Utilize GuideLLM (from Neural Magic) as an AI Workload suitable for multi-arch CPU-mode testing.

Use two inferencing engines: llama.cpp and vLLM which supports CPU-Mode as documented here

https://docs.vllm.ai/en/stable/getting_started/installation/cpu.html#set-up-using-docker

Document investigation findings along with setup procedures and initial test results, if possible.