Just when I thought the engine was finished, a friend sent me a snippet from a Russian physics textbook. It forced a complete rewrite, dropping the time complexity to O(K log N) and obliterating our benchmarks.

From a 14-year-old’s Turbo Pascal script on an 80286 to a multi-core supercomputing engine. We review the math, the code, and the ultimate lesson: hardware gets faster, but mathematics conquers.