AMD has released a comprehensive guide for running OpenClaw on Windows via two hardware configurations: “RyzenClaw” and “RadeonClaw.” Both setups utilize AMD’s silicon to enable local AI agent workloads, aligning with the company’s “Agent Computer” initiative, emphasizing data control, unlimited affordable AI, and local execution over cloud dependency. The installations leverage WSL2 and LM Studio for LLM inference through llama.cpp, facilitating memory management and real-time processing without relying on cloud services.
The RyzenClaw configuration features a Ryzen AI Max+ system with 128 GB of unified memory, capable of handling up to six concurrent agents, processing 10,000 input tokens in approximately 19.5 seconds. In contrast, RadeonClaw, with a Radeon AI PRO R9700 GPU, offers faster processing at 120 tokens per second but supports only two agents. Both setups target developers and early adopters, with entry costs ranging from $1,299 to $2,700, highlighting a niche yet promising opportunity for local AI experimentation.
Source link
