Estimating parallel execution time of loops with loop-carried dependences

T. Nakanishi, K. Joe, C. D. Polychronopoulos, K. Araki, A. Fukuda

Research output: Chapter in Book/Report/Conference proceedingConference contribution

3 Citations (Scopus)

Abstract

We propose a scheme to estimate exact minimum parallel execution time of the single loop with loop-carried dependences in medium and fine grain parallel execution. The minimum parallel execution time of a loop is given by the critical path length of the dependence graph which represents the code obtained from the fully unrolled loop. However, unrolling loops with a large number of iterations requires too much computation time and large storage space to be practical. The scheme proposed provides the minimum parallel execution time without unrolling the loop at all by reducing the problem into an integer linear programming problem and employing the simplex method and a branch-and-bound algorithm to solve it. We also show an experimental implementation of the proposed scheme with Livermore Benchmark Kernels to demonstrate that the computational complexity of our scheme is independent of the number of iterations of the given loop.

Original languageEnglish
Title of host publicationSoftware
EditorsK. Pingali
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages61-69
Number of pages9
ISBN (Electronic)081867623X
DOIs
Publication statusPublished - 1996
Event25th International Conference on Parallel Processing, ICPP 1996 - Ithaca, United States
Duration: Aug 12 1996Aug 16 1996

Publication series

NameProceedings of the International Conference on Parallel Processing
Volume3
ISSN (Print)0190-3918

Other

Other25th International Conference on Parallel Processing, ICPP 1996
Country/TerritoryUnited States
CityIthaca
Period8/12/968/16/96

All Science Journal Classification (ASJC) codes

  • Software
  • Mathematics(all)
  • Hardware and Architecture

Fingerprint

Dive into the research topics of 'Estimating parallel execution time of loops with loop-carried dependences'. Together they form a unique fingerprint.

Cite this