bool is_async = false, int in_flight_nodes_limit = 0);

  ~ParallelDevice();

  // Helper to copy a tensor handle from another device once for each component
  // of the ParallelDevice.
  //
  // Sets a bad status and returns a nullptr if `tensor` is already on the
  // ParallelDevice, or if the individual copies fail.

//
// Implicit copies onto `device_name` are allowed, replicating the value once
// per device in `underlying_devices`. Implicit copies off of the device throw
// an error.
//
// All component tensors must have the same dtype. Currently they must also have
// the same shape, although this requirement may be relaxed in the future.
//
// `device_name` must not name an existing physical or custom device (see

  //
  // -1 indicates an unknown axis length; this is unreachable for most standard
  // ImmediateExecutionTensorHandles, but comes up for example when computing
  // the shape of a parallel tensor with component shapes differing across
  // devices.
  virtual Status Dim(int dim_index, int64_t* dim) const = 0;

  // Returns the device which created the handle.
  virtual const char* DeviceName(Status* status) const = 0;