The conference was called to address the difficulty of programming distributed memory machines, which is inhibiting the spread of scalable parallel computers; the goal is high-level, application-oriented programming models that assign to the compiler the burdensome, low-level mechanisms of parallel