According to my timing experiments, sequential is still empirically faster. So being a CTF exercise is just about the only use for this now.

Run with pppp -n [number of threads] -i [input file] -o [output file]