PHAST: Hardware-accelerated shortest path trees

We present a novel algorithm to solve the non-negative single-source shortest path problem on road networks and graphs with low highway dimension. After a quick preprocessing phase, we can compute all distances from a given source in the graph with essentially a linear sweep over all vertices. Because this sweep is independent of the source, we are able to reorder vertices in advance to exploit locality. Moreover, our algorithm takes advantage of features of modern CPU architectures, such as SSE and multiple cores. Compared to Dijkstra’s algorithm, our method needs fewer operations, has better locality, and is better able to exploit parallelism at multi-core and instruction levels. We gain additional speedup when implementing our algorithm on a GPU, where it is up to three orders of magnitude faster than Dijkstra’s algorithm on a high-end CPU. This makes applications based on all-pairs shortest-paths practical for continental-sized road networks. Several algorithms, such as computing the graph diameter, arc flags, or exact reaches, can be greatly accelerated by our method.

► Novel single-source shortest path algorithm.
► Takes advantage of features of modern CPU architectures (SSE and multiple cores).
► Additional speedup when implemented on a GPU.
► Up to three orders of magnitude faster than Dijkstra’s algorithm.
► Makes applications based on all-pairs shortest-paths practical.