Interleaving requests from multiple users through pipeline stages to improve hardware utilization without waiting for one request to complete.