This book constitutes the thoroughly refereed post-proceedings of the 17th International Workshop on Languages and Compilers for High Performance Computing, LCPC 2004, held in West Lafayette, IN, USA in September 2004. The 33 revised full papers presented were carefully selected during two rounds of reviewing and improvement. The papers are organized in topical sections on compiler infrastructures; predicting and reducing memory access; locality, tiling, and partitioning; tools and techniques for parallelism and locality; Java for high-performance computing; high-level languages and optimizations; large-scale data sharing; performance studies; program analysis; and exploiting architectural features.

Rese�a del editor

The 17th International Workshop on Languages and Compilers for High P- formance Computing was hosted by Purdue University in September 2004 on Purdue campus in West Lafayette, Indiana, USA. The workshop is an annual international forum for leading research groups to present their current research activities and the latest results, covering languages, compiler techniques, r- time environments, and compiler-related performance evaluation for parallel and high-performance computing. Eighty-six researchers from Canada, France, Japan, Korea, P. R. China, Spain, Taiwan and the United States attended the workshop. A new feature of LCPC 2004 was its mini-workshop on Research-Compiler Infrastructures. Representatives from four projects, namely Cetus, LLVM, ORC and Trimaran, gavea 90-minute long presentation each. In addition, 29 research papers were presented at the workshop. These papers were reviewed by the p- gram committee. External reviewers were used as needed. The authors received additional comments during the workshop. The revisions after the workshop are now assembled into these ?nal proceedings. A panel sessionwasorganizedby Samuel Midki? onthe questionof "What is GoodCompilerResearch-Theory,PracticeorComplexity?"Theworkshopalso had the honor and pleasure to have two keynote speakers, Peter Kogge of the University of Notre Dame and David Kuck of Intel Inc., both pioneers in high performance computing. Peter Kogge gave a presentation titled "Architectures and Execution Models: How New Technologies May A?ect How LanguagesPlay on Future HPC Systems". David Kuck presented Intel's vision and roadmap for parallel and distributed solutions.

��ber diesen Titel� kann sich auf eine andere Ausgabe dieses Titels beziehen.