Python prefetching_ops.copy_to_device函数代码示例

  def testInsideFunction(self):
    if test_util.is_gpu_available():
      self.skipTest(
          "b/123899495: Colocation errors for critical sections in map on GPU")
    cs = critical_section_ops.CriticalSection()
    with ops.device("/gpu:0" if test_util.is_gpu_available() else "/cpu:0"):
      v = resource_variable_ops.ResourceVariable(1)
    def fn():
      return v.read_value()

    # map() creates a TensorFlow function.
    ds = dataset_ops.Dataset.range(1)
    if test_util.is_gpu_available():
      ds = (ds.apply(prefetching_ops.copy_to_device("/gpu:0"))
            .apply(prefetching_ops.map_on_gpu(lambda _: cs.execute(fn))))
    else:
      ds = ds.map(lambda _: cs.execute(fn))

    def get_first():
      if context.executing_eagerly():
        return self.evaluate(ds.make_one_shot_iterator().get_next())
      itr = ds.make_initializable_iterator()
      self.evaluate([v.initializer, itr.initializer])
      return self.evaluate(itr.get_next())

    self.assertEqual(1, get_first())

  def testCopyToDeviceWithReInitAndPrefetch(self):
    host_dataset = dataset_ops.Dataset.range(10)
    device_dataset = host_dataset.apply(
        prefetching_ops.copy_to_device("/cpu:1")).prefetch(1)

    with ops.device("/cpu:1"):
      iterator = dataset_ops.make_initializable_iterator(device_dataset)
      next_element = iterator.get_next()

    self.assertEqual(host_dataset.output_types, device_dataset.output_types)
    self.assertEqual(host_dataset.output_types, iterator.output_types)
    self.assertEqual(host_dataset.output_shapes, device_dataset.output_shapes)
    self.assertEqual(host_dataset.output_shapes, iterator.output_shapes)
    self.assertEqual(host_dataset.output_classes, device_dataset.output_classes)
    self.assertEqual(host_dataset.output_classes, iterator.output_classes)

    self.assertEqual(dtypes.int64, next_element.dtype)
    self.assertEqual([], next_element.shape)

    worker_config = config_pb2.ConfigProto(device_count={"CPU": 2})
    with self.test_session(config=worker_config):
      self.evaluate(iterator.initializer)
      for i in range(5):
        self.assertEqual(i, self.evaluate(next_element))
      self.evaluate(iterator.initializer)
      for i in range(10):
        self.assertEqual(i, self.evaluate(next_element))
      with self.assertRaises(errors.OutOfRangeError):
        self.evaluate(next_element)

  def testCopyToSameDevice(self):
    host_dataset = dataset_ops.Dataset.range(10)
    device_dataset = host_dataset.apply(
        prefetching_ops.copy_to_device("/cpu:0"))

    with ops.device("/cpu:0"):
      iterator = device_dataset.make_one_shot_iterator()
      next_element = iterator.get_next()

    self.assertEqual(host_dataset.output_types, device_dataset.output_types)
    self.assertEqual(host_dataset.output_types, iterator.output_types)
    self.assertEqual(host_dataset.output_shapes, device_dataset.output_shapes)
    self.assertEqual(host_dataset.output_shapes, iterator.output_shapes)
    self.assertEqual(host_dataset.output_classes, device_dataset.output_classes)
    self.assertEqual(host_dataset.output_classes, iterator.output_classes)

    self.assertEqual(dtypes.int64, next_element.dtype)
    self.assertEqual([], next_element.shape)

    worker_config = config_pb2.ConfigProto(device_count={"CPU": 2})
    with self.test_session(config=worker_config) as sess:
      for i in range(10):
        self.assertEqual(i, sess.run(next_element))
      with self.assertRaises(errors.OutOfRangeError):
        sess.run(next_element)

  def testCopyToDeviceGpuWithMap(self):
    if not test_util.is_gpu_available():
      self.skipTest("No GPU available")

    def generator():
      for i in range(10):
        yield i, float(i), str(i)

    host_dataset = dataset_ops.Dataset.from_generator(
        generator, output_types=(dtypes.int32, dtypes.float32, dtypes.string))
    device_dataset = host_dataset.apply(
        prefetching_ops.copy_to_device("/gpu:0"))

    def gpu_map_func(x, y, z):
      return math_ops.square(x), math_ops.square(y), z

    device_dataset = device_dataset.apply(
        prefetching_ops.map_on_gpu(gpu_map_func))
    options = dataset_ops.Options()
    options.experimental_autotune = False
    device_dataset = device_dataset.with_options(options)

    with ops.device("/gpu:0"):
      iterator = device_dataset.make_initializable_iterator()
      next_element = iterator.get_next()

    with self.cached_session() as sess:
      sess.run(iterator.initializer)
      for i in range(10):
        x, y, z = sess.run(next_element)
        self.assertEqual(i**2, x)
        self.assertEqual(float(i**2), y)
        self.assertEqual(util_compat.as_bytes(str(i)), z)
      with self.assertRaises(errors.OutOfRangeError):
        sess.run(next_element)

  def testIteratorGetNextAsOptionalOnGPU(self):
    if not test_util.is_gpu_available():
      self.skipTest("No GPU available")

    host_dataset = dataset_ops.Dataset.range(3)
    device_dataset = host_dataset.apply(
        prefetching_ops.copy_to_device("/gpu:0"))
    with ops.device("/gpu:0"):
      iterator = device_dataset.make_initializable_iterator()
      next_elem = iterator_ops.get_next_as_optional(iterator)
      elem_has_value_t = next_elem.has_value()
      elem_value_t = next_elem.get_value()

    with self.cached_session() as sess:
      # Before initializing the iterator, evaluating the optional fails with
      # a FailedPreconditionError.
      with self.assertRaises(errors.FailedPreconditionError):
        sess.run(elem_has_value_t)
      with self.assertRaises(errors.FailedPreconditionError):
        sess.run(elem_value_t)

      # For each element of the dataset, assert that the optional evaluates to
      # the expected value.
      sess.run(iterator.initializer)
      for i in range(3):
        elem_has_value, elem_value = sess.run([elem_has_value_t, elem_value_t])
        self.assertTrue(elem_has_value)
        self.assertEqual(i, elem_value)

      # After exhausting the iterator, `next_elem.has_value()` will evaluate to
      # false, and attempting to get the value will fail.
      for _ in range(2):
        self.assertFalse(sess.run(elem_has_value_t))
        with self.assertRaises(errors.InvalidArgumentError):
          sess.run(elem_value_t)

  def testCopySparseTensorsToDeviceWithPrefetch(self):

    def make_tensor(i):
      return sparse_tensor.SparseTensorValue(
          indices=[[0, 0]], values=(i * [1]), dense_shape=[2, 2])

    host_dataset = dataset_ops.Dataset.range(10).map(make_tensor)

    device_dataset = host_dataset.apply(
        prefetching_ops.copy_to_device("/cpu:1")).prefetch(1)

    with ops.device("/cpu:1"):
      iterator = device_dataset.make_one_shot_iterator()
      next_element = iterator.get_next()

    self.assertEqual(host_dataset.output_types, device_dataset.output_types)
    self.assertEqual(host_dataset.output_types, iterator.output_types)
    self.assertEqual(host_dataset.output_shapes, device_dataset.output_shapes)
    self.assertEqual(host_dataset.output_shapes, iterator.output_shapes)
    self.assertEqual(host_dataset.output_classes, device_dataset.output_classes)
    self.assertEqual(host_dataset.output_classes, iterator.output_classes)

    self.assertEqual(dtypes.int64, next_element.dtype)

    worker_config = config_pb2.ConfigProto(device_count={"CPU": 2})
    with self.test_session(config=worker_config) as sess:
      for i in range(10):
        actual = sess.run(next_element)
        self.assertAllEqual([i], actual.values)
        self.assertAllEqual([[0, 0]], actual.indices)
        self.assertAllEqual([2, 2], actual.dense_shape)
      with self.assertRaises(errors.OutOfRangeError):
        sess.run(next_element)

  def testCopyDictToDeviceWithPrefetch(self):
    host_dataset = dataset_ops.Dataset.range(10).map(lambda x: {"a": x})
    device_dataset = host_dataset.apply(
        prefetching_ops.copy_to_device("/cpu:1")).prefetch(1)

    with ops.device("/cpu:1"):
      iterator = dataset_ops.make_one_shot_iterator(device_dataset)
      next_element = iterator.get_next()

    self.assertEqual(host_dataset.output_types, device_dataset.output_types)
    self.assertEqual(host_dataset.output_types, iterator.output_types)
    self.assertEqual(host_dataset.output_shapes, device_dataset.output_shapes)
    self.assertEqual(host_dataset.output_shapes, iterator.output_shapes)
    self.assertEqual(host_dataset.output_classes, device_dataset.output_classes)
    self.assertEqual(host_dataset.output_classes, iterator.output_classes)

    self.assertEqual(dtypes.int64, next_element["a"].dtype)
    self.assertEqual([], next_element["a"].shape)

    worker_config = config_pb2.ConfigProto(device_count={"CPU": 2})
    with self.test_session(config=worker_config):
      for i in range(10):
        self.assertEqual({"a": i}, self.evaluate(next_element))
      with self.assertRaises(errors.OutOfRangeError):
        self.evaluate(next_element)

  def __init__(self, dataset):
    """Creates a new iterator over the given dataset.

    For example:
    ```python
    dataset = tf.data.Dataset.range(4)
    for x in Iterator(dataset):
      print(x)
    ```

    Tensors produced will be placed on the device on which this iterator object
    was created.

    Args:
      dataset: A `tf.data.Dataset` object.

    Raises:
      TypeError: If `dataset` is an unsupported type.
      RuntimeError: When invoked without eager execution enabled.
    """
    if not context.context().device_spec.device_type:
      is_remote_device = False
    else:
      is_remote_device = context.context().device_spec.device_type != "CPU"
    if is_remote_device:
      with ops.device(None):
        # Let the placer figure out where to place the various functions etc.
        # created by the CopyToDeviceDataset.
        dataset = dataset.apply(prefetching_ops.copy_to_device(
            context.context().device_name))
        dataset = dataset.prefetch(1)
    super(Iterator, self).__init__(dataset)

def copy_to_device(target_device, source_device="/cpu:0"):
  """A transformation that copies dataset elements to the given `target_device`.

  Args:
    target_device: The name of a device to which elements will be copied.
    source_device: The original device on which `input_dataset` will be placed.

  Returns:
    A `Dataset` transformation function, which can be passed to
    `tf.data.Dataset.apply`.
  """
  return prefetching_ops.copy_to_device(target_device, source_device)

  def testCopyToDevicePingPongCPUGPU(self):
    if not test_util.is_gpu_available():
      self.skipTest("No GPU available")

    with compat.forward_compatibility_horizon(2018, 8, 4):
      host_dataset = dataset_ops.Dataset.range(10)
      device_dataset = host_dataset.apply(
          prefetching_ops.copy_to_device("/gpu:0", source_device="/cpu:0"))
      back_to_cpu_dataset = device_dataset.apply(
          prefetching_ops.copy_to_device("/cpu:0", source_device="/gpu:0"))

      with ops.device("/cpu:0"):
        iterator = back_to_cpu_dataset.make_initializable_iterator()
        next_element = iterator.get_next()

      with self.cached_session() as sess:
        sess.run(iterator.initializer)
        for i in range(10):
          self.assertEqual(i, sess.run(next_element))
        with self.assertRaises(errors.OutOfRangeError):
          sess.run(next_element)

  def testCopyToDeviceGpuStringsAndPrefetch(self):
    if not test_util.is_gpu_available():
      self.skipTest("No GPU available")

    host_dataset = dataset_ops.Dataset.from_tensors(["a", "b", "c"])
    device_dataset = host_dataset.apply(
        prefetching_ops.copy_to_device("/gpu:0"))

    with ops.device("/gpu:0"):
      iterator = device_dataset.make_initializable_iterator()
      next_element = iterator.get_next()

    with self.cached_session() as sess:
      sess.run(iterator.initializer)
      self.assertAllEqual([b"a", b"b", b"c"], sess.run(next_element))
      with self.assertRaises(errors.OutOfRangeError):
        sess.run(next_element)

  def testCopyToDeviceGpuWithPrefetch(self):
    if not test_util.is_gpu_available():
      self.skipTest("No GPU available")

    host_dataset = dataset_ops.Dataset.range(10)
    device_dataset = host_dataset.apply(
        prefetching_ops.copy_to_device("/gpu:0")).prefetch(1)

    with ops.device("/gpu:0"):
      iterator = device_dataset.make_initializable_iterator()
      next_element = iterator.get_next()

    with self.cached_session() as sess:
      sess.run(iterator.initializer)
      for i in range(10):
        self.assertEqual(i, sess.run(next_element))
      with self.assertRaises(errors.OutOfRangeError):
        sess.run(next_element)

  def testCopyToDeviceGpuStrings(self):
    if not test_util.is_gpu_available():
      self.skipTest("No GPU available")

    host_dataset = dataset_ops.Dataset.from_tensors(["a", "b", "c"])
    device_dataset = host_dataset.apply(
        prefetching_ops.copy_to_device("/gpu:0"))

    with ops.device("/gpu:0"):
      iterator = dataset_ops.make_initializable_iterator(device_dataset)
      next_element = iterator.get_next()

    with self.cached_session(
        config=config_pb2.ConfigProto(allow_soft_placement=False)):
      self.evaluate(iterator.initializer)
      self.assertAllEqual([b"a", b"b", b"c"], self.evaluate(next_element))
      with self.assertRaises(errors.OutOfRangeError):
        self.evaluate(next_element)

  def testCopyToDeviceGpuWithPrefetch(self):
    if not test_util.is_gpu_available():
      self.skipTest("No GPU available")

    host_dataset = dataset_ops.Dataset.range(10)
    device_dataset = host_dataset.apply(
        prefetching_ops.copy_to_device("/gpu:0")).prefetch(1)

    with ops.device("/gpu:0"):
      iterator = dataset_ops.make_initializable_iterator(device_dataset)
      next_element = iterator.get_next()

    with self.cached_session(
        config=config_pb2.ConfigProto(allow_soft_placement=False)):
      self.evaluate(iterator.initializer)
      for i in range(10):
        self.assertEqual(i, self.evaluate(next_element))
      with self.assertRaises(errors.OutOfRangeError):
        self.evaluate(next_element)

  def __init__(self, dataset):
    """Creates a new iterator over the given dataset.

    For example:
    ```python
    dataset = tf.data.Dataset.range(4)
    for x in Iterator(dataset):
      print(x)
    ```

    Tensors produced will be placed on the device on which this iterator object
    was created.

    Args:
      dataset: A `tf.data.Dataset` object.

    Raises:
      TypeError: If `dataset` is an unsupported type.
      RuntimeError: When invoked without eager execution enabled.
    """
    # pylint: disable=protected-access
    if (isinstance(dataset, prefetching_ops._PrefetchToDeviceDataset)
        or (isinstance(dataset, dataset_ops.DatasetV1Adapter)
            and isinstance(
                dataset._dataset, prefetching_ops._PrefetchToDeviceDataset))):
      raise TypeError(
          "`tf.data.experimental.prefetch_to_device()` is not compatible with "
          "`tf.contrib.eager.Iterator`. Use `for ... in dataset:` to iterate "
          "over the dataset instead.")
    # pylint: enable=protected-access

    if not context.context().device_spec.device_type:
      is_remote_device = False
    else:
      is_remote_device = context.context().device_spec.device_type != "CPU"
    if is_remote_device:
      with ops.device(None):
        # Let the placer figure out where to place the various functions etc.
        # created by the CopyToDeviceDataset.
        dataset = dataset.apply(prefetching_ops.copy_to_device(
            context.context().device_name))
        dataset = dataset.prefetch(1)
    super(Iterator, self).__init__(dataset)