The reason the above works is because since each element requires elements from the points around it, there would be more reuse of data (and thus better cache utilization) if we use blocks of data rather than just rows. If the cache can only hold one row, then in the first implementation the cache would have to be updated each time we try to access data from a row above or below an element.
The reason the above works is because since each element requires elements from the points around it, there would be more reuse of data (and thus better cache utilization) if we use blocks of data rather than just rows. If the cache can only hold one row, then in the first implementation the cache would have to be updated each time we try to access data from a row above or below an element.
This comment was marked helpful 0 times.