Comparison of the use of point locations with distance computations versus the use of the grid-based approach when searching for clusters. (Above): Setting where cases and controls data consists of points (identified by latitude and longitude). Determining the number of cases and controls in a (circular) region requires computing the distances from the center of the region to each case and each control. For cluster searches, millions of regions (the potential clusters) are considered which means billions of distance calculations for a medium size (case/control) study. (Below): Setting where cases and controls have been projected to a grid. Determining the number of cases and controls in a (rectangular) region requires only reading values from memory-mapped arrays and summing them. This method is very fast even for cluster searches where millions of regions are considered (the potential clusters).