diff --git a/libavcodec/x86/hpeldsp.asm b/libavcodec/x86/hpeldsp.asm
index d50bf5d39c..70066e4f45 100644
--- a/libavcodec/x86/hpeldsp.asm
+++ b/libavcodec/x86/hpeldsp.asm
@@ -27,11 +27,9 @@
 
 %include "libavutil/x86/x86util.asm"
 
-SECTION_RODATA
 cextern pb_1
 cextern pw_1
 cextern pw_2
-pb_interleave16: db 0, 8, 1, 9, 2, 10, 3, 11, 4, 12, 5, 13, 6, 14, 7, 15
 
 cextern pw_8192
 
@@ -503,51 +501,49 @@ SET_PIXELS_XY2 avg, pw_1, _no_rnd
 
 %macro SSSE3_PIXELS_XY2 1-2
 cglobal %1_pixels16_xy2, 4,5,%2
-    mova        m4, [pb_interleave16]
-    mova        m5, [pb_1]
-    movu        m0, [r1]
     movu        m1, [r1+1]
-    pmaddubsw   m0, m5
+    movu        m0, [r1]
+    mova        m5, [pb_1]
+    mova        m4, [pw_8192]
     pmaddubsw   m1, m5
+    pmaddubsw   m0, m5
     xor         r4, r4
     add         r1, r2
 .loop:
-    movu        m2, [r1+r4]
     movu        m3, [r1+r4+1]
-    pmaddubsw   m2, m5
+    movu        m2, [r1+r4]
     pmaddubsw   m3, m5
-    paddw       m0, m2
-    paddw       m1, m3
-    pmulhrsw    m0, [pw_8192]
-    pmulhrsw    m1, [pw_8192]
+    pmaddubsw   m2, m5
 %ifidn %1, avg
     mova        m6, [r0+r4]
-    packuswb    m0, m1
-    pshufb      m0, m4
+%endif
+    paddw       m1, m3
+    paddw       m0, m2
+    pmulhrsw    m1, m4
+    pmulhrsw    m0, m4
+    pslldq      m1, 1
+    por         m0, m1
+%ifidn %1, avg
     pavgb       m0, m6
-%else
-    packuswb    m0, m1
-    pshufb      m0, m4
 %endif
     mova   [r0+r4], m0
     add         r4, r2
 
-    movu        m0, [r1+r4]
     movu        m1, [r1+r4+1]
-    pmaddubsw   m0, m5
+    movu        m0, [r1+r4]
     pmaddubsw   m1, m5
-    paddw       m2, m0
-    paddw       m3, m1
-    pmulhrsw    m2, [pw_8192]
-    pmulhrsw    m3, [pw_8192]
+    pmaddubsw   m0, m5
 %ifidn %1, avg
     mova        m6, [r0+r4]
-    packuswb    m2, m3
-    pshufb      m2, m4
+%endif
+    paddw       m3, m1
+    paddw       m2, m0
+    pmulhrsw    m3, m4
+    pmulhrsw    m2, m4
+    pslldq      m3, 1
+    por         m2, m3
+%ifidn %1, avg
     pavgb       m2, m6
-%else
-    packuswb    m2, m3
-    pshufb      m2, m4
 %endif
     mova   [r0+r4], m2
     add         r4, r2