When processing large amounts of data, it’s important to choose a good algorithm, but “good” does not always mean “fancy”. It’s important to think about how your code will perform in practice. Constant factors drop out of “Big O” analysis, but they can be large enough to matter in the real world. Simpler algorithms are easier to optimize, and less vulnerable to bad edge-case behavior.
via Searching 20 GB/sec: Systems Engineering Before Algorithms | Scalyr Blog.