Literatura académica sobre el tema "Fortran"

Thesis advisor: Richard Spinello<br>Thesis advisor: Stephen Pope<br>The information revolution is the single most important change in society since the printing press. At no other time has our society benefited from nearly limitless access to information and communication, and the changes brought on by this access have changed, and are changing, society in fundamental ways. Our homes, our workplaces, and our democracy have all been transformed by information and communication technologies.However, our ethics have not kept pace with our technological progress, and the immense changes brought by this this revolution have posed some equally immense moral questions. Indeed, there has been an almost total lack of religious discourse regarding the problems which have arisen out of the information revolution. This thesis is an attempt to change that.The thesis itself is structured as a series of essays on four key problems:1. Intellectual Property, as it relates to Scripture2. Information Ethics, the novel moral theory arising from Computer Ethics3. Robotic Ethics, especially the ethics of robotic warfare4. Hacker Culture, and its implicit spiritualityEach of these essays attempts to tackle one of these key problems, and demonstrate how a religious perspective illuminates some aspect of it. As befits a thesis from a Jesuit, Catholic university, most of the essays are drawn from a Catholic, Christian perspective<br>Thesis (BA) — Boston College, 2011<br>Submitted to: Boston College. College of Arts and Sciences<br>Discipline: Theology Honors Program<br>Discipline: Philosophy Honors Program<br>Discipline: Theology

With the advent of Distributed Memory Machines (DMMs) numerous work have been undertaken to ease the work of a programmer these systems. Data parallel languages like Fortran D, Vienna Fortran, High Performance Fortran and C+ allow the user to specify data distribution across processor with some directives, and the compiler for these language use the directives to compile the programme in to an SPMD code. There are number of old program which are still in use and rewriting them in to new data parallel languages is a costly effort. Most of the work on these parallelizing compilers concentrate on efficient data communication between the processors.With the advancement in technology, data communication time is also decreasing.This allows bigger programs to execute in the same time span.The resources of a DMM being finite puts a limit on the size of the problem that can be run. Improving the memory usage for a problem will hence allow us run bigger size problems. Further, as communication speed increases, the overhead caused by house-keeping computations like global index to local index transformation, and owner processor computation will degrade the performance of the resultant code. Hence a uniform and efficient method for these computations also becomes a necessity. We have implemented parallelizing parts of a compiler using the SUIF compiler system, which accepts programs written in Fortran77 with directives to the compiler as comments. The output of the compiler is an SPMD C program, with embedded PVM calls for message communication between the processors. We have also proposed algorithms to improve data communications,and minimizing memory usage in the output code. A uniform method for performing owner processor computations and global-to-local transformations has also been implemented.

With the advent of Distributed Memory Machines (DMMs) numerous work have been undertaken to ease the work of a programmer these systems. Data parallel languages like Fortran D, Vienna Fortran, High Performance Fortran and C+ allow the user to specify data distribution across processor with some directives, and the compiler for these language use the directives to compile the programme in to an SPMD code. There are number of old program which are still in use and rewriting them in to new data parallel languages is a costly effort. Most of the work on these parallelizing compilers concentrate on efficient data communication between the processors.With the advancement in technology, data communication time is also decreasing.This allows bigger programs to execute in the same time span.The resources of a DMM being finite puts a limit on the size of the problem that can be run. Improving the memory usage for a problem will hence allow us run bigger size problems. Further, as communication speed increases, the overhead caused by house-keeping computations like global index to local index transformation, and owner processor computation will degrade the performance of the resultant code. Hence a uniform and efficient method for these computations also becomes a necessity. We have implemented parallelizing parts of a compiler using the SUIF compiler system, which accepts programs written in Fortran77 with directives to the compiler as comments. The output of the compiler is an SPMD C program, with embedded PVM calls for message communication between the processors. We have also proposed algorithms to improve data communications,and minimizing memory usage in the output code. A uniform method for performing owner processor computations and global-to-local transformations has also been implemented.

Um editor orientado a uma linguagem de programação explora conhecimentos sobre a mesma para conduzir o processo de edição e prover um auxílio mais efetivo ao usuário. Este auxílio geralmente é expresso por um maior controle sob a forma do texto editado e pela detecção de erros no mesmo. Estas capacidades requerem que editores desta natureza possuam uma estrutura mais complexa: usualmente eles resultam da composição entre um editor de textos comum e analisadores léxico, sintático e semântico para a linguagem alvo. Este trabalho trata da concepção e implementação de um editor orientado a FORTRAN 77. Este editor é uma das ferramentas componentes do Ambiente de Reestruturação de Programas do Projeto Computação Científica do CTA-IEAv. A necessidade de manutenção de informações atualizadas sobre a árvore sintática e sobre os erros detectados nos programas editados exige que o editor recompile-os freqüentemente. Mostra-se que a utilização de princípios convencionais de compilação deteriora o desempenho do editor na medida em que o tamanho dos programas editados aumenta. Como alternativa, são propostos algoritmos incrementais que reaproveitam informações de análises anteriores para reduzir o tempo gasto na compilação. Como conseqüência, impõe-se uma nova estrutura ao compilador, onde aspectos livres e dependentes de contexto são tratados separadamente.

Refactoring is a software engineering technique that aims at improving the internal structure of an application, in such a way that the changes do not interfere in the results produced by the software. This technique is permanently employed in the software life cycle and refers to non-functional characteristics as legibility and performance. Most refactoring techniques currently apply to object-oriented systems and are widely available in integrated development environments for this programming paradigm. In high performance computing applications, code refactoring is a little-explored technique, as a great amount of legacy high performance code was written before the widespread use of object-oriented languages. The Fortran language (FORmula TRANslator) is heavily used in high performance applications, but has a few and limited tools for code refactoring. In such context, our work explores this open research area through the development of new techniques for refactoring Fortran source code. Our development starts from identifying refactoring opportunities for Fortran code and extends the Photran tool, which is an Eclipse IDE plugin for Fortran programming. The new automated refactoring techniques are applied to third-party code written in Fortran, in order to evaluate their impact on the software performance.<br>Refatoração é uma técnica de engenharia de software que consiste em alterar a estrutura interna de uma aplicação sem que tais alterações interfiram nos resultados produzidos pela mesma. Trata-se de uma tarefa permanentemente presente no ciclo de vida de uma aplicação e está diretamente associada às características não funcionais do software, como legibilidade e desempenho. Técnicas de refatoração são amplamente utilizadas em sistemas desenvolvidos para o paradigma da orientação a objetos e estão presentes de forma automatizada em diversas ferramentas que atuam neste paradigma. Na computação de alto desempenho, a refatoração de código é pouco explorada, principalmente em função de que boa parte do código legado de programas de alto desempenho está escrita em linguagens anteriores ao paradigma da orientação a objetos. A linguagem Fortran (FORmula TRANslator), largamente utilizada em aplicações de alto desempenho, possui poucas e limitadas ferramentas para refatoração de código. Neste contexto, este trabalho explora essa deficiência através da automatização de técnicas de refatoração, utilizando-se do framework da ferramenta Photran (um plugin para edição de código Fortran integrado ao IDE Eclipse). Partindo-se da identificação de oportunidades de refatoração para código Fortran, algumas técnicas são desenvolvidas e integradas à ferramenta Photran. As técnicas automatizadas são utilizadas em aplicações escritas nesta linguagem, de forma a avaliar seu impacto no desempenho das mesmas.

The high-level array programming language MATLAB is widely used for prototyping algorithms and applications of scientific computations. However, its dynamically-typed nature, which means that MATLAB programs are usually executed via an interpreter, leads to poor performance. An alternative approach would be converting MATLAB programs to equivalent Fortran 95 programs. The resulting programs could be compiled using existing high-performance Fortran compilers and thus could provide better performance. This thesis presents techniques that are developed for our MATLAB-to-Fortran compiler, McFor, for extracting information from the high-level semantics of MATLAB programs to produce efficient and reusable Fortran code.<br>The McFor compiler includes new type inference techniques for inferring intrinsic type and shape of variables and uses a value-propagation analysis to precisely estimate the sizes of arrays and to eliminate unnecessary array bounds checks and dynamic reallocations. In addition to the techniques for reducing execution overhead, McFor also aims to produce programmer-friendly Fortran code. By utilizing Fortran 95 features, the compiler generates Fortran code that keeps the original program structure and preserves the same function declarations.<br>We implemented the McFor system and experimented with a set of benchmarks with different kinds of computations. The results show that the compiled Fortran programs perform better than corresponding MATLAB executions, with speedups ranging from 1.16 to 102, depending on the characteristics of the program.<br>Le langage de programmation de tableaux de haut niveau MATLAB est largement utilisé afin de faire du prototypage d'algorithmes et des applications de calculs scientifiques. Cependant, sa nature de type dynamique, ce qui veut dire que les programmes MATLAB sont habituellement exécutés par un interpréteur, amène une mauvaise performance. Une approche alternative serait de convertir les programmes MATLAB aux programmes Fortran 95 équivalents. Les programmes résultants pourraient être compilés en utilisant les compilateurs de haute performance Fortran, ainsi ils peuvent fournir une meilleure performance. Cette thèse présente les techniques qui sont développées pour notre compilateur MATLAB-à-Fortran, McFor, pour extraire l'information des hauts niveaux des sémantiques des programmes MATLAB afin de produire un code Fortran efficace et réutilisable.<br>Le compilateur McFor inclut de nouvelles techniques de déduction pour inférer les types et formes intrinsèques des variables et utilise une analyse à propagation de valeurs pour estimer avec précision la tailles des tableaux de variables et pour éliminer les vérifications des limites et les réallocations dynamiques superflues de ces tableaux. En plus de ces techniques de réduction des temps d'exécution, McFor vise aussi a génèrer du code Fortran convivial pour les développeurs. En utilisant les avantages de Fortran 95, le compilateur génère du code Fortran qui préserve la structure originale du programme ainsi que les mêmes déclarations de fonctions.<br>Nous avons mis en oeuvre le système McFor et l'avons expérimenté avec un ensemble de tests de performance avec différentes sortes de calculs. Les résultats montrent que les programmes de Fortran compilés offrent une meilleure performance que les exécutions MATLAB correspondantes, avec une cadence accélérée de l'ordre de 1.16 à 102, selon les caractéristiques du programme.