1# Copyright 2018 The TensorFlow Authors. All Rights Reserved.
2#
3# Licensed under the Apache License, Version 2.0 (the "License");
4# you may not use this file except in compliance with the License.
5# You may obtain a copy of the License at
6#
7#     http://www.apache.org/licenses/LICENSE-2.0
8#
9# Unless required by applicable law or agreed to in writing, software
10# distributed under the License is distributed on an "AS IS" BASIS,
11# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12# See the License for the specific language governing permissions and
13# limitations under the License.
14# ==============================================================================
15"""Experimental API for controlling threading in `tf.data` pipelines."""
16from __future__ import absolute_import
17from __future__ import division
18from __future__ import print_function
19
20
21from tensorflow.core.framework import dataset_options_pb2
22from tensorflow.python.data.util import options
23from tensorflow.python.util.tf_export import tf_export
24
25
26@tf_export("data.experimental.ThreadingOptions")
27class ThreadingOptions(options.OptionsBase):
28  """Represents options for dataset threading.
29
30  You can set the threading options of a dataset through the
31  `experimental_threading` property of `tf.data.Options`; the property is
32  an instance of `tf.data.experimental.ThreadingOptions`.
33
34  ```python
35  options = tf.data.Options()
36  options.experimental_threading.private_threadpool_size = 10
37  dataset = dataset.with_options(options)
38  ```
39  """
40
41  max_intra_op_parallelism = options.create_option(
42      name="max_intra_op_parallelism",
43      ty=int,
44      docstring=
45      "If set, it overrides the maximum degree of intra-op parallelism.")
46
47  private_threadpool_size = options.create_option(
48      name="private_threadpool_size",
49      ty=int,
50      docstring=
51      "If set, the dataset will use a private threadpool of the given size.")
52
53  def _to_proto(self):
54    pb = dataset_options_pb2.ThreadingOptions()
55    if self.max_intra_op_parallelism is not None:
56      pb.max_intra_op_parallelism = self.max_intra_op_parallelism
57    if self.private_threadpool_size is not None:
58      pb.private_threadpool_size = self.private_threadpool_size
59    return pb
60
61  def _from_proto(self, pb):
62    if pb.WhichOneof("optional_max_intra_op_parallelism") is not None:
63      self.max_intra_op_parallelism = pb.max_intra_op_parallelism
64    if pb.WhichOneof("optional_private_threadpool_size") is not None:
65      self.private_threadpool_size = pb.private_threadpool_size
66