PLH : uma linguagem de programação baseada estritamente em tarefas

• Ano de defesa: 2012
• Autor(a) principal: Hübner, Rodrigo
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Estadual de Maringá; Brasil; Departamento de Informática; Programa de Pós-Graduação em Ciência da Computação; UEM; Maringá, PR; Centro Tecnologia
• Palavras-chave em Português: Linguagem de programação PLH; Paralelismo; Tarefas; Ambiente de execução; Brasil.; Pogramming languages; Tasks; Parallelism, and execution environment; Brazil.; Ciências Exatas e da Terra; Ciência da Computação
• Link de acesso: http://repositorio.uem.br:8080/jspui/handle/1/2546

Currently the software mechanisms employed to provide parallel programming are presented in three forms: libraries, extended languages and specific languages. These mechanisms allow to specify units or threads further a mode of communication between such threads. Although each mechanism present their advantages, disadvantages and simplify the development of parallel programs, it hide from the programmer part of management of the execution environment, usually this mechanisms make some explicit parallel constructs. With the motivation of having a simple programming language was developed the programming language PLH, a general purpose language strictly task-based, in which part of related parallelism management is removed from developer. For this purpose, besides the programming language was also developed an execution environment with a series of strategies based on other environments. The execution environment is able to create and execute tasks in parallel, and potentially also deal with management issues, moving them from the programming language to execution model well-defined. The basic premise and the main management issue is imposed in PLH as think in parallel, so applications developed in PLH have fewer dependencies on tasks, therefore, can make the best resources provided by the hardware. A performance analysis conducted in this work showed that four applications for scalability was the ideal, getting an average of 50.47% to scalability as processors increase. Four other applications came close to ideal, with an average of 43.99% scalability. However, an application didn't get good performance with an average of 38.01% scalability. Regarding the use of computational resources, the average percentage obtained in CPU utilization for applications was 668.11% for a configuration with eight processors.