Brief announcement: efficient graph algorithms without synchronization

We give a graph decomposition technique that creates entirely independent subproblems for graph problems such as coloring and dominating sets that can be solved without synchronization on a distributed memory system. For coloring, evaluation shows a performance gain of a factor 3 to 5 at the price of using more colors. Categories and Subject Descriptors: E.1 Data structures, F.2.3 Tradeoffs among Complexity Measures General terms: graphs, algorithms, performance Key Words: parallel algorithms, concurrent data structures, coloring, dominating sets

• Type: Conference
• Year: 2010
• Where: PODC
• Authors: Johannes Schneider, Roger Wattenhofer
• Keywords: Data Structures, Distributed And Parallel Computing, Distributed Memory System, Graph Decomposition Technique, PODC 2010