[feature request] Can we add H200 in infer_cluster_key() method?

[dongluw]

https://github.com/NVIDIA/TensorRT-LLM/blob/main/tensorrt_llm/auto_parallel/cluster_info.py

here it handles a list of devices which doesn't include H200, can we add H200 (and potentially GB200) to that list as well?

[renjie0]

What is infer_cluster_key used for

[nv-guomingz]

Hi @dongluw thanks for suggestion. We'll add H200 into infer_cluster_key() menthod and this update will be avaiable in next release.