It is probably preferable to have an ultrasonic bath the size of your reaction flask, as you need to shake off the oxide layer of all the Mg turnings you are using, but in my opinion it is only necessary to sonicate all the magnesium with say 10% of the alkyl halide, the other 90% of the halide can be added in another flask, if you figure out a clever way of transferring a reacting grignard solution from one flask to another.
An alternative for large reactions is to use an ultrasonic probe, which you dip into the reaction flask, instead of having the whole flask in a sonicating bath.