Strategy for loading 16x16 tiles

[beneficii]

I would like to utilize 16x16 tiles, which contain data for the upper left 8x8 tile info, the lower left 8x8 tile info, the upper right 8x8 tile info, and the lower right 8x8 tile info. However, I'm not sure of the best way to approach loading these tiles into memory in the most efficient manner. Apparently, I guess I need to load the upper left and upper right first for a screenblock row, and then load the lower left and lower right? Then rinse and repeat for each subsequent row?