Gitlab runner with autoscaling

Hi, I wanna host my own Gitlab Runner which is able to do jobs from unique projects in parallel. I think that I need the Docker Machine autoscaling tutorial to do so (following this guide:

As stated in the following step: Optionally but recommended, prepare a proxy container registry and a cache server to be used with the autoscaled Runners. Running your own cache server and container registry is recommended. Given the fact the machine on which the gitlab runner will be running is a powerhouse in terms of CPU and SSD performance, id like to run the gitlab-runner, cache and registry on the same machine.

How should I setup my config.toml using the default ports (6000 and 9005)? Or, if I misunderstand a concept; What do I misunderstand and what should is the best solution given my situation?