OR	$0,R2,R2

	VPMSUMD	V20,const1,V12	// vpmsumd with constants
	LVX	(R4+off64),V20	// load next from buffer
	OR	$0,R2,R2

	VPMSUMD	V21,const1,V13	// vpmsumd with constants
	LVX	(R4+off80),V21	// load next from buffer
	OR	$0,R2,R2

	VPMSUMD	V22,const1,V14	// vpmsumd with constants
	LVX	(R4+off96),V22	// load next from buffer
	OR	$0,R2,R2

	VPMSUMD	V23,const1,V15	// vpmsumd with constants

	VPERM  IN0, IN0, LEMASK, IN0
#endif

	VPMSUMD XH, H4L, XL   // H^4.lo·Xi.lo
	VPMSUMD XH, H4, XM    // H^4.hi·Xi.lo+H^4.lo·Xi.hi
	VPMSUMD XH, H4H, XH   // H^4.hi·Xi.hi
	VPMSUMD IN1, H3L, XL1
	VPMSUMD IN1, H3, XM1
	VPMSUMD IN1, H3H, XH1

	VXOR  XL, XL3, XL
	VXOR  XM, XM3, XM
	VXOR  XH, XH3, XH
	VPERM IN2, IN3, LOPERM, T0
	VPERM IN2, IN3, HIPERM, T1

	VPMSUMD XL, XC2, T2   // 1st reduction phase