Re benchmarking/ finding an optimum config: Michael Goetz recommends to run specific tasks directly from the command line, not live tasks through the boinc-client. That way you can compare different configs directly on a single task and without having to wait for validation (if your rig works reasonably stable). I'll need to look into that.
Edit: Of course commandline benchmarking needs to be done outside of challenges, as you don't get credits as a byproduct.
Edit: Of course commandline benchmarking needs to be done outside of challenges, as you don't get credits as a byproduct.
Last edited: