TensorFlow distributed clustered Local Server

A simple distributed service (local server) feature is provided in TensorFlow, and a simple distributed program can be run with the local server service. The local Server implements the same interface as the distributed service, so it is more convenient for development testing.

Here's the simplest example.

Import TensorFlow as tf
server = Tf.train.Server.create_local_server () with
TF. Session (Server.target) as session:
    matrix1 = Tf.constant ([[3., 3.]])
    matrix2 = Tf.constant ([[[2.],[2.]])
    Product = Tf.matmul (matrix1, matrix2) Result
    = Session.run (product)
    print result

where the Tf.train.Server.create_local_server () method creates a single process cluster.

In addition, running this program is actually the server and client started in the same process.