Dictionaries or Numpy?

Hi, I'm writing an AI for a board game called Blokus, and I believe
that the quality of the program is going to greatly depend on the
efficiency of the algorithm I use to generate legal moves and rate
them.

I know sort of how I'm going to do this legal move generation, and I
know it is going to require lots of accesses to an array of some kind
which will represent the game board.

I'm fairly new to Python, and I am not sure whether it would be faster
for me to use a Dictionary data type to represent the board
(advantages: indices can be strings or tuples, and the dictionary can
store different types of data as well, so programming would be easier)
or the Numpy array type (don't know much about this, but I have heard
Numpy is faster).

So is Numpy faster for simple array accesses? That would be valuable
and I'd deal with having an integer-indexed integer array if I had to.

Code might look something like this:

for playable_corner in all_playable_corners:
    for piece in all_available_pieces:
        for each_square in piece:
            # if square is not in a legal position, don't add to list
of legal moves

So that if statement would be accessing the board array many many
times. I guess my question is: Is there any way to test Dictionary
access time vs. Numpy array access time?