Yes, Ollama is very plug-and-play when it comes to multi GPU.

llama.cpp probably is too, but I haven't tried it with a bigger model yet.