Okay, so now, relative to floating point arithmetic, you've sacrificed an enormous amount of dynamic range and have to worry about overflow and underflow much m

Okay, so now, relative to floating point arithmetic, you've sacrificed an enormous amount of dynamic range and have to worry about overflow and underflow much more than before. What did you gain from this, exactly? I'm sure there are some special situations where fixed point is better, but for general purpose scientific and technical computing, floating point is obviously the right choice.