Initializes Multi Client DTensor.
tf.experimental.dtensor.initialize_multi_client( enable_coordination_service: Optional[bool] = False ) -> None
The following environment variables controls the behavior of this function. If the variables are unset, DTensor will be configured to run in single-client mode.
{hostname}:{port}
and the items must be sorted in alphabet order. The implication is the RPC port numbers of the clients from the same host must be ordered by the client ID. Examples of valid DTENSOR_JOBS values: localhost:10000,localhost:10001,localhost:10002,localhost:10003
host1:10000,host1:10001,host2:10000,host2:10003
Args | |
---|---|
enable_coordination_service | If true, enable distributed coordination service to make sure that workers know the devices on each other, a prerequisite for data transfer through cross-worker rendezvous. |
© 2022 The TensorFlow Authors. All rights reserved.
Licensed under the Creative Commons Attribution License 4.0.
Code samples licensed under the Apache 2.0 License.
https://www.tensorflow.org/versions/r2.9/api_docs/python/tf/experimental/dtensor/initialize_multi_client