(Cvt32Fto64U ...) => (I64TruncSatF32U ...)
(Cvt64Fto32U ...) => (I64TruncSatF64U ...)
(Cvt64Fto64U ...) => (I64TruncSatF64U ...)

(Cvt32Fto64F ...) => (F64PromoteF32 ...)
(Cvt64Fto32F ...) => (F32DemoteF64 ...)

(CvtBoolToUint8 ...) => (Copy ...)

(Round32F ...) => (Copy ...)
(Round64F ...) => (Copy ...)

// Lowering shifts
// Unsigned shifts need to return 0 if shift amount is >= width of shifted value.

(Cvt32Fto32 ...) => (TRUNCFW ...)
(Cvt64Fto32 ...) => (TRUNCDW ...)
(Cvt32Fto64 ...) => (TRUNCFV ...)
(Cvt64Fto64 ...) => (TRUNCDV ...)
(Cvt32Fto64F ...) => (MOVFD ...)
(Cvt64Fto32F ...) => (MOVDF ...)

(CvtBoolToUint8 ...) => (Copy ...)

(Round(32|64)F ...) => (Copy ...)

// comparisons
(Eq8 x y)  => (SGTU (MOVVconst [1]) (XOR (ZeroExt8to64 x) (ZeroExt8to64 y)))

// float-int conversion
(Cvt32to(32|64)F ...) => (MOVW(F|D) ...)
(Cvt(32|64)Fto32 ...) => (TRUNC(F|D)W ...)
(Cvt32Fto64F ...) => (MOVFD ...)
(Cvt64Fto32F ...) => (MOVDF ...)

(CvtBoolToUint8 ...) => (Copy ...)

(Round(32|64)F ...) => (Copy ...)

// comparisons
(Eq8 x y)  => (SGTUconst [1] (XOR (ZeroExt8to32 x) (ZeroExt8to32 y)))
(Eq16 x y) => (SGTUconst [1] (XOR (ZeroExt16to32 x) (ZeroExt16to32 y)))

(Cvt32Fto32 ...) => (FCVTWS ...)
(Cvt32Fto64 ...) => (FCVTLS ...)
(Cvt64Fto32 ...) => (FCVTWD ...)
(Cvt64Fto64 ...) => (FCVTLD ...)

(Cvt32Fto64F ...) => (FCVTDS ...)
(Cvt64Fto32F ...) => (FCVTSD ...)

(CvtBoolToUint8 ...) => (Copy ...)

(Round(32|64)F ...) => (LoweredRound(32|64)F ...)

(Slicemask <t> x) => (SRAI [63] (NEG <t> x))

// Truncations

(Cvt32Fto32 ...) => (TRUNCFW ...)
(Cvt64Fto32 ...) => (TRUNCDW ...)
(Cvt32Fto64 ...) => (TRUNCFV ...)
(Cvt64Fto64 ...) => (TRUNCDV ...)
(Cvt32Fto64F ...) => (MOVFD ...)
(Cvt64Fto32F ...) => (MOVDF ...)

(CvtBoolToUint8 ...) => (Copy ...)

(Round(32|64)F ...) => (Copy ...)

// comparisons
(Eq8 x y)  => (SGTU (MOVVconst [1]) (XOR (ZeroExt8to64 x) (ZeroExt8to64 y)))

	{name: "Cvt32Fto64", argLength: 1},
	{name: "Cvt64Fto32", argLength: 1},
	{name: "Cvt64Fto64", argLength: 1},
	{name: "Cvt32Fto64F", argLength: 1},
	{name: "Cvt64Fto32F", argLength: 1},
	{name: "CvtBoolToUint8", argLength: 1},

	// Force rounding to precision of type.
	{name: "Round32F", argLength: 1},
	{name: "Round64F", argLength: 1},

	// Automatically inserted safety checks

(Cvt64Fto32  (Const64F [c])) => (Const32  [int32(c)])
(Cvt64Fto64  (Const64F [c])) => (Const64  [int64(c)])
(Round32F x:(Const32F)) => x
(Round64F x:(Const64F)) => x
(CvtBoolToUint8 (ConstBool [false])) => (Const8 [0])
(CvtBoolToUint8 (ConstBool [true])) => (Const8 [1])

(Trunc16to8  (ZeroExt8to16  x)) => x
(Trunc32to8  (ZeroExt8to32  x)) => x
(Trunc32to16 (ZeroExt8to32  x)) => (ZeroExt8to16  x)

(Cvt64Fto(32|64) x) => (MFVSRD (FCTI(W|D)Z x))

(Cvt32Fto64F ...) => (Copy ...) // Note v will have the wrong type for patterns dependent on Float32/Float64
(Cvt64Fto32F ...) => (FRSP ...)

(CvtBoolToUint8 ...) => (Copy ...)

(Round(32|64)F ...) => (LoweredRound(32|64)F ...)

(Sqrt ...) => (FSQRT ...)
(Sqrt32 ...) => (FSQRTS ...)
(Floor ...) => (FFLOOR ...)
(Ceil ...) => (FCEIL ...)
(Trunc ...) => (FTRUNC ...)

(Cvt32Fto64U ...) => (CLGEBR ...)
(Cvt64Fto32U ...) => (CLFDBR ...)
(Cvt64Fto64U ...) => (CLGDBR ...)

// Lowering float32 <-> float64
(Cvt32Fto64F ...) => (LDEBR ...)
(Cvt64Fto32F ...) => (LEDBR ...)

(CvtBoolToUint8 ...) => (Copy ...)

(Round(32|64)F ...) => (LoweredRound(32|64)F ...)

// Lowering shifts

// Lower bounded shifts first. No need to check shift value.

(Cvt64Fto32 ...) => (MOVDW ...)
(Cvt32Fto32U ...) => (MOVFWU ...)
(Cvt64Fto32U ...) => (MOVDWU ...)
(Cvt32Fto64F ...) => (MOVFD ...)
(Cvt64Fto32F ...) => (MOVDF ...)

(Round(32|64)F ...) => (Copy ...)

(CvtBoolToUint8 ...) => (Copy ...)

// fused-multiply-add
(FMA x y z) => (FMULAD z x y)

// comparisons
(Eq8 x y)  => (Equal (CMP (ZeroExt8to32 x) (ZeroExt8to32 y)))