(Com(64|32|16|8) ...) => (MVN ...)

// math package intrinsics
(Abs         ...) => (FABSD   ...)
(Sqrt        ...) => (FSQRTD  ...)
(Ceil        ...) => (FRINTPD ...)
(Floor       ...) => (FRINTMD ...)
(Round       ...) => (FRINTAD ...)
(RoundToEven ...) => (FRINTND ...)
(Trunc       ...) => (FRINTZD ...)
(FMA       x y z) => (FMADDD z x y)

(Sqrt32 ...) => (FSQRTS ...)

(Min(64|32)F ...) => (FMIN(D|S) ...)

(Div64F x (Const64F <t> [c])) && reciprocalExact64(c) => (Mul64F x (Const64F <t> [1/c]))

// rewrite single-precision sqrt expression "float32(math.Sqrt(float64(x)))"
(Cvt64Fto32F sqrt0:(Sqrt (Cvt32Fto64F x))) && sqrt0.Uses==1 => (Sqrt32 x)

(Sqrt (Const64F [c])) && !math.IsNaN(math.Sqrt(c)) => (Const64F [math.Sqrt(c)])

// for rewriting results of some late-expanded rewrites (below)
(SelectN [0] (MakeResult x ___)) => x

		{name: "POPCNTL", argLength: 1, reg: gp11, asm: "POPCNTL", clobberFlags: true},

		// SQRTSx computes sqrt(arg0)
		// S = float32, D = float64
		{name: "SQRTSD", argLength: 1, reg: fp11, asm: "SQRTSD"},
		{name: "SQRTSS", argLength: 1, reg: fp11, asm: "SQRTSS"},

		// ROUNDSD rounds arg0 to an integer depending on auxint
		// 0 means math.RoundToEven, 1 means math.Floor, 2 math.Ceil, 3 math.Trunc

  }
  return nullptr;
}

//===----------------------------------------------------------------------===//
// SqrtOp
//===----------------------------------------------------------------------===//

OpFoldResult SqrtOp::fold(FoldAdaptor adaptor) {
  auto operands = adaptor.getOperands();
  Type result_type = getType();
  // Only constant fold for tensor of f32 is implemented.

INFER_RETURN_TYPE_COMPONENTS_FROM_OPERANDS(SoftplusGradOp);
INFER_RETURN_TYPE_COMPONENTS_FROM_OPERANDS(SoftsignOp);
INFER_RETURN_TYPE_COMPONENTS_FROM_OPERANDS(SoftsignGradOp);
INFER_RETURN_TYPE_COMPONENTS_FROM_OPERANDS(SqrtOp);
INFER_RETURN_TYPE_COMPONENTS_FROM_OPERANDS(SqrtGradOp);
INFER_RETURN_TYPE_COMPONENTS_FROM_OPERANDS(SquareOp);
INFER_RETURN_TYPE_COMPONENTS_FROM_OPERANDS(StringStripOp);
INFER_RETURN_TYPE_COMPONENTS_FROM_OPERANDS(TanOp);

	case OpSignExt8to64:
		return rewriteValueWasm_OpSignExt8to64(v)
	case OpSlicemask:
		return rewriteValueWasm_OpSlicemask(v)
	case OpSqrt:
		v.Op = OpWasmF64Sqrt
		return true
	case OpSqrt32:
		v.Op = OpWasmF32Sqrt
		return true
	case OpStaticCall:
		v.Op = OpWasmLoweredStaticCall
		return true
	case OpStore:
		return rewriteValueWasm_OpStore(v)
	case OpSub16:
		v.Op = OpWasmI64Sub

		return true
	case OpSignmask:
		return rewriteValueMIPS_OpSignmask(v)
	case OpSlicemask:
		return rewriteValueMIPS_OpSlicemask(v)
	case OpSqrt:
		v.Op = OpMIPSSQRTD
		return true
	case OpSqrt32:
		v.Op = OpMIPSSQRTF
		return true
	case OpStaticCall:
		v.Op = OpMIPSCALLstatic
		return true
	case OpStore:
		return rewriteValueMIPS_OpStore(v)
	case OpSub16:
		v.Op = OpMIPSSUB

		return true
	case OpSignExt8to64:
		v.Op = OpLOONG64MOVBreg
		return true
	case OpSlicemask:
		return rewriteValueLOONG64_OpSlicemask(v)
	case OpSqrt:
		v.Op = OpLOONG64SQRTD
		return true
	case OpSqrt32:
		v.Op = OpLOONG64SQRTF
		return true
	case OpStaticCall:
		v.Op = OpLOONG64CALLstatic
		return true
	case OpStore:
		return rewriteValueLOONG64_OpStore(v)
	case OpSub16:
		v.Op = OpLOONG64SUBV

		return true
	case OpSignExt8to64:
		v.Op = OpMIPS64MOVBreg
		return true
	case OpSlicemask:
		return rewriteValueMIPS64_OpSlicemask(v)
	case OpSqrt:
		v.Op = OpMIPS64SQRTD
		return true
	case OpSqrt32:
		v.Op = OpMIPS64SQRTF
		return true
	case OpStaticCall:
		v.Op = OpMIPS64CALLstatic
		return true
	case OpStore:
		return rewriteValueMIPS64_OpStore(v)
	case OpSub16:
		v.Op = OpMIPS64SUBV

		return true
	case OpSignExt8to64:
		v.Op = OpRISCV64MOVBreg
		return true
	case OpSlicemask:
		return rewriteValueRISCV64_OpSlicemask(v)
	case OpSqrt:
		v.Op = OpRISCV64FSQRTD
		return true
	case OpSqrt32:
		v.Op = OpRISCV64FSQRTS
		return true
	case OpStaticCall:
		v.Op = OpRISCV64CALLstatic
		return true
	case OpStore:
		return rewriteValueRISCV64_OpStore(v)
	case OpSub16:
		v.Op = OpRISCV64SUB