I was already running the "#define Ma" optimized kernel, and this is good for another .5% on top of that.