// CHECK: tac.cost = 1.000000e+03
  %0 = "tfl.add"(%arg0, %arg1) {fused_activation_function = "RELU", tac.device = "CPU"} : (tensor<10x10x10xf32>, tensor<10xf32>) -> tensor<10x10x10xf32>
  // CHECK: tac.cost = 1.000000e+03

}

func.func @testAdd(%arg0: tensor<10x10x10xf32>, %arg1: tensor<10x10x10xf32>) -> tensor<10x10x10xf32> {
  // CHECK: _arithmetic_count = 1000 : i64
  %0 = "tfl.add"(%arg0, %arg1) {fused_activation_function = "NONE"} : (tensor<10x10x10xf32>, tensor<10x10x10xf32>) -> tensor<10x10x10xf32>
  func.return %0 : tensor<10x10x10xf32>
}

func.func @testAddBroadcast(%arg0: tensor<10x10x10xf32>, %arg1: tensor<10xf32>) -> tensor<10x10x10xf32> {

  // CHECK-NEXT: "tf.AssignVariableOp"([[ARG1]], [[PARALLEL_EXECUTE]]#0)
  "tf.AssignVariableOp"(%arg1, %0#0) : (tensor<!tf_type.resource>, tensor<i1>) -> ()
  // CHECK-NEXT: return [[PARALLEL_EXECUTE]]#0
  func.return %0#0 : tensor<i1>
}

// CHECK-LABEL: func @not_assign_var
// CHECK-SAME:  ({{.+}}: tensor<i1>, [[ARG1:%.+]]: tensor<!tf_type.resource>)
func.func @not_assign_var(%arg0: tensor<i1>, %arg1: tensor<!tf_type.resource>) {

// CHECK:}

func.func @add(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) -> tensor<1xf32> {
  %0 = "vhlo.add_v1"(%arg0, %arg1) : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>
  func.return %0 : tensor<1xf32>
}

// CHECK:func.func private @add(%arg0: tensor<1xf32>, %arg1: tensor<1xf32>) -> tensor<1xf32> {
// CHECK: %0 = "vhlo.add_v1"(%arg0, %arg1) : (tensor<1xf32>, tensor<1xf32>) -> tensor<1xf32>

		{name: "FNMSUBS", argLength: 3, reg: fp31, asm: "FNMSUBS"}, // -arg0 + (arg1 * arg2)
		{name: "FNMSUBD", argLength: 3, reg: fp31, asm: "FNMSUBD"}, // -arg0 + (arg1 * arg2)
		{name: "MADD", argLength: 3, reg: gp31, asm: "MADD"},       // +arg0 + (arg1 * arg2)
		{name: "MADDW", argLength: 3, reg: gp31, asm: "MADDW"},     // +arg0 + (arg1 * arg2), 32-bit
		{name: "MSUB", argLength: 3, reg: gp31, asm: "MSUB"},       // +arg0 - (arg1 * arg2)

  // CHECK: %[[MUL:.*]] = "tf.Mul"
  %0 = "tf.Mul"(%arg0, %arg1) : (tensor<i32>, tensor<i32>) -> tensor<i32>
  // CHECK: %[[RESULT:.*]]:4 = "tf_device.cluster"
  %1:4 = "tf_device.cluster"() ({
    // CHECK: %[[ADD:.*]] = "tf.AddV2"
    %2 = "tf.AddV2"(%arg0, %arg1) : (tensor<i32>, tensor<i32>) -> tensor<i32>

CHECK:  %[[LSTM:[a-z0-9]*]] = "tfl.unidirectional_sequence_lstm"(%arg2, %arg0, %arg0, %arg0, %arg0, %arg0, %arg0, %arg0, %arg0, %arg1, %arg1, %arg1, %arg1, %arg1, %arg1, %arg1, %arg0, %arg1, %[[CST_0]], %[[CST_1]], %arg0, %arg0, %arg0, %arg0) <{fused_activation_function = "NONE", time_major = true}> : (tensor<4x4x4xf32>, tensor<4x4xf32>, tensor<4x4xf32>, tensor<4x4xf32>, tensor<4x4xf32>, tensor<4x4xf32>, tensor<4x4xf32>, tensor<4x4xf32>, tensor<4x4xf32>, tensor<4xf32>, tensor<4xf32>, tensor<4xf32>,...

		{name: "DIVU", argLength: 2, reg: gp2hilo, asm: "DIVU", typ: "(UInt32,UInt32)"},                                                                  // arg0 / arg1, signed, results hi=arg0%arg1,lo=arg0/arg1

		{name: "ADDF", argLength: 2, reg: fp21, asm: "ADDF", commutative: true}, // arg0 + arg1

  %arg0: tensor<1x28x28x64xf32>,
  %arg1: tensor<64xf32>
) -> tensor<1x28x28x64xf32> {

  // CHECK: "tf.FusedBatchNormV3"
  // CHECK-SAME: (%[[X_TRANSPOSE:[0-9]*]], %arg1, %arg1, %arg1, %arg1)
  // CHECK-SAME: data_format = "NCHW"
  %y, %batch_mean, %batch_var, %reserve_1, %reserve_2, %reserve_3
    = "tf.FusedBatchNormV3"(%arg0, %arg1, %arg1, %arg1, %arg1)
       {
         data_format = "NHWC",

// CHECK-LABEL: bias_add
func.func @bias_add(%arg0: tensor<1x10x10x32xf32>, %arg1: tensor<32xf32>) -> tensor<1x10x10x32xf32> {
  %0 = "tf.BiasAdd"(%arg0, %arg1) {T = "tfdtype$DT_FLOAT", data_format = "NHWC"} : (tensor<1x10x10x32xf32>, tensor<32xf32>) -> tensor<1x10x10x32xf32>
  func.return %0 : tensor<1x10x10x32xf32>