func.func @main(%arg0: tensor<2xf32>) -> tensor<2xf32> {
  %0 = "tfl.custom"(%arg0) {custom_code = "stablehlo.rsqrt", custom_option = #tfl<const_bytes : "0x00000100002401">} : (tensor<2xf32>) -> tensor<2xf32>
  func.return %0 : tensor<2xf32>
}
}

// CHECK:       module
// CHECK-NEXT:  func @main(%arg0: tensor<2xf32>) -> tensor<2xf32> {
// CHECK-NEXT:  %0 = stablehlo.rsqrt %arg0 : tensor<2xf32>
// CHECK-NEXT:  return %0 : tensor<2xf32>
// CHECK-NEXT:  }

// tfl.log / tfl.logistic / tfl.max_pool_2d / tfl.mirror_pad / tfl.maximum /
// tfl.custom / tfl.mean / tfl.minimum / tfl.pad / tfl.pow / tfl.prelu /
// tfl.relu / tfl.relu6 / tfl.rsqrt / tfl.sin / tfl.slice / tfl.softmax /
// tfl.space_to_depth / tfl.sqrt / tfl.square / tfl.squared_difference /
// tfl.strided_slice / tfl.tanh / tfl.transpose / tfl.transpose_conv
class GpuBasicSupportedOpNoCost : public TargetHardwareOperation {

 Const 5
 GreaterEqual 1
 MatMul 1
 Mul 5
 Select 2
 Sigmoid 3
 Snapshot 1
 Split 1
 Tanh 2
cluster 29 size 9
 Add 1
 Mul 2
 RealDiv 2
 Sqrt 2
 Sub 2
cluster 30 size 9
 Add 1
 Mul 2
 RealDiv 2
 Sqrt 2
 Sub 2
cluster 31 size 4
 Mul 3
 UnsortedSegmentSum 1
cluster 32 size 4
 Mul 3

// and "bias" are found in src/math/bits.go

// Sqrt returns the square root of x.
//
// Special cases are:
//
//	Sqrt(+Inf) = +Inf
//	Sqrt(±0) = ±0
//	Sqrt(x < 0) = NaN
//	Sqrt(NaN) = NaN
func Sqrt(x float64) float64 {
	return sqrt(x)
}

// Note: On systems where Sqrt is a single instruction, the compiler
// may turn a direct call into a direct use of that instruction instead.

		}
		if real(x) < 0 {
			return complex(0, math.Copysign(math.Sqrt(-real(x)), imag(x)))
		}
		return complex(math.Sqrt(real(x)), imag(x))
	} else if math.IsInf(imag(x), 0) {
		return complex(math.Inf(1.0), imag(x))
	}
	if real(x) == 0 {
		if imag(x) < 0 {
			r := math.Sqrt(-0.5 * imag(x))
			return complex(r, -r)
		}
		r := math.Sqrt(0.5 * imag(x))
		return complex(r, r)
	}
	a := real(x)
	b := imag(x)

import (
	"math"
	"sync"
)

var threeOnce struct {
	sync.Once
	v *Float
}

func three() *Float {
	threeOnce.Do(func() {
		threeOnce.v = NewFloat(3.0)
	})
	return threeOnce.v
}

// Sqrt sets z to the rounded square root of x, and returns it.
//
// If z's precision is 0, it is changed to x's precision before the
// operation. Rounding is performed according to z's precision and

	FDIVD   F1, F0       // F0=q(=q/p), F1=p
	FMULD   F0, F0       // F0=q*q, F1=p
	FLD1                 // F0=1, F1=q*q, F2=p
	FADDDP  F0, F1       // F0=1+q*q, F1=p
	FSQRT                // F0=sqrt(1+q*q), F1=p
	FMULDP  F0, F1       // F0=p*sqrt(1+q*q)
	FMOVDP  F0, ret+16(FP)
	RET
	FMOVDP  F0, F1       // F0=0
	FMOVDP  F0, ret+16(FP)
	RET
not_finite:
// test bits for -Inf or +Inf

		return complex(math.Copysign(math.Pi/2, re), math.Copysign(re, im))
	}
	ct := complex(-imag(x), real(x)) // i * x
	xx := x * x
	x1 := complex(1-real(xx), -imag(xx)) // 1 - x*x
	x2 := Sqrt(x1)                       // x2 = sqrt(1 - x*x)
	w := Log(ct + x2)
	return complex(imag(w), -real(w)) // -i * w
}

// Asinh returns the inverse hyperbolic sine of x.
func Asinh(x complex128) complex128 {

	sink64[4] = math.RoundToEven(x)
}

func sqrt(x float64) float64 {
	// amd64:"SQRTSD"
	// 386/sse2:"SQRTSD" 386/softfloat:-"SQRTD"
	// arm64:"FSQRTD"
	// arm/7:"SQRTD"
	// mips/hardfloat:"SQRTD" mips/softfloat:-"SQRTD"
	// mips64/hardfloat:"SQRTD" mips64/softfloat:-"SQRTD"
	// wasm:"F64Sqrt"
	// ppc64x:"FSQRT"
	// riscv64: "FSQRTD"
	return math.Sqrt(x)
}

func sqrt32(x float32) float32 {
	// amd64:"SQRTSS"

	name string
	in   float64 // used for error messages, not an input
	got  float64
	want float64
}{
	{"sqrt0", 0, math.Sqrt(0), 0},
	{"sqrt1", 1, math.Sqrt(1), 1},
	{"sqrt2", 2, math.Sqrt(2), math.Sqrt2},
	{"sqrt4", 4, math.Sqrt(4), 2},
	{"sqrt100", 100, math.Sqrt(100), 10},
	{"sqrt101", 101, math.Sqrt(101), 10.04987562112089},
}

var nanTests = [...]struct {
	name string
	in   float64 // used for error messages, not an input