From Think Parallel to Think Sequential

This paper presents GRAPE , a parallel GRAPh Engine for graph computations. GRAPE differs from previous graph systems in its ability to parallelize existing sequential graph algorithms as a whole, without the need for recasting the entire algorithms into a new model. Underlying GRAPE are a simple programming model, and a principled approach based on fixpoint computation with partial evaluation and incremental computation. Under a monotonic condition, GRAPE guarantees to converge at correct answers as long as the sequential algorithms are correct. We show how our familiar sequential graph algorithms can be parallelized by GRAPE . In addition to the ease of programming, we experimentally verify that GRAPE achieves comparable performance to the state-of-the-art graph systems, using real-life and synthetic graphs.