DOCSLIB.ORG

Language Constructs for Modular Parallel Programs 1

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Language Constructs for Modular Parallel Programs 1 Preprint MCSP Mathematics and Computer Science Division Argonne National Lab oratory Argonne Ill January Language Constructs for Mo dular Parallel Programs Ian Foster Mathematics and Computer Science Division Argonne National Lab oratory Argonne IL USA Abstract We describ e programming language constructs that facilitate the application of mo dular design techniques in parallel programming These constructs allow us to iso late resource management and pro cessor scheduling decisions from the sp ecication of individual mo dules which can themselves encapsulate design decisions concerned with concurrency communication pro cess mapping and data distribution This ap proach p ermits development of libraries of reusable parallel program comp onents and the reuse of these comp onents in dierent contexts In particular alternative map ping strategies can b e explored without mo difying other asp ects of program logic We describ e how these constructs are incorp orated in two practical parallel programming languages PCN and Fortran M Compilers have b een developed for b oth languages allowing exp erimentation in substantial applications Keywords mo dularity parallel programming programming languages program com p osition co de reuse virtual computer Introduction In sequential programming mo dular and ob jectoriented design and programming tech niques are well understo o d and widely used to reduce complexity p ermit separate devel opment of comp onents and encourage reuse In parallel programming the situation is less advanced Parallel programs are for the most part developed in an adho c fashion as monolithic entities that cannot easily b e adapted to changing circum stances Co de reuse is rare outside the sp ecialized context of singleprogram multipledata SPMD programming where the same program is run on every pro cessor and libraries can b e called to p erform common global op erations This pap er presents programming language constructs that p ermit the b enets of mo d ularity to b e realized in parallel programs The central ideas are as follows First pro cess and data placement decisions within an individual mo dule are sp ecied with resp ect to a virtual computer the embedding of this virtual computer within a physical or another virtual computer is sp ecied when the mo dule is invoked This approach p ermits resource management and lo cality decisions to b e separated from the sp ecications of individual mo dules and developed in a hierarchical fashion by using stepwise renement techniques Second virtual computer constructs are incorp orated into compositional programming languages in which concurrency is sp ecied with explicit parallel constructs and inter actions b etween concurrent pro cesses are restricted so that neither physical lo cation nor execution schedule aects the result of a computation Languages with this prop erty include Strand and PCN in which pro cesses interact by reading and writing shared singleassignment variables and Fortran M in which interactions o ccur via singlereader singlewriter virtual channels Comp ositional languages p ermit design decisions concerned with mapping data distribution communication and scheduling to b e made separately and to b e mo died without changing other asp ects of a design In this pap er we show how virtual computer constructs are integrated into two com p ositional parallel programming languages PCN and Fortran M PCN is a Clike language that integrates ideas from concurrent logic programming and imp erative programming Fortran M is a small set of extensions to Fortran In each case virtual com puters and related constructs have b een introduced in a manner that is consistent with the base language hence simplifying b oth comprehension and compilation Both PCN and Fortran M have b een used to develop a range of substantial parallel applications these provide an empirical basis for evaluation of the concepts In summary the contributions of this pap er are as follows The denition of programming language constructs that facilitate the mo dular con struction of parallel programs The instantiation of the constructs in practical parallel programming languages Empirical evaluation in a range of applications The next three sections of this pap er address each of these issues in turn after which we review related work and present our conclusions Mo dularity and Parallel Programs We use the term mo dularity to refer to two related programstructuring techniques Stepwise renement the decomp osition of a complex problem into simpler subproblems each solved by a separate mo dule with a welldened interface Mo dular decomp osition the isolation in separate mo dules of design decisions that are dicult likely to change or common to several program comp onents The rst of these techniques is more naturally applied top down in the design pro cess and if care is taken to ensure generality pro duces mo dules that can b e reused in dierent contexts The second is more naturally applied bottom up and can reduce the cost of program mo dications Both techniques are central to ob jectoriented design It is instructive to examine how these techniques can b e applied in sequential and paral lel programming For illustrative purp oses we consider a simple example the convolution op eration used for example in motion estimation algorithms in image pro cessing Each pair of images is rst pro cessed by using a twodimensional fast Fourier transform FFT The resulting matrices are then multiplied and an inverse FFT is applied to the result The various op erations and the dep endencies b etween them are illustrated in Figure FFT MUL FFT -1 FFT Figure Convolution Algorithm Structure In a sequential implementation stepwise renement may identify FFT matrix multi plication input and output mo dules The data structures to b e input to and output from each mo dule are sp ecied in its interface internal data structures and algorithmic details are encapsulated and not visible to other comp onents The interface sp ecies the size and p erhaps the type of the data structures to b e op erated on hence allowing the mo dule to b e used in dierent contexts For example the same FFT mo dule can b e called three times in the convolution example The interface design may also address eciency issues for example by sp ecifying that data is to b e passed by reference rather than by value The same decomp osition can b e used in a parallel implementation Now however we need to encapsulate not only data and computation but also concurrency communication pro cess mapping and data distribution If we do not encapsulate this information two mo dules mapp ed to the same pro cessor might interfere for example if one received mes sages intended for the other When comp osing two mo dules we prefer not to b e aware of how each maps computation and data The mappings used in dierent mo dules can b e made conformant to improve p erformance but should not aect correctness Again eciency issues need to b e addressed in the interface design for example by allowing for the exchange of distributed data structures b etween parallel mo dules In a sequential implementation mo dular decomp osition may identify the representa tion of matrices as common to several program comp onents Hence we dene a mo dule that provides the required matrix op erations read an element write an element etc The rest of the program uses these functions to access matrices alternative matrix rep resentations eg linear storage or a linkedlist representation of sparse matrices can b e substituted without changes to other comp onents A design decision of equal signicance in a parallel implementation is how computation and data are mapp ed to pro cessors For example the various phases of the convolution problem illustrated in Figure can b e organized so as to execute a on disjoint sets of pro cessors as a pip eline b in sequence for each input data set or c as some hybrid of these two approaches Figure We wish to lo calize the changes required to explore these alternatives which do not aect the basic structure of the algorithm but may have dierent p erformance characteristics These changes concern the allo cation of computa tional resources pro cessors to mo dules and the scheduling of comp onents mapp ed to the same pro cessor (a) (b) (c) Figure Alternative Mappings for the Parallel Convolution Algorithm We have found that the design issues encountered in the convolution example are common to many parallel programming problems In summary the most imp ortant issues are as follows Parallel mo dules need to encapsulate not only computation and data as in sequential programming but also communication concurrency pro cess mapping and data distribution Resource allo cation and scheduling decisions need to b e isolated from mo dule sp ec ications Interfaces b etween parallel mo dules need to allow for the sharing of distributed data structures We now present the programming language concepts that we use to address these issues Virtual Computers We remove resource allo cation decisions from individual mo dules by sp ecifying pro cess and data mapping within a mo dule relative to a virtual computer The shap e size and mapping of this virtual computer to physical pro cessors are sp ecied externally to the mo dule This mapping need not b e one to one several virtual computer no des can b e lo cated on a single physical pro cessor or a virtual computer can corresp ond to a subset of a physical computer For example a parallel FFT mo dule might b e sp ecied with
Load more

Recommended publications
  • Object-Oriented Programming in Scheme with First-Class Modules
    Ob jectOriented Programming in Scheme with FirstClass Mo dules and Op eratorBased Inheritance Guruduth Banavar Gary Lindstrom Department of Computer Science University of Utah Salt LakeCityUT Abstract Wecharacterize ob jectoriented programming as structuring and manipulating a uniform space of rstclass values representing modules a distillation of the notion of classes Op erators over mo dules individually achieve eects such as encapsulation sharing and static binding A variety of idioms of OO programming nd convenient expression within this mo del including several forms of single and multiple inheritance abstract classes class variables inheritance hierarchy combination and reection Weshow that this programming style simplies OO programming via enhanced uniformity and supp orts a exible mo del of ob jectorientation that provides an attractive alternative to metaprogramming Finallyweshow that these notions of OO programming are language indep endent by implementing a Mo dular Scheme prototyp e as a completion of a generic OO framework for mo dularity Pap er Category Research Topic Area Language design and implementation Intro duction Classbased ob jectoriented programming is usually thought of as creating a graph structured inher itance hierarchy of classes instantiating some of these classes and computing with these instances Instances are typically rstclass values in the language ie they can b e created stored accessed and passed into and out of functions Classes on the other hand are usually not rstclass values and inheritance is
    [Show full text]
  • Modular Programming with Functions
    CHAPTER 4 MODULAR PROGRAMMING WITH FUNCTIONS Copyright © 2013 Pearson Education, Inc. Modularity •A program may also contain other functions, and it may refer to functions in another file or in a library. These functions , or modules , are sets of statements that perform an operation or compute a value •To maintain simplicity and readability in long and complex programs, we use a short main, and other functions instead of using one long main function. •By separating a solution into a group of modules, each module is easier to understand, thus adhering to the basic guidelines of structured programming Copyright © 2013 Pearson Education, Inc. Modularity •Braking a problem into a set of modules has many advantages: 1. Every module can be written and tested separately from the rest of the program 2. A module is smaller than a complete program, so testing is easier 3. Once a module has been tested, it can be used in new program without having to retest it ( reusability ) 4. Use of modules ( modularity ) usually reduces the overall length of programs 5. Several programmers can work on the same project if it is separated into modules Copyright © 2013 Pearson Education, Inc. Modularity Main Modules Copyright © 2013 Pearson Education, Inc. Function Definition •A function consists of a definition statement followed by declarations and statements. The general form of a function is: return_type function_name(parameter_declarations) { declarations; statements; return expression; } •The parameter declarations represent the information passed to the function •Additional variables used by the function are defined in declarations statement •All functions should include a return statement Copyright © 2013 Pearson Education, Inc.
    [Show full text]
  • Scope in Fortran 90
    Scope in Fortran 90 The scope of objects (variables, named constants, subprograms) within a program is the portion of the program in which the object is visible (can be use and, if it is a variable, modified). It is important to understand the scope of objects not only so that we know where to define an object we wish to use, but also what portion of a program unit is effected when, for example, a variable is changed, and, what errors might occur when using identifiers declared in other program sections. Objects declared in a program unit (a main program section, module, or external subprogram) are visible throughout that program unit, including any internal subprograms it hosts. Such objects are said to be global. Objects are not visible between program units. This is illustrated in Figure 1. Figure 1: The figure shows three program units. Main program unit Main is a host to the internal function F1. The module program unit Mod is a host to internal function F2. The external subroutine Sub hosts internal function F3. Objects declared inside a program unit are global; they are visible anywhere in the program unit including in any internal subprograms that it hosts. Objects in one program unit are not visible in another program unit, for example variable X and function F3 are not visible to the module program unit Mod. Objects in the module Mod can be imported to the main program section via the USE statement, see later in this section. Data declared in an internal subprogram is only visible to that subprogram; i.e.
    [Show full text]
  • A Parallel Program Execution Model Supporting Modular Software Construction
    A Parallel Program Execution Model Supporting Modular Software Construction Jack B. Dennis Laboratory for Computer Science Massachusetts Institute of Technology Cambridge, MA 02139 U.S.A. [email protected] Abstract as a guide for computer system design—follows from basic requirements for supporting modular software construction. A watershed is near in the architecture of computer sys- The fundamental theme of this paper is: tems. There is overwhelming demand for systems that sup- port a universal format for computer programs and software The architecture of computer systems should components so users may benefit from their use on a wide reflect the requirements of the structure of pro- variety of computing platforms. At present this demand is grams. The programming interface provided being met by commodity microprocessors together with stan- should address software engineering issues, in dard operating system interfaces. However, current systems particular, the ability to practice the modular do not offer a standard API (application program interface) construction of software. for parallel programming, and the popular interfaces for parallel computing violate essential principles of modular The positions taken in this presentation are contrary to or component-based software construction. Moreover, mi- much conventional wisdom, so I have included a ques- croprocessor architecture is reaching the limit of what can tion/answer dialog at appropriate places to highlight points be done usefully within the framework of superscalar and of debate. We start with a discussion of the nature and VLIW processor models. The next step is to put several purpose of a program execution model. Our Parallelism processors (or the equivalent) on a single chip.
    [Show full text]
  • The Best of Both Worlds?
    The Best of Both Worlds? Reimplementing an Object-Oriented System with Functional Programming on the .NET Platform and Comparing the Two Paradigms Erik Bugge Thesis submitted for the degree of Master in Informatics: Programming and Networks 60 credits Department of Informatics Faculty of mathematics and natural sciences UNIVERSITY OF OSLO Autumn 2019 The Best of Both Worlds? Reimplementing an Object-Oriented System with Functional Programming on the .NET Platform and Comparing the Two Paradigms Erik Bugge © 2019 Erik Bugge The Best of Both Worlds? http://www.duo.uio.no/ Printed: Reprosentralen, University of Oslo Abstract Programming paradigms are categories that classify languages based on their features. Each paradigm category contains rules about how the program is built. Comparing programming paradigms and languages is important, because it lets developers make more informed decisions when it comes to choosing the right technology for a system. Making meaningful comparisons between paradigms and languages is challenging, because the subjects of comparison are often so dissimilar that the process is not always straightforward, or does not always yield particularly valuable results. Therefore, multiple avenues of comparison must be explored in order to get meaningful information about the pros and cons of these technologies. This thesis looks at the difference between the object-oriented and functional programming paradigms on a higher level, before delving in detail into a development process that consisted of reimplementing parts of an object- oriented system into functional code. Using results from major comparative studies, exploring high-level differences between the two paradigms’ tools for modular programming and general program decompositions, and looking at the development process described in detail in this thesis in light of the aforementioned findings, a comparison on multiple levels was done.
    [Show full text]
  • Software Quality / Modularity
    Software quality EXTERNAL AND INTERNAL FACTORS External quality factors are properties such as speed or ease of use, whose presence or absence in a software product may be detected by its users. Other qualities applicable to a software product, such as being modular, or readable, are internal factors, perceptible only to computer professionals who have access to the actual software text. In the end, only external factors matter, but the key to achieving these external factors is in the internal ones: for the users to enjoy the visible qualities, the designers and implementers must have applied internal techniques that will ensure the hidden qualities. EXTERNAL FACTORS Correctness: The ability of software products to perform their tasks, as defined by their specification. Correctness is the prime quality. If a system does not do what it is supposed to do, everything else about it — whether it is fast, has a nice user interface¼ — matters little. But this is easier said than done. Even the first step to correctness is already difficult: we must be able to specify the system requirements in a precise form, by itself quite a challenging task. Robustness: The ability of software systems to react appropriately to abnormal conditions. Robustness complements correctness. Correctness addresses the behavior of a system in cases covered by its specification; robustness characterizes what happens outside of that specification. There will always be cases that the specification does not explicitly address. The role of the robustness requirement is to make sure that if such cases do arise, the system does not cause catastrophic events; it should produce appropriate error messages, terminate its execution cleanly, or enter a so-called “graceful degradation” mode.
    [Show full text]
  • Chapter-2 Object Oriented Programming Very Short/ Short Answer Questions 1
    CHAPTER-2 OBJECT ORIENTED PROGRAMMING VERY SHORT/ SHORT ANSWER QUESTIONS 1. Discuss major OOP concepts briefly. Ans. Following are the general OOP concepts: 1. Data Abstraction: Data abstraction means, providing only essential information to the outside word and hiding their background details i.e. to represent the needed information in program without presenting the details. 2. Data Encapsulation: The wrapping up of data and operations/functions (that operate o the data) into a single unit (called class) is known as Encapsulation. 3. Modularity: Modularity is the property of a system that has been decomposed into a set of cohesive and loosely coupled modules. 4. Inheritance: Inheritance is the capability of one class of things to inherit capabilities or properties from another class. 5. Polymorphism: Polymorphism is the ability for a message or data to be processed in more than one form. 2. What are programming paradigms? Give names of some popular programming paradigms. Ans. Programming Paradigm: A Programming Paradigm defines the methodology of designing and implementing programs using the key features and building blocks of a programming language. Following are the different programming paradigms: (i) Procedural Programming (ii) Object Based Programming (iii) Object Oriented Programming 3. What are the shortcomings of procedural and modular programming approaches? Ans. Following are the various shortcomings of procedural and modular programming approaches: Procedural Programming is susceptible to design changes. Procedural Programming leads to increased time and cost overheads during design changes. Procedural and Modular programming both are unable to represent real world relationship that exists among objects. In modular programming, the arrangement of the data can’t be changed without modifying all the functions that access it.
    [Show full text]
  • Modular Programming
    Modular Programming Prof. Clarkson Fall 2016 Today’s music: "Giorgio By Moroder" by Daft Punk Te Moog modular synthesizer Review Previously in 3110: • Functions, data • lots of language features • how to build small programs Today: • language features for building large programs: structures, signatures, modules Question What’s the largest program you’ve ever worked on, by yourself or as part of a team? A. 10-100 LoC B. 100-1,000 LoC C. 1,000-10,000 LoC D. 10,000-100,000 LoC E. 100,000 LoC or bigger Scale • My solution to A1: 100 LoC • My solution to A2: 300 LoC • OCaml: 200,000 LoC • Unreal engine 3: 2,000,000 LoC • Windows Vista: 50,000,000 LoC http://www.informationisbeautiful.net/visualizations/million-lines-of-code/ ...can’t be done by one person ...no individual programmer can understand all the details ...too complex to build with subset of OCaml we’ve seen so far Modularity Modular programming: code comprises independent modules – developed separately – understand behavior of module in isolation – reason locally, not globally Java features for modularity • classes, packages: organize identifiers (classes, methods, fields, etc.) into namespaces • interfaces: describe related classes • public, protected, private: control what is visible outside a namespace • subtyping, inheritance: enables code reuse OCaml features for modularity • structures: organize identifiers (functions, values, etc.) into namespaces • signatures: describe related modules • abstract types: control what is visible outside a namespace • functors, includes: enable
    [Show full text]
  • Modularity and Reuse of Domain-Specific Languages
    Modularity and reuse of domain-specific languages Citation for published version (APA): Sutii, A. M. (2017). Modularity and reuse of domain-specific languages: an exploration with MetaMod. Technische Universiteit Eindhoven. Document status and date: Published: 07/11/2017 Document Version: Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers) Please check the document version of this publication: • A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website. • The final author version and the galley proof are versions of the publication after peer review. • The final published version features the final layout of the paper including the volume, issue and page numbers. Link to publication General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying the publication in the public portal. If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement: www.tue.nl/taverne Take down policy If you believe that this document breaches copyright please contact us at: [email protected] providing details and we will investigate your claim.
    [Show full text]
  • A Module System for C++ (Revision 2)
    Document number: N4214 Date: 2014-10-13 Working group: Modules SG Reply to: [email protected] A Module System for C++ (Revision 2) Gabriel Dos Reis Mark Hall Gor Nishanov Abstract We present a design of and a specification of module system for C++. The proposal focuses on programmer’s view of modules (both production and consumption) and how to better support modular programming in the large, componentization, scalable compilation, and semantics-aware devel- oper tools. Contents 1 Introduction ......... 1 5 Tools Support ......... 20 2 The Problems ......... 2 6 Build Systems ........ 21 3 Goals and Principles ..... 4 7 Migration ........... 21 4 Design Choices ........ 7 A Standardese ......... 21 1 Introduction The lack of direct language support for componentization of C++ libraries and programs, combined with increasing use of templates, has led to serious impedi- ments to compile-time scalability, and programmer productivity. It is the source of lackluster build performance and poor integration with cloud and distributed build systems. Furthermore, the heavy-reliance on header file inclusion (i.e. copy-and- paste from compilers’ perspective) and macros stifle flowering of C++ developer tools in increasingly semantics-aware development environments. Responding to mounting requests from application programmers, library de- velopers, tool builders alike, this report proposes a module system for C++ with a handful of clearly articulated goals. The proposal is informed by the current state of the art regarding module systems in contemporary programming languages, past 1 suggestions [4, 6] and experiments such as Clang’s [2, 5], and practical constraints specific to C++ ecosystem, deployment, and use. The design is minimalist; yet, it aims for a handful of fundamental goals 1.
    [Show full text]
  • 006-31: Modular Programming Using SAS/AF® and SAS®
    SUGI 31 Applications Development Paper 006-31 Modular Programming using AF/SCL Kevin Graham, Montura, San Francisco, California ABSTRACT Build modular SAS Frame applications using advanced programming techniques and features available in SAS SCL. Easily separate critical logic and business rules from your code base without creating tangled source code. Isolate and integrate decision-logic from task-logic without separating your brain from your sanity. INTRODUCTION Object programming is the science of program structure and organization. The goal in object programming is to produce modular and reusable code. The problem is that applications can be object-oriented but not modular. Rules are added, removed, or modified frequently and some changes can affect multiple sections of the application. The implementation of rules is what usually kills program modularity. This tutorial shows how to make a modular business rule. WHAT IS OBJECT PROGRAMMING? Objects and SAS/Macro actually have something in common – they both serve as a structure for the source code that is a SAS program. Thinking of the object, and macro, as simple structures used to encapsulate a SAS program will make this tutorial easier to understand. The two most important aspects in object programming are “separation of concerns” and “integration of concerns”. During the separation phase, the logic programmers call “decision logic” and “task logic” are physically separated into different programs. During the integration phase, business rule programs are dynamically attached to the other modules. This allows a rule to perform complex operations on the application without adding complexity. STRUCTURES A structure is composed of statements that mark the beginning and ending of a program.
    [Show full text]
  • The Circumstances in Which Modular Programming Becomes the Favor Choice by Novice Programmers
    I.J. Modern Education and Computer Science, 2018, 7, 1-12 Published Online July 2018 in MECS (http://www.mecs-press.org/) DOI: 10.5815/ijmecs.2018.07.01 The Circumstances in which Modular Programming becomes the Favor Choice by Novice Programmers Ilana Lavy Department of Information Systems, Yezreel Valley College, Israel Email: [email protected] Rashkovits Rami Department of Information Systems, Yezreel Valley College, Israel Email: [email protected] Received: 16 May 2018; Accepted: 04 June 2018; Published: 08 July 2018 Abstract—One of the key indicators for testing code quality is the level of modularity. Nevertheless, novice I. INTRODUCTION programmers do not always stick to writing modular code. In this study, we aim to examine the circumstances in The principle of modular design is defined [1] as which novice programmers decide to do so. To address splitting up large computation into a collection of small, this aim, two student groups, twenty each, were given a nearly independent, specialized sub-processes. In other programming assignment, each in a different set-up. The words, modular programming is a design principle that first group was given the assignment in several stages, divides the functionality of a program into modules each each add complexity to the previous one, while the responsible to one aspect of the desired functionality. The second group was given the entire assignment at once. modules are independent and can be easily replaced or The students' solutions were analyzed using the dual- extended with minimal changes in other modules [2]. process theory, cognitive dissonance theory and content Modularity hence supports gradual development, easy analysis methods to examine the extent of modularity.
    [Show full text]