2011 Poster Sessions : Simple and Efficient Parallel Graph Processing

Student Name : Sungpack Hong
Advisor : Oyekunle Olukotun
Research Areas: Computer Systems
Abstract:
Graph is a fundamental data representation that is favored in many domains of computer science and engineering including computational biology, social network analysis, and artificial intelligence. However, fast processing of graph data are still highly demanded since these domains typically perform expensive computations on massive data-set represented as graphs.

Our approach is two-fold: First, we seek for efficient parallel implementation of key graph algorithms such as breadth first search (BFS) or single source shortest path on top of modern computation environments such as multi-core CPUs and GPUs. These key algorithms role as basic building blocks for user applications. Second, we take a domain-specific language (DSL) approach in order to facilitate the use of above implementation methods over multiple environments. Such an approach is critical since domain applications tend to adorn basic algorithms with a custom computation rather than to call the fixed functions in a graph library.

Bio:
Sungpack Hong is a Ph. D. candidate in Electrical Engineering at Stanford University. He is interested in how to exploit parallel and heterogeneous computer architectures in an easy and efficient manner. His current research topic is to find such a solution for graph problems.