If it's a straight 2 methods compare use ms although you might have to test it in a loop. Gab is right about 'silly' tests but take a good guess how often you're likely to call a method.
It's easy to keep adding 0s to a loop and be happy when there's a couple of ms difference when you get up to millions.
Alternatively, use/create a profiler like this .
Finally, taken into account things like readability, maintainability, portability and robustness. Unless you're REALLY looking for speed doing something a nice way can be worth a couple of ms when compared with doing something a quick but horribly complex way.
It's easy to keep adding 0s to a loop and be happy when there's a couple of ms difference when you get up to millions.
Alternatively, use/create a profiler like this .
Finally, taken into account things like readability, maintainability, portability and robustness. Unless you're REALLY looking for speed doing something a nice way can be worth a couple of ms when compared with doing something a quick but horribly complex way.