Evaluating Two Loop Transformations for Reducing Mutliple-Writer False Sharing

To simplify the programming of hierarchical and distributedmemory parallel systems, the notion of shared virtual memory (SVM) has been proposed. This abstraction provides the programmer with the illusion of a flat global address space and coherence is maintained at the page level. The success of this abstraction depends on the efficiency of page management. This in turn depends on the efficiency of handling false sharing and the resulting ping-pong effects that it can cause. In this paper, we evaluate two loop transformations for attacking this problem. The first is a simple, new, compile-time technique for reducing the pingpong effects that result from multiple-writer false sharing. The second is our previously-proposed technique for eliminating multiple-writer false sharing itself. These techniques have been implemented in the Fortran-S compiler, which generates code that runs on the iPSC/2 under the KOAN SVM. Preliminary performance results are presented