No, because you don't have additional IO overhead.
I think that overhead would be negligible, especially if the other examples are all based on micro-frameworks (i.e. loads of files).
Premature optimisation is the root of all evil, right?
Disk IO is a massive problem at scale. This obviously isn't "at scale", so it works fine here, but avoiding hitting the disk (edit: any slow IO, really) is Performance 101.