How does the learning process for a tensor flow driver and invoke update variables on ps?

Question

I am reading tensor flow code and find this answer tensorflow-using-parameter-servers-in-distributed-training

Would love to know the details?

Answer 1

Here are some details on how to configure and use distributed models at https://www.tensorflow.org/deploy/distributed .

, " ", , ps server.join(), ops. , , ps, , , , , with, op gpu cpu.

- , ?