Numa-Aware Scalable Graph Traversal On Sgi Uv Systems

Breadth-first search (BFS) is one of the most fundamental processing algorithm singraph theory. We previously presented a scalable BFS algorithm based on Beamer's direction optimizing algorithm forn on-uniform memory access(NUMA)-based systems, in which the NUMA architecture was care-fully considered. This paper presents our new implementation that reduces remote memory access in a top-down direction of direction-optimizing algorithm. We also discuss numerical results obtained on the SGI UV 2000 and UV 300 systems, which are shared-memory super computers based on a cache coherent (cc)-NUMA architecture that can handle thousands of threads on a single operating system. Our implementation has a chieved performance rates of 219 billion edges per second on a Kronecker graph with 2(34) vertices and 2(38) edges on arack of an SGI UV 300 system with 1,152 threads. This result exceeds the fast estentry for a shared memory system on the current Graph500 list presented in November 2015, which includes our previous implementation.