Automated smoke tests for CUDA cannot be executed on the following instance types in AWS because they use NVSwitch and require further configuration.

• p5.48xlarge
• p5e.48xlarge
• p5en.48xlarge
• p4d.24xlarge
• p4de.24xlarge

Acceptance criteria:

• RHAIIS container tests run successfully on these instance types
• Tests use all available GPUs
• At least two models work, e.g. ibm-granite/granite-3.3-8b-instruct and meta-llama/Meta-Llama-3-8B-Instruct

Notes
https://docs.nvidia.com/datacenter/tesla/fabric-manager-user-guide/index.html#nvidia-server-architectures