-movdqu (%3), %%xmm0 # Load 16 Y Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
-movdqa %%xmm0, %%xmm2 # Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
-punpcklbw %%xmm1, %%xmm2 # v1 Y3 u1 Y2 v0 Y1 u0 Y0 \n\
-movdqu %%xmm2, (%1) # Store low YUYV \n\
-punpckhbw %%xmm1, %%xmm0 # v3 Y7 u3 Y6 v2 Y5 u2 Y4 \n\
-movdqu %%xmm0, 16(%1) # Store high YUYV \n\
+movdqa %%xmm3, %%xmm4 # Y7 Y6 Y5 Y4 Y3 Y2 Y1 Y0 \n\
+punpcklbw %%xmm1, %%xmm4 # v1 Y3 u1 Y2 v0 Y1 u0 Y0 \n\
+movdqu %%xmm4, (%1) # Store low YUYV \n\
+punpckhbw %%xmm1, %%xmm3 # v3 Y7 u3 Y6 v2 Y5 u2 Y4 \n\
+movdqu %%xmm3, 16(%1) # Store high YUYV \n\