While there has to be some practical limit... if that 100W is rated for 12v at the lamp and you're running from a 13.nn V system, then you could probably stand to run at 1.nn drop -- if you're OK with the 100W (8.33A draw across a lighted filament resistance of something like 1.44 ohms) --
If you're looking to get 100W starting with a 12.0v system (and those are rated 100W@12V as above) then you should get to the no-drop situation ...
Let's guess that you prolly have like 13.5v available to work with...
If you end-up with a 1.5 volt drop total (for 12v at the lamp for the rated 100w output drawing 8.33A) then your wire will be dissipating some of the power that could'a been available to your lamps (about 12.5 watts (1.5 volts dropped at 8.33A) --
Go for what is most practical for you and your mounting situation - no need to run #00 welding cables for the lamps, but use some common sense as well -- watch the power dissipation of the wire. While the ~12.5W above seems large (afer all a curling iron, or small soldering iron only dissipates 15W or so), it's reasonable for a long run of wire and does give you 12V to these lamps so you'll get rated output...
A little bigger wire would allow you to provide a hotter line and produce more light, but who's to say how 13.5v on a 12v rated lamp (OK so not a HUGE overdriving, but), a 12.5% overvoltage condition , would shorte filament life -- we all know it'll be brighter, just for how long and how much extraordinary effort...?
IMHO, don't loose qty=2 volts and things should be fine (of course that implies a perfect ground as well! - the same current will have to move there)