Dynamic Load Distribution For Parallel Applications

This book is a state-of-the-art report on load distribution for parallel applications. Compute intensive applications are parallelized with the motivation to speed up program run time. One of the most challenging problems with parallel applications is the load distribution task. Application objects such as processes and data structures have to be distributed so that given system resources are used efficiently. The book provides a comprehensive overview on state-of-the-art load distribution techniques and systems. It is particularly intended for developers of parallel applications that want to obtain a survey in the field. It will also prove useful to both students and teachers of parallel programming courses and to those carrying out detailed research in the field. Contents 1 Introduction 2 Survey and Classifications / General Classification of Local Distribution / Classification and Survey of Strategies / Classification of Load Models / Migration Mechanisms / Exemplary Load Distribution Concepts: A Classification 3 Systems - Consistent Checkpointing / Load Distribution Strategies of the Distributed Thread Kernel DTK / Economic-Based Dynamic Load Distribution In Large Workstation Networks / ALDY - an Adaptive Load Distribution System 4 Applications / A Middleware-Based Architecture for Load Management / Cooperative Parallel Automated Theorem Proving / Automatic Test Pattern Generation / MPSIM - Parallel Event Driven Simulation of Logic Circuits by Time Warp 5 Summary