How can I optimize CPU usage on a multi-GPU mining rig with a weak CPU?

You can reduce CPU strain on a multi-GPU rig by running one separate miner instance for each GPU instead of one instance managing all GPUs. In a typical multi-GPU setup, a single miner process must coordinate every graphics card, which puts heavy demand on the CPU — a bottleneck if your processor has few cores or runs slowly. By using the --cuda-device option (for NVIDIA cards) or --opencl-device option (for AMD cards), you tell each miner instance to control exactly one GPU, spreading the workload across multiple lightweight processes. This matters for Kaspa miners building or running rigs on budget hardware, because it lets you add more GPUs without requiring a high-end CPU to keep up.