How is this dual setup's performance? Is it difficult to set-up everything with for example llama.cpp? I am asking since the dual setup would be way cheaper. I am very satisfied with a few new models and it would be nice to run Qwen 3.6 27B on higher quants. Thanks in advance!
