Aroma simplifies the task of parallelizing large applications on multicomputers by providing applications with a shared object space. Aroma supports both traditional monolithic objects and aggregate object that can be partitioned across multiple nodes. Aggregate objects support data parallelism efficiently. An Aroma program consists of tasks that operate on shared objects. Tasks typically execute on the node on which their input data is located, thus minimizing communication. Shared data objects have synchronization properties associated with them, making it possible to parallelize a large class of applications without using explicit locks and condition variables. In this paper, we present and justify the Aroma language features, and give examples of Aroma programs. Aroma has been implemented on the Nectar multicompnter and we give performance results for several applications.

No abstract

The goal of load balancing is to assign to each node a number of tasks proportional to its performance. On distributed-memory machines, it is important to take data dependencies into account when distributing tasks, since they have a big impact on the communication requirements of the distributed application. Many load balancers have been proposed that deal with applications with homogeneous tasks, but applications with heterogeneous tasks have proven to be far more complex to handle. In this paper we present a load balancing architecture that can deal with applications with heterogeneous tasks. The idea is to provide a set of load balancers that are effective for different types of homogeneous tasks, and to allow users to combine these load balancers for applications with heterogeneous tasks. We implemented this architecture on the Nectar multicomputer and we present performance results for several applications with homogeneous and heterogeneous tasks.

No abstract

No abstract