A Compiler and Runtime Infrastructure for Automatic Program Distribution

  • Authors:
  • Roxana E. Diaconescu;Lei Wang;Zachary Mouri;Matt Chu

  • Affiliations:
  • California Institute of Technology;University of California, Irvine;University of California, Irvine;University of California, Irvine

  • Venue:
  • IPDPS '05 Proceedings of the 19th IEEE International Parallel and Distributed Processing Symposium (IPDPS'05) - Papers - Volume 01
  • Year:
  • 2005

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents the design and the implementation of a compiler and runtime infrastructure for automatic program distribution.We are building a research infrastructure that enables experimentation with various program partitioning and mapping strategies and the study of automatic distribution's effect on resource consumption (e.g., CPU, memory, communication). Since many optimization techniques are faced with conflicting optimization targets (e.g., memory and communication), we believe that it is important to be able to study their interaction. We present a set of techniques that enable flexible resource modeling and program distribution. These are: dependence analysis, weighted graph partitioning, code and communication generation, and profiling. We have developed these ideas in the context of the Java language. We present in detail the design and implementation of each of the techniques as part of our compiler and runtime infrastructure. Then, we evaluate our design and present preliminary experimental data for each component, as well as for the entire system.