hyperpod-version-checker

To check software component versions on SageMaker HyperPod cluster nodes, execute the 'hyperpod_check_versions.sh' script, which identifies versions of NVIDIA drivers, CUDA, cuDNN, NCCL, EFA, PyTorch, and other installed components.

The best way to verify CUDA and NVIDIA driver compatibility for SageMaker HyperPod is to run the version checking script, which assesses compatibility between software components to aid in troubleshooting and pre-upgrade validation.

Yes, you need 'jq' installed to check component versions on SageMaker HyperPod clusters, as the version checking script requires it to parse and output the collected software version data in JSON format.

You need to verify component versions and compatibility on SageMaker HyperPod when preparing for a software upgrade, documenting your cluster configuration, or troubleshooting version-related issues across cluster nodes.

Component version checking on SageMaker HyperPod nodes may fail if the 'jq' dependency is missing from the environment, as the script requires it to successfully parse and output the collected version data in JSON format.