Process-Oriented Collective Operations
نویسندگان
چکیده
Distributing process-oriented programs across a cluster of machines requires careful attention to the effects of network latency. The MPI standard, widely used for cluster computation, defines a number of collective operations: efficient, reusable algorithms for performing operations among a group of machines in the cluster. In this paper, we describe our techniques for implementing MPI communication patterns in process-oriented languages, and how we have used them to implement collective operations in PyCSP and occam-π on top of an asynchronous messaging framework. We show how to make use of collective operations in distributed processoriented applications. We also show how the process-oriented model can be used to increase concurrency in existing collective operation algorithms.
منابع مشابه
Abstraction of Agent Cooperation in Agent Oriented Programming Language
ion of agent cooperation in agent oriented programming language Nguyen Tuan Duc and Ikuo Takeuchi The University of Tokyo, Japan [email protected], [email protected] Abstract. Collective operation is a concept of parallel programming in which many processes participate in an operation. Since collective operations are suitable for modeling the coordination of many processes, they can be used ...
متن کاملCollective Objects: An Object-Oriented Tool for Collective Operations in Distributed Parallel Computation
This paper describes the collective object, a new abstraction providing support for collective operations common in parallel programming. The collective object is introduced in the context of research aiming to produce a design of a distributed object-oriented environment suitable for parallel computation, and will constitute a part of the object model of this environment. We give a formal de n...
متن کاملMPI Collective Operations over IP Multicast
Many common implementations of Message Passing Interface (MPI) implement collective operations over point-to-point operations. This work examines IP multicast as a framework for collective operations. IP multicast is not reliable. If a receiver is not ready when a message is sent via IP multicast, the message is lost. Two techniques for ensuring that a message is not lost due to a slow receivin...
متن کاملIrregular data-parallelism in a parallel object-oriented language by means of Collective Replication
In parallel object-oriented languages it is hard to elegantly express efficient data-parallel operations on objects of an irregularly-shaped object structure that is spread across the parallel computing environment. This paper presents a new programming model that smoothly integrates both task and data parallelism in a distributed object-oriented context. So called Collective Replication combin...
متن کاملSparse Non-blocking Collectives in Quantum Mechanical Calculations
For generality, MPI collective operations support arbitrary dense communication patterns. However, in many applications where collective operations would be beneficial, only sparse communication patterns are required. This paper presents one such application: Octopus, a production-quality quantum mechanical simulation. We introduce new sparse collective operations defined on graph communicators...
متن کامل