A programmer on the Computer Chess Club forum reported beating Stockfish 18 in a 5-second hyperbullet game with their engine, Skipper_15_143. The win came after disabling the engine's output window, which was consuming processing time by displaying the principal variation tree. The programmer plans further optimizations, specifically removing the move-logging function that writes to games.pgn, as this file I/O also costs precious milliseconds.
The game itself followed standard Italian Game lines with a typical early middlegame pawn structure. After 1.e4 e5 2.Bc4 Nf6 3.d3 c6 4.Be3 d5, White captured on d5 and grabbed a small advantage. The position remained roughly equal throughout, suggesting time trouble rather than a position blunder cost Stockfish the match.
This continues a pattern of incremental improvements in Skipper's performance against Stockfish. The programmer is methodically identifying and removing overhead from move generation and display. In hyperbullet engines, every millisecond matters. Even basic I/O operations can prevent an engine from completing a full search depth before the clock runs out. The next iteration should be even faster.
THE TAKEAWAY: Speed optimization in engine development often matters more than raw playing strength in the hyperbullet time controls where milliseconds determine wins.