Presented here is a set of methods and tools developed to provide transportable measurements of performance in heterogeneous networks of machines operating together as a single virtual heterogeneous machine (VHM). The methods are work-based rather than time-based, and yield significant analytical information. A technique for normalizing the measure of useful work performed across a heterogeneous network is proposed and the reasons for using a normalized measure are explored. It is shown that work-based performance measures are better than time-based ones because they may be 1) taken while a task is currently executing on a machine, 2) taken without interrupting production operation of the machine network, 3) used to compare disparate tasks, and 4) used to perform second order analysis of machine network operation. This set of performance tools has been used to monitor the utilization of high-performance computing networks, provide feedback on algorithm design and determine the veracity...

The idea of Architecture Adaptive Algorithms was introduced in order to reduce porting overhead in parallel algorithms. The AAA presents a way for algorithms to adapt themselves to the computer system on which they are running without the assistance of the application programmer. For the development of these algorithms it is, nevertheless, necessary to obtain reliable and reproducible execution time information. A good method for achieving these results is to simulate the hardware. In this paper a new simulator (the Machine Independent Simulation System for PVM 3 ) is presented and its usefulness for this type of application is investigated. Contents 1 Introduction 3 2 AAA 3 3 Performance Prediction using Simulation 5 4 MISS-PVM 8 5 The Model 13 6 The Determination of Simulation Parameters 14 7 Validation 20 8 Conclusion 23 References 23 1 Introduction Since parallel programming is more complicated than sequential programming it is only advisable to use it when the performance of s...

The SUPRENUM project was a major effort in Germany to develop a national expertise in massively parallel processing both at the hardware and software level. We review the development of SUPRENUM and introduce the twelve later papers in this proceedings that deal in more detail with SUPRENUM. Taken together these papers present an overview of the project goals, its historical development, successes and failures of the project, and indicate ways in which the project has influenced MPP developments elsewhere. GENESIS was a European ESPRIT project focused on massively parallel processing. It included both hardware and software architecture aspects. GENESIS was a natural development from SUPRENUM and used the Suprenum-1 machine as its primary hardware platform. We briefly discuss GENESIS and introduce the nine papers in this volume that represent the GENESIS effort. GENESIS has had a significant effect on many other MPP efforts. Several of the papers address these influences in some detail....