Abstract— Priority queues are essential for various network processing applications, including per-flow queueing with Quality-of-Service (QoS) guarantees, management of large fast packet buffers, and management of statistics counters. In this paper, we propose a new data structure for implementing highperformance priority queues based on a pipelined version of the van Emde Boas tree. We show that we can achieve O(1) amortized time operations using our architecture, but we can achieve this algorithmic efficiency using only O(log log u) number of pipelined stages, where u is the size of the universe used to represent the priority keys.