A GPU approach to fortran legacy systems

A large number of Fortran legacy programs are still running in production environments, and most of these applications are running sequentially. Multi- and Many- core architectures are established as (almost) the only processing hardware available, and new programming techniques that take advantage...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Méndez, Mariano, Tinetti, Fernando Gustavo
Formato: Articulo Reporte
Lenguaje:Inglés
Publicado: 2012
Materias:
Ciencias Informáticas
CUDA
legacy systems
GPU
openACC
Acceso en línea:http://sedici.unlp.edu.ar/handle/10915/23114
Aporte de:
SEDICI (UNLP) de Universidad Nacional de La Plata
id I19-R120-10915-23114
record_format dspace
institution Universidad Nacional de La Plata
institution_str I-19
repository_str R-120
collection SEDICI (UNLP)
language Inglés
topic Ciencias Informáticas
CUDA
legacy systems
GPU
openACC
spellingShingle Ciencias Informáticas
CUDA
legacy systems
GPU
openACC
Méndez, Mariano
Tinetti, Fernando Gustavo
A GPU approach to fortran legacy systems
topic_facet Ciencias Informáticas
CUDA
legacy systems
GPU
openACC
description A large number of Fortran legacy programs are still running in production environments, and most of these applications are running sequentially. Multi- and Many- core architectures are established as (almost) the only processing hardware available, and new programming techniques that take advantage of these architectures are necessary. In this report, we will explore the impact of applying some of these techniques into legacy Fortran source code. Furthermore, we have measured the impact in the program performance introduced by each technique. The OpenACC standard has resulted in one of the most interesting techniques to be used on Fortran Legacy source code that brings speed up while requiring minimum source code changes.
format Articulo
Reporte
author Méndez, Mariano
Tinetti, Fernando Gustavo
author_facet Méndez, Mariano
Tinetti, Fernando Gustavo
author_sort Méndez, Mariano
title A GPU approach to fortran legacy systems
title_short A GPU approach to fortran legacy systems
title_full A GPU approach to fortran legacy systems
title_fullStr A GPU approach to fortran legacy systems
title_full_unstemmed A GPU approach to fortran legacy systems
title_sort gpu approach to fortran legacy systems
publishDate 2012
url http://sedici.unlp.edu.ar/handle/10915/23114
work_keys_str_mv AT mendezmariano agpuapproachtofortranlegacysystems
AT tinettifernandogustavo agpuapproachtofortranlegacysystems
AT mendezmariano gpuapproachtofortranlegacysystems
AT tinettifernandogustavo gpuapproachtofortranlegacysystems
bdutipo_str Repositorios
_version_ 1764820465669373954

Ejemplares similares