{ACMPPS, yxcmpi, Pm, opBytes{0xc2, 0}},
	{ACMPQ, ycmpl, Pw, opBytes{0x83, 07, 0x3d, 0x81, 07, 0x39, 0x3b}},
	{ACMPSB, ynone, Pb, opBytes{0xa6}},
	{ACMPSD, yxcmpi, Px, opBytes{Pf2, 0xc2}},
	{ACMPSL, ynone, Px, opBytes{0xa7}},
	{ACMPSQ, ynone, Pw, opBytes{0xa7}},
	{ACMPSS, yxcmpi, Px, opBytes{Pf3, 0xc2}},
	{ACMPSW, ynone, Pe, opBytes{0xa7}},

(Load <t1> p1 (Store {t2} p2 x _))
	&& isSamePtr(p1, p2)
	&& t1.Compare(x.Type) == types.CMPeq
	&& t1.Size() == t2.Size()
	=> x
(Load <t1> p1 (Store {t2} p2 _ (Store {t3} p3 x _)))
	&& isSamePtr(p1, p3)
	&& t1.Compare(x.Type) == types.CMPeq
	&& t1.Size() == t2.Size()
	&& disjoint(p3, t3.Size(), p2, t2.Size())
	=> x

			opset(AVPOPCNTD, r0)

		case AVCMPEQ: /* vcmpequb[.], vcmpequh[.], vcmpequw[.], vcmpequd[.] */
			opset(AVCMPEQUB, r0)
			opset(AVCMPEQUBCC, r0)
			opset(AVCMPEQUH, r0)
			opset(AVCMPEQUHCC, r0)
			opset(AVCMPEQUW, r0)
			opset(AVCMPEQUWCC, r0)
			opset(AVCMPEQUD, r0)
			opset(AVCMPEQUDCC, r0)

		case AVCMPGT: /* vcmpgt[u,s]b[.], vcmpgt[u,s]h[.], vcmpgt[u,s]w[.], vcmpgt[u,s]d[.] */

(Leq64U x y) => (LessEqualU (CMP x y))

// Optimize comparison between a floating-point value and 0.0 with "FCMP $(0.0), Fn"
(FCMPS x (FMOVSconst [0])) => (FCMPS0 x)
(FCMPS (FMOVSconst [0]) x) => (InvertFlags (FCMPS0 x))
(FCMPD x (FMOVDconst [0])) => (FCMPD0 x)
(FCMPD (FMOVDconst [0]) x) => (InvertFlags (FCMPD0 x))

// CSEL needs a flag-generating argument. Synthesize a TSTW if necessary.

		rel := obj.Addrel(c.cursym)
		rel.Off = int32(c.pc)
		rel.Siz = 8
		rel.Sym = p.From.Sym
		rel.Add = 0
		rel.Type = objabi.R_ARM64_GOTPCREL

	case 72: /* vaddp/vand/vcmeq/vorr/vadd/veor/vfmla/vfmls/vbit/vbsl/vcmtst/vsub/vbif/vuzip1/vuzip2/vrax1 Vm.<T>, Vn.<T>, Vd.<T> */
		af := int((p.From.Reg >> 5) & 15)
		af3 := int((p.Reg >> 5) & 15)
		at := int((p.To.Reg >> 5) & 15)
		if af != af3 || af != at {