Languages and Compilers for Parallel Computing: 6th International Workshop, Portland, Oregon, USA, August 12 - 14, 1993. Proceedings

by Utpal Banerjee (editor)

Early experiences with Olden.- Arbitrary order operations on trees.- Analysis of dynamic structures for efficient parallel execution.- On automatic data structure selection and code generation for sparse computations.- Synchronization issues in data-parallel languages.- An array sublanguage.- Event-based composition of concurrent programs.- Adaptive parallelism on Preliminary experience with Piranha on the CM-5.- Slicing analysis and indirect accesses to distributed arrays.- Do& Integrating parallel loops and reductions.- Automatic support for data distribution on distributed memory multiprocessor systems.- A compilation approach for Fortran 90D/HPF compilers.- A framework for exploiting data availability to optimize communication.- The alignment-distribution graph.- An overview of a compiler for scalable parallel machines.- Toward a compile-time methodology for reducing false sharing and communication traffic in shared virtual memory systems.- Program transformation for locality using affinity regions.- Maximizing loop parallelism and improving data locality via loop fusion and distribution.- Align and distribute-based linear loop transformations.- Extending software pipelining techniques for scheduling nested loops.- A methodology for generating efficient disk-based algorithms from tensor product formulas.- Loop transformations for Prolog programs.- A multithreaded implementation of Id using P-RISC graphs.- Acceleration of first and higher order recurrences on processors with instruction level parallelism.- Efficient compile-time/run-time contraction of fine grain data parallel codes.- The Visual Interface for Scheduling Transformations and Analysis.- Efficiently computing ?-nodes on-the-fly.- Construction of thinned gated single-assignment form.- Automatic array privatization.- A framework for interprocedural analysis and transformation.- An exact method for analysis of value-based array data dependences.- Symbolic A basis for parallelization, optimization, and scheduling of programs.- Towards a non-intrusive approach for monitoring distributed computations through perturbation analysis.- Efficient computation of precedence information in parallel programs.- Trace size vs parallelism in trace-and-replay debugging of shared-memory programs.- Parallel Program Graphs and their classification.

676 pages, Paperback

First published April 1, 1992