Siemens AG Private Networks

Distributed shared objects are a well known approach to achieve independence of the memory model for parallel pro-gramming. The illusion of shared (global) objects is a con-venient abstraction which leads to ease of programming on both kinds of parallel architectures, shared memory and dis-tributed memory machines. In this paper, we present several different implementation variants for distributed shared ob-jects on distributed platforms. We have considered these variants while implementing a high level parallel program-ming model known as coordinators [10]. These are global objects coordinating accesses to the encapsulated data ac-cording to statically defined access patterns. Coordinators have been implemented on both shared memory multipro-cessors and networks of workstations (NOWs). In this pa-per, we describe their implementation as distributed shared objects and give basic performance results on a NOW.