On Methods for Fast and Efficient Parallel Memory Access