A Data Reorganization Technique for Improving Data Locality of Irregular Applications in Software Distributed Shared Memory

Irregular applications are characterized by highly irregular and ne-grained data referencing patterns. When there is poor locality between the ne-grained data, serious false sharing can occur which has largely contributed to poor performance of irregular applications on page-based software distributed shared memory (DSM) systems. Partitioning data in irregular applications to improve data locality has been routinely practiced in message passing programming to reduce inter-processor communication, but data reordering is often ignored in shared memory programming. In this paper we show that data reordering is also important in programming irregular applications under software DSMs, in particular because of the false sharing eeect. We describe a simple yet highly eeec-tive data reordering technique which can signiicantly improve data locality and reduce false sharing in many irregular applications. We evaluate the eeectiveness of this technique on a set of ve irregular applications under both Princeton's home-based protocol and Rice's TreadMarks protocol on a 16-processor platform. We nd that with a simple data ordering during the initial-ization, the performance of these irregular applications is improved by 30% { 366% under TreadMarks and 14%-269% under the home-based protocol.