Memory access patterns are one of the key factors for writing efficient code that runs over large data sets. In this blog post, you’ll see why there might be a perf difference of almost 2.5x (in both directions!) when working with a vector of pointers versus a vector of value types.