A loop parallelization algorithm for HPF compilers

by: Kazuaki Ishizaki, Hideaki Komatsu

Languages and Compilers for Parallel Computing (1996), pp. 176-190.

View FullText article

Reviews [Write a review of this article]

There are no reviews of this article

Find related articles from these CiteULike users

rob_quill

Find related articles with these CiteULike tags

no-tag

Abstract

This paper presents a formalized loop parallelization algorithm for effectively extracting parallelism from data to be allocated to processors with array decomposition directives, in languages such as High Performance Fortran (HPF). We define a communication dependence vector that shows data dependence among processors, and use it in our algorithm to formalize and unify the detection of vector prefetch communication and vector pipeline communication for loop parallelization. The paper also presents a method, based on our algorithm for generating vector communications. We implemented the algorithm in our HPF compiler and carried out experiments with two applications on an IBM RS/6000 Scalable POWERparallel System.

@incollection{citeulike:2743394, abstract = {This paper presents a formalized loop parallelization algorithm for effectively extracting parallelism from data to be allocated to processors with array decomposition directives, in languages such as High Performance Fortran (HPF). We define a communication dependence vector that shows data dependence among processors, and use it in our algorithm to formalize and unify the detection of vector prefetch communication and vector pipeline communication for loop parallelization. The paper also presents a method, based on our algorithm for generating vector communications. We implemented the algorithm in our HPF compiler and carried out experiments with two applications on an IBM RS/6000 Scalable POWERparallel System.}, author = {Ishizaki, Kazuaki and Komatsu, Hideaki }, citeulike-article-id = {2743394}, doi = {http://dx.doi.org/10.1007/BFb0014199}, journal = {Languages and Compilers for Parallel Computing}, pages = {176--190}, posted-at = {2008-05-01 21:35:43}, priority = {3}, title = {A loop parallelization algorithm for HPF compilers}, url = {http://dx.doi.org/10.1007/BFb0014199}, year = {1996} }

RIS record

TY - CHAP ID - citeulike:2743394 L3 - citeulike-article-id:2743394 N2 - This paper presents a formalized loop parallelization algorithm for effectively extracting parallelism from data to be allocated to processors with array decomposition directives, in languages such as High Performance Fortran (HPF). We define a communication dependence vector that shows data dependence among processors, and use it in our algorithm to formalize and unify the detection of vector prefetch communication and vector pipeline communication for loop parallelization. The paper also presents a method, based on our algorithm for generating vector communications. We implemented the algorithm in our HPF compiler and carried out experiments with two applications on an IBM RS/6000 Scalable POWERparallel System. JF - Languages and Compilers for Parallel Computing EP - 190 TI - A loop parallelization algorithm for HPF compilers SP - 176 AU - Ishizaki, Kazuaki AU - Komatsu, Hideaki PY - 1996/// UR - http://dx.doi.org/10.1007/BFb0014199 ER -

RIS BibTeX