Description

For the supernode mode networking of the Atlas A3 training product/Atlas A3 inference product, if the rank table file is not used to configure cluster resource information, you can use this environment variable to specify the ID of the supernode to which the process running on the current node belongs. In this way, a physical supernode can be divided into multiple logical supernodes.

This environment variable is set to a string of fewer than 128 characters. The default value is an empty string.

If it is not set, the value of Super Pod ID in the environment is used as the supernode ID. You can run the npu-smi info -t spod-info -i id -c chip_id command to view the value of Super Pod ID.

Example

export HCCL_LOGIC_SUPERPOD_ID=super_pod_id_1

Constraints

This environment variable applies only to the scenario where the rank table file is not used to configure cluster information in supernode mode. If the rank table file is used, the configuration in the file is used preferentially.
This environment variable is used to divide a physical supernode into multiple logical supernodes. Ranks that belong to different physical supernodes cannot be configured in the same logical supernode.

Applicability

Atlas A3 training product/Atlas A3 inference product