Functional Programming For Loosely Coupled Multiprocessors

This book proposes a new high level approach to programming that addresses the pragmatic issue of how a computation is distributed across a machine. Recent progress in VLSI provides massive parallelism but general purpose parallel computers remain elusive due to limited communications performance. This book proposes a new high level approach to programming that addresses the pragmatic issue of how a computation is distributed across a machine. The book's approach is based on functional programming and has significant advantages over existing comparable approaches, extending the domain of functional programming to include computer architectures in which communication costs are not negligible. It looks at how high-level functional programming languages can be used to specify, reason about, and implement parallel programs for a variety of multiprocessor systems, but in particular a class of loosely coupled multiprocessors whose operation can be described by a process network In these networks the nodes correspond to processes and the arcs to communications channels. A simple language called Caliban is described in which the functional program text is augmented with a declarative description of how processes are partitioned and mapped onto a network of processing elements. The notation gains expressive power by allowing these annotations to be generated by predicates defined in the functional language. Thus, common communications structures have simple and concise definitions as "network forming operators." The main objective of these annotations is to provide an abstract description of the process network specified by the program so that an efficient mapping of processes to processors can be carried out by the compiler. Functional Programming for Loosely Coupled Multiprocessors is included in the series Research Monographs in Parallel and Distributed Computing, copublished with Pitman Publishing.

Programming is hard. Building a large program is like constructing a steam locomotive through a hole the size of a postage stamp. An artefact that is the fruit of hundreds of person-years is only ever seen by anyone through a lOO-line window. In some ways it is astonishing that such large systems work at all. But parallel programming is much, much harder. There are so many more things to go wrong. Debugging is a nightmare. A bug that shows up on one run may never happen when you are looking for it - but unfailingly returns as soon as your attention moves elsewhere. A large fraction of the program's code can be made up of marshalling and coordination algorithms. The core application can easily be obscured by a maze of plumbing. Functional programming is a radical, elegant, high-level attack on the programming problem. Radical, because it dramatically eschews side-effects; elegant, because of its close connection with mathematics; high-level, be cause you can say a lot in one line. But functional programming is definitely not (yet) mainstream. That's the trouble with radical approaches: it's hard for them to break through and become mainstream. But that doesn't make functional programming any less fun, and it has turned out to be a won derful laboratory for rich type systems, automatic garbage collection, object models, and other stuff that has made the jump into the mainstream.

This book constitutes the refereed proceedings of the 8th International Conference on Algorithms and Architectures for Parallel Processing, ICA3PP 2008, held in Agia Napa, Cyprus, in June 2008. The 31 revised full papers presented together with 1 keynote talk and 1 tutorial were carefully reviewed and selected from 88 submissions. The papers are organized in topical sections on scheduling and load balancing, interconnection networks, parallel algorithms, distributed systems, parallelization tools, grid computing, and software systems.

This book constitutes the thoroughly refereed post-workshop proceedings of the 9th International Workshop on Implementation of Functional Languages, IFL'97, held in St. Andrews, Scotland, UK, in September 1997. The 21 revised full papers presented were selected from the 34 papers accepted for presentation at the workshop during a second round of thorough a-posteriori reviewing. The book is divided in sections on compilation, types, benchmarking and profiling, parallelism, interaction, language design, and garbage collection.

Massively Parallel Systems (MPSs) with their scalable computation and storage space promises are becoming increasingly important for high-performance computing. The growing acceptance of MPSs in academia is clearly apparent. However, in industrial companies, their usage remains low. The programming of MPSs is still the big obstacle, and solving this software problem is sometimes referred to as one of the most challenging tasks of the 1990's. The 1994 working conference on "Programming Environments for Massively Parallel Systems" was the latest event of the working group WG 10.3 of the International Federation for Information Processing (IFIP) in this field. It succeeded the 1992 conference in Edinburgh on "Programming Environments for Parallel Computing". The research and development work discussed at the conference addresses the entire spectrum of software problems including virtual machines which are less cumbersome to program; more convenient programming models; advanced programming languages, and especially more sophisticated programming tools; but also algorithms and applications.

These proceedings are devoted to communicating significant developments in all areas pertinent to Parallel Symbolic Computation.The scope includes algorithms, languages, software systems and application in any area of parallel symbolic computation, where parallelism is interpreted broadly to include concurrent, distributive, cooperative schemes, and so forth.

This is the first comprehensive account of this new approach to the fundamentals of parallel programming.