&xorps ($Xhi,$Xhn);
&movdqu ($Xhn,&QWP(0,$inp)); # Ii
&pxor ($T1,$Xi); # aggregated Karatsuba post-processing
&movdqu ($Xn,&QWP(16,$inp)); # Ii+1
&pxor ($T1,$Xhi); #
&xorps ($Xhi,$Xhn);
&movdqu ($Xhn,&QWP(0,$inp)); # Ii
&pxor ($T1,$Xi); # aggregated Karatsuba post-processing
&movdqu ($Xn,&QWP(16,$inp)); # Ii+1
&pxor ($T1,$Xhi); #