With the appearance of massively parallel computers, increased attention has been paid to algorithms which rely upon analogies to natural processes. This development defines the scope of the PPSN conference at Dortmund in 1990 whose proceedings are presented in this volume. The subjects treated include: - Darwinian methods such as evolution strategies and genetic algorithms; - Boltzmann methods such as simulated annealing; - Classifier systems and neural networks; - Transfer of natural metaphors to artificial problem solving. The main objectives of the conference were: - To gather theoretical results about and experimental comparisons between these algorithms, - To discuss various implementations on different parallel computer architectures, - To summarize the state of the art in the field, which was previously scattered widely both among disciplines and geographically.

The innovative progress in the development of large- and small-scale parallel computing systems and their increasing availability have caused a sharp rise in interest in the scientific principles that underlie parallel computation and parallel programming. The biannual Parallel Architectures and Languages Europe (PARLE) conferences aim at presenting current research on all aspects of the theory, design and application of parallel computing systems and parallel processing. PARLE '91, the third conference in the series, again offers a wealth of high-quality research material for the benefit of the scientific community. Compared to its predecessors, the scope of PARLE '91 has been broadened so as to cover the area of parallel algorithms and complexity, in addition to the central themes of parallel architectures and languages. The two-volume proceedings of the PARLE '91 conference contain the text of all contributed papers that were selected for the programme and of the invited papers by leading experts in the field.

Past, Present, Parallel is a survey of the current state of the parallel processing industry. In the early 1980s, parallel computers were generally regarded as academic curiosities whose natural environment was the research laboratory. Today, parallelism is being used by every major computer manufacturer, although in very different ways, to produce increasingly powerful and cost-effec- tive machines. The first chapter introduces the basic concepts of parallel computing; the subsequent chapters cover different forms of parallelism, including descriptions of vector supercomputers, SIMD computers, shared memory multiprocessors, hypercubes, and transputer-based machines. Each section concentrates on a different manufacturer, detailing its history and company profile, the machines it currently produces, the software environments it supports, the market segment it is targetting, and its future plans. Supplementary chapters describe some of the companies which have been unsuccessful, and discuss a number of the common software systems which have been developed to make parallel computers more usable. The appendices describe the technologies which underpin parallelism. Past, Present, Parallel is an invaluable reference work, providing up-to-date material for commercial computer users and manufacturers, and for researchers and postgraduate students with an interest in parallel computing.

Logic programming refers to execution of programs written in Horn logic. Among the advantages of this style of programming are its simple declarativeand procedural semantics, high expressive power and inherent nondeterminism. The papers included in this volume were presented at the Workshop on Parallel Logic Programming held in Paris on June 24, 1991, as part of the 8th International Conference on Logic Programming. The papers represent the state of the art in parallel logic programming, and report the current research in this area, including many new results. The three essential issues in parallel execution of logic programs which the papers address are: - Which form(s) of parallelism (or-parallelism, and-parallelism, stream parallelism, data-parallelism, etc.) will be exploited? - Will parallelism be explicitly programmed by programmers, or will it be exploited implicitly without their help? - Which target parallel architecture will the logic program(s) run on?

This monograph concentrates on the implementation of concurrent logic programming languages on loosely-coupled parallel computer architectures. Two different abstract models are also considered: a graph reduction based model and a more imperative approach. The design of a specialized instruction set is presented that is general enough to be used for the whole family of concurrent logic programming languages. In particular the language Parlog is concentrated upon because it is the most efficiently implementable of all the non-flat logic programming languages. Other issues arising from the implementation of a programming language are discussed, in particular the design of a load balancing scheme which deals with the large number of short-lived processes inherent in a correct language implementation. Detailed descriptions of all the data structures and synchronisation algorithms are also presented. Resource exhaustion is also discussed and handled within the implementations. Detailed performance results for both models of computation are given. From the material it should be possible for the reader to create an implementation.

Parallel architectures are no longer pure research vehicles, as they were some years ago. There are now many commercial systems competing for market segments in scientific computing. The 1990s are likely to become the decade of parallel processing. CONPAR 90 - VAPP IV is the joint successor meeting of two highly successful international conference series in the field of vector and parallel processing. This volume contains the 79 papers presented at the conference. The various topics of the papers include hardware, software and application issues. Some of the session titles best reflect the contents: new models of computation, logic programming, large-grain data flow, interconnection networks, communication issues, reconfigurable and scalable systems, novel architectures and languages, high performance systems and accelerators, performance prediction / analysis / measurement, performance monitoring and debugging, compile-time analysis and restructurers, load balancing, process partitioning and concurrency control, visualization and runtime analysis, parallel linear algebra, architectures for image processing, efficient use of vector computers, transputer tools and applications, array processors, algorithmic studies for hypercube-type systems, systolic arrays and algorithms. The volume gives a comprehensive view of the state of the art in a field of current interest.

This volume presents papers from the 2nd Scandinavian Workshop on Algorithm Theory. The contributions describe original research on algorithms and data structures, in all areas, including combinatorics, computational geometry, parallel computing, and graph theory. The majority of the papers focus on the design and complexity analysis of: data structures, text algorithms, and sequential and parallel algorithms for graph problems and for geometric problems. Examples of tech- niques presented include: - efficient ways to find approximation algorithms for the maximum independent set problem and for graph coloring; - exact estimation of the expected search cost for skip lists; - construction of canonical representations of partial 2-trees and partial 3-trees in linear time; - efficient triangulation of planar point sets and convex polygons.

Advances and problems in the field of compiler compilers are considered in this volume, which presents the proceedings of the third in a series of biannual workshops on compiler compilers. Selected papers address the topics of requirements, properties, and theoretical aspects of compiler compilers as well as tools and metatools for software engineering. The 23 papers cover a wide spectrum in the field of compiler compilers, ranging from overviews of new compiler compilers for generating quality compilers to special problems of code generation and optimization. Aspects of compilers for parallel systems and knowledge-based development tools are also discussed.

This volume presents the proceedings of a workshop on parallel database systems organized by the PRISMA (Parallel Inference and Storage Machine) project. The invited contributions by internationally recognized experts give a thorough survey of several aspects of parallel database systems. The second part of the volume gives an in-depth overview of the PRISMA system. This system is based on a parallel machine, where the individual processors each have their own local memory and communicate with each other over a packet-switched network. On this machine a parallel object-oriented programming language, POOL-X, has been implemented, which provides dedicated support for database systems as well as general facilities for parallel programming. The POOL-X system then serves as a platform for a complete relational main-memory database management system, which uses the parallelism of the machine to speed up significantly the execution of database queries. The presentation of the PRISMA system, together with the invited papers, gives a broad overview of the state of the art in parallel database systems.

This volume is a collection of papers on topics focused around concurrency, based on research work presented at the UK/Japan Workshop held at Wadham College, Oxford, September 25-27, 1989. The volume is organized into four parts: - Papers on theoretical aspects of concurrency which reflect strong research activities in the UK, including theories on CCS and temporal logic RDL. - Papers on object orientation and concurrent languages which reflect major research activities on concurrency in Japan. The languages presented include extensions of C, Prolog and Lisp as well as object-based concurrent languages. - Papers on parallel architectures and VLSI logic, including a rewrite rule machine, a graph rewriting machine, and a dataflow architecture. - An overview of the workshop including the abstracts of the talks and the list of participants. The appendix gives a brief report of the first UK/Japan Workshop in Computer Science, held at Sendai, Japan, July 6-9, 1987.

This book includes the papers presented at the Third International Workshop on Distributed Algorithms organized at La Colle-sur-Loup, near Nice, France, September 26-28, 1989 which followed the first two successful international workshops in Ottawa (1985) and Amsterdam (1987). This workshop provided a forum for researchers and others interested in distributed algorithms on communication networks, graphs, and decentralized systems. The aim was to present recent research results, explore directions for future research, and identify common fundamental techniques that serve as building blocks in many distributed algorithms. Papers describe original results in all areas of distributed algorithms and their applications, including: distributed combinatorial algorithms, distributed graph algorithms, distributed algorithms for control and communication, distributed database techniques, distributed algorithms for decentralized systems, fail-safe and fault-tolerant distributed algorithms, distributed optimization algorithms, routing algorithms, design of network protocols, algorithms for transaction management, composition of distributed algorithms, and analysis of distributed algorithms.

This work relates different approaches for the modelling of parallel processes. On the one hand there are the so-called "process algebras" or "abstract programming languages" with Milner's Calculus of Communicating Systems (CCS) and the theoretical version of Hoare's Communicating Sequential Processes (CSP) as main representatives. On the other hand there are machine models, i.e. the classical finite state automata (transition systems), for which, however, more discriminating notions of equivalence than equality of languages are used; and secondly, there are differently powerful types of Petri nets, namely safe and general (place/transition) nets respectively, and predicate/transition nets. Within a uniform framework the syntax and the operational semantics of CCS and TCSP are explained. We consider both, Milner's well-known interleaving semantics, which is based on infinite transition systems, as well as the new distributed semantics introduced by Degano et al., which is based on infinite safe nets. The main part of this work contains three syntax-driven constructions of transition systems, safe nets, and predicate/transition nets respectively. Each of them is accompanied by a proof of consistency. Due to intrinsic limits, which are also investigated here, neither for transition systems and finite nets, nor for general nets does a finite consistent representation of all CCS and TCSP programs exist. However sublanguages which allow finite representations are discerned. On the other hand the construction of predicate/transition nets is possible for all CCS programs in which every choice and every recursive body starts sequentially.

This volume presents the proceedings of a workshop at which major Parallel Lisp activities in the US and Japan were explained. Work covered includes Multilisp and Mul-T at MIT, Qlisp at Stanford, Lucid and Parcel at Illinois, PaiLisp at Tohoku University, Multiprocessor Lisp on TOP-1 at IBM Tokyo Research, and concurrent programming in TAO. Most papers present languages and systems of Parallel Lisp and are in particular concerned with: - Language constructs of Parallel Lisp and their meanings from the standpoint of implementing Parallel Lisp systems; - Some important technical issues such as parallel garbage collection, dynamic task partitioning, futures and continuations in parallelism, automatic parallelization of Lisp programs, and the kernel concept of Parallel Lisp. Some performance results are reported that suggest practical applicability of Parallel Lisp systems in the near future. Several papers on concurrent object-oriented systems are also included.

It was the aim of the conference to present issues in parallel computing to a community of potential engineering/scientific users. An overview of the state-of-the-art in several important research areas is given by leading scientists in their field. The classification question is taken up at various points, ranging from parametric characterizations, communication structure, and memory distribution to control and execution schemes. Central issues in multiprocessing hardware and operation, such as scalability, techniques of overcoming memory latency and synchronization overhead, as well as fault tolerance of communication networks are discussed. The problem of designing and debugging parallel programs in a user-friendly environment is addressed and a number of program transformations for enhancing vectorization and parallelization in a variety of program situations are described. Two different algorithmic techniques for the solution of certain classes of partial differential equations are discussed. The properties of domain-decomposition algorithms and their mapping onto a CRAY-XMP-type architecture are investigated and an overview is given of the merit of various approaches to exploiting the acceleration potential of multigrid methods. Finally, an abstract performance modeling technique for the behavior of applications on parallel and vector architectures is described.

Each week of this three week meeting was a self-contained event, although each had the same underlying theme - the effect of parallel processing on numerical analysis. Each week provided the opportunity for intensive study to broaden participants' research interests or deepen their understanding of topics of which they already had some knowledge. There was also the opportunity for continuing individual research in the stimulating environment created by the presence of several experts of international stature. This volume contains lecture notes for most of the major courses of lectures presented at the meeting; they cover topics in parallel algorithms for large sparse linear systems and optimization, an introductory survey of level-index arithmetic and superconvergence in the finite element method.

The papers collected in this volume are most of the material presented at the Advanced School on Mathematical Models for the Semantics of Parallelism, held in Rome, September 24- October 1, 1986. The need for a comprehensive and clear presentation of the several semantical approaches to parallelism motivated the stress on mathematical models, by means of which comparisons among different approaches can also be performed in a perspicuous way.

"WOPPLOT 86 - Workshop on Parallel Processing: Logic, " "Organization and Technology" - gathered together experts from various fields for a broad overview of current trends in parallel processing. There are contributions from logic (e.g., the connection between time and logic, or non-monotonic reasoning); from organizational structure theory (of great importance for pyramid architecture) and structure representation; from intrinsic parallelism and problem classification; from developments in future technologies (3-D Silicon technology, molecular electronics); and from various applications (pattern storage in adaptive memories, simulation of physical systems). The proceedings show clearly that progress in parallel processing is an interdisciplinary goal; they present a cross section of the state of the art as well as of future trends. Furthermore, some contributions (in particular, those from logic and organization) deserve a broader interest also outside the field of parallel processing.

The programming language Fortran dates back to 1957 when a team of IBM engineers released the first Fortran Compiler. During the past 60 years, the language had been revised and updated several times to incorporate more features to enable writing clean and structured computer programs. The present version is Fortran 2018. Since the dawn of the computer era, there had been a constant demand for a "larger" and "faster" machine. To increase the speed there are three hurdles. The density of the active components on a VLSI chip cannot be increased indefinitely and with the increase of the density heat dissipation becomes a major problem. Finally, the speed of any signal cannot exceed the velocity of the light. However, by using several inexpensive processors in parallel coupled with specialized software and hardware, programmers can achieve computing speed similar to a supercomputer. This book can be used to learn the modern Fortran from the beginning and the technique of developing parallel programs using Fortran. It is for anyone who wants to learn Fortran. Knowledge beyond high school mathematics is not required. There is not another book on the market yet which deals with Fortran 2018 as well as parallel programming. FEATURES Descriptions of majority of Fortran 2018 instructions Numerical Model String with Variable Length IEEE Arithmetic and Exceptions Dynamic Memory Management Pointers Bit handling C-Fortran Interoperability Object Oriented Programming Parallel Programming using Coarray Parallel Programming using OpenMP Parallel Programming using Message Passing Interface (MPI) THE AUTHOR Dr Subrata Ray, is a retired Professor, Indian Association for the Cultivation of Science, Kolkata.

Build your expertise in the BPF virtual machine in the Linux kernel with this practical guide for systems engineers. You'll not only dive into the BPF program lifecycle but also learn to write applications that observe and modify the kernel's behavior; inject code to monitor, trace, and securely observe events in the kernel; and more. Authors David Calavera and Lorenzo Fontana help you harness the power of BPF to make any computing system more observable. Familiarize yourself with the essential concepts you'll use on a day-to-day basis and augment your knowledge about performance optimization, networking, and security. Then see how it all comes together with code examples in C, Go, and Python. Write applications that use BPF to observe and modify the Linux kernel's behavior on demand Inject code to monitor, trace, and observe events in the kernel in a secure way-no need to recompile the kernel or reboot the system Explore code examples in C, Go, and Python Gain a more thorough understanding of the BPF program lifecycle

In view of the growing presence and popularity of multicore and manycore processors, accelerators, and coprocessors, as well as clusters using such computing devices, the development of efficient parallel applications has become a key challenge to be able to exploit the performance of such systems. This book covers the scope of parallel programming for modern high performance computing systems. It first discusses selected and popular state-of-the-art computing devices and systems available today, These include multicore CPUs, manycore (co)processors, such as Intel Xeon Phi, accelerators, such as GPUs, and clusters, as well as programming models supported on these platforms. It next introduces parallelization through important programming paradigms, such as master-slave, geometric Single Program Multiple Data (SPMD) and divide-and-conquer. The practical and useful elements of the most popular and important APIs for programming parallel HPC systems are discussed, including MPI, OpenMP, Pthreads, CUDA, OpenCL, and OpenACC. It also demonstrates, through selected code listings, how selected APIs can be used to implement important programming paradigms. Furthermore, it shows how the codes can be compiled and executed in a Linux environment. The book also presents hybrid codes that integrate selected APIs for potentially multi-level parallelization and utilization of heterogeneous resources, and it shows how to use modern elements of these APIs. Selected optimization techniques are also included, such as overlapping communication and computations implemented using various APIs. Features: Discusses the popular and currently available computing devices and cluster systems Includes typical paradigms used in parallel programs Explores popular APIs for programming parallel applications Provides code templates that can be used for implementation of paradigms Provides hybrid code examples allowing multi-level parallelization Covers the optimization of parallel programs

Almost all software solutions are developed through academic research and implemented only in prototype machines leaving the field of software techniques for maintaining the cache coherence widely open for future research and development. This book is a collection of all the representative approaches to software coherence maintenance including a number of related efforts in the performance evaluation field.
The book presents a selection of 27 papers dealing with state-of-the-art software solutions for cache coherence maintenance in shared-memory multiprocessors. It begins with a set of four introductory readings that provides a brief overview of the cache coherence problem and introduces software solutions to the problem. The text defines and illustrates static and dynamic software schemes, techniques for modeling performance evaluation mechanisms, and performance evaluation studies.
The book is intended for the experienced reader in computer engineering but possibly a novice in the topic of cache coherence. It also provides an in-depth understanding of the problem as well as a comprehensive overview for multicomputer designers, computer architects, and compiler writers. In addition, it is a software coherence reference handbook for advanced undergraduate and typical graduate students in multiprocessing and multiprogramming areas.

"This book is required reading for anyone working with accelerator-based computing systems." -From the Foreword by Jack Dongarra, University of Tennessee and Oak Ridge National Laboratory CUDA is a computing architecture designed to facilitate the development of parallel programs. In conjunction with a comprehensive software platform, the CUDA Architecture enables programmers to draw on the immense power of graphics processing units (GPUs) when building high-performance applications. GPUs, of course, have long been available for demanding graphics and game applications. CUDA now brings this valuable resource to programmers working on applications in other domains, including science, engineering, and finance. No knowledge of graphics programming is required-just the ability to program in a modestly extended version of C. CUDA by Example, written by two senior members of the CUDA software platform team, shows programmers how to employ this new technology. The authors introduce each area of CUDA development through working examples. After a concise introduction to the CUDA platform and architecture, as well as a quick-start guide to CUDA C, the book details the techniques and trade-offs associated with each key CUDA feature. You'll discover when to use each CUDA C extension and how to write CUDA software that delivers truly outstanding performance. Major topics covered include Parallel programming Thread cooperation Constant memory and events Texture memory Graphics interoperability Atomics Streams CUDA C on multiple GPUs Advanced atomics Additional CUDA resources All the CUDA software tools you'll need are freely available for download from NVIDIA. http://developer.nvidia.com/object/cuda-by-example.html

Focusing on algorithms for distributed-memory parallel architectures, Parallel Algorithms presents a rigorous yet accessible treatment of theoretical models of parallel computation, parallel algorithm design for homogeneous and heterogeneous platforms, complexity and performance analysis, and essential notions of scheduling. The book extracts fundamental ideas and algorithmic principles from the mass of parallel algorithm expertise and practical implementations developed over the last few decades.

In the first section of the text, the authors cover two classical theoretical models of parallel computation (PRAMs and sorting networks), describe network models for topology and performance, and define several classical communication primitives. The next part deals with parallel algorithms on ring and grid logical topologies as well as the issue of load balancing on heterogeneous computing platforms. The final section presents basic results and approaches for common scheduling problems that arise when developing parallel algorithms. It also discusses advanced scheduling topics, such as divisible load scheduling and steady-state scheduling.

With numerous examples and exercises in each chapter, this text encompasses both the theoretical foundations of parallel algorithms and practical parallel algorithm design.