Experimental Study of High Performance Priority Queues

The priority queue is a very important and widely used data structure in computer science, with a variety of applications including Dijkstra’s Single Source Shortest Path algorithm on sparse graph types. This study presents the experimental results of a variety of priority queues. The focus of the experiments is to measure the speed and performance of highly specialized priority queues in out-of-core and memory intensive situations. The priority queues are run in-core on small input sizes as well as out-of-core using large input sizes and restricted memory. The experiments compare a variety of well-known priority queue implementations such as Binary Heap with highly specialized implementations, such as 4-ary Aligned Heap, Chowdhury and Ramachandran’s Auxiliary Buffer Heap, and Fast Binary Heap. The experiments include Cache-Aware as well as Cache-Oblivious priority queues. The results indicate that the high-performance priority queues easily outperform traditional implementations. Also, overall the Auxiliary Buffer Heap has the best performance among the priority queues considered in most in-core and out-of-core situations.