Empirical study of load distribution strategies on multicomputers
Nazief, Bobby Achirul Awal
This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/21154
Description
Title
Empirical study of load distribution strategies on multicomputers
Author(s)
Nazief, Bobby Achirul Awal
Issue Date
1991
Doctoral Committee Chair(s)
Reed, Daniel
Department of Study
Computer Science
Discipline
Computer Science
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
Computer Science
Language
eng
Abstract
In this thesis, we examine an important issue in the execution of parallel programs on multicomputers. We consider parallel programs that consist of dynamically created tasks. These tasks communicate via message passing. The important issue is: how to assign the tasks of a parallel program on multicomputer nodes, such that the program execution time is minimal. This is but one facet of the class of resource allocation problems known as load distribution.
We have identified two groups of load distribution strategies on multicomputers. The first contains task placement strategies that assign newly created tasks to nodes. Once a task starts executing, it cannot be reassigned to a different node. The second group contains task migration strategies that manage initial task placement, and potential task relocation. Tasks may now be relocated to different nodes even after they have begun execution. To determine the performance of these strategies, we must study their behavior for a given range of machine architectures and a set of computational models.
In this thesis, we study the behavior of task placement and task migration strategies on multicomputers by comparing their static and dynamic performance through parametric simulations. In particular, we study the interaction between task placement strategies and parallel programs under different hardware and machine architectures. We also study whether task migration strategies can improve the performance obtained by task placement strategies alone. Because the cost of task migration may potentially be more expensive than that of task placement, we also study the possibility of reducing the task migration overhead.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
