Splitting a computation across many processors that work at once. Graphics cards and AI training depend on it.