CM_WORKER_IP

Description

In TensorFlow distributed training or inference scenarios, you can choose not to use the rank table file. By jointly configuring environment variables CM_CHIEF_IP, CM_CHIEF_PORT, CM_CHIEF_DEVICE, CM_WORKER_SIZE, and CM_WORKER_IP, resource information can be automatically generated to complete the initialization of the collective communication component.

Configures the NIC IP address used for information exchange between the current device and the master node.

The value is a character string in IPv4 or IPv6 format.

Example

export CM_WORKER_IP=192.168.0.1

Constraints

This environment variable cannot be used together with RANK_TABLE_FILE, RANK_ID, or RANK_SIZE.

Applicability

Atlas training product

Atlas A2 training product/Atlas A2 inference product