Managing A Real-Time Massively-Parallel Neural Architecture