Figures of "efficiency" do not really apply for (MOSFET) gate driver circuits: as the gate itself does not "consume" any power the efficiency is generally 0 %.
If you want to optimize the power consumption of the gate driver, other influences are important: desired switch on/switch off (transition) times, gate characteristics (primarily gate capacitance), switching frequency etc.
Nice to see you considering efficiency of gate driver circuit.
The total power input in any converter may be considered as power input to the converter along with power drawn from all bias supplies. whereas, gate-driver is concerned, as U. Dreher said, it does not consume power, if it is so, then efficiency should be 100%, but gate driver consumes power from bias supply and deliver to switch, so we can not say its efficiency to be 0% or 100%.
A garte driver consumes power. The minum power consummed by a gate driver is Qg x Vg x fs where Qg is the gate charge, Vs the gate voltage and fs the switching frequency.
Then you have to add the power consumed by the gate circuit driver itself, which usually is much less.
I think this question is somehow general, we don't know what is the rate of switching frequency and what is your application as well, For example in multilevel inverters that we expect to apply simple switching strategy on switches, an isolator transformer can solve the efficiency problems whereas in other applications like high-step up dc-dc converters that we need to accomplish soft switching algorithm, we need to use a snubber for both anti-parallel switches. Design of snubber is also diiffferent in each application for example in z-source converters there is a great voltage stress on switches during the shoot-through states. I think in this general case, make sure your input voltage to gate-source is isolated and smooth ( through using a trans with separated output and use of large capacitors) . In hard switching nature, the major loss of gate-drivers is related to smooth required dc link voltage (25V), I think.