func BenchmarkAnd8(b *testing.B) {
	var x [512]uint8 // give byte its own cache line
	sink = &x
	for i := 0; i < b.N; i++ {
		atomic.And8(&x[255], uint8(i))
	}
}

func BenchmarkAnd(b *testing.B) {
	var x [128]uint32 // give x its own cache line
	sink = &x
	for i := 0; i < b.N; i++ {
		atomic.And(&x[63], uint32(i))
	}
}

func BenchmarkAnd8Parallel(b *testing.B) {
	var x [512]uint8 // give byte its own cache line
	sink = &x

        attr_map = ""
      } : (tensor<*xf32>, tensor<*x!tf_type.qint8>, tensor<*xf32>, tensor<*xi32>) -> tensor<*xf32>


    func.return %out : tensor<*xf32>
  }

  // Currently only 4-d case is supported
  func.func @quantized_depthwise_conv2d_fn(
                         %input : tensor<*xf32>, %weight : tensor<*x!tf_type.qint8>,

	Dev_idx          uint32
	Target_idx       uint32
	Nfc_protocol     uint32
	Dsap             uint8
	Ssap             uint8
	Service_name     [63]uint8
	Service_name_len uint64
}

type RawSockaddr struct {
	Family uint16
	Data   [14]int8
}

type RawSockaddrAny struct {
	Addr RawSockaddr
	Pad  [96]int8
}

type Iovec struct {
	Base *byte
	Len  uint64
}

type Msghdr struct {

				a := (m - (sa * ma / m)) * 0x101

				d[0] = uint8((dr*a + sr*ma) / m >> 8)
				d[1] = uint8((dg*a + sg*ma) / m >> 8)
				d[2] = uint8((db*a + sb*ma) / m >> 8)
				d[3] = uint8((da*a + sa*ma) / m >> 8)

			} else {
				d[0] = uint8(sr * ma / m >> 8)
				d[1] = uint8(sg * ma / m >> 8)
				d[2] = uint8(sb * ma / m >> 8)
				d[3] = uint8(sa * ma / m >> 8)
			}
		}
		i0 += dy * dst.Stride
	}
}

    // CHECK-NEXT: %[[UDQ:.*]] = "tf.UniformDequantize"(%[[UQ]], %[[CST_0]], %[[CST_1]]) <{quantization_axis = -1 : i64, quantization_max_val = 127 : i64, quantization_min_val = -128 : i64}> : (tensor<1x!tf_type.qint8>, tensor<f32>, tensor<i32>) -> tensor<1xf32>
    // CHECK-NEXT: return %[[UDQ]] : tensor<1xf32>

    func.func @main(%arg0: tensor<3x3x!tf_type.qint8>, %arg1: tensor<3x3x!tf_type.qint8>) -> tensor<6x3x!tf_type.qint8> {
      %axis = "tf.Const"() { value = dense<0> : tensor<i64> } : () -> tensor<i64>
      %1 = "tf.ConcatV2"(%arg0, %arg1, %axis) : (tensor<3x3x!tf_type.qint8>, tensor<3x3x!tf_type.qint8>, tensor<i64>) -> tensor<6x3x!tf_type.qint8>
      func.return %1 : tensor<6x3x!tf_type.qint8>
    }
  })";
  CreateModule(kMlirModuleStr);

	if (uint64(uint16(x)) == x || uint64(uint16(^x)) == ^x) && (uint64(uint16(y)) == y || uint64(uint16(^y)) == ^y) {
		checkuint16(uint16(x), uint16(y))
	}
	if (uint64(uint8(x)) == x || uint64(uint8(^x)) == ^x) && (uint64(uint8(y)) == y || uint64(uint8(^y)) == ^y) {
		checkuint8(uint8(x), uint8(y))
	}
	
	
	sx := int64(x)
	sy := int64(y)
	checkint64(sx, sy)

	Pno     int32
	Devname [64]uint8
	Volname [64]uint8
	_       [4]byte
}

const (
	BLKPG = 0x1269
)

type CryptoUserAlg struct {
	Name        [64]int8
	Driver_name [64]int8
	Module_name [64]int8
	Type        uint32
	Mask        uint32
	Refcnt      uint32
	Flags       uint32
}

type CryptoStatAEAD struct {
	Type         [64]int8
	Encrypt_cnt  uint64

	Dev_idx          uint32
	Target_idx       uint32
	Nfc_protocol     uint32
	Dsap             uint8
	Ssap             uint8
	Service_name     [63]uint8
	Service_name_len uint32
}

type RawSockaddr struct {
	Family uint16
	Data   [14]int8
}

type RawSockaddrAny struct {
	Addr RawSockaddr
	Pad  [96]int8
}

type Iovec struct {
	Base *byte
	Len  uint32
}

type Msghdr struct {

					b >>= 16
				} else {
					b = ^(b >> 31)
				}

				// use a temp slice to hint to the compiler that a single bounds check suffices
				rgba := dpix[x : x+4 : len(dpix)]
				rgba[0] = uint8(r)
				rgba[1] = uint8(g)
				rgba[2] = uint8(b)
				rgba[3] = 255
			}
		}

	case image.YCbCrSubsampleRatio422:
		for y, sy := y0, sp.Y; y != y1; y, sy = y+1, sy+1 {
			dpix := dst.Pix[y*dst.Stride:]