Automatic determination of grain size for efficient parallel processing

Journal Article · Fri Sep 01 00:00:00 EDT 1989 · Communications of the ACM; (USA)

DOI:https://doi.org/10.1145/66451.66454· OSTI ID:5587746

McCreary, C ^[1]; Gill, H ^[2]

---

1. Auburn Univ., AL (USA)
2. Mitre Corp., McLean, VA (USA)

Efficiently exploiting parallelism in programs is one of today's most critical problems in the use of parallel systems. The difficulty lies in determining how to partition a task into modules that can be executed concurrently and scheduling those modules in such a way as to minimize execution time. Two different approaches are currently being researched: a programmer-based approach which relies on special languages or language features that allow the programmer to indicate sections for simultaneous execution; and the automatic detection of parallelism from a conventional sequential language. Both approaches have merit, and research must be continued along both lines. In this article the authors propose a method for automatic determination and scheduling of modules from a sequential program.

• OSTI ID: 5587746
• Journal Information: Communications of the ACM; (USA), Vol. 32:9; ISSN 0001-0782
• Country of Publication: United States
• Language: English

Related Subjects

99 GENERAL AND MISCELLANEOUS//MATHEMATICS, COMPUTING, AND INFORMATION SCIENCE
PARALLEL PROCESSING
PERFORMANCE
AUTOMATION
DESIGN
EFFICIENCY
PROGRAMMING LANGUAGES
RESEARCH PROGRAMS
USES
PROGRAMMING
990210* - Supercomputers- (1987-1989)