Firstly, try whatever works then go on from there! As for the memory consumption of a 2d array, in your instance its only a few bytes for the array data and the compiler may decide to optimise the access to it rather than use multiplies etc. You could use a switch statement, but that would probably chew up as many if not more bytes than using an array. Using an array makes sense when you have a non-linear mapping from input to output - FPGAs use fast ram lookup tables to implement their logic!

If in doubt of the amount of memory required in a certain instance - write some test code and look at the results and your question will be answered with hard evidence.