Random tangential question given this is about llama, but how do you get llama.cpp or kobold (or whatever tool you use) to make use of multiple GPUs if you don'

Random tangential question given this is about llama, but how do you get llama.cpp or kobold (or whatever tool you use) to make use of multiple GPUs if you don't have NVlink in place? I got a bridge, but it was the wrong size. Thanks, in advance.