The reason why the 2-d procedure takes longer is because the 2-d dot operator has a broader scope than the 1-d dot operator. The dot operator is used to represent a number of different operations, including vector calculus operations. At parse time, it is not always possible to tell which meaning is the intended one. Thus, the 2-d dot operator translates into a procedure that examines the arguments after they are passed in and, using the types of the arguments, deduces which of the various "dot" commands is appropriate.
In my own Maple use, I like 2-d math for entering complicated math expressions and to get input and output that looks like math. However, for programming purposes, I tend to use 1-d math input because within procedures, context information that's useful for parsing isn't available until the procedure is called. For writing complicated procedures, especially when efficiency is very important, I suggest using 1-d input.
I will add, though, that we are aware of the issues related to the 2-d dot operator that have been brought up on MaplePrimes, and we do have long-term plans to improve this situation.