mask for first four bytes
-
         do
         {
             y1_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1);
@@ -942,13 +904,11 @@ void ih264_weighted_bi_pred_luma_sse42(UWORD8 *pu1_src1,
             y1_2_16x8b = _mm_srli_si128(y1_0_16x8b, 8);
             y1_3_16x8b = _mm_srli_si128(y1_0_16x8b, 12);
 
-            _mm_maskmoveu_si128(y1_0_16x8b, mask_ll4B_16x8b, (char*)pu1_dst);
-            _mm_maskmoveu_si128(y1_1_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + dst_strd));
-            _mm_maskmoveu_si128(y1_2_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + (dst_strd << 1)));
-            _mm_maskmoveu_si128(y1_3_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + dst_strd * 3));
+            *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(y1_0_16x8b);
+            *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(y1_1_16x8b);
+            *((WORD32 *)(pu1_dst + (dst_strd << 1))) = _mm_cvtsi128_si32(y1_2_16x8b);
+            *((WORD32 *)(pu1_dst + dst_strd * 3)) = _mm_cvtsi128_si32(y1_3_16x8b);
+
 
             ht -= 4;
             pu1_src1 += src_strd1 << 2;
@@ -1187,11 +1147,6 @@ void ih264_weighted_bi_pred_chroma_sse42(UWORD8 *pu1_src1,
     {
         __m128i y1_0_8x16b, y2_0_8x16b;
 
-        __m128i mask_full_16x8b, mask_ll4B_16x8b;
-
-        mask_full_16x8b = _mm_set1_epi8(0xff);
-        mask_ll4B_16x8b = _mm_srli_si128(mask_full_16x8b, 12);
-
         do
         {
             y1_0_16x8b = _mm_loadl_epi64((__m128i *)pu1_src1);
@@ -1218,9 +1173,8 @@ void ih264_weighted_bi_pred_chroma_sse42(UWORD8 *pu1_src1,
             y1_0_16x8b = _mm_packus_epi16(y1_0_8x16b, y1_0_8x16b);
             y1_1_16x8b = _mm_srli_si128(y1_0_16x8b, 4);
 
-            _mm_maskmoveu_si128(y1_0_16x8b, mask_ll4B_16x8b, (char*)pu1_dst);
-            _mm_maskmoveu_si128(y1_1_16x8b, mask_ll4B_16x8b,
-                                (char*)(pu1_dst + dst_strd));
+            *((WORD32 *)(pu1_dst)) = _mm_cvtsi128_si32(y1_0_16x8b);
+            *((WORD32 *)(pu1_dst + dst_strd)) = _mm_cvtsi128_si32(y1_1_16x8b);
 
             ht -= 2;
             pu1_src1 += src_strd1 << 1;