// CHECK: {{%.*}} = tfrt_fallback_async.executeop {{.*}} device("/device:CPU:0") "tf.MatMul"
  // CHECK-SAME: {T = f32, transpose_a = false, transpose_b = false}

  // CHECK-NEXT: [[ch1:%.*]], [[var:%.*]] = tfrt_fallback_async.executeop.seq([[in_chain]]) {{.*}} "tf.ReadVariableOp"([[arg1]])
  // CHECK-NEXT: [[r0:%.*]] = tfrt_fallback_async.executeop {{.*}} "tf.MatMul"([[arg0]], [[var]])
  %2 = "tf.MatMul"(%arg0, %1) {T = f32, _output_shapes = ["tfshape$dim { size: 3 } dim { size: 3 }"], device = "/device:CPU:0", transpose_a = false, transpose_b = false} : (tensor<3x1xf32>, tensor<1x3xf32>) -> tensor<3x3xf32>

 BiasAdd 1
 BiasAddGrad 1
 Cast 3
 Const 357
 Conv2D 53
 Conv2DBackpropFilter 53
 Conv2DBackpropInput 52
 DivNoNan 1
 Equal 1
 FusedBatchNorm 53
 FusedBatchNormGrad 53
 Identity 2
 MatMul 3
 MaxPool 1
 MaxPoolGrad 1
 Mean 1
 Mul 164
 Pad 1
 ReadVariableOp 646
 Relu 49
 ReluGrad 49
 Reshape 2
 ResourceApplyKerasMomentum 161
 ShapeN 50
 Softmax 1

==============================================================================*/
#ifndef TENSORFLOW_C_EAGER_C_API_REMOTE_TEST_UTIL_H_
#define TENSORFLOW_C_EAGER_C_API_REMOTE_TEST_UTIL_H_

// Run a function containing a MatMul op and check its output.
// If heavy_load_on_streaming_rpc is true, send some rpc requests before the one
// which creates a remote input, to simulate a scenario that the remote input

func.func @device_test(%arg0: tensor<3x1xf32>) -> (tensor<3x3xf32>) {

  // CHECK: device = "gpu"
  %0 = "tf.Const"() {value = dense<[[1.0, 2.0, 3.0]]> : tensor<1x3xf32>} : () -> tensor<1x3xf32>
  // CHECK: device = "gpu"
  %1 = "tf.MatMul"(%arg0, %0) {T = f32, _output_shapes = ["tfshape$dim { size: 3 } dim { size: 3 }"], device = "", transpose_a = false, transpose_b = false} : (tensor<3x1xf32>, tensor<1x3xf32>) -> tensor<3x3xf32>
  // CHECK: device = "cpu"

// CHECK-NOT: func private @internal_quantize_i8
// CHECK-NOT: func private @internal_matmul_fn
// CHECK: func private @quantized_matmul_fn
// CHECK-SAME: tf_quant.quantized_ops = ["MatMul"]
// CHECK: func private @quantized_conv2d_fn
// CHECK-SAME: tf_quant.quantized_ops = ["Conv2D"]
// CHECK: func private @quantized_depthwise_conv2d_fn
// CHECK-SAME: tf_quant.quantized_ops = ["DepthwiseConv2D"]

 ArgMax 1
 AssignAddVariableOp 1
 BiasAdd 1
 BiasAddGrad 1
 Cast 115
 Const 407
 Conv2D 53
 Conv2DBackpropFilter 53
 Conv2DBackpropInput 52
 Equal 1
 FusedBatchNormGradV2 53
 FusedBatchNormV2 53
 MatMul 3
 MaxPool 1
 MaxPoolGrad 1
 Mean 1
 Mul 218
 Pad 2
 ReadVariableOp 538
 Relu 49
 ReluGrad 49
 Reshape 2
 ResourceApplyKerasMomentum 161
 Slice 1
 Softmax 1

  // CHECK: %[[TRANSPOSE:.*]] = "tf.Transpose"(%[[DEQUANT]], %[[CST]]) : (tensor<3x4xf32>, tensor<?xi32>) -> tensor<*xf32>
  // CHECK: %[[MATMUL:.*]] = "tf.MatMul"(%arg0, %[[TRANSPOSE]]) <{grad_a = false, grad_b = false, transpose_a = false, transpose_b = true}> : (tensor<2x3xf32>, tensor<*xf32>) -> tensor<2x4xf32>
  // CHECK: return %[[MATMUL]] : tensor<2x4xf32>
}

  %x = "tf.TensorArrayReadV3"(%handle_0, %index, %flow_0) {device = "/job:localhost/replica:0/task:0/device:CPU:0"} : (tensor<2x!tf_type.resource<tensor<?x100xf32>>>, tensor<i32>, tensor<f32>) -> tensor<?x100xf32>
  %y = "tf.MatMul"(%x, %cst) {device = "/job:localhost/replica:0/task:0/device:CPU:0"} : (tensor<?x100xf32>, tensor<100x512xf32>) -> (tensor<?x512xf32>)

namespace tensorflow {
namespace ops {

//
Status Neg(AbstractContext* ctx, AbstractTensorHandle* const x, AbstractTensorHandle** y, const char* name = nullptr, const char* raw_device_name = nullptr);

//