std::vector<TFE_TensorHandle*> handles_task2;
  for (auto* h_task0 : handles_task0) {
    handles_task2.push_back(
        TFE_TensorHandleCopyToDevice(h_task0, ctx, task2_name, status));
    ASSERT_EQ(TF_GetCode(status), TF_OK) << TF_Message(status);
  }

  auto* h1_task2 =
      TFE_TensorHandleCopyToDevice(h1_task0, ctx, task2_name, status);
  ASSERT_EQ(TF_GetCode(status), TF_OK) << TF_Message(status);

  const char remote_device_name[] =
      "/job:localhost/replica:0/task:1/device:CPU:0";
  auto* h0_task1 =
      TFE_TensorHandleCopyToDevice(h0_task0, ctx, remote_device_name, status);
  ASSERT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);
  auto* h1_task1 =
      TFE_TensorHandleCopyToDevice(h1_task0, ctx, remote_device_name, status);
  ASSERT_EQ(TF_OK, TF_GetCode(status)) << TF_Message(status);

                                      TF_Status* status, void* device_info) {
  LoggingDevice* dev = reinterpret_cast<LoggingDevice*>(device_info);
  TFE_TensorHandle* t = TFE_TensorHandleCopyToDevice(
      tensor, context, dev->underlying_device.c_str(), status);
  if (TF_GetCode(status) != TF_OK) return nullptr;
  auto dst = std::make_unique<LoggedTensor>(t);
  *(dev->arrived_flag) = true;